You are on page 1of 502

Distributed in North and South America

by Transaction Books
ISBN 978-3-86838-174-0

9 783868 381740

Malcolm Budd, FBA,


Emeritus Grote Professor of the
Philosophy of Mind and Logic,
University College London

In this highly ambitious, wide ranging,


immensely impressive and ground-breaking
work Fabian Dorsch surveys just about every
account of the imagination that has ever been
proposed. He identifies five central types of
imagining that any unifying theory must
accommodate and sets himself the task of
determining whether any theory of what
imagining consists in covers these five
paradigms. Focussing on what he takes to be the
three main theories, and giving them each equal
consideration, he faults the first two and
embraces the third. The scholarship is
immaculate, the writing crystal clear and the
argumentation always powerful.

PhR 9

The Unity of Imagining

Fabian Dorsch

verlag

ontos

The Unity of Imagining

Fabian Dorsch

Herausgegeben von / Edited by


Johannes Brandl Andreas Kemmerling
Wolfgang Knne Mark Textor

Philosophische Forschung
Philosophical Research

Fabian Dorsch
The Unity of Imagining

Philosophische Forschung
Philosophical Research
Herausgegeben von / Edited by
Johannes Brandl Andreas Kemmerling
Wolfgang Knne Mark Textor
Band 9 / Volume 9

Fabian Dorsch

The Unity of Imagining

Bibliographic information published by the Deutsche Nationalbibliothek


The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliographie;
detailed bibliographic data is available in the Internet at http://dnb.ddb.de

North and South America by


Transaction Books
Rutgers University
Piscataway, NJ 08854-8042
trans@transactionpub.com
United Kingdom, Ireland, Iceland, Turkey, Malta, Portugal by
Gazelle Books Services Limited
White Cross Mills
Hightown
LANCASTER, LA1 4XS
sales@gazellebooks.co.uk

Livraison pour la France et la Belgique:


Librairie Philosophique J.Vrin
6, place de la Sorbonne ; F-75005 PARIS
Tel. +33 (0)1 43 54 03 47 ; Fax +33 (0)1 43 54 48 18
www.vrin.fr

2012 ontos verlag


P.O. Box 15 41, D-63133 Heusenstamm
www.ontosverlag.com
ISBN: 978-3-86838-174-0
2012
No part of this book may be reproduced, stored in retrieval systems or transmitted
in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise
without written permission from the Publisher, with the exception of any material supplied specifically for the
purpose of being entered and executed on a computer system, for exclusive use of the purchaser of the work
Printed on acid-free paper
ISO-Norm 970-6
FSC-certified (Forest Stewardship Council)
This hardcover binding meets the International Library standard
Printed in Germany
by CPI buch bcher.de GmbH

Table of Contents

Detailed Table of Contents

ix

Acknowledgements

xv

Introduction

Part One The Nature and Variety of Imagining


Introduction to Part One

23

Unified Accounts of Imagining

27

Key Features of Imaginative Episodes

59

Key Features of Sensory Imaginings

79

Imagination and Knowledge

115

Other Theories of Imagining

149

Part Two The Epistemological Account


Introduction to Part Two

191

OShaughnessys View

197

OShaughnessys Arguments

221

Critical Assessment

253

viii

Part Three The Dependency Account


9

Introduction to Part Three

275

The Representational Account

281

10 Visual Imagining As Experiential Imagining

313

11 Emotional Imagining As Experiential Imagining

337

12 Semantic Dependency, Simulation, and Pretense

365

Part Four The Agency Account


Introduction to Part Four

381

13 Imaginative Agency

385

14 Meeting the Desiderata

409

Conclusion

431

Bibliography

443

Index

455

Detailed Table of Contents

Acknowledgements

xv

Introduction

1
5
7
11
17

(i) The Need for a Unified Account


(ii) Three Main Theories of Imagining
(iii) Other Theories of Imagining
(iv) The Structure of the Book

Part One The Nature and Variety of Imagining


1

Introduction to Part One

23

Unified Accounts of Imagining

27
28
32
37
41
42
44
46
47
51
52
53
54
55

1.1 Two Desiderata


1.2. Five Central Cases of Imagining
(i) Sensory Imagining
(ii) Affective Imagining
(iii) Intellectual Imagining
(iv) Experiential Imagining
(v) Imaginative Projects

1.3. Mental Episodes and Complexes Thereof


1.4. Non-Central Cases of Imagining
(i) Imaginative Dispositions
(ii) Dream Representations
(iii) Unusual or Pathological Cases
(iv) Fanciful Expectation, Anticipation, Etc.

x
2

Key Features of Imaginative Episodes


2.1. Imaginative Attitude
2.2. No Epistemic Function
2.3. Reference to Reality

Key Features of Sensory Imaginings


3.1. Diminished Vivacity
3.2. No Immediacy
3.3. Indeterminacy
3.4. Quasi-Observationality
3.5. Immunity to Error through Misidentification
3.6. Intellectuality
(i) Peacockes View
(ii) Sartres View
(iii) Imaginative Thought vs. Intention-in-Action
(iv) Explanatory Power

Imagination and Knowledge


4.1. Two Examples of Cognising Imaginings
4.2. The Absence of Alternative Grounds
4.3. Similarities to Perception (and Memory)
4.4. Differences from Perception (and Memory)
4.5. Imagination and Information
4.6. Spontaneity, Attention, and Filling-In
4.7. Voluntariness and Cognitive Constraints

Other Theories of Imagining


5.1. The Quantitative Account
5.2. The Phenomenal Account
5.3. The Ontological Account
5.4. The Modal Account
5.5. The Attitude Account
5.6. The Spontaneity Account

59
61
68
73
79
81
83
85
87
94
96
96
100
103
106
115
116
122
126
128
133
138
142
149
150
152
154
161
168
176

xi

Part Two The Epistemological Account


6

Introduction to Part Two

191

OShaughnessys View

197
199
202
213

6.1. A Brief Synopsis


6.2. The Three Main Claims
6.3. The Argumentative Strategy

OShaughnessys Arguments
7.1. The Argument from Origin
(i) Standard Constraints on Cognitions
(ii) Wakefulness as a Further Constraint
(iii) The Origin of Imaginings

7.2. The Argument from Attention

Critical Assessment
8.1. The Explanatory Power of Negation Claims
(i) General Limitations
(ii) The Limitations of (NC*)
(iii) The Limitations of (NI) and Similar Claims
(iv) The Limitations of (NC)

8.2. The Extensional Adequacy of Negation Claims

221
222
223
232
242
247
253
254
254
256
258
264
266

Part Three The Dependency Account


9

Introduction to Part Three

275

The Representational Account

281
282
290
292
294
302
309

9.1. Humes Copy Principle


9.2. Objections to Humes View
9.3. Causal Echo Claims
9.4. Representational Echo Claims
9.5. Intellectual Imagining
9.6. Imaginative Projects

xii
10 Visual Imagining As Experiential Imagining
10.1. Martins Argument: Perspectivalness
10.2. A Second Argument: Egocentric Orientation
10.3. Replies to Objections
10.4. The Nature of Experiential Imagining

11 Emotional Imagining As Experiential Imagining


11.1. The Puzzle of Fiction
11.2. Waltons Account of Emotional Imagining
11.3. Moran on the Various Forms of Imagining
11.4. Morans Account of Emotional Imagining
11.5. The Experiential Nature of Emotional Imagining

12 Semantic Dependency, Simulation, and Pretense


12.1. Semantic Dependency
12.2. Imagination as Simulation
12.3. Imagination as Pretense
12.4. Conclusion and Outlook

313
314
322
325
332
337
340
346
351
356
360
365
366
368
373
377

Part Four The Agency Account


Introduction to Part Four
13 Imaginative Agency
13.1. Mental vs. Bodily Agency
13.2. The Nature of Imaginative Agency
13.3. Qualifications
(i) Concreteness
(ii) Voluntariness
(iii) Directness

13.4. The Essentiality of Imaginative Agency


13.5. The Subjection to the Will Account

14 Meeting the Desiderata


14.1. Explanatory Power
(i) Informativity
(ii) Fundamentality

381
385
386
387
391
391
392
394
396
404
409
409
410
413

xiii
14.2. Extensional Adequacy
(i) Non-Imaginative Phenomena
(ii) Imaginative Phenomena

14.3. Non-Representationality: the Case of Imagined Pains


14.4. Passivity: the Case of Spontaneous Images & Thoughts

Conclusion
(i) The Agency Account
(ii) The Epistemological and the Dependency Account
(iii) Other Theories of Imagining
(iv) Some Loose Ends

414
414
417
421
423

431
431
434
437
439

Bibliography

443

Index

455

Acknowledgements

I am grateful to the following journals for permission to reproduce material: Enrahonar for parts of my paper Emotional Imagining and Our Responses to Fiction (Enrahonar, vol. 46, 2011; reprinted by permission of
the publisher, Universitat Autnoma de Barcelona); Rivista di Estetica for
parts of my paper The Recreative Imagination (Rivista di Estetica, vol.
54, 2013; reprinted by permission of the publisher, Rosenberg & Sellier);
and Philosophical Explorations for parts of my paper Transparency and
Imagining Seeing (Philosophical Explorations, vol. 13, 2010; reprinted by
permission of the publisher, Taylor & Francis).
My special thanks are due to Malcolm Budd, without whom I would not
have started and continued to work on the imagination; Mike Martin, who
has challenged and pushed me further on the topic than anyone else; Lucy
OBrien, who made sure that I finished my PhD thesis on the imagination
in time; Sebastian Gardner and Gianfranco Soldati, who taught me a lot
about the phenomenological approach to the mind and the imagination; and
Rob Hopkins, who has always been most helpful in his criticism.
In addition, I am very grateful, both for their constructive criticisms and
for their sympathetic encouragements, to Jiri Benovsky, Davor Bodrozic,
Johannes Brandl, Josep Corbi, Peter Goldie, Henning Hahn, David Harris,
Frank Hofmann, Eduard Marbach, Aaron Meskin, Richard Moran, Kevin
Mulligan, Matt Nudds, Martine Nida-Rmelin, Matt Soteriou, Kathleen
Stock, Juan Suarez, Mark Textor, Gian-Andri Toendury, Cain Todd, Lambert Wiesing, Marcus Willaschek, Richard Wollheim, several anonymous
referees, as well as audiences at University College London, the University

xvi
of Valencia and the University of Fribourg.
Finally, I would like to extend my gratitude to Johannes Brandl, Andreas
Kemmerling, Wolfgang Knne and Mark Textor, for their personal assistance and editorial guidance; the Graduate School of University College
London and the Swiss National Science Foundation, for their financial support of my research; and to my family and friends, for their uncomplaining
patience during the last years.

For Evgenia

Introduction

Introduction

The principal aim of this monograph is to present and compare the main
theories of imagining in a systematic and opinionated way.
The presentation is systematic in so far as it is structured by the concern
with the unity and common nature of the many forms of imagining, such as
visualising, supposing or daydreaming. Accordingly, the book involves
both a close study of the distinct kinds of imagining and of their main features, similarities and differences; and a critical discussion of the principal
accounts of imagining put forward in the literature. The book therefore
serves as an introduction to the main ideas and positions that have been adopted with respect to imaginative phenomena. But it does so in a strictly
problem-oriented manner, namely from the perspective of the possibility of
providing an account of imagining that unifies its different incarnations. As
a consequence, the discussion is organised around the possible sources of
the unity of imagining and centres less on particular philosophers and their
specific views.1
The comparison of the corresponding principal theories of imagining is
1

Another reason for this is that with some notable exceptions, such as the works by
Husserl (2006), Sartre (1940), Casey (1976), White (1990), OShaughnessy (2000)
and McGinn (2004) many philosophers have not presented fully developed and focused theories of imagining, but instead embedded their claims in discussions of
other, often more general issues (see, e.g., Collingwood (1938), Ryle (1949/1963),
Scruton (1974), Peacocke (1985), Walton (1990, Hopkins (1998) or Martin (2002a)).
Even of the noted exceptions, who devote whole books or lecture series to the topic,
some have not aimed at formulating a cohesive account of all forms of imagining,
but rather focused on discussing particular kinds of imagining, or on developing a
list of more or less loosely connected ideas and observations.

Introduction

opinionated, on the other hand, in so far as a more specific goal of the book
is to defend the Agency Account of imagining as the best unified account of
imagining available. The main claim of this theory is that imaginings are,
essentially and fundamentally, mental actions of a certain kind. Accordingly, imagining is something that we actively and voluntarily do.
A unified account of imagining is characterised by the fact that it captures the common nature of the central cases of imagining (e.g., visualising, supposing, or daydreaming) in fundamental terms and is able to distinguish them from the central cases of non-imaginative mental phenomena, notably cognitive representations (e.g., perceptions, judgements, or
memories). The minimal goal of any theory intended as a unified account
of imagining should therefore be to achieve extensional as well as explanatory adequacy: it should be valid for the paradigm cases of imaginative
and non-imaginative phenomena; and it should correctly describe and explain the distinctive and fundamental nature of the former. A main theme of
the discussion is therefore the elucidation of primary examples of imagining, and of what distinguishes them from other, non-imaginative mental
phenomena. Another and closely related guiding issue is whether it is possible to account for the specific character of these imaginings by identifying a set of features distinctive of them and responsible for their imaginative status (i.e., their property of being instances of imagining). That is, the
book addresses the issue of whether it is possible to specify and elucidate
this status in terms of individually necessary and jointly sufficient conditions for something to be an imagining, at least with respect to the
paradigm cases.
The issue of formulating a unified account of imagining has often been
neglected in the philosophical tradition. Many of the discussions of imagining in the past and the present have focused, not on the nature of imaginings, but on their role in our mental lives and our interactions with other
people and the world. It has been widely acknowledged that imagining is
very prominent in and significant for various parts of our lives, ranging
from our emotional engagement with other people (e.g., Goldie (2000):
194ff.) and our moral evaluation of actions (e.g., Johnson (1993)) to the
aesthetic appreciation of artworks (e.g., Walton (1990) and its many fol-

Introduction

lowers) and even the acquisition of knowledge about the external world.2
Moreover, even when philosophers address the question of what it
means for a mental phenomenon to be imaginative (i.e., possess the property of being an imagining), they usually concentrate exclusively on specific forms of imagining, notably on sensory or visual imaginings. 3 Indeed,
the neglect of the issue of what all imaginings have in common as imaginings may be closely linked to the neglect of non-sensory or complex kinds
of imagining. Finally, philosophers writing about the imagination have not
always succeeded, or been particularly interested, in developing their ideas
into a proper theory of what is distinctive of imaginings.
As a result, most discussions of imaginings have been concerned either
2

Although imaginings are typically held not to constitute knowledge about the external world or provide relevant epistemic justification by themselves (though see
Chapter 4 for an opposing view), it seems beyond doubt that they are often involved
in other ways in the acquisition of knowledge. See, for instance, the discussions on
thought experiments (e.g., Sorensen (1992) and Gendler (2000a)), the link between
conceivability and possibility (e.g., Gendler & Hawthorne (2002)), the role of mental imagery in geometry (e.g., Giaquinto (1992)) and in counterfactual reasoning
(e.g., Williamson (2008)), or the involvement of visualisation in memory-based
cognitive projects, such as the project of determining whether frogs have lips, or
how many windows are in ones own house (e.g., Kosslyn (1980): 1, or Pylyshyn
(2002): especially 164).
See Collingwood (1938): chs. 9f., Sartre (1940), Peacocke (1985), Hopkins (1998):
ch. 7, and, it seems, Wittgenstein (1984c): vol. II, sec. 63-147, to name just a few of
those who focus more or less exclusively on sensory or visual imaginings. Kind
(2001) does the same: while she argues that all instances of imagining involve
sensory imagery, she distinguishes the resulting sensory imagining from intellectual supposition. Scruton (1974), White (1990), Casey (2000), OShaughnessy
(2002), Currie & Ravenscroft (2002) and McGinn (2004) address both sensory and
intellectual imaginings, but no emotional imaginings (in contrast to emotional responses to imagining), and imaginative projects only in passing and not in relation to
their commonality with imaginative episodes. Besides, although McGinn discusses
various forms of imagining, he does not aim to provide a unified account of them.
Instead, he argues merely that they form an imagination spectrum which extends
from the most simple and temporally and conceptually prior imaginative phenomena
(e.g., those involved in sensory representation) to the most complex and developed
ones (e.g., those involved in creativity; see McGinn (2004): 13). See also the discussion of the five main forms of imagining in Section 1.2 for further references to accounts which focus on particular kinds of imagining.

Introduction

with aspects of imaginings other than (though perhaps dependent on or


otherwise related to) their imaginativeness, or with the imaginativeness of
only certain kinds of imagining. Just a few philosophers have attempted to
provide a satisfactory account of imagining in its (more or less) full variety.4 The less comprehensive philosophical discussions have certainly shed
light on important aspects of many different kinds of imagining, such as
sensory imaginings (e.g., visualising a face), intellectual imaginings (e.g.,
supposing or imagining that it rains), affective imaginings (e.g., imagining
an itch), experiential imagining (e.g., imagining having the experiences of
someone else) or imaginative projects (e.g., daydreaming about being
rich).
But the imaginative nature common to all forms of imagining has remained largely uninvestigated. A good example for this and the underlying
way of thinking is Waltons account of imagining (1990: ch. 1). He spends
considerable time on specifying a number of dimensions along which imaginings can vary, but also insists that we cannot spell out what they have
in common and have instead to be content with an intuitive understanding
of what it is to imagine (ibid., 19). Indeed, it is rather common in discussions about imaginings and their role in our mental lives to take for granted
4

See Scruton (1974): chs. 7f., Casey (1976), OShaughnessy (2000): chs. 11f., and
McGinn (2004) for clear examples. But none of them discusses imaginative projects
(fantasies, daydreams, etc.) in any detail. Whether other proposals are intended or
able to constitute a unified account of imagining is less clear. Humes account of all
imaginings as forming a certain sub-class of ideas suffers from its treatment of the
differences between sensory and intellectual representations and between imaginative and cognitive representations as quantitative (i.e., a matter of vivacity) rather
than qualitative (see Chapter 9 for discussion and references). Although Ryle discusses mainly sensory cases, his account of imagining as a form of internal pretending or pretending in ones head might be applied to intellectual imagining as
well (see Section 12.3, and Ryle (1949/1963): chs. 7f., especially sec. 8.6). White
analyses both visualising and intellectual imagining in terms of thinking of the possible, but does not explicitly connect the two analyses (see Section 5.4, and White
(1990): 122f; 184). Despite the fact that Currie and Ravenscroft treat both sensory
and intellectual imaginings as simulations of their respective cognitive counterparts
(see Section 12.2, and Currie & Ravenscroft (2002): 11 and 49), their main concern
is with the involvement of the imagination in adopting a perspective on the world
different from ones current one, rather than with the general nature of imagining
(ibid: 8f. and 11).

Introduction

that it is sufficiently clear what it means to imagine something as it is


likewise often assumed that we have a good grasp of, say, what it means to
perceive or believe something, or to desire it.
One particular difficulty with this approach is, however, that imagining
perhaps in contrast to perception, belief or desire does not obviously
constitute a natural kind (more on this point in Section 1.1). In fact, the
suspicion that imagining understood as the class of visualising, supposing, empathizing, daydreaming, and so on is not among the most basic or
primitive mental phenomena may explain why it has by no means been the
subject of philosophical investigation to the same extent, and with the same
exclusivity, as perception, belief and desire. Similarly, it is not evident that
there is unity in imagining; nor, if so, what it consists in or comprises. This
may lead to a situation in which a certain kind of imagining is postulated or
appealed to in the context of explaining a given phenomenon; while the
lack of any further elucidation of the nature of this type of imagining raises
in others considerable skepticism about the proposed role or even existence
of the kind of imagining in question.5
Such complications as well as, more generally, the significance and
prominent position of imagining in our lives and interactions with each
other and the world provide sufficient motivation for the investigation of
the possibility of a unified account of imagining. In particular, it will be
helpful and illuminating to learn more about what it means for a
representation to be imaginative, and how this relates to or influences the
various forms of engagement or project involving imagining.

(i) The Need for a Unified Account


The nature and unity of imagining is of theoretical interest both in itself
and in relation to many important aspects of our lives. But the general interest inherent in the question of whether we can provide a unified theory
of imagining and the significant role of imaginings in our mental lives are
5

See, for instance, the skepticism expressed in Budd (1992), Hopkins (1998): ch. 1,
Wollheim (2003) and Dorsch (2012c) about the existence of the specific form of
imagining seeing the depicted, which Walton refers to in his account of pictorial experience (see Walton (1990): ch. 8, and (2002)).

Introduction

not the only motivations for the search of a unified account. The expectation that the central cases of imaginings share a common nature arises also
from the need to explain two facts about our actual treatment of such representations.
The first is simply that, even after discounting the less obvious cases,
we do group together a large variety of mental occurrences in the class of
imaginings, while excluding many others. Thus we accept visualising, supposing, daydreaming, being engaged with fictions, empathizing, and so on,
as paradigm instances of imagining, or at least as essentially involving
such instances; but not seeing, judging, deliberating, or feeling an emotion
or desire. If such imaginings had nothing in common with each other, but
shared features with the non-imaginative phenomena, this tendency in how
we categorise mental phenomena would be rather mysterious: there has to
be something about the imaginative mental phenomena which causes us to
treat them but not other mental phenomena as members of one and the
same class (see Scruton (1974): 91f.).
The second relevant observation is that our corresponding classifications are stable, and that we usually have a good grasp of whether though
not necessarily of why a given mental state is imaginative or not. Three
observations are particularly relevant here. First of all, we repeat the same
categorisations and do not locate certain representations today on one side
of a divide and tomorrow on the other. There is no change in opinion about
whether, say, visualising is an instance of imagining. Then, it normally
does not take us much effort or thought to come to a classification of a given mental phenomenon as imaginative or non-imaginative. Identifying an
instance of daydreaming as an example of imagining, say, is typically very
straightforward. Finally, we are seldom unsure about how to treat a certain
mental phenomenon. This may happen with borderline cases (e.g., spontaneously occurring images and thoughts), or when we are confronted with an
unusual or rare kind of representation (e.g., Macbeths awareness of Banquo, or of the dagger). But, in most cases, we are not in doubt about whether some mental episode is an instance of imagining. These three aspects of
the stability of our relevant classification, too, strongly suggest the existence of a certain kind of unity among imaginings, to which we furthermore
have easy epistemic access, but which is still in want of further elucidation.

Introduction

Otherwise, it would be very difficult to explain the firmness and ease with
which we take the sometimes very different instances of imagining to be of
the same kind, and why it is that they, but no other phenomena, count for
us as imaginative. Any satisfactory analysis of imagining needs to provide
an elucidation of this unity of the paradigm instances of imaginings.
The idea pursued in this book is that the unity under consideration is due
to some features shared by and distinctive of imaginings. This is the
simplest and most straightforward explanation of their homogeneity; and it
is adopted by all theories of imagining that have the ambition to provide a
unified account. But it is not the only possible account of the two noted
facts and the resulting apparent unity of imagining. One could, for instance, maintain that our grouping together of the variety of phenomena
described as instances of one and the same type of phenomenon is merely
accidental or conventional, without any substantial grounding in shared
features of the mental phenomena concerned. But such a claim would be
difficult to support in view of the facility and assurance with which we categorise mental states as either imaginative or non-imaginative. Until it has
been confirmed that such a strong form of skepticism is inevitable, the
realistic hope for a more positive theory of the common nature of imaginings should outweigh any doubts about the possibility of a unified characterisation of imagining. It is thus reasonable to demand from a theory of
imagining that it account for the fact that we classify a large variety of phenomena as imaginative; and the fact that this classification is not a pure coincidence or convention. A unified theory of imagining promises to provide
such an explanation by treating these facts as a consequence of the presence of certain features shared exclusively by all imaginings.

(ii) Three Main Theories of Imagining


This raises the question of which proposals for a unified account of imagining are on offer and should be considered. When looking at the theories
of imagining put forward in the philosophical tradition, three major recurring themes can be identified: the relationship (or lack thereof) of imaginative representations to reality, their relationship to cognitive representations, and their relationship to the will. These motives identify the three

Introduction

principal ways in which the common and distinctive nature of the core instances of imagining may be elucidated: in terms of their irrelevance for
our epistemic interaction with the world, in terms of their dependency on
perceptions and (occurrent) beliefs, or in terms of their origin in a specific
kind of mental agency.6 The resulting theories of imagining may be labelled the Epistemological, the Dependency and the Agency Account. The
prominence in the literature of these approaches and their central ideas
about the nature of imagining is indicative of the fact that they are the main
contenders for a unified theory of imagining. Accordingly, I concentrate
my discussion primarily on these three proposals and devote to each a
whole part of the book.
The Epistemological Account specifies the difference between imaginings and cognitions by reference to the idea that only the latter are concerned with the representation of reality at least in a particular sense still
to be specified. Our minds interact with the world by means of cognition
and action. Both forms of interaction are primarily the domain of cognitive
representations, such as perceptions, memories or beliefs. In accordance
with this observation, the versions of the Epistemological Account to be
found in the literature identify a lack of cognitive concern with reality and,
especially, the failure to play a substantial role in cognition as the distinctive feature of imaginings. Different versions of this approach may vary in
how they characterise the assumed lack of cognitive concern.7
It is conceivable that a unified account of imagining may also be formulated in terms of their specific insignificance for our active engagement
6

Although it might be possible to endorse a view that characterises imaginings in


terms of their relationship both to reality and to agency (without also tracing back
one relationship to the other), such an approach to imagining would seem to be overcomplex. As will become clear in the subsequent chapters, reference to the presence
and nature of one kind of relationship suffices to account for the distinctive nature of
imagining and, if necessary, for the presence and nature of the other kind of relationship. The general idea is that imaginings will turn out either to lack a cognitive concern with reality precisely because they are voluntarily formed by us; or instead to
(be able to) be mental actions precisely because they are not cognitively constrained
by how reality is like.
The main proponent of this view is OShaughnessy (2000). But very similar ideas
can be found in the writings of Hume (1739), Sartre (1940), Wittgenstein (1984c)
and McGinn (2004), among others (see also the introduction to Part Two).

Introduction

with the world. The idea is that imaginings in contrast to, say, desires, intentions, beliefs or perceptions cannot motivate us to act or guide us in
our actions (e.g., by providing us with information about our relevant environment, or about adequate means to achieve our goals). However, the
claim that imaginings lack a guiding role in agency can be traced back to
the idea that they lack a cognitive concern with the world: they cannot
guide us in action because they do not provide us with knowledge about
the relevant aspects of reality (i.e., the environment and the means). The
idea that imaginings cannot move us to act, on the other hand, is not only
controversial (see Note 16 in Chapter 1); it also does not distinguish them
from many non-imaginative phenomena. Hence, it is not very promising to
formulate a unified account of imagining specifically in terms of their
seeming unimportance for our actions; and not very surprising that the
focus in the literature on imagining has rather been on the apparent lack of
cognitive concern.
The Dependency Account, by contrast, puts emphasis on the asymmetric
dependency of imaginings and cognitions. It argues that imaginative phenomena are constitutively dependent on and, hence, have to be elucidated
by reference to their cognitive counterparts. While cognitions exist and can
be made sense of independently of imaginings, the opposite is false. Proponents of the Dependency Account disagree about the nature of the kind
of dependency concerned.8 Philosophers in the Humean tradition, for example, focus on the idea that imaginings depend on cognitions in so far as
they represent them (possibly due to a suitable causal link). Another prominent approach is to assume that, in some sense, imaginings imitate or
mimic cognitions for instance, in the shape of off-line simulation or inner
pretense. Besides, it is widely accepted that there is also some form of semantic dependency: that we acquire much of our ability to represent something in an imaginative way by means of engaging in cognition (i.e., in per8

Hume (1739): 1.1.1.6ff. and 1.3.7.7), and OShaughnessy (2000): 363, endorse the
representational version of the Dependency Account for all forms of episodic imagining; while Peacocke (1985): 22, Martin (2001): 273 and (2002): 404, and Dorsch
(2010c) accept this view for sensory imaginings; and Dorsch (2011b) which is
more or less identical with Chapter 11 for emotional imaginings. Currie &
Ravenscroft (2002): 49, and Ryle (1949/1963): 250ff., may be read as defending a
version of the Dependency Account formulated in terms of simulation or imitation.

10

Introduction

ceiving, sensing, judging, and so on) and developing the relevant recognitional and conceptual capacities.
Finally, the already introduced Agency Account proposes mental activity
as the characteristic element of imagining. It maintains that imaginings are,
in a particular way, essentially active; while cognitions are either passive,
or active in a different way. The general idea is thus that imaginings are
mental actions of a special kind: they are imaginative determinations or expressions of the will. They differ in this respect from mental passions,
among them: judgemental thoughts or occurrent beliefs that are formed in
response to epistemic reasons; episodic memories that are determined by
past experiences; perceptions and sensations that are determined by their
external objects; and deviantly caused instances of the each of the preceding cognitive kinds of episode. Again, distinct versions of the Agency Account vary in how exactly they individuate the active character distinctive
of imaginings.9
The particular version, that I intend to put forward in this book, maintains that imaginings are mental actions, that aim at the active and direct
formation of mental representations with specific contents. The requirement of representational specificity demands that the underlying motivational states (e.g., tryings, desires or intentions) determine which particular
features are to be represented as being instantiated by which particular objects; while the directness requirement is meant to ensure that the motivational states end up determining the content of the formed representations
without making use of epistemic or merely causal mechanisms of content
determination and, notably, those mechanisms involved in the actively
triggered manifestation of mnemonic or other mental dispositions, or those
relied on during the voluntary formation of beliefs on the basis of evidence
as means. My defense of this version of the Agency Account of imagining is paired with a rejection of the Epistemological and the Dependency
Account.
9

The Agency Account (or something very much like it) has been endorsed by, among
others, Wollheim (1973): 69, Scruton (1974): 95 (see in general 94-100), Levinson
(1998): p. 232, n. 3, Kind (2001): 90ff., and McGinn (2004): 12ff.; 131f. Most of
these endorsements have been inspired by Wittgensteins comments on sensory imaginings (see, for instance, his (1984b): vol. II, sec. 63 and 627; see also Budd
(1989): 104ff.).

Introduction

11

However, since the present book is not only finite in its length, but also
primarily centred on the issue of the unity of imagining and aimed at giving a balanced overview of the various approaches to this issue, the discussion of the Agency Account has to be confined to roughly the same number
of pages as the two other main theories (or, in the case of the Dependency
Account, as its most promising namely representational version). This
means, unfortunately, that there is not enough space for a comprehensive
elaboration and vindication of the Agency Account, which definitely merits
a monograph on its own. The chapters devoted to this theory therefore concentrate just on the principal aspects of this theory, such as its characterisation of imaginings as mental actions of a certain kind, the essentiality of
the imaginative activity involved in voluntary imaginings, or the special
status of spontaneously occurring images and thoughts in relation to imaginative agency. Other necessary elements of a full defense of the Agency
Account such as its application to daydreams and other imaginative projects, or a discussion of pictorial experience, association, guessing, and
other problematic cases have to be left for another occasion.

(iii) Other Theories of Imagining


In addition to the three parts dealing with the main theories of imagining,
there is a fourth and introductory part of the book in which, among other
things, I briefly consider and argue against less important contenders for a
satisfactory account of the unity of imagining. Some of these additional
views start from the same or at least very similar basic ideas as the Epistemological, the Dependency or the Agency Account. This is part of the
explanation of why there are some considerable overlaps in convictions,
arguments and motivations among the individual major and minor views.
But these shared elements do not undermine the fact that the various accounts of imagining exclude each other as unified theories in so far as they
identify different features as the fundamental constituents of the common
nature of the otherwise heterogeneous forms of imaginings.
One alternative proposal the Quantitative Account takes imaginative
and non-imaginative phenomena to differ solely in degree (e.g., a difference in vivacity), rather than in kind. As a result, imaginings are assumed

12

Introduction

to possess the same essence as their non-imaginative counterparts.10


The Phenomenal Account assumes the most basic differences between
imaginative and non-imaginative phenomena to be phenomenal in nature:
that is, to be differences in what the respective episodes are subjectively
like. Accordingly, theories of imagining should focus on the phenomenological description of imaginings and their non-imaginative counterparts,
rather than on the formulation of more fundamental explanations of the
noted phenomenal differences. The Phenomenal Account therefore takes it
to be distinctive of imaginings that they subjectively seem to have a certain
kind of object, attitude or origin in contrast to some of the following theories which assume that it is distinctive of imaginings that they indeed
have a certain kind of object, attitude or origin.
The Ontological Account, for instance, elucidates the distinctive character of imaginings in terms of the specific nature of the imagined entities.
The central idea is that there is a categorical ontological difference between
objects which are perceived, remembered or judged to be a certain way and
objects which are visualised or supposed to be a certain way. In particular,
it may be said that imagined and cognised objects differ in their type of existence, their spatio-temporality, their determinacy or (as also proposed by
one version of the following account) their modal status.11
The central idea of the Modal Account is that, while it is characteristic
of imaginings that they represent possibilities, it is characteristic of cognitions that they represent actualities.12 The most promising version of this
theory maintains that the difference is one in attitude, rather than content.
This means that both types of representation are concerned with the same
kind of objects and states of affairs. But while cognitions purport to represent them as part of the actual world, imaginings lack such a commitment
and replace it with a stance of non-neutrality towards some possible world
(or situation). That is, imaginings are said to make a claim, not about how
The idea that perceptions and imaginings differ merely in vivacity is central to
Humes view (see Section 9.1). Compare also Hopkins (2011a) approach to sensory
memories, according to which they belong to the class of sensory imaginings and
differ merely contextually from its non-mnemonic members.
11
Some of these ideas can be found in Sartre (1940) and Casey (1976), for instance.
12
The view presented in White (1990) comes close to the Modal Account, although
White presumably does not intend to restrict imaginings only to mere possibilities.
10

Introduction

13

things actually are, but about how things might possibly be.
The resulting modal theory of imagining is, in fact, nothing but a specific version of the Attitude Account, according to which the difference
between imaginative and cognitive episodes is fundamentally attitudinal.
Minimally, the idea is that cognitions involve a cognitive attitude that is,
the commitment that reality is just like they represent it to be while imaginings lack it. Versions of the Attitude Account may stay neutral on the issue of whether this means that imaginings possess instead a distinctively
imaginative attitude. However, the need to distinguish imaginative episodes not only from cognitive episodes, but also from other kinds of episode
has moved proponents of the Attitude Account to assume that imaginings
are non-neutral as well, and to specify the commitment distinctive of imaginings in more positive terms, such as by reference to an act of positing
entities as non-existent, absent or, indeed, non-actual.13
Next comes the Spontaneity Account which maintains that imaginings
differ from cognitions in their specific origin in the voluntary or involuntary spontaneity of the mind, rather than in the impression of the external world onto the mind. In its most promising incarnation, the view
comes close to the Agency Account especially in its insistence that the
origin of imaginings is not merely causal, but essentially involves some
subjective response or doing. But the two theories differ in that the Spontaneity Account extends the realm of the imagination to the involuntary,
thereby loosely drawing on a distinction between perceiving and imagining
that is similar to the Kantian distinction between receptivity and spontaneity.
The Subjection to the Will Account is also similar to the Agency Account. While the last-mentioned takes imagining to always involve voluntary control, the first-mentioned requires merely that there is the possibility
of such control, without the need for its actual exertion. The Subjection of
the Will Account has the advantage of being able to capture what seem to
be involuntary instances of imagining, while distinguishing them none the
less strictly from, say, perceptions and impression, the passivity of which
13

The idea of a specifically imaginative attitude has been put forward notably in the
writings of the phenomenologists, such as Husserl (2006), Sartre (1940) and Casey
(1976).

14

Introduction

resists any attempts to bring them under voluntary control.14


Finally, the Subpersonal Account the only one that I do not return to at
one point or another in the subsequent chapters characterises imaginings
in terms of the specific nature or use of the subpersonal neurofunctional
processes or modules involved in imagining, as they are studied by the
cognitive sciences. In other words, proponents of this approach maintain
that what is distinctive about imaginings is that they are subpersonally
formed in a special way, whether by means of the stimulation of certain
imaginative areas in the brain, or by means of certain imaginative ways
in which standard cognitive mechanisms are employed. It is of course an
empirical and, especially, experimental question whether it is possible
to identify neurofunctional areas or operations that are unique to imagining
and, if so, what their nature is. Accordingly, the formulation of the Subpersonal Account is primarily a matter of the empirical sciences, and only secondarily a matter of philosophy, which is by and large a non-experimental
discipline. In this respect, the Subpersonal Account differs sharply from the
much more philosophical theories of imagining listed above. 15 This difference is closely related to the three main reasons of why I do not discuss the
Subpersonal Account any further.
The first is that the cognitive sciences are yet simply not advanced
enough to formulate a satisfactory unified theory. Central to the explanation of this fact is that the unity of imagining has not been of much scientific interest, and that, hence, the relevant studies have typically focused
The noted similarities among the Agency, Spontaneity and Subjection to the Will
Accounts may also be captured by reference to the fact that they are all instances of
the more general Origin Account: that is, the view that imaginative and non-imaginative phenomena differ essentially in what does or can determine their occurrence
and representationality.
15
Of course, philosophy should take into account experimental findings; and philosophers may be engaged in experimental investigations. But philosophical methodology is not itself experimental at least not in the sense of being concerned with the
designing and carrying out of repeated observations or experiments with replicable
results. This is true even of the so-called empirical or experimental philosophy: it
simply incorporates some scientific discipline (e.g., cognitive psychology) and applies its experimental methodology to empirical phenomena or problems, that are selected because of their relevance for certain philosophical issues. Hence, there is still
a division of labour between philosophy and the empirical sciences.
14

Introduction

15

on particular forms of imagining, notably visualising and belief-like imagining.16 What this may reflect is that the mental category of imagining is
perhaps not a very interesting category with respect to the subpersonal
level; while, as noted above, it plays an important role on the personal
level. Indeed, at least to my knowledge, there has been no serious attempt
so far to spell out, in terms of the neurofunctional entities and processes
recognised by the cognitive sciences, what the sensory, affective,
intellectual, experiential and more complex forms of imagining have
fundamentally in common.
Moreover, it is unclear whether it is reasonable to expect to be able to
formulate such a theory. Perhaps there are several distinct and fairly independent neural or otherwise subpersonal correlates of the feature distinctive of all imaginings on the personal level for instance, one correlate for
each of the basics form of imagining. That is, it is unclear whether such
correlates need to share a scientifically interesting feature in virtue of
which they count as correlates of one and the same personal level phenomenon (i.e., imagining). For instance, green or, more generally, coloured objects arguably do not have any physical property in common that
is significant from the perspective of physics. 17 Perhaps the same is true of
Tye (1991), Thomas (1999) and (2010) provide an overview of recent empirical
studies on sensory imagining, while Nichols (2006) and especially Gendler (2011)
do the same with respect to different kinds of intellectual imagining and related phenomena, such as pretense, empathy or delusions. Harris (2000) offers a developmental perspective on intellectual imagining and pretense, while Currie &
Ravenscroft (2002), Heal (2003) and Goldman (2006) look at it from a simulationist
perspective. More specifically, Currie & Ravenscroft (2002) and Gendler (2011) discuss evidence linking intellectual imagining to empathy, autism and delusion; while
Weinberg & Meskin (2006) develop an empirically-based account of intellectual
imagining partly based on the work by Nichols & Stich (2003) and apply it to
various philosophical problems namely the problems of emotional responses to
fiction (see Chapter 11), of imaginative resistance, and of distinguishing imagining
from supposing (see Note 18 in Chapter 1). By contrast, the empirically informed
debate about the nature of sensory imagining has been largely dominated by the
dispute between Kosslyn (see, for instance, his 1980) and Pylyshyn (see, for
instance, his 2002).
17
If at all, colour hues of surfaces should be identified with certain classes of reflectance properties, the members of which are individuated and grouped together in
terms of their dispositional power to bring about a certain pattern of stimulation in
16

16

Introduction

the neurofunctional correlates of imagining. Besides, the prospects of securing a satisfactory unified theory of imagining are likely to dwindle further, the more removed from the personal level an account becomes. Perhaps, generality is to be had only very close to or, indeed, only at the
level of theorising about persons.
My second reason for not further investigating the prospects of the Subpersonal Account is that it is not really in competition with the unified theories aiming to capture the personal level of imagining. Rather, the two
kinds of account should be seen as complementing each other. While the
previously introduced theories aim to capture the nature of imagining in relation to subjectively accessible features (e.g., representationality, rational
role, origin, passivity, and so on) and in contrast to other conscious mental
phenomena (i.e., perception, feeling, judgement, action, and so on), the
Subpersonal Account intends to identify the neurofunctional foundation of
whichever aspects of our conscious mental lives are essential to imagining.
That is, the Subpersonal Account is concerned with how imagining is realised in the brain, and not what it means to imagine something from the
point of view of persons. Accordingly, personal and subpersonal theories
that is, philosophy and the cognitive sciences have different goals and
employ distinct notions of fundamentality. The same is true of physics or
biochemistry and their concern with happenings in the mind or brain.
My third and final motivation for not spending more time on discussing
the Subpersonal Account is that the more philosophically minded theories
show a certain priority over it most notably with respect to issues of epistemic authority. If philosophy comes up with a satisfactory account of
imagining that makes good sense of the unity of the various forms of imagining on the personal level, it would be no reason to revise or give up this
philosophical view that the cognitive sciences were to reveal that there are
the relevant opponent processing channels and, hence, to bring about certain phenomenally individuated colour experiences. However, from the perspective of physics, these classes of reflectance properties in contrast to the reflectance properties
themselves are of no scientific interest. Physicists have been concerned with
developing and experimentally testing theories of how much of the incoming light
surfaces reflect, but not of the extent to which surfaces stimulate the opponent processing system or cause phenomenally different colour experiences (see Dorsch
(2009a) and (2010b)).

Introduction

17

no significant or interesting commonalities among all the different instances of imagining. Instead, and in the absence of further reasons to
question the philosophical view, it would be more reasonable to conclude
that imagining, as a personal-level phenomenon, has no single correlate on
the subpersonal level. Similarly, the fact that physics is not really concerned with colours does not undermine the attempt to formulate an adequate philosophical theory of colours (see Notes 17f. above). Of course,
philosophical theories of imagining need to be given up if they contradict
empirically well-supported versions of the Subpersonal Account (if there
are any). But, on the whole, the best philosophical theories of imagining
tell the cognitive sciences what to look for (e.g., for the neurofunctional
correlate of a particular kind of mental agency), and not the other way
round.18 Indeed, whether a given version of the Subpersonal Account is
successful should in part be evaluated in respect of whether it captures the
feature, or set of features, identified as being distinctive of all instances of
imagining by the best philosophical unified theory.

(iv) The Structure of the Book


As already touched upon, the book consists of four parts. Part One has the
function to set the stage for the subsequent critical discussions of the Epistemological, the Dependency and the Agency Account. It does so in three
steps.
First, it specifies the two main desiderata for any unified account of
imagining: extensional adequacy and explanatory power (see Chapter 1).
As part of this characterisation, Part One involves a characterisation of the
five central forms of imagining to be unified: sensory imagining, affective
imagining, intellectual imagining, experiential imagining, and imaginative
projects (see Chapter 1). It also includes a description of the main features
of imaginative episodes to be explained most notably their lack of a cognitive attitude and of an epistemic function and, indeed, provides a first
tentative reason for the endorsement of the Agency Account, namely that it
18

See Dorsch (2010b) and Dorsch (2010a) for a similar division of labour between
philosophy and the empirical sciences in the case of colours and of hallucinations,
respectively.

18

Introduction

promises the best explanation of the quasi-observationality of sensory imaginings (see Chapters 2f.).
Second, Part One is concerned with the relationship between imagining
and knowledge. In particular, it argues that certain instances of sensory
imagining can constitute knowledge in a way very similar to that of perceptual experiences and, in response to this observation, asks in which
sense (if any) imaginings can still be said to be uninformative with respect
to the external world (see Chapter 4).
Third, Part One finishes with a discussion of the less important unified
accounts of imagining listed above (see Chapter 5). The only exceptions
are the Causal Account and the Subjection to the Will Account, which are
briefly discussed in the Chapters 9 and 13, respectively; as well as the Subpersonal Account, which is, as mentioned above, not further discussed at
all.
Part Two is concerned with the characterisation (see Chapters 6f.) and
rejection (see Chapter 8) of the various versions of the Epistemological Account of imagining, formulated in terms of the alleged inability of imaginings to be reliable or rationally integrated with our picture of reality, to inform us or provide us with knowledge about the external world, or to possess the commitment, justificatory power or epistemic function essential to
cognitions. The central component of this discussion is a detailed and
scholarly examination of OShaughnessys version of the Epistemological
Account.
Part Three deals with the various versions of the Dependency Account.
Chapter 9 highlights the relevant causal and representational elements in
Humes conception of imaginings and reveals its continuity with more contemporary neo-Humean views that understand the assumed dependency
of imaginings on cognitions in purely representational terms. Together with
the following two chapters, it is also devoted to the applicability of the resulting Representational Account that is, the representational version of
the Dependency Account to intellectual imaginings (see Section 9.5),
imaginative projects (see Section 9.6), sensory imaginings (see Chapter 10)
and emotional imaginings (see Chapter 11). The conclusion to be put forward is that, while the Representational Account is probably true of sensory and emotional imaginings, it fails to pay justice to the nature of the

Introduction

19

other two forms of imagining. Part Three concludes with a negative assessment of the prospects of devising a unified account of imagining by reference to the idea that imaginings simulate or otherwise imitate cognitions
(see Chapter 12).
The final Part Four develops and defends the Agency Account of imagining. It does so, first of all, by spelling out the main thesis of the theory,
distinguishing it from the Subjection to the Will Account and presenting an
argument in favour of the endorsement of the Agency Account (see
Chapter 13). The central claim of this account is that imaginings are to be
identified with mental actions that have the purpose of producing mental
mental representations and of applying direct control over which states of
affairs are thereby represented. And it is crucial for the truth of this view is
that the kind of mental agency involved is essential to the formed
imaginative representations, thus ruling out the possibility of passive
imaginings. In addition, Part Four illustrates how the Agency Account can
meet the two desiderata for unified theories of imagining established
earlier and deal with potential counterexamples, notably the imaginative
counterparts to non-representational phenomena, as well as spontaneously
occurring images and thoughts (see Chapter 14).
Many of the parts, chapters and sections of the book are more or less
self-contained and can be read fairly independently of the others. Engagement with Part Two and the Epistemological Account requires mainly
knowledge of Chapter 1 and Sections 2.1f., although reading at least the
first half of Chapter 4 will be helpful as well. In addition, the reader may
decide to skip most of Chapter 6 and all of Chapter 7 if (s)he is not interested in the details of OShaughnessys theory of imagining and his more
general approach to conscious awareness. For Chapter 6 begins with a brief
synopsis of the main claims and arguments of OShaughnessys, which
provides all the information necessary to proceed directly with Chapter 8.
Part Three (the Dependency Account) and Part Four (the Agency Account)
presuppose even less than Part Two, namely really not much more than acquaintance with the content of Chapter 1, as well as perhaps Sections 2.1f..
Both Chapter 10 and Chapter 11 on sensory and on emotional imagining,
respectively can be read largely on their own, which is in part due to the
fact that they are based on previously published material. Similarly, the dis-

20

Introduction

cussion of the (neo-)Humean approach to imagining and its rejection as a


unified account in Chapter 9 is relatively self-contained; as is the discussion of the idea that imaginings imitate cognitions in Chapter 12. Chapter
13 constitutes a reasonably free-standing vindication of the Agency Account, with some additional support by Chapter 14. Besides, the Chapters
1, 2, 3 and 5 in Part One may be used as introductory readings to the topic
of imagining, given that they highlight the main forms and features of imagining and portray many of the theories of imagining to be found in the
literature.
Although the outlook and structure of the book is primarily systematic
in its focus on the issue of the unity of imagining, it also includes substantial critical studies of the views of particular philosophers. These views do
not always constitute proper theories or provide comprehensive accounts
of imagining, or of some of its specific forms (the views of Hume and Wittgenstein are good examples). But even if not, they are informative and instructive with respect to the ambition to provide a unified theory of imagining and, more generally, to understand what it means to imagine something. Discussions of the views of the following philosophers, among others, are scattered across the book: Hume (Chapter 9); Husserl (Sections
2.1, 3.1f., 5.2f. and 5.5); Sartre (Sections 3.4ff., 4.5f., 5.2f. and 5.4f.); Wittgenstein (Sections 3.4f., 4.7 and 5.3); Ryle (section: 12.3); Casey (Sections 3.3, 5.2, 5.5f. and 13.5); White (Section 5.4); Peacocke (Sections 3.2,
3.5f. and 9.4); Walton (Chapter 11 and Section 13.5); Moran (Chapter 11);
OShaughnessy (Chapters 6ff. and Sections 9.4ff.); Martin (Sections 2.3,
3.2, 9.4 and Chapter 10); and Currie and Ravenscroft (Sections 12.1f.). Especially Husserls theory of imagining would deserve more attention. But it
is some compensation that most of his relevant and important ideas still receive due attention in so far as they can be found again in the Husserlian
account defended by Sartre (1940).
Finally, due to the partially modular structure of the book, many passages depend on or allude to discussions in other parts, chapters, sections
or notes of the book. Throughout the text, I use capital letters (and no
abbreviations) to distinguish these internal cross-references from external
references to particular elements or portions of cited works.

Part One
The Nature and Variety of Imagining

Introduction to Part One

This opening part of the book has two main functions. First, it is intended
to provide an overview of three different things: (i) the different forms of
imagining; (ii) their main features; and (iii) various unified accounts of
imagining (i.e., those that differ from the three main theories to be discussed later on in being less well-developed, less plausible, or less prominent in the literature). In short, it is meant to serve as an introduction to the
philosophy of imagining. Second, this part also aims to set the stage for the
rest of the book. In particular, its purpose is to specify the main desiderata
for a unified account of imagining and to highlight some features of imaginative episodes that are to become crucial during the upcoming discussions.
In the first chapter, I spell out the two main desiderata and list the five
central forms of imagining. The next two chapters are devoted to a description of the main differences between imaginative and cognitive episodes
and, in particular, their sensory kinds. In Chapter 2, I highlight the differences in attitude and epistemic function between imaginings and cognitions and argue that both may refer to aspects of reality, though in virtue of
different referential mechanisms. Chapter 3, by contrast, focuses on important characteristics of sensory and, especially, visual imaginings,
which distinguish them from perceptual experiences. The following
chapter addresses the relationship between imagining and the acquisition
of knowledge. More specifically, I argue that sensory imaginings can, under the right circumstances, lead to knowledge in a way that is very similar
to the way in which perceptual experiences may result in knowledge. And I

24

The Nature and Variety of Imagining

question Sartres and Wittgensteins claim that imaginings strictly differ


from cognitions in being uninformative. The final chapter critically discusses six unified accounts of imagining that are both because of their inherent weaknesses and because of their lesser popularity in the recent debates about the nature of imagining of lesser importance than the Epistemological, the Dependency and the Agency Account to be considered in
the subsequent parts. The six accounts in question are the Quantitative, the
Phenomenal, the Ontological, the Modal, the Attitude and the Spontaneity
Account. The discussion of a seventh minor view the Subjection to the
Will Account has to wait until Section 13.5.
Before I move on to the main desiderata for a unified account of imagining, three notes on my terminology are in order. First, when I speak of
knowledge, I intend to refer both to intellectual (or belief-constituted)
knowledge and to successful perceptions, episodic memories and other
grounds for such intellectual knowledge. There is some debate about
whether, say, perception constitutes a form of knowledge, or rather just a
ground for knowledge (see, e.g., Williamson (2000) on the equation of
evidence with knowledge). Without intending to prejudice the rightful conclusion to this debate, I assume mainly in order to simplify the exposition
of respective positions and arguments that knowledge can be realised not
only by intellectual episodes, but also by non-intellectual episodes. In line
with this stipulation, I will say, for instance, that perceptual experiences or
episodic memories may constitute knowledge. 1 Second, instead of always
writing that episodes constitute or realise knowledge of reality, I will also
say that they cognise reality. This is not meant to imply that the respective
episodes are cognitive episodes (i.e., perceptions, memories, judgements,
and so on). As will become clear, there may very well be non-cognitive
and, more specifically, imaginative episodes that may realise knowledge
in a way very similar to, say, successful perceptions or episodic memories.
Third, I normally use the term imaginative to denote the property of being
an instance of imagining, and not the property of being inventive or ori1

I also intend this not to imply disjunctivism about episodes or states of knowledge.
In particular, I stay neutral on whether belief is a proper constituent of knowledge
(see Williamson (2000) for further discussion), or whether perceptions and
hallucinations share some or all aspects of their nature (see Martin (2006) and
Dorsch (2012a) for further discussion).

Introduction to Part One

25

ginal. The imaginativeness of a mental episode is therefore identical with


its general status as an instance of imagining, rather than with a specific
quality of its representationality: all and only imaginings are imaginative in this sense.

CHAPTER 1
Unified Accounts of Imagining

For the assessment of the main candidates for a unified account of imagining, it is necessary to say first a bit more about what is required of such a
theory. An important part of this task is to clarify which phenomena should
definitely be captured by a unified account of imagining, and which definitely not. Accordingly, the three sections of this chapter are concerned with
the particular demands on, and the mandatory scope of, theories of imagining.
In the first section, I specify two desiderata for a unified account of imagining: (a) extensional adequacy with respect to all central imaginative and
non-imaginative phenomena; and (b) explanatory power with respect to the
distinctive nature of imaginings. The following section lists the paradigm
instances of imagining and their non-imaginative counterparts. In particular, I describe five main forms of imagining which any unified account has
to capture: (i) sensory imaginings; (ii) affective imaginings; (iii) intellectual imaginings; (iv) experiential imaginings; and (v) imaginative projects.
In addition, I introduce the main examples of non-imaginative phenomena,
notably cognitive states and projects. The third and final section briefly
discusses the status of unclear, controversial or borderline instances of imagining (e.g., dispositional imaginings, or representations symptomatic of
psychological disorders) and their relative irrelevance for the evaluation of
unified accounts of imagining.

28

The Nature and Variety of Imagining

1.1 Two Desiderata


As already noted, a unified theory of imagining is concerned with characterising the imaginativeness common and essential to all central cases of
imaginings. It fulfils this task by identifying a feature (or set of features),
the exemplification of which is both necessary and sufficient for something
to be a paradigm instance of imagining, and reference to which helps to illuminate the imaginative character of the respective phenomena. But the
particular demands on a unified account of imagining can be formulated
more precisely in terms of two desiderata which any candidate theory has
to satisfy.
The first desideratum consists in the demand for extensional adequacy.
Accordingly, the theory has to be true of all central cases of imagining, and
false of all paradigm instances of non-imaginative phenomena. This means
that the class of mental phenomena delineated by means of reference to the
set of necessary and sufficient conditions identified by the theory in question should contain all primary examples of imagining, but no primary examples of non-imaginative phenomena. In the next section, I specify in
more detail which mental phenomena should be taken to be central cases of
imagining, and which not. But it should already be clear that a theory
which focuses exclusively on a particular form of imagining (e.g., visualising) cannot on its own constitute a unified account of imagining. At best,
it may hope to make some contribution to, or figure as a starting-point for,
the formulation of such a theory. Due to their (often deliberately) limited
scope, this consideration applies to many discussions about the nature of
imaginings found in the literature (see, for instance, Note 3 in the Introduction).
The second desideratum for a unified account of imagining is that it
should be explanatorily powerful. This desideratum has several aspects.
First, it requires that the theory in question says something illuminating
about why imaginings are essentially imaginative, in the sense of being an
instance of imagining. The task of the theory is thus to clearly identify and
characterise the essential feature (or set of features) responsible for and
perhaps identical with the feature of being imaginative. In addition, it
may though need not elucidate some other important aspects of imagining.

Unified Accounts of Imagining

29

Second, a theory of imagining is explanatorily powerful only if it


provides one and the same account for all central cases of imaginings. In
other words, it has to be true of all imaginings for the same reason and in
virtue of the same aspects of their nature. A theory which traces back the
imaginativeness to some set of features with respect to one kind of imagining and to another set of features with respect to a different kind of imagining is not unified, since it does not concern the nature common to all
paradigm instances of imagining, but instead provides two independent accounts for two distinct phenomena.
Third, it is required of a unified theory of imagining that it identifies the
most basic essential feature (or set of features) responsible for the imaginativeness which the respective phenomena have in common. I thereby take
a property to be responsible for another if and only if the exemplification
of the first, and nothing else, explains the exemplification of the second (as
well as any immediate consequences of the exemplification of the second).
I intend this characterisation of explanatory responsibility to be compatible
with various metaphysical relations which may hold between the two properties in question: they may be identical; the first may constitute the second
in another way (e.g., by means of inter-level realisation, as in the case of
heat and average kinetic energy); the first may figure as the supervenience
or emergence base for the second; and there may be other options. Hence,
whichever feature explains the imaginativeness of the central cases of imagining, but is not explained itself by a more basic feature, is the most fundamental feature which a unified account of imagining should identify.
The two desiderata specify what a unified account of imagining has to
accomplish: it has to identify the feature (or set of features) of the central
cases of imagining, which is responsible for their essential imaginativeness
and distinguishes them from central cases of non-imaginative phenomena. 2
2

It is of course a further advantage of a theory of imagining if it is able to identify and


elucidate also certain features of imaginings which some or all of them possess as a
consequence of their imaginativeness (e.g., their specific way of referring to reality).
And failure on this account might equally undermine the plausibility of a theory of
imagining. But such failure need not necessarily rule out the account concerned as a
candidate theory, in particular if it is not immediately obvious how other accounts of
imagining might be able to fulfil the task in question. Equally, it should not be expected that a unified account of imagining provides a theory of features of imagin-

30

The Nature and Variety of Imagining

But it will also be helpful to mention what a unified account of imagining


does not have to achieve.
First of all, a unified theory of imagining need not include the claim that
imaginings form or, alternatively, do not form a natural mental kind. It
may instead stay neutral with respect to whether the class of imaginings is
something for us to discover in nature, or something that is delimited by
our purposes or by our discourse about the mind, say. It is not obvious how
phenomena as varied as visualisings, suppositions and daydreams could
constitute a natural kind (see OShaughnessy (2000): 357ff.). Hence, the
idea that imagining is among the most basic and primitive mental phenomenon (as, arguably, perceiving and believing are) requires more argument.
Indeed, the major views seem to be skeptical about this idea. For example, while the Agency Account elucidates the nature of imaginings by
making reference to their instantiation of the natural kind linked to being a
(mental) action, it identifies the class of imaginings with just a subset of
that kind (see Section 13.2).3 Similarly, the Representational Account,
which assumes imaginings to be, essentially and fundamentally,
representations of cognitive episodes and is thus the representational version of the Dependency Account, does not claim that all such representations are necessarily imaginative episodic memories being a likely further candidate (see Section 9.4). Finally, OShaughnessys Epistemological
Account assumes that the different forms of imaginings do not even share a
common essence because they count as imaginative in virtue of different
properties (see Section 6.2).
In this respect, OShaughnessy weakens the ambition of a unified theory
of imagining even further. While, for instance, the Representational and the
Agency Account take all imaginings to share the same essence, and other
unified theories may even assume such a shared essence to individuate a
natural kind, OShaughnessy assumes different forms of imagining to possess different essences that just lead to the same consequence (i.e., to the

ings independent of their imaginative character (e.g., their being conscious, or their
being representational).
See OBrien ((2007): ch. 8) for a defense of the claim that actions form a natural
mental kind or are basic elements of the psychological repertoire, much as beliefs
are (ibid: 130).

Unified Accounts of Imagining

31

possession of the property of being an instance of imagining). Accordingly,


for him, the various types of imagining do not even belong to the same natural kind.
Then, a unified account of imagining need not be able to provide an
analysis of our ordinary concept of imagining. The set of necessary and
sufficient conditions supplied by such a theory may perhaps be understood
as demarcating some core notion of imagining. It may thus be concerned
not only with what imaginings are or how they are constituted, but also
with how we may conceive of them. It is, however, unrealistic to expect of
a unified theory of imagining that it has the ambition to provide a full and
satisfactory analysis of our ordinary concept of imagining. For not only
does it in general seem doubtful whether ordinary concepts at least if
they do not capture a natural kind allow for definition in terms of necessary and sufficient conditions.4 But the many ambiguities and uncertainties
in our use of the notion of imagining as partly reflected in the existence
of borderline or other unclear cases (see Section 1.3) suggest that this is
especially the case with respect to how we commonly conceive of imaginings. Hence, a unified account of imagining should not try to aim at
providing a definitional analysis of our ordinary concept of imagining. It
should instead focus on the phenomena and, more specifically, on the
central cases of imagining and the clarification of how we may be able to
conceive of them as imaginings. The core notion of imagining provided by
such an account might comprise only certain aspects of our ordinary
concept (and might perhaps deviate in others from it).
Finally, a unified theory of imagining need provide neither an account of
the faculty of imagination (if there is any), nor an account of what it means
for people to be imaginative, say, in what they think, say or do. It may
turn out that we possess a certain capacity which we employ just when we
are engaged in imagining. In particular, if the Agency Account is correct
and imaginings are really mental actions of a certain kind, it seems plausible to assume the existence of such an imaginative ability, namely the capacity to act in the required way. But it is not necessary that this ability
should be taken to amount to a distinct and independent faculty of ours. In4

See, for instance, the various attempts to define knowledge in terms of justified true
belief; or Wittgensteins related skepticism and his alternative idea of family resemblance concepts.

32

The Nature and Variety of Imagining

stead, it may simply consist in the capacity to use, in a certain way, one or
more of our more basic mental faculties (e.g., to form mental representations). In any case, the question of whether there is a faculty or capacity involved in all and only cases of imagining need not be answered by a unified account of imagining.5 Similarly, a theory of what it means to imagine
something may lead to a better understanding of the imaginativeness exhibited by creative or inventive people. But again, it is not clear whether
there is a particularly close link between imaginativeness and instances of
imagining. Accordingly, a unified account of imagining does not have to
address the issue of the imaginativeness of people or their works. It needs
to be concerned solely with states, acts and projects of imagining.

1.2. Five Central Cases of Imagining


One important issue that has been left open so far is which mental phenomena are actually paradigm instances of imagining, and which not. In this
section, I present my detailed answer to this question. Most of the phenomena concerned are mental representations. It is therefore helpful before I
begin to engage with the task just outlined to clarify what I have in mind
when speaking of representations. That a mental state is a representation
or representational means that it is directed at or about certain entities
and represents them as being a particular way, or represents them under
certain aspects.6 This notion of representational states allows for various
types of representation, whether they are propositional or object-directed,
5

There has been a long tradition of assigning the faculty of imagination a particular
role in our cognition of the world (see Hume (1739) and Kant (1781); for further
discussions, see Strawson (1970), Warnock (1976): Part I, and Sellars (1978)). However, the faculty in question is not exclusively related to the formation of imaginative representations, but usually said to be also centrally involved in the occurrence
of, say, perceptions or memories. Hence, it cannot be interpreted as the faculty or capacity to imagine; and its discussion need not be part of theory of imagining.
OShaughnessy ((2000): ch. 11) provides a general discussion of the relationship
between imaginative representations and the capacity to imagine.
Representations may also have a referent, and be correct relative to whether how
they represent entities as being satisfies a certain standard of correctness (e.g., the
relevant facts in reality, or ones intentions). But I would like to stay neutral here
concerning the issue of whether all types of representation can refer or be correct.

Unified Accounts of Imagining

33

conceptual or non-conceptual, sensory or intellectual,7 visual or auditory,


and so on. It also permits that representations may involve a certain
attitude towards what is represented (such as the attitudes of belief, desire,
hope or fear, as well as the related directions of fit), or a certain affective
element (such as a particular instance of emotional arousal).
The proposed notion of representations is thus broad enough to cover all
sorts of mental phenomena: perceptions, thoughts, beliefs, memories, emotions, desires, hopes, imaginings, and many others. In particular, it captures
what is commonly labelled intentional states (see Husserl (1901): part V,
Searle (1983): ch. 1, and Crane (2001): ch. 1). 8 But the outlined class of
representations need not be exhausted by intentional states. It is natural to
maintain that intentionality implies the possibility of genuine
misrepresentation: that is, the possibility of representing an object incorrectly, while still representing it and perhaps even referring to it (see Dretske (1986)). And there might be mental representations which do not allow
for this possibility of error. That is, there might be mental states that are
about certain objects and ascribe certain features to them, but which cannot
misrepresent or misrefer and, in this sense, lack intentional content (e.g.,
perceptions, as conceived by some disjunctivists, such as Martin (2002a) or
Travis (2004)).
That a mental state is a representation implies that it possesses a content.
7

My contrast between sensory and intellectual episodes is meant to track the basic
difference in representationality between perceptions and judgements (or beliefs),
given that they may be about the same objects and features. Although I would like to
stay neutral towards the issue of whether the difference in question is one of degree
or of kind, I assume that there is some distinction to be drawn, as it has been accepted in the tradition at least since Kant. In addition, I take this difference to be more
fundamental than but presumably overlapping and perhaps even coextensional
with some of the various differences to be found in the more recent literature,
which have been characterised in terms of the difference between conceptual and
non-conceptual (e.g., Peacocke (1992): ch. 2; Crane (1992)), digital and analogue
(Dretske (1981): ch. 6) or propositional and non-propositional (or object-directed)
representation (OShaughnessy (2000): ch. 10).
My notion of a representation is thus very close to Lockes notion of an idea,
Humes notion of a perception, Kants notion of a representation (i.e., Vorstellung), or Husserls notion of intentional act. It is perhaps less close to certain contemporary uses of the term representation, in particular to those employed in cognitive sciences.

34

The Nature and Variety of Imagining

Its content may be taken to consist in what it represents, that is, in the represented entities and features. Accordingly, the judgement that it rains, the
supposition that it rains, the desire that it rains and joy about the fact that it
rains all have the same content, while the judgement that it snows has a
different content. But we sometimes also refer to the type of representation
in question when we talk about the contents of representations for instance, when we note that the visual content of the perception that it rains
differs from the intellectual content of the corresponding judgement. I follow both usages of the term content and hope that the context will clarify
which one I have in mind.
In either case, the content of a mental representation strictly contrasts
with any potential attitude towards what is represented (e.g., whether it is
believed or desired), as well as with any affective element (e.g., any emotional feeling) possibly involved. Besides, I understand the representational
element of a mental representation as precisely that concrete aspect of it
which realises its content and its type of representation. Two representations, which represent the same entities and features and are of the same
type of representation, involve the same representational element for instance, a certain visual image, or the entertainment of a particular proposition.9
Now, that certain kinds of mental phenomena are central cases of imaginative or non-imaginative phenomena, respectively, means that their possession or lack of an imaginative character is not subject to serious disagreement, neither in daily discourse nor in philosophical debate. It is important to note, however, that whether something is an instance of imagining is not merely and presumably not even primarily a matter of whether we label it as such. Our use of terms like imagining can vary greatly.
9

My broad use of the term visual image (or of the related expression visual imagery) is perhaps unorthodox, given that others reserve it for visual memories and
imaginings alone and do not apply it to visual perceptions (see Sartre (1940)). In addition, it is not meant to imply the presence of any kind of mental or internal pictures, as they have been criticised by Sartre (1936): part 2, Sartre (1940): ch. 1, Wittgenstein (1984d): sec. 620ff., especially 638 and 642, and Ryle (1949): ch. 8, esp.
sec. 3. Perhaps some of the positions which endorse what I call an echo thesis
(roughly, the claim that imaginings depend on cognitions in such a way that they are
both similar to and different from them) may be taken to embrace the idea that visualising involves internal pictures.

Unified Accounts of Imagining

35

And they are not always used to denote imaginative representations. For
instance, the statement could you really have imagined that can, in some
situations, question the imaginative skill of someone (e.g., in response to
hearing a great story, or to the assertion of a person that she had no problem visualising a complex scene); but in others, it can cast doubt on the capacity to have previously believed or expected something (e.g., a surprising turn of events). Hence, the agreement about the paradigms of imaginative and non-imaginative phenomena, that I have in mind here and aim at,
concerns the nature of the respective mental phenomena, which is taken to
be largely independently of how we refer to them.
There are very many mental phenomena which are clearly not imaginative. Most prominent among them (especially in discussions about imagining) are cognitive phenomena: on the one hand, cognitive states, such as
perceptions, judgements10, beliefs or memories; and, on the other hand,
cognitive projects, such as trying to come to understand another person, or
finding the best answer to the mind-body problem. Other central cases of
non-imaginative phenomena are desires, emotions, intentions, moods, inclinations, the practical projects of deciding on a course of action, or of
calming oneself down, and so on. Any theory of imagining which cannot
distinguish such cognitive and other non-imaginative phenomena from
imaginative ones should not count as providing a unified account of imagining.
What I take to be central cases of imagining, on the other hand, is perhaps best illustrated by considering an example. Suppose I ask you to imagine that Rome is covered by a thick layer of snow. In case you intend to
follow my instruction, it is quite likely that you start to visualise how certain parts of the city might look, their appearance transformed by the
masses of snow. You may thus picture the Piazza Navona, or one of the
narrow streets in the old centre, as brightly white, and void of most of the
people and all kinds of traffic. You may also auditorily imagine how the
10

I intend to use the term judgements to refer to episodes of judgemental thought,


rather than to acts of judging, that is, instantaneous events which constitute the onset
or formation of such judgemental thoughts (see Dorsch (2009b)). Moreover, I take
occurrent beliefs (i.e., beliefs manifest in phenomenal consciousness) to be judgemental thoughts as well and apart from issues which bear on consciousness assume that what I say about judgements applies also to beliefs (and vice versa).

36

The Nature and Variety of Imagining

snow swallows and mutes the sounds, or how it makes a distinctive sound
when pressed down by your feet while you begin, in your imagination, to
walk through the city. You are thereby perhaps able to imagine feeling the
resistance of the snow to your feet, the texture and thickness of its layer,
the harshness and freshness of the cold, how it slightly hurts your almost
numb skin, and the pleasure and exhilaration at experiencing this rare and
beautiful scenery. Or you may begin to imagine what you would like to do
in such a situation, or where you would feel like going.
But instead of, or in addition to, summoning up certain experiential
qualities in your mind, you may also simply begin to hypothesise about the
consequences of the highly unusual setting for the citys completely unprepared inhabitants, on the basis of imagining that the streets and squares of
Rome have disappeared beneath large amounts of snow. Thus, you may
wonder how they might cope with the unexpected situation and, in particular, how they might manage to move around and keep the basics of daily
life running. You may imaginatively think through different options of
dealing with the rare and surprising circumstances, and you may imagine
how it would feel for the inhabitants of Rome to handle the situation,
whether they would be taken up with the same enjoyment that you have
imagined yourself to feel (or that you may actually feel while imagining),
or instead would despair with or be annoyed by the unexpected disruption
to their normal life which leaves them stranded in their homes. Finally, you
may engage in imagining these things about Rome and its inhabitants, not
in response to my invitation to do so, but in response to reading a work of
fiction in which a scene is, in some detail, described in terms similar to
those used above.
This complex example of imaginative engagement on your behalf contains all five cases of imagining that I consider to be paradigmatic: (i) sensory imaginings; (ii) affective imaginings; (iii) imaginative thoughts; (iv)
experiential imaginings; and (v) imaginative projects. Being a member of
one of the five groups should not be taken to necessarily exclude being a
member of another one. As will become clear, it is very likely that there are
considerable overlaps between the five different types of imagining. In addition, this list is not necessarily meant to be exhaustive: it might not capture all cases we would agree on as central cases of imagining. But to my

Unified Accounts of Imagining

37

mind, there is no obvious candidate for a further central type of imagining.


In particular, the list seems to be able to cover all the forms of imagining
involved in the example just presented each of which I now discuss in
turn.

(i) Sensory Imagining


First of all, there are instances of sensory imagining: such as visualising
something and its counterparts in other sense modalities (including some
forms of bodily sensation), such as hearing something in ones mind or
imagining the texture of an object (e.g., see Casey (1976): 43ff.; Peacocke
(1985): 22; Martin (2002a): 403). While following my instructions, you
may in your imagination have seen the snow in the streets, heard the
sounds of your steps in it, felt how it resisted but eventually gave way to
your feet, and sensed the coldness of the air. In accordance with these examples, sensory imaginings may be characterised as mental episodes that
possess a sensory content which is imaginatively entertained (whatever
that may mean).
But the content of sensory imaginings may include more than the sensory representation of objects and their features. For at least some sensory
imaginings may amount to imagining an episode of perceiving, that is,
have a perceptual experience as part of their content (from which they inherit their sensory representationality). There is an ongoing debate about
whether instances of sensory imagining are in fact experiential in the sense
of having episodes of perception as their direct object.11 I present two argu11

This idea Berkeleian in spirit is defended by Peacocke (1985) and, at least for
some central cases, in Chapter 10 of this book and by Martin (2002a): sec. 3 and
Martin (2001): sec. 3, as well as Dorsch (2010c). It comes under attack in Williams
(1966), White (1990): chs. 13-15, Hopkins (1998): ch. 7, Noordhof (2002) and Currie & Ravenscroft (2002): sec. 2.2. Note that it may seem natural to one to describe
visualising as imagining seeing without being committed to the claim that seeing is
part of what is imagined (see Wollheim (1973): 55; Hopkins (1998): ch. 7). Besides,
there is also the (independent) debate about whether sensory imagining or experiential imagining, for that matter sometimes or even always involves an intellectual
element, such as an imaginative thought or intention-in-action. This debate is closely
related to the question of how best to understand the multiple use thesis, according

38

The Nature and Variety of Imagining

ments in support of this claim in Chapter 10. But, for the time being, I intend to stay neutral on the possibility of non-experiential sensory imaginings, that is, of sensory imaginative episodes that do not have perceptions
or similar episodes as part of their contents.
Many discussions of sensory imaginings have focused on visual instances (e.g., see Peacocke (1985); Sartre (1940)), and I mainly follow
them in that respect. Instances of visualising a face or a landscape thus figure as paradigm cases of sensory imagining. Of course, there are important
differences between, say, visualising something and hearing or touching it
in ones imagination, as there are important differences between seeing it
and hearing or touching it in real life. But I presume that these differences
do not bear much on the issues linked to the possibility of formulating a
satisfactory unified account of imagining. Also, it is interesting to ask
whether there are kinaesthetic or proprioceptive imaginings; whether these
are similar in structure to cases like visualising; and whether they amount
to imagining kinaesthetic or proprioceptive experiences. But my hope is
that they do not differ in important respects from the much more investigated example of visual imagining. So far, there does not seem to be a good
reason to doubt that a unified account of imagining capable of capturing
visual imaginings is also able to capture other types of sensory imagining.
Let me thus say a bit more about the nature and variety of visual imagining comments which I hope will apply, to the relevant extent and with
the necessary modifications, to instances of imagining in other modalities.
One way of visualising something and presumably the easiest is to stop
seeing (e.g., by closing ones eyes) and to concentrate on the production of
the visual image. But we can also visualise something while seeing (as
well as thinking about) something completely different (see Wittgenstein
(1984c): vol. II, sec. 65). For example, while walking down the street, we
may visualise the face of a friend whom we plan to surprise with something in the near future. However, this may be more difficult than the first
way of visualising something (e.g., the result may be less rich in detail or
in determinacy), and require more effort in particular, in that we have to
to which the same imaginative visual representation may figure in different instances
of imagining (see Noordhof (2002), from whom I borrow the label, as well as Martin
(2002a)). I address this question and the respective debate in the literature in Section
3.6.

Unified Accounts of Imagining

39

sufficiently shift our attention from the perception (or the thinking) to the
imagining.
Furthermore, there are cases in which the content of a perception
provides us with the material for the sensory imagining. Visualising a perceived face as it would look in old age, or when looked at via a distorting
mirror, are examples. But the two episodes are still metaphysically independent of each other: each could occur without the other, and their contents are distinct. This remains true even if there is for us an explicit link
between the objects of the two representations for instance, if we think
them to be spatially linked by visualising an object as being a part of our
currently perceived environment (what Hopkins (2011c) calls visualising
into seen space). This is the case, say, when we see a white wall before us
and visualise how it would look like if a certain painting were hanging
there. Again, the two representations and their contents are distinct and do
not merge in any sense. In particular, we still see the whiteness of the wall
exactly where we visualise the painting to be. The case would be different
if we stopped seeing the white wall (e.g., by closing our eyes) and recalled
its appearance in order to visualise it together with the painting. 12 These
considerations are in fact related to the sometimes endorsed idea that we
cannot have a visualisation with the same sensory content as a simultaneous seeing. For instance, Wittgenstein writes that:
While seeing an object, I cannot imagine it. (Wittgenstein (1984c): vol. II,
sec. 63)13

The idea is that when we already see the National Gallery in front of us we
That perceptual and imaginative representational elements do not or at least not so
easily merge also constitutes a problem for imagination-based accounts of pictorial
experience which take our visual awareness of what pictures depict to be imaginative and to be merged with our perceptual awareness of the depicting surface. For the
proponents of this view have problems to explain why perceptual and elements
merge in the pictorial case, but not in examples like the one discussed here (see
Hopkins (1998): 3 and Dorsch (2012c)).
13
The translation is mine. The original reads: Whrend ich einen Gegenstand sehe,
kann ich ihn mir nicht vorstellen. The same point is stressed by Mulligan (1999):
sec. 3.
12

40

The Nature and Variety of Imagining

cannot also visualise it at the same time (and, presumably, from the same
point of view). Similarly, the claim rules out the possibility of first visualising the National Gallery and then coming to also and simultaneously see
it (again, presumably from the same perspective). According to one possible way of understanding this thesis, it seems to be true: the sensory representation involved in seeing the National Gallery cannot also be the representation involved in visualising this building; the two do not simply
share a token visual representation. It may also be true in another
interpretation: it may be that (as already suggested above) visualising
something while seeing something else presupposes sufficient attention to
the task of visualising, and that we may lack this required degree of attention when we, say, observe very carefully how the National Gallery really
looks. But a third reading is at least questionable, if not false: namely that
we cannot simultaneously have two independent visual episodes with the
same content and point of view, when one is perceptual and the other imaginative.14 If we can visualise a face (or anything else) while looking at the
National Gallery, it should also be possible that we can instead visualise
the building while continuing to have our perception. For if there is any
feature of visualising which prevents it under all circumstances from occurring at the same time as seeing, it should better not depend on what is
the particular content of the imaginative episode.15
See Hopkins (2011a). Hopkins also observes that visualising something is incompatible with simultaneously visually remembering something (else), as well as that we
cannot have two simultaneous, but separate episodes of visualising (or visual recall)
that is, two episodes that occur at the same time, but visually present their objects as
part of distinct imagined spaces or worlds. His explanation of these observed facts is
that visual recall is a form of visualising that is constrained by the determination of
what is visually represented by what has been visually perceived in the past. My alternative explanation is that both visualising and visual recall involve the same kind
of experiential representation of visual perception (see Section 9.4 and Chapter 10).
15
For example, it might be argued that we generally cannot enjoy an episode of seeing
and an episode of visualising at the same time because the same neurons or regions
in the brain would be involved in the realisation of the two episodes. But claiming
that such a conflict on the neural level prevents us only from simultaneously seeing
and visualising the National Gallery, but not from simultaneously seeing the National Gallery and visualising a face would mean claiming that it is possible to
identify neurons or regions in the brain that are responsible for the visual representation of very specific objects or features in such a way that no other neuron or region
14

Unified Accounts of Imagining

41

(ii) Affective Imagining


The second paradigm case of imagining, that your imagining about Rome
may have typified, is affective imagining: that is, the case of imaginative
experiences of emotion, desire or sensation which are characterised by
their involvement of an affective element (i.e., some form of emotional,
conative or similar feeling), often coupled with a related content. 16 You
may have imaginatively felt the pleasure of walking in the snow and looking at it, the annoyance that less relaxed inhabitants of the city might have
in reaction to the sudden and heavy onset of winter, or the longing to walk
over the untouched field of snow in a square and to start a snowball fight.
Other examples of affective imaginings are the imaginative counterparts to
bodily sensations, such as imagined or imaginatively felt itches and pains
(see Martin (2002a): 406f. on the former). Whether there are also desirelike imaginings that have occurrent desires as their counterparts and do not
merely consist in the imaginative thought that one has a certain desire is
more controversial.17
Affective imaginings are apart from their affective character not that
different from sensory imaginings. This concerns, in particular, the issue of
whether affective imaginings are always experiential imaginings of affective experiences, or whether there can also consist in simple imaginative
episodes which do not have real feelings or experiences as their intentional
objects. Again, I present some considerations in favour of the idea that affective imaginings are imaginings of real feelings in Chapter 11.
It is important to note, however, that our real emotional responses to fictional entities and situations (i.e., what Walton (1990) calls quasi emocould realise the same type of visual representation something which is perhaps
not very likely from an empirical point of view.
16
Although affective episodes may be picked out by the term feeling (and I will
sometimes do this), it is important to have in mind that this does not mean that they
comprise all phenomenally experienced or consciously felt states (i.e., all mental
episodes). Besides, I stay neutral on the content of affective and, especially, conative
states for instance, whether they are propositional or instead agentive (i.e., of the
imperative form (to) do X).
17
The postulation of such desire-like imaginings is defended by Velleman (2002), Currie (2002) and Currie & Ravenscroft (2002). See Carruthers (2003), Nichols (2004b)
and Soldati (2006) for good objections.

42

The Nature and Variety of Imagining

tions) are not instances of affective imagining. In particular, although


these responses are based on imaginings, they are not imaginative themselves. When we watch a horror movie and begin to feel afraid and have
some corresponding bodily reactions, thoughts and inclinations to act
(though usually not the same as when confronted with real danger), our
feeling of fear and the respective bodily changes (e.g., a rising heartbeat)
or inclinations to act (e.g., to cover ones eyes) are real, not imagined
even though, typically, they are responses to imagined entities and are
prompted by imaginative representations. There is a debate about whether
such quasi emotions should count as real emotions. But it should be beyond doubt that these states involve real emotional (or emotion-like)
arousal, and not only an imagined or imaginatively felt one. Besides, as the
relevant discussions reveal, affective imaginings are not necessary for our
emotional engagement with fictions.18

(iii) Intellectual Imagining


Your imaginings in the example are also likely to have included imaginative thoughts, such as the general assumption that Rome is covered in snow,
or hypothetical thoughts about the possible consequences that this might
have for the life of its inhabitants. Supposing, imagining-that and makingbelieve with a propositional content are also cases of intellectual imagining.19 Although there presumably are cases in which we experientially imaI return to these issues in Chapter 11. See also Walton (1990): sec. 5.3 and ch. 7,
Moran (1994), the excellent survey in Levinson (1997): especially 22ff., Walton
(1997), and the other essays in Hjort & Laver (1997).
19
Perhaps some or all of these kinds of intellectual imagining are identical with each
other (see Currie & Ravenscroft (2002): sec. 2.4; and Kind (2011)) but see Gendler
(2000b) and (2006), as well as Weatherson (2004), for arguments to the contrary. For
the claim that supposition is imaginative, see Furlong (1961): chs. 2 and 3, Casey
(1976): 42ff., Peacocke (1985): 20, and Currie & Ravenscroft (2002): 33ff.. For the
claim that making-believe is a form of imagining, see Walton (1990): ch. 1, and
OShaughnessy (2000): 342ff.. There are uses of imagining, assumption or supposition which clearly do not refer to intellectual imaginings, but instead to (false)
beliefs, expectations, and so on (e.g., he only imagined it, everyone is assumed to
be willing to pay for dinner or old people are supposed to become forgetful; see
White (1990): ch. 16, especially p. 136; see also Scruton (1974): 95). When I use the
18

Unified Accounts of Imagining

43

gine intellectual episodes (e.g., we can imagine the experience of drawing


a certain conclusion), intellectual imaginings do not usually seem to be of
this kind (see Section 9.5).
Intellectual imaginings have in common with other intellectual representations that they may be understood as relations to propositions and can
stand in inferential relations (see Currie & Ravenscroft (2002): 11ff.).
However, hypothetical or imaginative inferences (i.e., inferences which
start off from at least one imaginatively entertained premiss) differ from
cognitive inferences (i.e., those involving only judgements or beliefs as
starting-points) in one significant respect: the former cannot lead to judgements or beliefs, but only to intellectual imaginings (see Currie &
Ravenscroft (2002): 14). On the basis of supposing that Milan is the capital
of Italy and knowing that the president of Italy resides in the capital, I cannot acquire the belief that the president of Italy resides in Milan (at least
not, if I am rational and do not mistake my supposition for a judgement).
That means there is a logical restriction on the conclusions that we can
draw from imaginatively entertained premisses.
Relatedly, there are important differences between isolated intellectual
imaginings and those which occur in close relation to other (intellectual)
imaginative and non-imaginative phenomena say, in response to fictions,
or in the context of story-telling, games of make-believe, thought experiments, hypothetical reasoning, and so on. For instance, we may encounter
logical or moral restrictions on what we can imagine when we are engaged
in building a system of rationally linked intellectual imaginings (see Wollheim ((1973): 69ff.) and Eco (1994) for the influence of logical and Gendler (2000) for the influence of moral forces). Furthermore, these differences may warrant the use of different terms for isolated and embedded intellectual imaginings.20 But this does not undermine the fact that all the respective episodic representations are instances of intellectual imagining,
whether they occur on their own or in conjunction with other episodes.
Whether I simply assume that it rains or imagine it as part of reading or
20

terms, I mean to denote only cases of intellectual imagining.


For instance, Gendler distinguishes between supposition and imagination (Gendler (2000b): 80f.); while Velleman suggests a similar distinction between fancifully
imagining and tentatively or hypothetically assuming something (Velleman
(2000): 183; 251f.)).

44

The Nature and Variety of Imagining

hearing a story does not bear on the issue of which basic category of mental episode the intellectual representation in question belongs to, even if its
formation may be constrained in different ways, depending on its mental
context.21

(iv) Experiential Imagining


The fourth central type of imagining is experiential imagining: it captures
instances of imagining which have mental episodes as their directly represented objects.22 You may not have simply visualised the streets of Rome,
but imagined yourself walking along them and seeing them from a particular point of view; you may have imagined feeling uncomfortable in the
cold, or imagined having the feelings and thoughts of children playing in
the snow.
Experiential imaginings contain the contents of the imagined episode(s)
and hence involve sensory, affective or intellectual representations, depending on the nature of those contents. Imagining seeing something, for
instance, involves having a visual image of it. But experiential imaginings
make also essential reference in their content to the imagined type of episode. Imagining seeing something thus represents visual perception (or,
even more specifically, veridical or hallucinatory visual perception). The
representational link may thereby be intentional or relational, depending on
ones account of experiential imagining (see Section 9.4).
In many (if not all) cases, experiential imagining involves also an
ascription of the imagined episode to a particular person; and it may even
combine the imagined episode with other imagined episodes of (presumably) the same person. This is the case, for example, when we imagine being (or what it would be like to be) Napoleon as he watches the battle of
See also Curries and Ravenscrofts argument that the resistance to imaginatively adopt certain morally evaluative points of view (e.g., when engaging with particular
stories) pertains not only to the respective intellectual imaginings, but also to connected desires or desire-like imaginings (Currie & Ravenscroft (2002): 34ff.).
22
Experiential imaginings are often described in terms of imagining from the inside
(e.g., see Peacocke (1985): 21, and Walton (1990): 29f.). Referring to them as experiential imaginings also seems an option, but it may give the false impression that
only such imaginings are experienced and phenomenally conscious.
21

Unified Accounts of Imagining

45

Jena and thinks of Josephine, or about the next tactical moves (see Wollheim (1984): ch. 3, for an extensive discussion of such cases).
Apart from such cases of imagining experiencing something or imagining being someone, experiential imagining also includes cases of imagining doing something, assuming that it is imagined by means of imagining
the respective experiences which one would have while doing it (see Peacocke (1985): 22; Budd (1989): 100; Walton (1990): 29f.). Experientially
imagining (the experience of) playing the piano involves, for instance, imagining seeing some hands moving over the keyboard, imagining hearing
the corresponding sounds of a piano, and linking the imagined episodes to
each other and to oneself. It may also involve intellectually imagining that
one plays the piano, but cannot be simply reduced to it (see, for instance,
the discussion of Waltons view on pictorial experience in Section 11.2 and
in Dorsch (2012c)).
Sensory and, in particular, visual instances of experiential imagining
have been the subject of extensive discussions which have addressed at
least two important issues: what it means to imagine an episode, and to
imagine it to be of a certain kind and belong to a particular person; and
whether it is possible to imagine an episode (especially if it involves a perspectival representation) without also imagining it as being the episode of a
particular person (who, for instance, is imagined as occupying the respective point of view; for references, see Note 10 above). For my purposes, it
does not matter much how these two issues about the nature of experiential
imagining are settled. But I return to the first question in the Chapters 10
and 11. What I defend there are two ideas. The first is that experiential
imagining is an instance of object awareness, rather than propositional
awareness. That is, it makes us aware of concrete objects and their features
(i.e., mental episodes and their phenomenal characters), instead of the bearers of a truth-value (i.e., propositions). The second idea is that experiential
imagining may but need not be accompanied by an imaginative thought
(or thought-like element) that identifies the type or nature of the imagined
episode and, if necessary, relates it to a particular person as its bearer, or to
other imagined episodes (see also Section 3.6, as well as Peacocke (1985):
25f., and Noordhof (2002)).

46

The Nature and Variety of Imagining

(v) Imaginative Projects


When imagining Rome in winter you will have combined some or all of
the previous four types of imagining in more complex imaginative projects. You may have daydreamed about walking along the streets of the
city; you may have imagined yourself to be in the position of its inhabitants, perhaps resulting in your empathizing with them, or in your imaginatively living through their experiences; or you may have imaginatively engaged with the fictional (or perhaps also historical) description of these or
similar scenarios. This group of imaginative phenomena is the most varied
and comprehensive, not the least because there is not much of a limit to
which and how many instances of the other forms of imagining (as well
as cognitive and other non-imaginative episodes) we can combine with
each other in imaginative projects. Examples of such projects are daydreams or fantasies about oneself or other people, the imaginative invention in ones head of a story or melody, empathy and similar phenomena, or
our imaginative appreciation of representational artworks.23
Many imaginative projects are revolves around instances of experiential
imagining (see Peacocke (1985): 24ff., Hopkins (1998): ch. 7, and Noordhof (2002)). Daydreaming about being Napoleon is likely to involve
23

As already noted (see Notes 3f. in the Introduction), most comprehensive discussions of imagining focus exclusively on imaginative episodes or (simple) experiential imagining, while neglecting imaginative projects. Notable exceptions are Currie
& Ravenscroft (2002: 38ff.) on fantasies and Walton (1990) on our extended imaginative engagement in games of make-believe, or with representational artworks.
OShaughnessy (2002: ch. 5) discusses daydreams, but without linking this discussion to his account of imagining. Scruton (1974), Casey (2000) and McGinn (2004)
mention imaginative projects (e.g., daydreams, reveries, or fantasies) only in
passing. By contrast, there are also some texts which focus more or less exclusively
on imaginative projects, rather than imaginative episodes. Williams (1966), Wollheim (1973), Wollheim (1984: ch. 3) and Velleman (2006) are concerned with the
various ways in which we can imagine (or daydream) something about ourselves, as
well as with the question of what it means and to which extent it is possible to
imagine being someone else. Relatedly, Goldie (2000: ch. 6) provides a detailed
elucidation of empathy (which he understands, roughly, as experientially imagining
oneself as having the thoughts, feelings and emotions of another person) and its relatives, such as imaginatively putting oneself in the situation or the shoes of another person.

Unified Accounts of Imagining

47

experientially imagining the sensations, perceptions, and so on, of Napoleon in specific situations; and empathising with another person involves
her thoughts and feelings. But not all imaginative projects include experiential imagining, or at least not as their primary ingredient as, for instance, in the case of inventing in ones mind, or of imaginatively telling, a
third-personally narrated story. Besides, the simplest instances of experiential imagining such as imagining seeing (or visually experiencing) a tree
are mereologically too simple to count as projects (see Section 10.3f.,
Hopkins (1998): ch. 7, and Dorsch (2010c)).

1.3. Mental Episodes and Complexes Thereof


What all of the instances of these five types of imagining have in common
is that they occur in phenomenal consciousness. It may thus be helpful to
consider for a moment what it means for mental phenomena to be phenomenally conscious (or occurrent), namely: (i) to be a constitutive and continuous part of the stream of consciousness; (ii) to have a specific phenomenal character (i.e., a what-it-feels-like aspect) which characterises the way
we consciously and subjectively experience in the Husserlian sense of
Erleben the mental phenomena in question; and (iii) to be actually or
potentially subject to attention and introspection.24
But the elucidation of conscious mental phenomena and of their phenomenal characters should not stop here. First of all, the complexity or detailedness of the latter may differ. For instance, the experience of conscious
thoughts may be less rich or less specific than the experience of visual perceptions (e.g., because of the type of representation involved).
Then, phenomenologically salient differences need not coincide with, or
be as specific as, our distinctions of different types of mental phenomena.
Disjunctivists with respect to perceptions, for example, claim that veridical
24

It has been denied that there is any such thing as phenomenal consciousness (e.g.,
see Dennett (1988)). If this turned out to be true, my comments should be regarded
as concerning the phenomenal characters that the respective mental phenomena
seem to have. I discuss the nature of phenomenal consciousness in more detail in
Dorsch (2012a). For characterisations of phenomenal character similar to mine, see,
for instance, Block (1994), Carruthers (2000): 13f., Siewert (2000): sec. 3.10, and
OShaughnessy (2000): 534.

48

The Nature and Variety of Imagining

and non-veridical perceptions are different kinds of states, despite their


first-personal indistinguishability (see Martin (2002a: sec. 5); and Dorsch
(2011a) for an overview of different kinds of disjunctivism). On the other
hand, two visual perceptions, one representing a car as blue and the other
representing a car as red, are phenomenally distinct but none the less at
least on a fundamental level are of the same type, namely visual perception.
In addition, there may be limits to the distinctiveness of phenomenal
characters. It may be doubted, for instance, that there is any difference in
how we experience the judgement that it rains and the judgement that it
snows. In particular, it may be argued that there is no distinctive aspect
which the phenomenal characters of, say, judging that it rains, supposing
that it rains, and wondering that it rains have in common as there is, say,
a distinctive aspect shared by the phenomenal characters of perceptual,
mnemonic and imaginative red-experiences.25
Furthermore, even if it is accepted that all instances of two kinds of
mental phenomena have different phenomenal characters, it may still be
maintained that this difference is not due to a distinctive way in which we
experience all the instances of each kind. How we experience watching
television may always differ from how we experience reading a book, but
that does not imply that there is a characteristic and identifiable phenomenal aspect shared by all our experiences of watching television, and by no
other episodes. Similarly, there may be no such thing as a distinctive phenomenal aspect common to all visual representations or even all visual
perceptions of cars.
Finally, the phenomenal character of a given mental phenomenon (and
the related introspective evidence) may not always adequately reflect the
nature of that phenomenon. For instance, it may seem to us that we are
visually remembering a face of a long-forgotten friend, while we are in fact
(partly) visualising it. It may even be that some of the errors involved in
25

Considerations in favour of such a position may be found in Wittgensteins private


language argument (Wittgenstein (1984a): sec. 243-315; see also Budd (1989): ch.
3), or may derive from an externalist account of semantic content (as it can be found
in, say, Putnam (1975) or Burge (1979)). For a defense of a distinctive phenomenal
difference between judging that it rains and judging that it snows, see Soldati &
Dorsch (2005).

Unified Accounts of Imagining

49

how we experience (or introspect) the phenomena occurring in our stream


of consciousness are of a systematic nature. Theories of perceptual experience, for instance, may have to say this about how we experience perfect
hallucinations (i.e., hallucinations that are subjectively indistinguishable
from perceptions), given that it seems to be part of the phenomenal character of such experiences that they are phenomenally marked as veridical
perceptions presenting us with mind-independent, physical objects.26
Now, it is distinctive of mental episodes that they are the simplest distinct and self-contained units which can occur independently of each other
in phenomenal consciousness. In contrast, if a distinct and self-contained
conscious phenomenon allows for an analysis into elements which could
occur in phenomenal consciousness on their own, it is not episodic, but
complex.27 Such mental complexes unify in themselves a set of mental
episodes. Since nothing in what follows depends on whether episodes
and, thus, complexes should be construed as mental events or mental
states, I intend to stay neutral on this issue (even when choosing to talk of
events rather than states).
What precisely gives unity to particular mental complexes and distinguishes them from mere temporal series of episodes may depend on the
kind of mental complex in question. One example of mental complexes are
aesthetic or similar experiences which may involve many different episodes, stretch over a long period of time and even be interrupted (e.g., when
In Dorsch (2010a), (2012a) and Soldati & Dorsch (2011), I argue that this is the
right description of the phenomenal character of hallucinations, and that, as a consequence, theories of perception including disjunctivist accounts have to accept
that this character misleads us about the true nature of hallucinations: they are essentially non-relational, while being phenomenally given as perceptual relations. Martin
(2002a) restricts the same conclusion to intentionalist and sense-data views, partly
because he assumes that, while hallucinations are subjectively indistinguishable
from relational perceptions, they do not possess the same phenomenal character as
the latter and, therefore, are not phenomenally given as relational. I object to this
phenomenological description and its consequences in the works cited.
27
There may be some unclear borderline cases between mental episodes and mental
complexes for instance, visualising a moving object, visually imagining moving
around, or the most simple instances of experiential imagining (e.g., imagining seeing a static face). In most cases, however, it should be clear whether a given mental
phenomenon can be further analysed into independent elements.
26

50

The Nature and Variety of Imagining

we read a novel bit by bit, or explore a city or landscape over several days
or months). Such experiences are unified by their common object, plus other factors (e.g., the appreciators intentions and memories; see Dorsch
(2000): ch. 1). Another, here more important type of mental complexes are
mental projects such as the imaginative ones discussed above which are
unified by their intrinsic purpose and the corresponding underlying motivational states (see Section 14.2). However, mental episodes can
temporarily overlap, occur simultaneously or form temporal series with
other mental episodes without also thereby forming mental complexes,
such as aesthetic experiences or projects.
Applying these considerations to the five central forms of imagining
identified above, we get the following results. The first three forms of imagining are clearly episodic in character. Intellectually imagining that it
rains is a mental episode; as is visualising a tree, imagining an itch or imaginatively feeling pleasure. Imaginative projects, on the other hand, are
mental complexes: they are unities made up of mental episodes. Experiential imaginings are not so easily classified, partly because their nature is
still unclear. The more sophisticated instances (such as the case involving
imagining being Napoleon) are presumably mental complexes as well.
They involve more than one imagining of a mental episode, and typically
also more than one imaginative thought. But elementary examples of experiential imagining (such as imagining seeing a tree, or even visualising a
tree if the latter is to be identified with the former) may very well be episodic.
Whether simple instances of experiential imagining are indeed episodes
rather than complexes of episodes depends, for instance, on whether
imagining experiencing something may fail to involve an imaginative
thought-like element; and if not, whether this element could occur independently of imagining of the experience (see Section 3.6). If experiential
imagining involves an independent episode of thought, then it is complex,
consisting of the thought and some other, presumably non-intellectual imaginative element (e.g., a sensory imagining). If, on the other hand, experiential imagining involves no or just a dependent thought, then a good case
can be made for the claim that it is episodic, given that it forms an indivisible whole. But irrespective of how this dispute is to be settled, it is true

Unified Accounts of Imagining

51

that all the central cases of imagining put forward are phenomenally conscious.

1.4. Non-Central Cases of Imagining


As noted, a theory of imagining should capture all the paradigm instances
of imagining if it is to count as providing a unified account of imagining.
However, there are many other kinds of mental states or complexes which
have at one time or another been treated as imaginative, or as involving
imagining, but never without controversy. I therefore take them to be at
best candidates for non-central instances of imagining. No unified theory
of imagining is required to be able to capture or classify them. In particular, if an otherwise satisfactory account of imaginings is in tension with
some of our opinions about whether such non-central cases are imaginative
or not, we should revise our opinions rather than the account in question.
In other words, whether a theory assumes such candidates for non-paradigmatic instances of imagining to be imaginative or not should not matter for
its evaluation as a unified account of imagining.
Still, it does count as an advantage for a given theory of imagining it if
it can say something about the either imaginative or non-imaginative
nature of these cases. Hence, it is worthwhile to take a brief look at some
borderline or otherwise controversial cases. This also helps to clarify the
extensional scope required of unified accounts of imagining. Thus, in this
final section of the chapter, I address the possibility of dispositional imaginings and, in turn, the possible imaginativeness of dream representations,
of psychologically unusual or pathological cases, and of states concerned
with the non-actual in a rather fanciful manner (e.g., unrealistic hopes,
wishes, anticipations, or speculations).28
28

I discuss two other candidates for non-central cases of imagining in much more detail in Dorsch (2012c) and Section 14.4: namely, respectively, pictorial experiences
and spontaneous representations. The reason for spending more time on the former
is that the postulated imaginativeness of pictorial experiences has been very prominent in recent discussions (see the impact of the theories expounded in Scruton (1974)
and Walton (1990), as witnessed, for instance, in Stock (2008)); while the reason for
spending more time on the latter is that they are very similar to imaginative phenomena, despite not being instances of imagining themselves.

52

The Nature and Variety of Imagining

(i) Imaginative Dispositions


While all instances of imaginative thought are episodic, they might not exhaust the class of intellectual imaginings. For instance, we can store the
piece of information that Sherlock Holmes lives at 221b Baker Street
without having to be constantly conscious of it. We may not have thought
about Sherlock Holmes for years, but still be able to readily answer the
question of where he lives. Moreover, when we do come to think that Sherlock Holmes lives at 221b Baker Street (e.g., because we are responding to
the question), our thought is presumably imaginative: we do not really believe that Sherlock Holmes lives there and that we could visit him in his
flat. This may be taken to suggest that there is a mental state that stores the
piece of information in question, and which, despite being itself phenomenally unconscious, can become manifest in phenomenal consciousness in
the form of an imaginative thought. Following common terminology, I take
such a state to be a mental disposition (though standing state may perhaps
be a more appropriate label: see Martin (2001): p. 265, n. 10). Beliefs, desires, emotions, and so on, are typical examples of dispositional states;
while, here, we are apparently concerned with an imaginative disposition.
Whether the interpretation of the state under consideration as an imaginative dispositional state is, indeed, adequate depends primarily on whether
it should really be understood as a state that stores information about Sherlock Holmes and Baker Street, rather than as a state that stores information
about the nature of the writings of Conan Doyle. If the first interpretation is
to be preferred, it seems unavoidable to treat it as an imaginative dispositional state. But, in the second case, the state in question turns out to be a
However, I do not have the time here to say anything about another important
potential candidate for imaginativeness: namely the phenomenon (or cluster of
phenomena) often referred to as seeing-as or seeing an aspect. It seems beyond
doubt that the instances of this phenomenon involve some kind of agency. But it
may be argued that the agency involved is closer to the activity of attention than to
imaginative activity: that is, that it concerns less the active and direct formation of a
representation with a specific content and more a shift of focus from one perceivable
aspect of an object to another. The view that seeing-as or seeing an aspect
involves imagining is defended by McGinn (2004): ch. 3, and Scruton (1974): 92;
ch. 8. I argue against this in Dorsch (2012c). See OShaughnessy (2000): ch. 14, for
an account of the active (and non-imaginative) character of attention.

Unified Accounts of Imagining

53

belief for instance, the belief that the writings describe Sherlock Holmes
as living at 221b Baker Street; or the belief that they are such that a suitable appreciation of them requires us to imagine that Sherlock Holmes
lives at 221b Baker Street; or simply the belief that we ought to imagine
that Sherlock Holmes lives at 221b Baker Street.
Here is not to the place to settle this debate. 29 But it is important to note
that many cases, that initially may appear to be instances of dispositional
imaginings, are in fact rather examples of occurrent imaginative thoughts
that stay in the background of our consciousness or attention, after having
been formed in the foreground. When I am daydreaming that I am rich, I
continue to have the respective premiss in mind, even if I stop constantly
paying attention to it; just as my thought that I find the person opposite me
very attractive stays in my consciousness during the whole duration of my
rendezvous with her, despite often receding to the background.
Independent of this, even if dispositional imaginings turn out to exist,
they are bound to come into being only as a consequence of first having
had the respective imaginative thought notably in response to some other
existing entity (a story, an artwork, a game of make-believe, a daydream,
and so on) that determines what to imagine in the situation in question. At
least, it is difficult to conceive of other ways in which we might be said to
form and to have reason to form dispositional imaginings. Hence, it is
natural to expect that any satisfactory account of dispositional imaginings
would be dependent on an account of episodic imaginings; and that any
unified theory of imagining capturing the latter would also be able to capture the former.

(ii) Dream Representations


Now, if there really are dispositional states like the ones just discussed,
they should clearly be taken to be imaginative (though not paradigm examples of imagining). By contrast, with most other kinds of mental phenomena that might be taken to constitute non-central cases of imagining, it
29

There are good reasons to be skeptical about the possibility of dispositional imaginings. For instance, Nichols (2004a) argues against their existence; while Walton
(1990: 17f.) is undecided, but with some skeptical undertones.

54

The Nature and Variety of Imagining

is very unclear, or even unlikely, that they should count as instances of


imagining. This happens, for instance, if we do not have a clear grasp of
the nature of the respective phenomena.
Dream representations are good examples. Although they seem to be
similar to imaginings (and memories, for that matter), we do not really
know what they are, or whether they can be compared in their non-cognitive character to the relevant elements of our waking mental lives. In
particular, since they occur in a state of consciousness which is substantially different from wakefulness, they lack two closely related features that
all (central) instances of imagining have in common: the feature of being
phenomenally conscious (in the sense of wakeful consciousness), and the
feature of being introspectively accessible (see OShaughnessy (2000):
64ff.; see also Section 1.2). This difference in how we experience imaginings and dream representations is already to be sufficient to distinguish
them in nature.30

(iii) Unusual or Pathological Cases


The other type of phenomena, the imaginativeness of which is controversial because of their unclear nature, are psychologically unusual or pathological cases, such as obsessive thoughts or images, delusions, instances of
hearing voices or inserted thoughts, or the hallucinations induced by hallucinogenic substances.31 In these and similar cases, we often speak of the
See Scruton (1974: 100f.). OShaughnessy none the less defends the view that
dreaming involves imagining, given that he takes non-cognitivity to be distinctive of
both (see OShaughnessy (2000): 107 and ch. 11, and his (2002)). Armstrong argues
against a position like this by maintaining that dream representations differ from
imaginings in that they are actually cognitive (Armstrong (1968): 303ff.). McGinn,
on the other hand, maintains that at least the intellectual representations involved in
dreams involve a cognitive attitude (McGinn (2004): ch. 7). In Chapter 8, I try to undermine OShaughnessys idea that non-cognitivity is distinctive of imagining.
31
The term hallucination is often used in two very different ways (see (Dorsch
(2010a)): as denoting perceptual experiences which are subjectively indistinguishable from veridical perceptions, but not adequately caused or constituted by the respective aspects of reality (see Armstrong (1968): 297f.; Lewis (1980); Martin
(2006)); or as denoting sensory experiences which (like sensory imaginings) differ
phenomenally from perceptions in that they do not present themselves as being per30

Unified Accounts of Imagining

55

people concerned as imagining something (e.g., when someone imagines that everyone around him is part of a conspiracy against him, or
when someone imagines that the objects he sees are becoming fluid).
But it is not always clear what we mean by this; and if it is clear, we seldom have an instance of imagining in mind. In particular, we use the term
imagining to say that the subjects concerned make something up against
their will; that they wrongly believe it; that they do not realise their own involved (imaginative) activity; or that they are simply in a state which is
different from our normal thoughts, feelings, perceptions, judgements and
memories.
As a consequence, some of the examples such as delusions are better
taken to constitute forms of belief, rather than forms of imagining (see
Bayne & Pacherie (2005)). In contrast, other cases such as inserted or
compulsive thoughts are perhaps more like imaginings in that they involve the active formation of representations, although the subject does not
fully acknowledge them as the product of their own doing (see Roessler
(2001)), or cannot easily banish or terminate them. I briefly return to obsessive and similar representations in Section 13.3 and the Conclusion,
where I indicate which of them should count as imaginative, and how the
Agency Account of imagining can accommodate them.

(iv) Fanciful Expectation, Anticipation, Etc.


A last group of cases, that may be partly imaginative or involve imagining but surely do not constitute uncontroversial or paradigm instances of
imagining, is made up of phenomena such the following: hoping that
Switzerland will win the next World Cup in football (while knowing that
their chances are very slim); wishing to become the king or queen of the
United Kingdom; anticipating or fearing some terrible, but very improbable disaster; or freely speculating about why a person has behaved or receptual and concerning reality, and in that they are symptoms of some psychological
disorder, or of the use of certain drugs or stimulants (such as magic mushrooms or
LSD; see OShaughnessy (2000): 346ff.; 479). I typically use perceptual hallucination (or hallucinatory perceptions) to refer to the first, and hallucination to refer
to the second, but also hope that the context makes clear which is intended.

56

The Nature and Variety of Imagining

acted in an unexpected way, for which he or she does not seem to have any
apparent reason. These cases are closer to imagination than to cognition in
so far as they are concerned, not with real or close possibilities supported by ones evidence, but with possibilities unlikely to be realised, or more
or less unrelated to (what one knows about) reality. But, as I argue further
below in Section 5.4, their concern with possibilities whether they are
realistic or fanciful is as such not enough to mark them as instances of
imagining.
However, what may signal that they do involve imagining in one form
or another is that the states at issue are usually accompanied by a belief
about their fancifulness. If this were not the case, they would come turn out
to be delusions of some sort and would have to be treated accordingly (see
the preceding section). It is clear to most of us that only a handful of specific people have a realistic chance of becoming the next monarch of the
United Kingdom. Hence, if one of us wishes to become the next king or
queen without taking this possibility to be far-fetched, he or she will, at
least to some extent, be deluded. Now, that we take our own unrealistic expectations, hopes, and so on, to be fanciful may indicate that they are partly
imaginative in so far as they involve, as one of their elements, one or more
of the five central forms of imagining described. Wishing to become the
sovereign of the United Kingdom, for instance, is likely to involve some
imaginative thoughts about what it means to be in that position; while the
excitement involved in the desire for Switzerlands victory may be founded
on a vivid imaginative representation of how it would be like to experience
such an event.
To recapitulate, a unified account of imagining has to satisfy two important requirements: (i) it has to get the extension right with respect to the
central instances of imagining and with respect to the paradigm cases of
non-imaginative phenomena (notably cognitive ones); and (ii) it has to be
able to elucidate the imaginativeness essential to all imaginings and, possibly, also some of their other shared features. In the next few chapters, I
spell out the second desideratum in more detail by highlighting important
properties of imaginings that distinguish them from cognitions, and which
are in need of explanation by a unified theory of imagining. After that,
beginning with Chapter 5, I start to to present and assess the various ways

Unified Accounts of Imagining


in which the unity of imagining may be accounted for.

57

CHAPTER 2
Key Features of Imaginative Episodes

The debates about imagining in the literature have focused primarily on the
opposition between imaginative and cognitive phenomena.32 Since this particular contrast is of great significance in the following chapters notably
for the discussion of the Epistemological and the Dependency Accounts
it is helpful to identify and elucidate important differences between imagining and cognition, at least in their episodic form. The elucidation of
these differences also gives further support to, and deepens the understanding of, the distinction between central cases of imaginative phenomena and
central cases of cognitive phenomena, as outlined in the last chapter.
Two particular differences in kind between imaginings and cognitions
have often been noted in the philosophical literature and are of special importance in the subsequent discussion of the various accounts of the unity
of imagining. This is due to the fact that they are relatively fundamental
and concern all episodic forms of imagining. Moreover, the second difference parallels closely that between imaginative and cognitive projects.
The first difference is phenomenologically salient and consists in a difference in attitude, that is, in a difference in the stance or commitment
which the episodes in question adopt towards their content. The idea is that
only cognitive states or cognitions, as I also often say involve a cognitive attitude: only they make a claim about how things actually are.
The second difference becomes manifest in the different justificatory
and motivational power of imaginative and cognitive episodes: while we
32

See Budd (1989): 100, Currie & Ravenscroft (2002): 11, and OShaughnessy
(2000): 362ff. for the claim that the various kinds of imaginings have cognitive
counterparts.

60

The Nature and Variety of Imagining

usually are entitled and moved to endorse cognitions in belief (and, relatedly, in action), the same is normally false of imaginings. This difference
in rational role is a reflection of a difference in epistemic function or aim:
cognitive episodes possess and imaginative episodes lack the function
to ground or constitute knowledge (and, subsequently, to influence action).
The second difference is therefore comparable to the fact that cognitive
projects but not imaginative ones have the purpose to lead to the acquisition of knowledge (see Section 14.2).
The two differences have in common that they are strict differences, that
is, separate all imaginings from all cognitions. Moreover, they are perhaps
closely connected to each other in the sense that the phenomenologically
salient difference in attitude may count as a conscious reflection or manifestation of their difference in epistemic function (see Section 2.2 below).
For example, that there is a subjective difference in endorsement between
seeing and visualising something may indicate that only the first, but not
the second, has the function to inform us about reality. Such a link between
the phenomenal and the epistemic dimension of the episodes concerned
may also help to explain why we typically rely on cognitions, but not on
imaginings, when forming beliefs and intentions. The thought is that we
(do not) do so because we are phenomenally aware of the (lack of an) epistemic function (see Dorsch (2012a) and (2013)). In any case, given that
the two strict differences characterise all forms of episodic imagining, any
unified account of imagining should as part of the second desideratum introduced in the last chapter be expected to be able to accommodate and
illuminate them.
In the first two sections of this chapter, I zoom in on the specific contrast between imaginative and cognitive episodes and, in particular, on the
significant differences in phenomenal character and epistemic function,
which play an important role throughout this book. The third and final section of the chapter is then concerned with the fact that imaginative episodes
none the less share with cognitive ones the capacity to refer to real (past or
present) entities although it is still true that the two kinds of episode possess this capacity for different reasons, and that a lack of such reference is
the default case for imaginings, but not for cognitions.

Key Features of Imaginative Episodes

61

2.1. Imaginative Attitude


Let me begin by outlining the difference in attitude. Imaginative and cognitive episodes need not differ in their type of representation. Both our
imaginings and our cognitions can have visual contents, or intellectual
ones, or involve some affective elements, and so on. Nor do they have to
differ in what they represent. For instance, we can see or visualise a blue
car, judge or suppose a car to be blue, have a real or an imagined itch, or
perhaps can also really or imaginatively feel despair. However, representations can do more than be of a certain type and represent entities as having
certain features. In addition, they can and perhaps must take up a particular stance towards their content. To see what I have in mind, it is perhaps best to consider a specific example.
Intellectually imagining that my car is blue represents the car as being
blue. But it does not really claim that my car is blue: it makes no claim
about how things actually are (see Scruton (1974): 97f.). 33 Furthermore, it
is not as such to be evaluated in terms of whether my car is in fact blue:
there is nothing wrong with imagining the car to be blue simply because it
is in reality red. In these two respects, intellectually imagining that my car
is blue is similar to wondering whether my car is blue. Asking this question
to myself does not involve any claim about the specific nature of my car. In
contrast, judging that my car is blue not only represents my car as being
blue, it is also non-neutral about its really being blue: it purports to tell me
33

That I speak of perceptions, memories, judgements or beliefs as making a phenomenologically salient claim about how things actually is not intended to imply
that they represent the modal status of the relevant states of affairs (i.e., that they obtain in the actual world), and that the formalisation of their content requires the use
of a modal operator. Rather, what is meant is that their representation of the relevant
states of affairs (i.e., that they obtain) purports to be concerned with the actual world
and, hence, is to be assessed with respect to that world. The same applies to imaginings and their commitment to some imaginative (or possible) situation or world being a certain way. Rather than saying that cognitive and imaginative episodes differ
in whether they make a claim about how things actually (in contrast to possibly or
imaginatively) are, I may perhaps also have said that they differ in whether they
make a claim about how things are (in contrast to how they could be). One advantage of the way of talking chosen in the main text, however, is that it stays neutral on
whether a commitment to some imagined situation being a certain way needs to be
or even can be spelled out in terms of possibility (see Section 5.4).

62

The Nature and Variety of Imagining

how things actually are (i.e., which colour my car in fact has). Correspondingly, it is to be evaluated with regard to whether my car is actually blue:
my judgement is incorrect if the car is in fact red. Similar considerations
apply to the corresponding sensory episodes (see Martin (2002a): 391).
Visualising my car as being blue represents it as being blue, but is not committed to its actually being blue. Seeing my car as being blue, on the other
hand, involves such a commitment about the colour of my car. Accordingly, while perceptions are straightforwardly veridical or not with respect
to the actual world, sensory imaginings need not count as faulty if they fail
to match reality.
This difference between imaginative and cognitive representations is a
difference in their attitude or stance, commitment or non-neutrality towards their content.34 More specifically, it can be described in terms of the
attitude distinctive of cognitive states: a mental representation involves a
cognitive attitude if and only if it makes a claim about how things actually
are. Accordingly, the two types of mental representation differ in that cognitive states involve a cognitive attitude, while imaginative episodes do
not.35 I take it that Husserl and Sartre intend to refer to the same difference
when they say that only cognitive states involve a thetic quality (see
Husserl (1901): part V) or a positing attitude (see Sartre (1940): ch. 1).
In the case of judgements, their cognitive commitment or attitude may also be described in terms of their purporting to be true, or their involvement of an endorsement or acknowledgement of their propositional content as true, or of the corresponding state of affairs as being the case. Whether this way of talking extends naturally to perception partly depends on whether perceptual experiences are themselves
propositional, or at least possess a content with a structure very similar to that of
propositional contents (see Crane (1992)). OShaughnessy seems to be among the
very few who do not assume that cognitions and imaginings differ necessarily in
their phenomenal character or their attitude (see OShaughnessy (2000): 345; 349;
355). I discuss this apparent aspect of his view in the last part of Section 7.1. Even
Hume acknowledges that cognitions and imaginings differ both phenomenally and
in their rational or functional role (see Chapter 9).
35
This would remain true even if it would turn out that there are some non-representational imaginings (see Section 14.3 for a discussion and rejection of this possibility): since such imaginings would not have any content, they could not involve any
attitude or commitment relating to a content. In contrast, there should be no doubt
that cognitions are, indeed, always representational. Otherwise, they could not possess their epistemic function or play their actual role in cognition.
34

Key Features of Imaginative Episodes

63

Likewise, the cognitive attitude of cognitions parallels the assertoric


force involved in assertions: the latter also make a claim about how things
actually are (see Williams (1970): 137). And just as cognitions can misrepresent reality, assertions can do so as well.36
Moreover, whether a phenomenally conscious mental state possesses a
cognitive attitude towards its content is phenomenologically salient: it is
part of the phenomenal character of that state (see Martin (2002a): especially 391, on perceptions). This means that we can normally tell, simply
by reflecting on how we experience a mental episode, whether it involves a
cognitive attitude. This is crucial, for instance, for our ability to subjectively and, notably, introspectively distinguish imaginative from cognitive episodes. We can usually tell from our first-personal perspective whether we are currently perceiving a blue car, or instead visualising one. Similarly, we normally do not have any problems in telling apart our conscious
judgements and suppositions, even if they have the same content. As with
other aspects of phenomenal character, our introspective judgements and
beliefs about them need not always get the attitude of a given mental representation right. But this does not mean that we do not experience a phenomenal difference between kinds of representation such as imaginings
and cognitions that differ only in their attitude.37
Partly because we enjoy voluntary control over our assertions, and partly because
they are public statements, assertions can be misleading in a second way: we may
use their feature of claiming to represent how things actually are (or, if one prefers,
the feature of claiming to express our beliefs, that is, our take on how things actually
are) to camouflage (what we know to be) falsehoods as truths. In short, we may use
them to lie. Beliefs, by contrast, can lie at best inwardly and indirectly for instance, when we manipulate our beliefs by avoiding unpleasant evidence, or let them
be influenced by our emotions (see Papineau (1999) and Dorsch (2009b) for more
on such cases).
37
See Dorsch (2012a) for more on the distinction between phenomenal awareness and
introspection as two forms of first-personal access (or access from the inside), as
well as on the possibility of introspectively erring about the phenomenal character of
a mental episode. That we may make an error in introspection with respect to the
nature of an introspected mental representation is compatible with the idea that we
are unable to make an error with respect to the kind of state which we take the introspected representation (rightly or wrongly) to be an instance of. While we may mistake one of our visualisings for a visual memory, we cannot be mistaken about
(mis)taking it to be a visual memory rather than, say, a visual perception (see Sec36

64

The Nature and Variety of Imagining

Now, the claim that imaginings lack a cognitive attitude is neutral on


whether they possess some imaginative attitude instead (and, if they do so,
whether there is more than one imaginative attitude). But there are good
reasons to assume that imaginative episodes involve a general and noncognitive attitude that is distinctive of them and differentiates them from
cognitive episodes and their attitude. The idea is that imaginings are committed to things being a certain way, even though not with respect to the
actual world. More specifically, they are non-neutral about how things are
in the relevant imagined situation or world. Visualising an elephant in the
city means taking the imagined situation to contain an elephant and, not
say, a tiger.38
Before I consider the reasons for assuming such an imaginative attitude,
three qualifications are in order. First, this imaginative commitment may be
countered or cancelled (just as in the case of cognitions). For instance, in
addition to visualising an elephant, we may imagine as part of the same
imaginative project and, in particular, with respect to the same imagined
situation that the resulting visual image forms part of a hallucination.
That is, we may imagine hallucinating an elephant in the city in which
case our overall project is not committed to the presence of an elephant in
the imagined situation (see Dorsch (2010c)). Second, it is important to note
that the commitment involved in imagining is actual, pertaining to an actual mental phenomenon. It thus differs from merely imagined commitments for instance, the imagined cognitive commitment involved in imagining seeing or believing something (see Martin (2002a)). Third, it is perhaps possible that there are more than one kind of imaginative attitude (as
indicated in Sartre (1940): 12f., and discussed in Section 5.5 below). For
instance, the distinction between isolated and embedded intellectual imaginings elucidated above may perhaps consist in a difference in imaginative attitude: merely imagining (or thinking) something may be less committal than assuming it for heuristic purposes, or in the context of inventing
or appreciating a story (see Velleman (2000): 183; 251f.).
38

tion 5.1).
See Husserl ((1912): 172f.) on the idea that imaginings involve the attitude of
Quasi-Setzung (i.e., quasi-positing). A very similar kind of commitment is involved in pictorial experience and, in some sense, also in depiction itself this time,
with respect to the depicted situation or world (see Dorsch (2010c)).

Key Features of Imaginative Episodes

65

Now, several facts speak in favour of the assumption of imaginative


non-neutrality, primarily because they are difficult to explain without any
reference to an imaginative commitment. The first fact is that (at least)
visual imagining is transparent with respect to its objects. 39 What this
means is that introspection on episodes of visualising reveals their objects
but no other entities, notably no representational vehicles or intermediate
objects of awareness (such as pictures or sense-data). Now, transparency
understood in this way includes non-neutrality (see Martin (2002a) and
Dorsch (2010c)). Compare visualising with the perception of a picture.
Seeing a picture of an elephant is committed to the presence of a picture in
our actual environment, but not to the presence there of an elephant. 40 In
accordance with this, if we introspect on our experience, we do not find an
actual elephant, but just an actual picture of an elephant. So, it is phenomenologically salient to us that seeing a picture of an elephant is not
committed to the presence of an elephant in the seen situation. Analogously, if visualising an elephant were not committed to the presence of an
elephant in the imagined situation, we should expect to find, when firstpersonally reflecting on our imaginative experience, not an elephant, but
instead some substitute for it such as an (internal) picture of an elephant.
But this does not seem to be the case. When we introspect an episode of
visualising, what is revealed to us is the visualised object, and no representational vehicle or intermediary. In other words, visualising (just like
seeing) is transparent: it is committed to the presence of the visualised object in the imagined situation.
The second fact to be explained is that our actual imaginings stand in rational relations to other mental representations. In particular, they tell us
what else to imagine about the relevant situation, and what to believe to be
part of that situation. For instance, visualising an elephant in the city comes
with the demand not to imagine that there is no elephant in the city (at least
See Martin (2002a) and Dorsch (2010c) on the transparency of sensory imagining,
and Hopkins (2011b) on the transparency of both sensory and intellectual imagining.
40
Indeed, it may very well be non-neutral about its absence, rather than neutral about
its presence or absence. If we see inside an empty box, notice that it is empty and are
asked about whether there is a stone in the box, we should say no, rather than refrain
from judgement. Besides, I ignore here that pictorial experience is committed to the
presence of an elephant in the depicted scene.
39

66

The Nature and Variety of Imagining

not as part of the very same imaginative project), as well as with the demand to believe that there is an elephant present in the imagined situation
(if suitably prompted). Both demands, however, depend on the respective
episode of visualising being non-neutral about the presence of an elephant
in the imagined situation. For example, there is a tension between visualising an elephant in the city and, as part of the very same imaginative project, imagining that there is no elephant in the city precisely because both
imaginative episodes are committed to the imagined situation being a certain way (and because these two ways are incompatible with each other).
Similarly, it is right to believe that the imagined world contains an elephant
(rather than a tiger) because this is how the episode of visualising presents
the imagined situation to be like.
Finally, the third fact in need of explanation is that imagining may be
central to our acquisition of knowledge say, about possibilities or conditional truths.41 In his most recent book, Timothy Williamson argues that
visualising is one of the many empirical capacities that we may employ in
order to acquire modal knowledge or, in his concrete example, knowledge of some conditional, which forms the first step to modal knowledge.
Considering a situation in the mountains, he describes the largely non-inferential process involved in coming to know that a certain rock would
have landed in a lake, if its path had not been blocked by a bush:
You notice one rock slide into a bush. You wonder where it would have
ended if the bush had not been there. A natural way to answer the question
is by visualising the rock sliding without the bush there, then bouncing
down the slope. You thereby come to know this counterfactual: [...] If the
bush had not been there, the rock would have ended in the lake. (Williamson (2008): 142)

Acquiring knowledge in this way is possible only if, by visualising the


rock, the slope and the lake, we take them all to be part of the same ima41

The explanans has also to include the assumption that imaginings can refer directly or indirectly to aspects of the actual world. Besides, I argue in Chapter 4 that
imagining may also be central to the acquisition of knowledge about actual and contingent facts (e.g., whether a sofa is smaller than a door) for instance, when mentally rotating a piece of furniture in a shop in order to see whether it is possible to
get it through ones front door at home (see also Martin (2002a): 412 and 414).

Key Features of Imaginative Episodes

67

gined situation. But this requires, again, that our relevant episodes of imagining are committed to their presence or at least to the presence of some
objects instantiating the respective properties in that situation.
The issue of which property instances our episodes of sensory imagining can actually represent and be non-neutral about is thereby controversial. One option is to maintain that we are able, say, to visualise (and not
just think of) an object as being a rock; another to insist that we are instead
limited to merely visualising it as having an approximately round shape
and grey colour (and to taking it to be a rock in additional imaginative
thought). The very same issue arises with respect to perception and depiction for instance, whether we can see or visually depict also the property
of being a rock, or only the properties of being (approximately) round and
grey. Here is, however, not the place to settle this issue. 42 In what follows, I
assume that sensory representations including sensory imaginings can
represent and be non-neutral about the instantiation of certain higher-level
properties, such as being a bottle, a tiger, a lemon, or a lake. If the respective examples turn out to be problematic, it should be easy to replace them
with less controversial cases, such as the sensory representation of triangularity or sphericality.
Now, according to the preceding considerations, what perceiving,
judging and imagining have in common is that they are non-neutral about
how things are. They just differ in whether they are concerned with the actual world or, alternatively, with some imagined situation. In other words,
they differ in which particular theoretical attitude they posses: a cognitive
or an imaginative one. By contrast, conative and similar mental representations (such as desires and intentions) show a practical attitude. Although it
might perhaps be suggested that they, too, are non-neutral about how
things are in some non-actual (i.e., desired) situation or world, this is false
in the light of the fact that they are normally concerned with the actual
world. When desiring an ice-cream or that there is peace, we are not concerned with what is true in some possible situation. Rather, we want reality
to be a certain way: that is, we want real ice-cream and real peace. But, of
course, conative representations do not make a claim about how things are
in reality. Instead, they involve a claim about how things should be in the
42

See, for instance, Millar (2010), Hawley and Macpherson (2011) and Dorsch
(2012b) for further discussion.

68

The Nature and Variety of Imagining

actual (or perhaps also some other) world. That is, they possess a practical
rather than a theoretical attitude.43
Imagined (or, if one prefers, fictional) worlds resemble possible worlds,
not the least in consisting of propositions or states of affairs, and in being
non-actual or not matching reality. But there are two reasons why it is perhaps better to resist identifying imagined worlds with possible worlds.
First, imagined worlds may be incomplete or indeterminate in ways that
possible worlds cannot. For instance, an imaginative project concerned
with the adventures of Sherlock Holmes (as described in Conan Doyles
stories) may stay neutral on the colour of his eyes. As a result, the imagined world contains the state of affairs that he has eyes, and even the state
of affairs that his eyes have a determinate colour (e.g., this may be an implicit assumption of the stories and, hence, the project that responds to
them), but no state of affairs linking his eyes with the instantiation of a particular colour. All possible worlds that contain Sherlock Holmes, on the
other hand, are determinate with respect to which colour his eyes posses.
Second, it seems possible to imagine impossibilities if not within a single
episodic representation, then at least as part of an imaginative project or
another complex of imaginative representations (see Section 5.4).44

2.2. No Epistemic Function


The second difference between cognitive and imaginative episodes that I
would like to focus on is a difference in epistemic function: while cognitions have the function (or aim) to lead to first-order knowledge about realVarious attempts have been made at specifying the difference in question in more
detail for instance in terms of a difference in direction of fit (see Searle (1983):
7f.), in terms of the distinction between regarding as true and regarding as to be
made true (see Velleman (2000): 248f.; see also 182), or in terms of the distinction
between stative and other kinds of representation (see Martin (2001): 386f.).
44
Of course, imagined worlds can also not simply be understood as impossible worlds,
given that, in many cases, the explicit or implicit purpose of our imaginative activity
is to imagine precisely a coherent (i.e., minimally logically possible) situation or
world for instance, when being imaginatively engaged with a realistic novel. Accordingly, imagined worlds seem to be sometimes possible, and sometimes not.
43

Key Features of Imaginative Episodes

69

ity, imaginings lack this function.45 In particular, cognitive episodes but


not imaginative ones have the function to move and entitle us to form
first-order beliefs about reality (and also, in the case of judgements or beliefs, to be responsive to relevant epistemic reasons). Since our picture of
the world (i.e., how we take things to be) is shaped by, and again shapes,
our epistemic (and practical) interaction with the world, it is largely and,
ideally, exclusively constituted by our cognitions, and not our imaginings.
One central aspect of the epistemic function of cognitions is that they
possess prima facie justificatory power. More specifically, perceptions,
memories and judgements provide us with prima facie reasons for first-order belief. This means that we are justified to endorse these cognitive episodes in belief, at least in the absence of defeaters (see, e.g., Pollock & Cruz
(1999): ch. 2). Seeing or recalling rain normally suffices to be warranted in
believing that it is currently raining, or did rain in the past. By contrast,
imaginings lack any prima facie justificatory power with respect to first-order beliefs.46 Instances of visualising or supposing are not by themselves
capable of rendering the formation of first-order beliefs justified, even in
the absence of any relevant counterevidence. We are not entitled to come to
believe that it is raining simply in response to visualising rain, or supposing that it is raining.
The other key element of the epistemic function of cognitions is that
they incline us to form the respective first-order beliefs at least, again, as
long as we are unaware of any defeaters. This inclination may then become
manifested and lead to belief for instance, if we consider the subject matter in question. Even when there is no evidence to the contrary, seeing rain
See Williamson (2000), OShaughnessy (2000): ch. 11, and Dorsch (2009b). Some
philosophers prefer to characterise the epistemic function (or aim) at issue in terms
of truth, rather than knowledge (e.g., Burge (2003) and Velleman (2000)). Although
this difference in view is substantial (see Williamson (2000)), it does not matter for
the current purposes.
46
Like all conscious mental episodes, they do, however, provide us with a reason for
second-order belief, that is, for ascribing them to ourselves by means of introspective judgements (see Peacocke (1998), Martin (1998) and Dorsch (2012a)). I largely
ignore this fact in what follows, since what is distinctive about the epistemic function of cognitions is precisely its concern with first-order knowledge about the external world.
45

70

The Nature and Variety of Imagining

need not lead to the belief that it is raining. But it disposes us to form this
belief. Hence, as soon as you ask me how the weather is like, my perception of the rain will move me to believe that it is raining (and to express
this belief to you by saying it is raining). Visualising rain, on the other
hand, does not have the same effect: it does not by itself incline us to believe that it is raining. More generally, imaginings do not on their own motivate us to endorse them in belief.
The two aspects of the epistemic function of cognitions are intimately
connected. That cognitive episodes incline us to form certain first-order beliefs is a manifestation of their epistemic authority over us in theoretical
deliberations. But this does not mean that we cannot sometimes fail to endorse a cognition. For cognitive episodes enjoy this authority only as long
as we take them to provide us with reasons for the respective beliefs. Once
we start to seriously doubt the justificatory power of cognitions or the suitability of the circumstances of their occurrence (e.g., because we receive
evidence about their non-veridicality), we stop following them as our
guides in belief formation. For example, we cease to rely on our perceptual
experience as of rain when we begin to question its status as veridical perception.47 The same applies to a belief, the rational grounding of which we
mistrust for whatever reason. On the other hand, the difference in epistemic
function does not rule out the possibility that we may endorse some imaginings in belief and subsequently rely on them in action in roughly the
same way in which we endorse and rely on perceptual experiences (I discuss this possibility in Chapter 4).
Hence, an imaginative episode, which lacks any proper epistemic function, may none the less move and entitle us to form certain beliefs under
special circumstances; while a cognitive episode, which possesses the epistemic function described, may sometimes fail to live up to its function and
not motivate or justify us. The corresponding difference between cognitions and imaginings is still strict, however: every cognitive episode but
47

There is an important debate about whether epistemic doubt leads to a loss of


inclination, or merely overwhelming or masking of it. Relatedly, it is controversial
whether defeaters are undercutting or overriding (i.e., rebutting) the justificatory
power of cognitions (see, e.g., Pollock & Cruz (1999)). I aim to stay neutral on both
questions, even though it may not always be possible or practical to come up with
neutral descriptions of the epistemic situations concerned.

Key Features of Imaginative Episodes

71

no imaginative episode has the function to move and entitle us to form


first-order beliefs about reality.
Epistemic function is also closely connected to practical function. Cognitions have the function to guide our behaviour, both by informing us
about the means to our ends, and by giving us (not necessarily conscious)
feedback about the progress or success of our actions (see, e.g., BraddonMitchell & Jackson (1996)). If I desire to drink some milk, seeing a glass
of milk on the table or believing that there is milk in the fridge will enable
me to achieve my goal by telling me what to do about it in more specific
way (i.e., by telling me to pick up the glass or open the fridge). Similarly,
my perception of the glass on the table lets me know in which direction to
stretch my arm, and what kind of shape to form with my hand. Finally,
feeling the satisfaction of tasting the milk and of ceasing to desire to drink
more informs me about the success of my action that is, of my attempt to
act on the intention to drink some milk. Imaginings, by contrast, do not
play the same kind of role in practical deliberation and agency because
they lack the function to cognise the world.48
One prominent and plausible way of understanding the epistemic function of cognitions makes reference to evolution and adaption.49 According
to this view, perceptions and the like possess as part of and relative to a
larger organism the biological function of leading to the cognition of
reality. Furthermore, they are said to normally be able to fulfil this function, at least in the kind of environment in which they have acquired this
function and which they are adapted to. The introduction of such a biological function has several explanatory advantages, over and above of making sense of the presence of the capacity to motivate and justify first-order
beliefs. One is that it may help to explain why mental representations can
represent and refer to reality in the first place. More important for the current discussion, however, it provides an account of why we possess episodes (i.e., cognitions), which reliably lead to knowledge and help our speThis is not to deny that imaginings have other practical functions, for instance in
games of make-believe or our engagement with fictional works (see, especially,
Walton (1990); and also Chapter 11).
49
Teleological or similar accounts of mental representations are defended by Millikan
(1989), Papineau (1993), Dretske (1995) and Burge (2003), among others. They are
challenged by, for example, Fodor (1990) and Braddon-Mitchell & Jackson (1997).
48

72

The Nature and Variety of Imagining

cies to successfully interact with the world and survive in it. Compare
again the analogy with assertions: that we usually trust the assertions of
other people may be due to the fact that it is their underlying (social) function to express reliable and sincere views which ensures that (or which the
assertions possess only if) they usually are trustworthy. If most or all
people began to lie, assertions would lose both their trustworthiness and
their (social) function to express sincere opinions. But the introduction of a
biological function of cognitions also faces some difficulties, notably the
problem of specifying precisely when a cognition fails to fulfil its function
(e.g., which environments are relevant for the acquisition and utilisation of
the function). Hence, I aim to stay neutral on whether the epistemic function of cognitions may be accounted for in evolutionary terms.
Finally, it seems natural to assume that the strict difference in epistemic
function is closely linked to the difference in cognitive attitude. Otherwise,
their coextensionality would remain unexplained. This issue arises, in the
first place, because the two differences are not identical. And they are not
identical because the epistemic function of an episode and its cognitive attitude are different kinds of properties. Looking outside through a window
and seeing rain has the power both to move us to believe that it is raining,
and to justify the resulting belief. But this power need not become realised
under all circumstances. We may not form the belief because a trustworthy
friend may inform us that there is someone pouring water from above (e.g.,
for a film which is currently shot outside of the building), or because we
simply fail to consider the issue of what the weather is like (e.g., all our attention may be occupied by a problem or a reverie). And, even if we form
the belief that it is raining, it may not be justified by our perception because we ignore the defeating testimony of our friend, or because we have
formed the belief completely independently of the perception (e.g., in response to hearing the weather forecast in the radio). Making a claim about
how things are, on the other hand, does not allow for the same distinction
between a power and its realisation. Our perception tells us that it is raining
outside independently of the context for instance, independently of
whether we consider the issue, and of what other evidence is available to
us.
But if the two differences are not identical, we need to identify another

Key Features of Imaginative Episodes

73

intimate connection between them. The most promising proposal is perhaps to maintain that the difference in attitude renders the difference in epistemic function to some extent phenomenologically salient. We already encountered a way in which part of the epistemic function becomes manifest:
namely in its concrete realisation, that is, in the actual formation (and justification) of a first-order belief in response to the cognitive episode concerned. But this need not be the only way in which an epistemic function
of an episode may become salient. Another possibility is that it contributes
to the determination of the phenomenal character of the episode. As a result, episodes with different attitudes will also differ in phenomenal character. In the case of cognitions and imaginings, this phenomenal difference
may be identical with the difference in attitude. That is, the possession of a
cognitive attitude may be a phenomenologically salient reflection of the
possession of the function to constitute knowledge.50

2.3. Reference to Reality


Although many instances of imagining do not refer to real entities and their
real features, at least some do make reference to reality. If someone visualises or supposes that a friend of his is at this moment sitting in the Opera
de Bastille listening to Wozzeck, we say of him that he imagines something
about real and cognisable entities (his friend, the opera house, Bergs composition), and not about some merely imagined or otherwise unreal ones. 51
In particular, he may fall back on his memory or knowledge of these entitSee Martin (2002a): 391f. See also Dorsch (2009b), (2010c) and (2012a), as well as
Soldati & Dorsch (2011) for an elucidation and defense of the idea of a phenomenal
reflection of an episodes rational power to provide and to respond to reasons.
51
See Sartre (1940); Casey (1976): 113; Peacocke (1985): 26f.; Walton (1990); Martin
(2001): 275; OShaughnessy (2000): 166f.). It is also widely accepted that the referents of our visual imaginings are ultimately determined by our accompanying desires, intentions or thoughts about our visual imaginings (see Wittgenstein (1984c):
vol. II, sec. 115; Sartre (1940): 8ff.; Ishiguro (1966): 162; Peacocke (1985): 26f.;
Budd (1989): 114f.; Martin (2001): 275). I return to this issue in the Sections 3.6 and
14.1f.. However, it might be true that imaginings can refer to actual entities only in
an indirect way, namely by referring to the relevant counterparts of those entities in
some of the merely possible worlds (see Lewis (1980) for more on such counterparts
of actual entities).
50

74

The Nature and Variety of Imagining

ies and use it for, or include it in, his imaginative representation. Perhaps
one might want to maintain that only those aspects of the content of his
representation, which are not remembered or known, constitute the imaginative content. However, the imaginative attitude and lack of epistemic
function involved characterise the representation as a whole, and not only
some parts of it. This also fits the fact that, in our practice and talk, we take
the complex representation to constitute a single and imaginative mental phenomenon: we assume that the person imagines his friend as being at
the opera.
Furthermore, there is the possibility of cases in which an imagining is
made up entirely of material provided by past memories or knowledge. 52
The result of, say, using ones visual memory of the friend in order to picture him sitting in an equally remembered opera house is clearly not a visual memory, but a visual imagining (e.g., it differs from the case of the two
memories occurring separately, but temporally close to each other). If,
however, the mnemonic parts of the representation would not count as constituting the imagined content, the imagining would not have any, given
that the imaginative combination of the two remembered contents does not
add any further sensory material. Besides, it is not merely the combination
which is imaginative, but also the resulting episodic representation not
only the act of combining the image of the friend and the opera house, but
also the resulting sensory image of the friend in the opera house. Hence,
the complexity of imaginings, or their involvement of cognitive elements,
is not a good reason to deny their potential reference to reality.53
In addition, we can imagine existing objects as having features which
they really have. The subject may imagine a truth by imagining that his
friend is visiting the opera; and he may visualise the friends and the opTo some extent, this is in fact true of all sensory imaginings though only with respect to each of the represented features taken on its own, and not with respect to the
objects which these features may characterise or constitute. That is, we may not be
able to visualise something red without having seen red before; but we can visualise
a red cloud, even though we may have not already seen one.
53
Compare also the discussion in the Sections 3.1 and 3.3, in particular of the point
that the obtaining of a direct causal link between the referent and the representation
is not a necessary condition on reference or factivity: while this may be true in the
case of perceptions, it is not true in the case of thoughts, judgements, or feelings of
desire or emotion.
52

Key Features of Imaginative Episodes

75

eras appearances as they actually are. Indeed, it seems to make perfect


sense to say that imaginings can be veridical with respect to reality (see
Peacocke (1985): 27, n. 12; OShaughnessy (2000): 345). If it happens that
the friend of the imagining person is, at the moment of the imagining, sitting in that opera house in Paris and enjoying a performance of Bergs opera (and in the same way in which it is imagined), there is no reason to
deny that the imagining matches the relevant aspect of reality, even if only
accidentally. Indeed, that imaginative episodes are representational and
non-neutral ensures that they may be assessed with respect to whether they
match the imagined world. For instance, we may explicitly imagine hallucinating something, meaning that we visualise some state of affairs and
then imagine this state of affairs not to obtain in the imagined world (see
Dorsch (2010c)). Veridicality with respect to the actual world is just a special case of this comparative assessment, namely if the imagined world is
(imagined to be) identical with the real one.
One might prefer not to call such a correspondence veridicality for
instance, because there is no causal connection between the representation
and the represented; or because the representation does not have the function to provide correct information about the world; or because the representation does not purport to be true. However, the causal element may not
be decisive since imaginings may be caused, albeit in some wayward fashion, by what they represent. The sitting of the person in the opera may,
after all, trigger a process which again affects in the imagining person the
respective act of visualisation (see Peacocke (1985): n. 12; see also Lewis
(1980) on veridical hallucinations). The function to be veridical, on the
other hand, does not seem to be necessary for veridicality. It is doubtful,
for instance, that perceptual hallucinations really possess this function, given that it seems to be part of their nature to be false (see Martin (2004),
Dorsch (2012a) and Soldati & Dorsch (2011)). But such hallucinations can
none the less be veridical in the sense that they can match reality (see
Lewis (1980)). Finally, there is no good reason to assume that the fact that
some representation does not purport or commit itself to be veridical has
any influence on whether it actually can be veridical. Someone may not
purport to be nice, say, but still show some kindness.
Indeed, if imaginings could not refer to and match reality, it would be

76

The Nature and Variety of Imagining

impossible for us to make use of them during the acquisition of knowledge.


Consider, again, Williamsons example discussed above in Section 2.1.
Visualising the likely path of the rock under the condition that there had
not been the bush preventing it from sliding down the slope enables us to
come to learn that the rock would have landed in the lake only if the visual
representations of the rock, slope and lake involved refer to the real things.
Our cognitive project is concerned with where the rock before us in our actual environment would have landed, and not some rock in some merely
imagined situation. What all these examples show is that imaginings
should be assumed to possess the capacity to make reference to and be in
congruence with reality.
In the case of imaginative thoughts, they enjoy this ability in virtue of
involving the same concepts as judgemental thoughts or beliefs (e.g.,
names, natural kind concepts, or other direct-referential concepts; see also
Section 12.1). Similarly, sensory imaginings may include parts or aspects
of episodic memory that suffice to ensure direct reference to real entities.
But the semantic connection of imagining to the actual world may also be
less direct, especially in the case of its sensory form. The perceptual
experience of a tree and the perceptual experience of a picture of a tree
may refer to one and the same object (e.g., the tree in my garden). But the
referential link between the pictorial experience and the tree is mediated by
the representationality of the perceived picture: the experience of the picture refers to the tree in virtue of representing and referring to a picture
which, again, represents and refers to the tree. Similarly, experiential imaginings may count as indirect in so far as they make us aware of objects
like trees by representing, say, perceptions of them. This would fit well
with the observed lack of immediacy that distinguishes sensory imaginings
from perceptions (see Section 3.2). Perhaps, imaginings and, notably,
visualisings refer to reality in a similar roundabout manner. That is, what
we directly imagine in the examples discussed above may be not the actual
entities concerned, but instead merely some proxies which stand in for
those real objects whether these proxies are something like inner pictures
or, alternatively, cognitive representations. I return to either possibility in
the context of discussing the Ontological and the Representational Account, respectively (see Section 5.3 and Chapters 9ff., especially Section

Key Features of Imaginative Episodes

77

10.4).
Besides, sensory or at least visual imaginings are special in that they
allow for two kinds of reference to objects, namely not only to particular
objects (e.g., that friend), but also to types or sorts of objects (e.g., some
man or another):
If asked to visualise a green apple, you may well succeed in bringing to
mind an image of an apple. But, in many contexts, it is simply inappropriate
to press the question which apple you have imagined. In visualising an
apple, there need be no particular apple which is imagined.
This is not to say that particularity cannot be injected into our imagery.
One can, if one wants, imagine the very green apple now nestling in A. A.
Gills pocket. For we can use imagery for particular imaginative projects,
for particular tasks; and in such cases it is entirely appropriate to take the
imagery to be the imagining of the particular objects or events that one sets
out to have in mind. (Martin (2001): 275)

Indeed, the first type of sensory imagining is more fundamental than the
latter54, given that the default case is to imagine something about entities
which we do not assume to have a specific identity (let alone to exist).
Simply visualising an apple means visualising some apple of a certain type,
but no apple in particular. By contrast, visualising a specific apple such
as the one in A. A. Gills pocket requires some additional representation
of numerical identity of the visualised apple. This raises the question of
how precisely particularity becomes injected into our imagery and,
more generally, how the reference of our sensory imaginings is fixed.55 I
return to this issue in Section 3.6 on the presence of an intellectual element
in sensory imagining and its significance for the latter. Before that, it is
helpful to discuss two other, closely related features of sensory imaginings
that distinguish them strictly from perceptions (and, to some extent, also
sensory memories).
Just as depictions of types of object are more fundamental than depictions of
particular objects (see Wollheim (2003) and Dorsch (2012c); see also Note 11 in
Chapter 9). The related idea that sensory imaginings are representations (or
echoes) of types of perception and, therefore, also of types of perceived objects, is
discussed in Chapter 10.
55
The same question is not pressing in the case of intellectual imaginings since they do
not differ from other kinds of thought in their referentiality.
54

CHAPTER 3
Key Features of Sensory Imaginings

The features noted so far are central to and, to some extent, also distinctive
of episodic imagining, independently of whether it is sensory, affective or
intellectual. But there are also important differences between sensory imaginings and perceptions, which have notably been identified and described
by philosophers in the phenomenological tradition. These differences have
no direct relevance for the formulation of a unified account of imagining,
since most of them do not pertain to intellectual or non-episodic forms of
imagining, and since some group sensory memories together with sensory
imaginings, rather than with perceptions (see below; and, for an explicit
example, Sartre (1940): 10).56 None the less, it is worthwhile to gain some
understanding of them not only because they are philosophically interesting in their own right and figure prominently in the debates on imagining,
but also because the ability to capture and explain these features may separate otherwise equally adequate unified accounts of imagining and identify those among them eventually to be preferred over the others.
Perhaps following Hume in this, Husserl has noted two respects in
56

I stay largely neutral on the issue of whether the features to be discussed distinguish
sensory memories from sensory imaginings, or rather from perceptions. Sartre
(1940: ch. 1), for instance, treats memories as one kind of image, and imaginings
as another, contrasting both with perceptions; and this seems to be true with respect
to vivacity or immediacy (see Section 9.4). But it is more controversial to assume,
say, that we are immune to error through misidentification with respect to the identity of sensorily recalled objects, or that sensory memories are quasi-observational
(e.g., do not allow for knowledge-increasing changes in perspective onto the remembered objects). However, since nothing in what follows depends on settling
these issue, I do not pursue them here any further.

80

The Nature and Variety of Imagining

which sensory imaginings appear to differ from perceptions (see Husserl


(1901): part 5; see also Husserl (2006)). First, while the latter seem to put
us into direct contact with objects actually present in our environment, the
former seem to lack this form of immediacy and instead represent objects
in a more indirect way (perhaps similar to that involved in depiction), given that they are not presented as being currently before our eyes. This may
be glossed as a difference in immediacy (see Martin (2002a) and Dorsch
(2010a)). Second, perceptions appear to be more lively in their
representation than sensory imaginings thus pointing to a difference in
vivacity (see Husserl (1901): 455ff.). These further characterisations of
how the two kinds of sensory representation differ are suggestive of the
Humean idea that, first, sensory imaginings are copies of perceptions and,
second, the former are less vivid than the latter (see Chapter 9 for a detailed discussion of Humes views).57 That sensory imaginings seem to be
less determinate than their cognitive counterparts is a closely related idea.
Another important set of intimately connected features that sensory
imaginings are merely quasi-observational (i.e., unable to reveal previously hidden aspects of objects when involving a shift of our point of view
on them, such as during mental rotation), that they are immune to error
through misidentification concerning the identity of the imagined objects,
and that they involve some intellectual element specifying what they are
representing has been comprehensively (though not always absolutely
clearly) described by Sartre and is also mentioned in the writings of other
philosophers, especially Husserl, Wittgenstein and Casey (see below for
references). Sartre and Wittgenstein also defend the idea that, partly as a
consequence of the other features just listed, sensory imaginings are uninformative with respect to the external world. The uninformativity of sensory imaginings, perhaps together with their lack of a cognitive attitude and
57

Interestingly, both Hume and Husserl do not always seem to clearly distinguish the
differences in vivacity and immediacy perhaps because of the possibility of explaining the former in terms of the latter (see below), or perhaps also because of
their implicit acknowledgment that the difference between imaginings and cognitions is qualitative, and not quantitative. Accordingly, a difference in vivacity can
distinguish the two kinds of episode only if it does not come in degrees, or if it can
at least be traced back to a difference which does not come in degrees, such as immediacy (see Section 9.1f. for further discussion).

Key Features of Sensory Imaginings

81

an epistemic function, may finally be taken to indicate their inability to


cognise reality.
This chapter concentrates on most of these features pertaining specifically to sensory imaginings, and on the extent to which (if at all) they help
to establish the contrast between this form of imagining and its cognitive
counterpart, perception. I continue and conclude this discussion in the next
chapter, where I specifically address the seeming uninformativity of imaginings and their related apparent inability to constitute knowledge. It turns
out that, while the lack of vivacity, determinacy, informativity and power
to cognise aspects of the world distinguishes sensory imaginings only typically from their cognitive counterparts, the other features quasi-observationality, immunity to error through misidentification and the involvement
of some imaginative thought or intention-in-action concerning our
imaginative engagement constitute strict differences. Furthermore, while
sensory memories are sometimes closer to sensory imaginings (notably in
their lack of immediacy), they are also sometimes closer to perceptions
(e.g., in their lack of immunity to error through misidentification). In Part
Three, I give further substance to the difference in immediacy between
sensations and perceptions, on the one hand, and sensory or affective imaginings (and memories), on the other, while also arguing there that it does
not pertain to intellectual imaginings.

3.1. Diminished Vivacity


It is an often voiced idea that sensory imaginings are typically, if not always, less vidid than perceptions (see e.g., Husserl (1901): 455ff.; Collingwood (1938): 178ff.; see also Chapter 9 on Humes conception of imaginings). Vivacity (or vividness) is thereby usually taken to be an aspect of the
phenomenal character of the respective episodes (with the exception of
Collingwood: see below). As such, it can pertain either to the representational aspect of the episodes, or to some of their other phenomenologically
salient features (see Martin (2001): 270).
If vivacity is taken to be a modification of the representationality of
episodes, it presumably concerns the vagueness or repleteness of the
provided information (or some similar feature). But it is possible for sens-

82

The Nature and Variety of Imagining

ory imaginings to be as precise or vivid as normal perceptions or sensory


memories, while it is equally possible for perceptions or, especially,
memories to be as vague or faint as many sensory imaginings. An image of
a very familiar or simple object may show a high saturation and brightness
of colours, as well as a strong definiteness of outline, clarity and sharpness
of focus. In contrast, a perception in an almost dark room at night may lack
all these features (see Budd (1989): 104).
If, on the other hand, vivacity is interpreted as a non-representational aspect of episodes, it may be understood as characterising the determinacy or
intensity of some of the non-representational aspects of the phenomenal
character of episodes. But again, it is not clear whether there could not be,
on the one hand, perceptions or memories which are faint and, on the other
hand, sensory imaginings which are vivid in this sense. Assume, for instance, that the sensory quality of pain is non-representational. Nothing
seems to rule out the possibility of there being actual or, especially, remembered pains which are not very intense or determinate, or of there being imagined pains which are rather strong and specific. Another worry is
surely whether there are, in the first place, aspects of the phenomenal character of episodes which allow for modification in determinacy or intensity,
while being non-representational (see Dretske (1995); Tye (2000)).
Hence, although it seems reasonable to assume that the three kinds of
sensory representation (i.e., perception, memory and imagination) may often differ in vivacity in either of the two senses, these differences are at
best typical. There may certainly be other non-representational aspects of
the phenomenal character of episodes which do not come in degrees of
richness, specificity or strength and therefore may still be able to establish
a strict difference. And indeed, Collingwood has identified vivacity with
such a phenomenal aspect in an interesting, but also daring interpretation
of Berkeleys and Humes take on the notion (see Collingwood (1938):
178-185). There, he suggests that a vivid episode is one which is experienced as passive, or forced upon us; while faint episodes are experienced
as being subject to the will. However, this does not seem to be a very natural use or elucidation of the expression vivacity and it probably does
not capture how the empiricists have conceived of it (see, e.g., Section 9.1
on Humes understanding of the notion). Of course, I am very sympathetic

Key Features of Sensory Imaginings

83

with Collingwoods underlying idea that agency, and its phenomenological


salience, is essential to the formulation of a unified theory of imagining.
But I think that this proposal should be discussed under the heading of the
Agency Account of imagining, and not under the heading of a third way of
understanding the notion of vivacity.

3.2. No Immediacy
That sensory imaginings (as well as sensory memories) lack the immediacy
of perceptual experiences means, first of all, that they do not present their
objects as being there before us in our actual environment. When we see a
tree, it seems to be right there before our eyes. But when we visualise a
tree, we do not have a similar impression of its presence in our actual environment. This is true even if we imagine the tree to be part of the real
landscape in front of us: the perceptual experience of the landscape and the
imaginative experience of the tree do not merge in such a way that the tree
appears to be actually present in the same way as the landscape (see
Dorsch (2012c)). In Husserls words, the tree is not bodily present, but
instead merely sensorily re-presented (see Husserl (1905): sec. 42; see
also Martin (2001)).
It is important to distinguish immediacy from the possession of a cognitive attitude. While memories involve a cognitive attitude, they lack immediacy: when I recall the appearance of the tree that used to stand in our old
garden, but got cut down in the meantime, the visually represented tree
does not thereby begin to appear to be present in my actual environment.
Similarly, judgemental thoughts make a claim about how things are in the
actual world, but do not relate us in an immediate way to the external objects concerned. Hence, the possession of a cognitive attitude does not require immediacy. The explanation of this is that while the last-mentioned
feature concerns the issue of whether we are in direct contact with the represented entities, the first-mentioned feature pertains to the issue of whether the mental representation in question involves a claim about how things
actually are.
Whether immediacy presupposes or implies a cognitive attitude is more
controversial. While some think that immediacy gives rise to an epistemic

84

The Nature and Variety of Imagining

commitment about how things (really) are (see Martin (2002a)), or at least
contributes to such an attitude in so far as it identifies the world, that the
commitment is concerned with, as the actual world (see Dorsch (2010a)
and (2010c)), others deny any such link to theoretical non-neutrality (see
Husserl (1901): 442). One underlying issue is whether immediacy is part of
the attitude (i.e., phenomenologically salient commitment) of the mental
episode concerned, or instead part of its content (i.e., the phenomenologically salient manner in which things are presented). In this respect, its status
is similarly unclear as that of, say, the visual or auditory character of sensory representations (see Section 1.2).
Completely independently of this issue, the difference in immediacy is
perhaps best understood in terms of the idea that sensory imaginings are
re-presentations or re-productions in so far as they inherit their sensory
character and content from an underlying perception (i.e., a presentation
or original production), to which they stand in some constitutive and representational relation. Indeed, the just noted participants in the debate concerning the relationship between immediacy and commitment agree on this
point (see Husserl (2006) and Martin (2002a), as well as Chapter 10). That
sensory imaginings are re-presentations of perceptual experiences may
mean, for instance, that they are instances of experiential imagining: that
is, indirect representations of objects and their features in virtue of being
direct representations of (types of) perceptions with a corresponding content. This representational indirectness can then explain why, or in which
sense, imaginings do not mark their external objects as actually present in
our environment. I discuss this proposal in more detail in the context of the
discussion of the representational version of the Dependency Account and,
notably, its idea of a dependency of imaginings on cognitions.58
In addition, the idea of a difference in immediacy may be used to account for the typical difference in vivacity. The core idea is that, because
sensory imaginings and perhaps also sensory memories, for that matter
are taken to be some kind of reproduction of the original sensory impres58

The representational echo thesis with respect to sensory imaginings is endorsed by


Peacocke (1985), OShaughnessy ((2000): ch. 12), Martin (2001) and (2002), and
Dorsch (2010c); while Hopkins ((1998): ch. 7), Currie & Ravenscroft (2002) and
Noordhof (2002) are critical of it. I discuss the idea of understanding the imagination as a representational echo of cognition in the Chapters 9ff..

Key Features of Sensory Imaginings

85

sions, they may be expected to be normally less vivid (in either of the two
senses distinguished above) than the latter. This may, again, be said to be
due to the fact that sensory imaginings do not literally possess the kind of
experiential qualities pertaining to perceptions and bodily sensations (e.g.,
the phenomenal qualities constitutive of being a painful experience, or an
experience of something red), but instead merely represent instances of
these qualities (see Sections 10.1 and 11.5). Besides, the idea of immediacy
is also important for the ability of the Agency Account to accommodate
imaginings which might seem to be non-representational (see Section
14.3).

3.3. Indeterminacy
Casey proposes another though, as it seems, closely related strict difference between sensory imaginings and cognitions. He maintains that imaginings are necessarily indeterminate in the sense that they can never be
perfectly determinate, nor allow for an improvement in determinacy
(see Casey (1976): 104-106, 110). Although Casey does not explicitly restrict his comments to sensory imaginings, all his examples concern only
sensory cases. Indeed, it would be difficult to try to apply his ideas to intellectual representations. His description of what he means by indeterminacy is not always entirely clear, despite its extensiveness. Most of his examples concern representational and possibly also non-representational 59
aspects of sensory imaginings that are phenomenologically salient, and
which involve differences in vivacity (see Casey (1976): 36f., 105-109).
This suggests that indeterminacy may be the same as a lack of vivacity.
But then, his claim that sensory imaginings are always indeterminate becomes untenable in the light of the previous considerations. Moreover, Caseys only apparent argument in support of this claim (see Casey (1976):
105f.) turns out to be unsuccessful.
He acknowledges that perception can be as indeterminate as sensory
imaginings for instance, if occurring, in the visual case, under bad lighting conditions. His example consists of a perception of a statue of a woman
59

For instance, being blurry is not obviously representational (see Casey (1976): 106;
Peacocke (1983): ch. 1; Soldati & Dorsch (2011)).

86

The Nature and Variety of Imagining

at dusk which does not represent where exactly her arm emerges from her
torso. But, he maintains, we can none the less investigate the perceived object further (e.g., in improved conditions) to find out about its determinate nature or at least to get a determinate perception of it. That is, we
can see (or feel or otherwise discover) at which precise place the statues
arm begins. In contrast, the same is not possible with respect to a given
sensory imagining and its object: neither can be made more determinate
as it is. Caseys main idea is here, of course, that we cannot wait for better
conditions, or use other cognitive means, to further investigate the nature
of a sensorily imagined statue.60 Therefore, as Casey concludes, an
imagined object is inherently indeterminate and can never be apprehended
as perfectly definite (see Casey (1976): 106).
Apart from the problematic switch between the representations and their
objects as the bearers of the feature of indeterminacy, the main problem
with this argumentation is that it does not sufficiently distinguish the claim
that we cannot make an already indeterminate sensory imagining more
determinate from the thesis that there cannot be determinate sensory imaginings. The latter is what he really needs to argue for; but his argument
can at best support the former. In particular, it still remains a possibility to
deliberately try to improve the determinacy of ones sensory imaginings.
For instance, we can decide that the visualised statues arm should begin at
a certain point of its torso and then act on the resulting intention by sensorily imagining it to start there. We may perhaps not always succeed in such
a task for some contingent reasons for instance, because we are distracted, or because the task is too ambitious for our ability to visualise. But we
can at least in principle fill in the details missing in a given image. Casey
might presumably rightly insist that the alteration brings about a new
episode and hence cannot count as the determination of a previously indeterminate imagining. But he does not show that we cannot produce determinate sensory imaginings in the first place at least not if determin60

Caseys indeterminacy is therefore perhaps also similar to Sartres quasi-observationality. In this context, Casey actually quotes Sartre as writing: A hare which is
vaguely perceived is in itself a determinate hare. But a hare which is the object of a
vague image is an indeterminate hare (see Casey (1976): 106; see Sartre (1940):
16). But again, this observation does not show that there cannot be precise or determinate images.

Key Features of Sensory Imaginings

87

acy is understood as equivalent to vivacity, as specified above. However,


Casey seems sometimes to intend the expression to (also) denote immediacy:
In perceiving, we bodily confront, and are given directly, the object perceived. In imagining, there is no such direct encounter, but only a characteristic indefiniteness in the comparative directness with which imagined content is given. Although the imaginative presentation is frontal in character,
the imaginer does not feel that he is genuinely confronting it; it is not
present to him with the same determinate and forceful directness that is experienced in the case of perceptual objects. (Casey (1976): 109)

This second interpretation of what Casey means with indeterminate is


also supported by the fact that he claims that a lack of indeterminacy
would render sensory imaginings indistinguishable from perceptions (or
memories; see Casey (1976): 110). Furthermore, this reading that sensory
imaginings are indeterminate in the sense that they are merely some kind
of reproduction of perceptions and hence less directly related to entities in
reality (as well as typically less vivid) is more plausible than the first. Finally, if indeterminacy is understood as characterising the absence of a
presentation of the sensorily imagined objects as real (and present), Caseys claim amounts at least in this respect simply to the already defended thesis about the strict attitudinal difference between the two kinds of
sensory representation.

3.4. Quasi-Observationality
Two further characteristic features of sensory imaginings, which play a
central role in Sartres and in Wittgensteins writings, differentiate them
strictly from perceptions (and, arguably, sensory memories). The first feature is that sensory imaginings are, as Sartre calls it, merely quasi-observational (see Sartre (1940): 8ff.; see also Wittgenstein (1984d): sec. 63, 80
and 88).61 Sensory imaginings appear to be observational in so far as they
61

As with his general discussion, Sartre focuses solely on cases of visualising. But I
assume here that his considerations may apply equally to other form of sensory imagining.

88

The Nature and Variety of Imagining

resemble perception in being perspectival: both present their objects from


one particular perspective with the result, say, that only the frontside, but
not the backside, of the objects are sensorily given (see Macpherson, NidaRmelin & Dorsch (2012)). Thoughts lack this kind of observationality
since they are not limited to a single perspective (e.g., we may think of objects by means of concepts that simultaneously capture their front- and
their backside) and, indeed, may present objects completely independently
of any perspective (e.g., by means of concepts that capture abstract or other
imperceivable aspects of objects). None the less, sensory imaginings are
not observational in the same full sense as perceptions: they are merely
quasi-observational. Here is how Sartre characterises this difference:62
[I]t is characteristic of perception that the object never appears except in a
series of profiles, of projections. The cube is indeed present to me, I can
touch it, see it; but I can never see it except in a certain way, which calls for
and excludes at the same time an infinity of other points of view. One must
learn objects, which is to say, multiply the possible points of view on them.
The object itself is the synthesis of all these appearances. The perception of
an object is therefore a phenomenon of an infinity of aspects. [...]
[...] When I say the object I perceive is a cube, I make a hypothesis that
the later course of my perceptions may oblige me to abandon. When I say
the object of which I have an image at this moment is a cube, I make here
a judgement of obviousness: it is absolutely certain that the object of my
image is a cube. What does this say? In perception, knowledge is formed
slowly; in the image, knowledge is immediate. [...] An image is not learned:
it is organized exactly as the objects that are learned, but, in fact, it is given
62

One thing to note in this passage is that, since Sartre refers with [turning] a cubeimage in thought to the mental rotation of a visualised cube, thought includes conscious operations of the sensory imagination. The other thing to be highlighted is
that he uses the term image to establish a contrast both with perceptual experiences
and with the perceived objects (see also his comment that image is ambiguous in
simultaneously [signifying] a consciousness and its transcendent correlate (Sartre
(1940): 148)). That is, image refers both to sensory imagining and to its object,
thus reflecting Sartres view that the sensorily imagined objects are entirely determined (or even constituted) by our imaginative awareness of them (see Section 3.6). It
is also worthwhile to point out that he generally uses the term image to cover not
only sensory imaginings, but also sensory memories (and that the latter are also
quasi-observational and immune to error through misidentification; see Sartre
(1940): 8f.).

Key Features of Sensory Imaginings

89

whole, for what it is, in its appearance. If you turn a cube-image in thought
to amuse yourself, if you pretend that it presents its various faces to you,
then you will not be more advanced at the end of the operation: you will not
have learned anything. (Sartre (1940): 8f.)

The sense in which sensory imaginings are merely quasi-observational is


therefore that sensorily imagining an object from a single perspective always puts us already into the position to acquire indefeasible knowledge of
what kind of object we are thereby imagining. 63 That is, we know for certain whether what we are visualising at a certain moment is a closed cube
or a box with an opening at the backside, say.
In particular, we cannot gain additional knowledge about the kind of object imagined, nor be forced to revise our relevant, already acquired beliefs, by imaginatively changing our perspective onto the object. That is,
there are no yet unrepresented aspects of the object waiting to be discovered by us through the adoption of new points of view. This explains
why we cannot determine, by mentally rotating a visualised object that appears from one perspective to be a closed cube, whether it is really a closed
cube and not, say, a box open at the back, or three planks with meeting
corners and attached to each other in rectangular angles.
By contrast, having a perceptual experience of something from a single
point of view does not (always) suffice for such indefeasible knowledge
either because perceptual justification of this kind is generally defeasible,
or because perceptual justification of this kind may be indefeasible, but
cannot always be subjectively discriminated from the lack of such perceptual justification (e.g., in the case of illusory or hallucinatory
experiences).64 A closed cube and a box open at the back may look the
I ignore here the fact that there are certain limits on this knowledge of kinds (and on
the corresponding knowledge of the related properties; see also Section 3.6). For instance, it is unclear whether we can be certain that what we are sensorily visualising
is a heavy cube, a cube with lead inside, a cube on planet Mars, a valuable cube, or a
cube owned by me, say. This issue seems to be closely related to if not identical
with the issue of which properties or kinds are perceivable, or at least knowable by
perception alone (i.e., without the need to engage in reasoning or inferences; see
Dorsch (2012b)).
64
Which option is to be chose depends primarily on whether one endorses epistemological disjunctivism, that is, the view that perceptions and hallucinations differ essen63

90

The Nature and Variety of Imagining

same when seen from a particular point of view. Hence, coming to know
whether the seen object is one or the other requires the adoption of a
different perspective onto it.
It is important to note that the quasi-observationality of sensory imaginings extends beyond features that are open to sensory representation and
the related sensory non-neutrality. Real lemons and perfect wax replicas of
lemon share the same visual appearance, at least from the outside. What it
means for the replicas to be perfect is exactly that there is no point of view
from which we can visually discriminate them from real lemons by looking
at their surfaces. There should be no doubt that we can perceive the property of having the visual appearance of a lemon, given that the latter
amounts to nothing more than a (complex) set of of colours, shapes, sizes,
textures, and so on (see Martin (2010) and Dorsch (2010d)). Whether the
property of being a (genuine) lemon is perceivable, that is, visually or otherwise sensorily representable, on the other hand, is debatable (see Section
2.1)). But even presuming that we can indeed see the lemon-hood of an object, the property of being a perfect replica of a real lemon should not be
assumed to be visually perceivable.
The reason for this is that the visual appearance shared by both the lemons and their replicas can be indicative of the presence of only one of
them. That is, recognising this appearance can help us to visually discriminate at best one of the two kinds of object (see Millar (2010), Martin
(2010) and Dorsch (2010d)). If you encounter an object with a light yellowish colour and an egg-like shape, of the size of an apple and with a porous texture, seeing this visual appearance may be enough to see it as being
a lemon primarily because this appearance is characteristic of lemons in
the sense that most objects bearing it are in fact lemons. But it cannot also
be enough to see it as being a perfect replica of a lemon partly because
hardly any objects with this appearances are such replicas, but also partly
because one and the same visual appearance cannot be indicative of two incompatible properties (i.e., one and the same perception cannot present an
object as having two properties that exclude each other).
Now, these considerations about perceivability apply also to imaginabiltially in whether they provide us with (indefeasible) justification for perceptual belief (see Dorsch (2011a)). While Sartre seems to reject this kind of disjunctivism
(see the quoted passage), McDowell (1994) endorses it, for instance.

Key Features of Sensory Imaginings

91

ity (and depictability, for that matter). We may be able to imagine an object
as being a lemon simply by visually representing it as having the visual appearance of a lemon (i.e., as being yellow, egg-shaped, apple-sized, and so
on) meaning that the resulting representation of lemon-hood would be
entirely visual. But we cannot represent something as being a perfect replica of a lemon in a similarly visual manner. Instead, what we have to do is
to visually represent it as having the appearance of a lemon and then intellectually representing it to be merely a replica, rather than the real thing
(see Dorsch (2010c)). None the less, visually imagining a replica of a lemon is as much quasi-observational as visually imagining a lemon. In both
cases, our knowledge of the kind of object that we are visualising that it
is a wax replica or, alternatively, a real lemon is not open to defeat, especially not by evidence gained through visualising the object from a different perspective. To some extent, this claim is trivial: given that genuine
lemons and their replicas look exactly the same from the outside, we cannot tell, by changing our spatial point of view on it, whether something belongs to one kind of object or the other. But the claim is true even with respect to cases in which we, say, imaginatively slice open the objects and
begin to visually represent their insides. I return to this example in Section
3.6.
However, perceptual observation is not limited to looking at objects
from different points of views. We may also perceptually observe an object
by paying more attention to its details, which may happen without any
change in our perspective onto the object. For instance, we may come to
notice the slight cracks in a wall while continuing to stare at it from one
and the same point of view. Correspondingly, Sartre assumes that our
knowledge about which kinds of object we are sensorily imagining is indefeasible not only with respect to changes in perspective, but also with respect to paying closer attention to detail. That is, he assumes that there is a
further way in which sensory imaginings count as quasi-observational,
namely relative to changes in attention. For instance, our belief that what
we are visualising is a wall without any cracks could not be undermined,
say, by us happening to discover some cracks in the visualised wall. For,
according to Sartre, we could not make such a kind of discovery with respect to visualised objects. Here is how he describes this contrast between

92

The Nature and Variety of Imagining

perceptions and sensory imaginings:


This is not all. Let us consider this sheet of paper on the table. The more we
look at it, the more it reveals to us of its characteristics.
Each new orientation of my attention, of my analysis, reveals to me a
new detail: the upper edge of the sheet is slightly warped, the end of the
third line is dotted, etc. But I can keep an image in view as long as I want: I
will never find anything there but what I put there. [...]
In a word, the object of perception constantly overflows consciousness;
the object of an image is never anything more than the consciousness one
has of it; it is defined by that consciousness: one can never learn from an
image what one does not know already. [...] If I perceived this patch of
grass, I should study it for some time to know where it comes from. In the
case of the image, I know it immediately: it is the grass of such-and-such a
meadow, at such-and-such a place. And this origin cannot be deciphered
from the image: in the very act that gives me the object as imaged is included the knowledge (connaissance) of what it is. (Sartre (1940): 9f.)

In the last paragraph, Sartre also hints at an explanation of the indefeasibility of our knowledge of which kinds of object we are sensorily imagining:
it is due to the fact that sensory imagining involves an intellectual element
which provides us with indefeasible knowledge about what is imagined. 65 I
return to this proposal in Section 3.6, which is specifically devoted to the
discussion of the nature and role of the intellectual element involved in
sensory imagining.
For the time being, it is worthwhile to point out that there is good reason
to think that the quasi-observationality of sensory imaginings does not extend to shifts in attention (contrary, say, to what Hopkins (2011b) argues).
In Sections 4.6 and 7.2, I argue that sensory imagining allows for the involvement of object-directed attention and for the acquisition of new evidence concerning the nature of the represented objects in the same way as
65

This difference between visualising and seeing is, for Sartre, furthermore reflected in
the different metaphysical status of the two kinds of represented objects. The link is
established by Sartres view that sensorily imagined objects are constituted (i.e., determined both in their nature and in their existence) by our imaginative awareness of
them with the result, for instance, that they are not presented as being in the same
space or world as perceived objects (see Sartre (1940): 11ff., and also Sections 3.6
and 5.3 below).

Key Features of Sensory Imaginings

93

perception does at least to the extent to which how we sensorily imagine


objects as being is determined not by conscious mental agency, but instead
by some passive and subpersonal mechanisms, or by some sub-intentional
agency. For example, that the wall that we are visualising ends up as being
visually represented by our imaginative episode as having some cracks
may be due to association or generic memory, say, or to some unnoticed
imaginative activity rather than to our deliberate intention to visualise it
as being cracked.
This opens up the possibility of a mismatch between our sensory imaginings and our beliefs or intentions concerning the nature of the sensorily
imagined objects. We may, for instance, intend or take the imagined object
to be a wall without cracks, but then discover through attention to details
that we in fact have visualised a wall with cracks. In other words, we may
discover in this way that we have partly failed in acting on our intention, or
in identifying the specific kind to which the imagined object belongs. But
both kinds of failure should not be troubling. We may certainly fail when
trying to visualise a twenty-sided regular object (e.g., by coming up only
with a visual representation of a twelve-sided regular object); and we may,
moreover, remain ignorant about this intentional failure (e.g., by mistaking
this twelve-sided object for a twenty-sided one, or by being generally unable to visually discriminate objects of these kinds).
Hence, I assume in what follows that sensory imaginings are quasi-observational only with respect to perspectival changes: our knowledge of
which kinds of object we are sensorily imagining from one perspective
cannot be defeated by means of sensory imaginings that present the objects
from other perspectives. That sensory imaginings are quasi-observational
in this sense with respect to perceivable features is simply due to the fact
that they sensorily represent, and are sensorily non-neutral about, the presence of those features. Just as our knowledge concerning whether we are
thinking about something yellow or something green is not open to defeat,
so is our knowledge concerning whether we are visualising something yellow or something green not open to defeat. For, in both cases, we are concerned with indefeasible introspective knowledge of the conceptual or
sensory representational elements that make up the contents of our respective mental episodes. How the quasi-observationality of sensory imaginings

94

The Nature and Variety of Imagining

might be accounted for with respect to imperceivable features will be addressed towards the end of Section 3.6.

3.5. Immunity to Error through Misidentification


The second important feature that distinguishes sensory imaginings from
perceptions (and episodic memories), and which is discussed by Wittgenstein and Sartre, is that we are immune to error through misidentification
with respect to what we sensorily imagine (see Wittgenstein (1984b): 68;
Sartre (1940): 8ff.; Peacocke (1985): 20 and 26f.). What this means is that
it is absurd to question or doubt our opinions about what we are visualising: they cannot be mistaken about the identity of the visualised objects.
In particular, which objects we are visualising is not a matter of the
causal origins of the visual representations involved, or of the (lack of) resemblance between the represented visual appearance of the visualised objects and the real visual appearance of particular objects in the world. 66 For
example, the fact that we are visualising Kings College, and not another
college, is perfectly compatible with the fact that our image resembles
Hertford College rather than Kings College; or with the fact that we have
perhaps even knowingly formed our image of Kings College on the
basis of our visual memory of the appearance of Hertford College; or with
the fact that there is an exact replica of Kings College (say, in our close
environment) which might be considered as an alternative candidate for the
visualised object. Hence, the two main (and perhaps only) sources for an
erroneous identification of what we are sensorily representing namely
that we err about which external object has caused our representation, or
which external object resembles the represented object can be ruled out
in the case of imagination. It is therefore no surprise that we cannot err
about the identity of the represented object (or referent) of our visual imagining say, that it is, in this case, Kings College, and not the replica or
66

The absurdity, the irrelevance of likeness and the example about Kings College
have been introduced in Wittgenstein (1984b): 68. They are discussed and supplemented with the irrelevance of causal origin in Peacocke (1985): 20 and 26f.. See
also Sartre (1940): 8ff., although he does not always clearly distinguish between the
immunity to error and the quasi-observationality of sensory imaginings.

Key Features of Sensory Imaginings

95

Hertford College.
The same is obviously not true of perceptions (as well as sensory
memories). First, we often are not sure about the identity of the objects that
we are seeing (e.g., in the dark, or in heavy rain). And second, even if we
are sure about their identity (e.g., perhaps, when they seem to tell us
straightaway what they represent), it is not guaranteed that we can never be
mistaken. This is due to the fact that which objects we are perceiving is determined by which objects have caused our perceptions, or by which objects possess the properties that we are perceiving as being instantiated by
the perceived objects (see Dretske (1986)). Hence, given that we may err
about which external objects have actually caused our perception to occur,
or which external objects actually possess the features that we see the perceived objects as having, we may also go wrong in our identification of the
perceived objects. Such misidentifications occur, for instance, when we fail
to recognise: (i) that we are merely hallucinating the presence of a particular person; or (ii) that we mistake a particular person for her twin (or fall
victim to a cleverly disguised imposter); or (iii) that we simply misperceive
the properties of the person in front of us (e.g., her height, hair colour,
shape of face, etc.) and, as a result, take her to be someone else.
It is important to note that sensory imaginings are immune to error
through misidentification of the represented objects only to the extent to
which they do represent particular objects. If it is indeed true that sensory
imagining is, by default, concerned with the representation of types of objects rather than specific tokens (i.e., possesses a general rather than a particular content; see Section 2.3), then the issue of whether we can err with
respect to the identity of the sensorily imagined objects does not arise in
default cases (i.e., with respect to the sensory imagination as such). Instead, it becomes an issue only in non-default cases, in which the sensory
representation is supplemented by some intellectual element which determines the particular identity of the imagined objects. When we visualise
some man, but no man in particular, it is a trivial observation that there is
no question of whether we might be mistaken with respect to the identity
of what we are visualising there is nothing to be wrong about in the first
place. Only once we visualise a particular man perhaps because we think
of what we are visualising as being a specific man does it become a sub-

96

The Nature and Variety of Imagining

stantial claim to maintain that we are immune to error through misidentification with respect to which object we are visualising.

3.6. Intellectuality
According to both Peacocke and Sartre, this immunity to error through
misidentification and the related absurdity of asking whether one might be
mistaken about what one is currently sensorily imagining can be properly
accounted for only by assuming that sensory imagining involves an intellectual element specifying what is sensorily represented.67

(i) Peacockes View


Peacocke assumes the intellectual element to be a dependent imaginative
thought: it is a supposition-like thought aptly labelled S-imagining by
Peacocke that differs from normal supposition (or intellectual imagining)
in that it could not occur without an imaginative (or similar) sensory representation (see Peacocke (1985): 25f.). The reason for this is that S-imagining is partly de re about the the sensory representation concerned
either in a direct way (e.g., this visual representation is of Kings
College), or by way of referring to the sensorily represented objects in a
demonstrative or otherwise experience-dependent way (e.g., this is Kings
College or the visually represented object is Kings College). 68 In addition, S-imagining may but need not further identify or describe the repAlthough I am generally sympathetic to the postulation of such an intellectual element and do not question it here in the main text, I wonder whether it is necessary to
take the element to be intellectual. For the fact that it is not sensory (after all, it is
meant to determine what is sensorily represented) does not entail that it is intellectual. For instance, perceptual experiences present their objects as actually present
and experience-independent, but they do not do so in a sensory manner and at least
according to non-conceptualists about perception (see, for instance, Peacocke (1983)
and Crane (1992)) also not in an intellectual manner. This indicates that there is a
third kind of presentationality and, hence, also a third kind of aspects of phenomenal
character (see Dorsch (2012a) and (2013), as well as Soldati & Dorsch (2011)).
68
See Dorsch (2012c) where I discuss this kind of dependence due to de re representationality in more detail.
67

Key Features of Sensory Imaginings

97

resented object (e.g., that it is Kings College, or a suitcase with a cat behind it).
Another aspect of S-imagining is that it ascribes if not explicitly, then
at least implicitly (see ibid.: 26) the sensory content of the real imaginative episode to an imagined perceptual experience of a certain kind (e.g., a
veridical or hallucinatory one; one had by a person different from oneself;
etc.). Moreover, it may but need not further identify or describe this
imagined experience, as well as the subject which is imagined to undergo
it. For example, it may be S-imagined that the imaginative sensory representation of Kings College is in fact identical with one of Wittgensteins
perceptions of the college that he (could have) had in the past. But the two
noted aspects of S-imagining the identification and description of the
sensorily imagined objects and the identification and description of an imagined experience of these objects are independent of each other. One
may, for instance, agree with Peacocke that the identity of sensorily imagined objects is due to some additional imaginative thought, without assuming this thought to be concerned with some imagined perceptual experience.69
The claim that sensory imagining involves imagining a perceptual experience in virtue of S-imagining the sensory content of the real imaginative episode to be the content of an imagined perceptual experience might
seem closely related to or even identical with the claim that sensory
imagining involves imagining a perceptual experience in virtue of being an
instance of experiential imagining (see Chapter 10). But such an impression would be misleading and Peacocke, who endorses both, distinguishes them clearly (see Peacocke (1985): 22f. and 25f.), More specifically, the two claims are independent of each other (i.e., one may be true or
accepted without the other being true or accepted), not the least because
they concern different aspects of sensory imagination: while the first states
something about the content of the intellectual representation supposedly
69

It would be much more difficult, however, to eschew the de re aboutness and the
consequential dependence of the thought element on the episode of sensory imagining. For there is no other obvious way of ensuring that the sensory and the intellectual representations are concerned with the same object (and the same possible or
fictional world) and, indeed, that the sensory representation is particular in the first
place.

98

The Nature and Variety of Imagining

involved in sensorily imagining something (i.e., that it propositionally represents the identity of a real with an imagined experience), the second
states something about the content of the sensory representation involved
in sensorily imagining something (i.e., that it experientially or objectually
represents a perceptual experience with a certain sensory content). For instance, Noordhof (2002) is sympathetic to the idea that sensory imaginings
may include or be accompanied by imaginative thoughts that assume them
to share their sensory content with some imagined perceptual experience,
but denies that the sensory representation itself is a matter of imagining or
representing a perceptual experience.
Returning to Peacockes view, it is not absolutely clear whether Peacocke maintains that all instances of sensory imagining involve S-imagining even if they are all instances of experiential imagining.70 The crucial
issue here is what, for him, imagining having a visual experience as of a
tree, say, requires. According to Peacocke, both imagining seeing a tree and
imagining visually hallucinating a tree have to involve a dependent imaginative thought specifying that the imagined experience is, respectively,
perceptual or hallucinatory. By contrast, when we imagine having a visual
experience as of a tree, the status of the imagined experience is left open
(ibid.: 25). But Peacocke does not say whether he takes this neutral stance
to require a corresponding imaginative thought characterising the imagined
experience simply as a (possibly veridical or hallucinatory) perceptual experience; or whether he instead takes this neutral stance to be the default
case of experiential imagining (i.e., what happens when we experientially
imagine an experience without any additional dependent thought about it in
mind).
This unclarity about Peacockes view is closely related to a similar unclarity about Noordhofs multiple use thesis, according to which [t]he
same mental image may be used to fulfil different imaginative projects
(Noordhof (2002): 428). Indeed, Noordhof introduces this thesis in order to
70

Remember that the first function of S-imagining namely to identify the sensorily
imagined objects does not ensure that all instances of sensory imagining involve
S-imagining, given that sensory imagining may be concerned merely with types of
objects, rather than particular objects (see Section 2.3). Hence, any universality of
the involvement of S-imagining would have to be due to its second function
namely to identify the (experientially) imagined perceptual experience.

Key Features of Sensory Imaginings

99

shed more light on Peacockes example of the different projects of seeing,


of hallucinating and of visually experiencing something (e.g., a tree), as
well as on his example of the different projects of merely visualising a suitcase and of visualising a suitcase with a cat hidden behind it (see Peacocke
(1985): 19 and 25). What happens in this two sets of cases is that the imaginative projects concerned involve exactly the same imaginative visual
representation (e.g., of a tree, or of a suitcase), but different imaginative
thoughts about some of the imperceivable aspects of the represented objects (e.g., whether they are imagined to be perceived, hallucinated or
neither, or whether they are imagined to hide something invisible from
the specific spatial perspective adopted on them). 71 But, just like Peacocke,
Noordhof is not explicit about whether the simplest imaginative projects
consist solely of an episode of visualising (i.e., a self-contained mental
image), or whether they also always contain a (dependent or independent)
imaginative thought specifying further the imperceivable nature of what is
visualised, or how it is imagined to be experienced.72
In Dorsch (2010c), I argue that it is more plausible to identify the
simplest instances of visual imagining whether they should be understood in terms of experiential imagining or not as being exhausted by instances of visualising that do not involve some imaginative thought about
the imperceivable aspects of the represented objects (see also Section
10.3). The main reason for this is that the sensory representation is by itself
already non-neutral about the perceivable nature of the represented objects
which is further reflected in the fact that the sensory representation puts
some constraints on what can in addition be sensibly thought about those
objects (and not vice versa). I return to this point below, when discussing
See Dorsch (2010c). If visualising is indeed an instance of experiential imagining,
the thought is also likely to concern the identity and nature of the imagined type of
experience (see Chapter 10).
72
Indeed, one problem is that Noordhof is silent about whether mental images (in his
sense) constitute self-sufficient mental episodes which may occur on their own, or
whether they are instead merely dependent parts of such episodes which require the
presence of other (i.e., thought-like) parts for their occurrence. Note, however, that
his multiple use thesis takes the mental images to be part of imaginative projects,
and not of imaginative episodes suggesting that he, after all, assumes mental images to be self-contained mental episodes, thus favouring my preferred reading of
the multiple use thesis.
71

100

The Nature and Variety of Imagining

the case of visualising a closed cube. For the time being, it suffices to note
that the multiple use thesis is best understood as the claim that the same
(mereologically simple) episodes of visualising that is, self-sufficient
imaginative visual representations may figure in different (complex) imaginative projects.73 It is therefore to be distinguished from the claim that
different (simple) forms of sensory imagining, which always combine a
imagistic with a thought-like element, may involve the same
(mereologically dependent) imagistic element. Similarly, the thesis that all
instances of sensory imagining or of experiential imagining, for that
matter involve S-imagining seems to be too strong.74

(ii) Sartres View


Sartre who describes the intellectual element as knowledge agrees
with Peacocke that this element is central to the determination of what we
end up to imagine in a sensory manner, as well as that this element may
specify the identity or nature of the imagined objects (note that Sartres
term consciousness is roughly equivalent to my term mental episode):75
In the image, indeed, a certain consciousness gives itself a certain object.
The object is therefore correlative with a certain synthetic act, which includes among its structures a certain knowledge and a certain intention.
The intention is at the centre of consciousness: it is the intention that aims
at the object, which is to say, that constitutes it for what it is. The knowAlthough they do not explicitly discuss this issue, their description of the relevant
examples makes it clear that Peacocke ((1985): sec. 1), Martin ((2002a): sec. 3) and
perhaps also Noordhof (2002) share this (or a very similar) understanding of the
multiple use thesis.
74
In Dorsch (2010c), I also defend Martin (2002a)s view that the default case of experiential imagining in the visual mode is in fact imagining seeing and not imagining visually experiencing, as Peacocke suggests.
75
Note that Sartre uses the term intention primarily to refer to the representational (or
intentional) element of episodes, rather than to their motivational aspects (see
Sartre (1940): 11). But since, in the case of sensory imaginings, intentions are, for
him, manifestations of the minds spontaneity, they seem to acquire a conative element. I return to this issue below, when asking whether the intellectual element
should be understood as an imaginative thought or an intention-in-action.
73

Key Features of Sensory Imaginings

101

ledge, which is indissolubly linked to the intention, specifies that the object
is such or such, adds determinations synthetically. To constitute as an image
in oneself a certain consciousness of the table is at the same time to constitute the table as an object of imaging consciousness. The object as imaged
is therefore contemporary with the consciousness that I have of it and it is
exactly determined by that consciousness: it includes in itself nothing but
what I am conscious of; but, inversely, everything that constitutes my consciousness finds its correlate in the object. My knowledge is nothing other
than knowledge of the object, knowledge concerning the object. In the act
of consciousness, the representative element and the knowledge element are
linked in a synthetic act. The correlative object of this act is therefore constituted as a concrete, sensible object and at the same time as an object of
knowledge. (Sartre (1940): 11)

Accordingly, sensory imaginings do not only present their objects in a


sensory manner, but also determine their identity or nature in an intellectual manner. Indeed, this intellectual determination is crucial to the constitution of the imagined object, rendering it completely mind-dependent.
Furthermore, like Peacocke, Sartre assumes the intellectual element to be
dependent on the sensory representation minimally by being de re about
the sensorily represented object. But, in contrast to Peacocke, Sartre does
not assume that sensory imagining involves imagining a perceptual experience in any way. And his position also differs from Peacockes in that it is
not absolutely clear whether Sartre takes the intellectual element to be fundamentally cognitive, rather than conative. Here is how Sartre describes
the involvement of this element both in sensory imaginings and in pictorial
experiences, which he assumes to involve the same basic form of object
awareness as sensory imaginings (see also Hopkins (1998): ch. 7):
But I can keep an image in view as long as I want: I will never find anything there but what I put there. (Sartre (1940): 9)
[I]nvoluntary and voluntary images represent two very closely related types
of consciousness, of which one is produced by a voluntary spontaneity and
the other by a spontaneity without will. One must in no way confuse intention, in our sense of the term, and will. To say that there can be an image
without will in no way implies that there can be an image without intention.
In my opinion, it is not only the mental image that needs an intention to
constitute it: an external object functioning as an image cannot exercise that

102

The Nature and Variety of Imagining


function without an intention that interprets it as such. If someone suddenly
shows me a photo of Pierre, the case is functionally the same as when an
image appears in my consciousness suddenly and without being willed.
However this photograph, if it is simply perceived, appears to me as a paper
rectangle of a special quality and colour, with shades and clear spots distributed in a certain way. If I perceive that photograph as photo of a man
standing on steps, the mental phenomenon is necessarily already of a
different structure: a different intention animates it. And if that photo appears to me as the photo of Pierre, if, in some way, I see Pierre behind it, it
is necessary that the piece of card is animated with some help from me, giving it a meaning it did not yet have. If I see Pierre in the photo, it is because
I put him there. And how could I have put him there if not by a particular
intention? And if this intention is necessary, what does it matter whether the
image was presented unexpectedly or voluntarily sought? (Ibid.: 19)
I want to be facing Pierre, I want to believe that he is there, my entire consciousness is aimed towards him, it is charmed in some way. And this
spontaneity, this intention towards Pierre makes a new phenomenon spring
forth, to which nothing else is similar: the image consciousness. (Ibid.: 92)

According to these passages, our knowledge of the identity and nature of


what we are sensorily imagining (or experiencing as being depicted by a
photograph) is a matter of our voluntary or involuntary spontaneity,
which produces the imaginative representation and determines the identity
and nature of the imagined object. It is not necessary here to spell out the
details of Sartres conception of spontaneity (see Sections 4.6 and especially 5.6 for more on this). Instead, it suffices to point out that he contrasts
it with the passivity of perception, most notably in its involvement of the
subject concerned: we and not, say, the world are central to, and responsible for, the determination of which objects we are sensorily imagining. Accordingly, although spontaneity need not amount to voluntary
agency, it still involves the subject as some kind of agent, rather than as
some kind of object acted upon. Moreover, the knowledge involved in
sensory imaginings is identified as a consequence of the underlying (voluntary or involuntary) spontaneous activity and, more specifically, of the
intention at the heart of that activity. This contrast with perception and
comparison with voluntary agency suggests that Sartre takes the intellectual element in sensory imagining ultimately to be closer to (voluntary or

Key Features of Sensory Imaginings

103

involuntary) conation than to (perceptual) cognition.76

(iii) Imaginative Thought vs. Intention-in-Action


However, independently of how best to understand Sartres view, it may be
doubted that it is really an option to identify the intellectual element with
an intention (or a similarly conative state), rather than with an imaginative
thought (e.g., like S-imagining). Indeed, Peacocke seems to be right about
the fact that the intellectual element cannot be identified with a prior intention (or a similar prior motivational state), since this would not suffice to
explain our related immunity to error through misidentification. 77 For our
infallible knowledge of what we are visualising and our (possibly also infallible) non-observational and non-inferential knowledge of the content of
our relevant prior intentions may come apart in cases of intentional failure.
If we intend to visualise Kings College, but fail to do so and end up visualising Hertford College instead (e.g., because of some last-minute whim
or change of heart), our knowledge of what we have been intending to
visualise does not help to explain our knowledge of what we are visualising, let alone the latters immunity to error through misidentification.
But these considerations leave open the possibility that the intellectual
element at issue consists in an intention-in-action, which accompanies the
sensory representations and guides us in their ongoing sustainment or development.78 What explains this possibility is the fact that the specific inTrue, Sartre maintains that also thoughts including judgemental ones originate in
spontaneity (see, for instance, Sartre (1940): 153-59). But instead of likening
spontaneity and sensory imaginings again more to cognition, this has merely the
opposite effect of stressing the (rational) activity involved in intellectual cognition
(see Section 5.6).
77
See Peacocke (1985): 26. His argument is directed at the attempt of Ishiguro (1966)
to explain the immunity to error through misidentification with respect to which objects are sensorily imagined by reference to the relevant prior intentions of the visualising subject.
78
Intentions-in-action are the mental elements of actions that guide our actions during
our performance of them (see Searle (1983)). They may but need not be determined by some prior intention on which we act and, in this sense, mediate between
the motivational force of the prior intention and the actual performance of the corresponding action. Other philosophers call intentions-in-actions or something very
76

104

The Nature and Variety of Imagining

tentional failure relevant for Peacockes argument namely that we end up


visualising Hertford College, despite intending to visualise Kings College
is ruled out with respect to intentions-in-action.
The crucial issue is which factor could interfere with our attempt to act
on the intention-in-action to visualise Kings College in such a way as to
determine that the object, that we actually end up visualising, is in fact
Hertford College. One suitable candidate would be the competing intention-in-action to visualise Hertford College. But Peacocke cannot refer to
this option, since it would just strengthen the claim that the identity of
visualised objects is determined by our operational intentions. Instead, Peacocke is likely to suggest that the interfering factor is an imaginative
thought to the effect that the visualised object is identical with Hertford
College. However, this suggestion is begging the question against his or
her opponents, since it assumes that the imaginative thought is more decisive with respect to the determination of which object is visualised than the
equally present occurrent intention.79 Finally, Peacocke cannot point to other potential factors (e.g., some associative or other subpersonal causes),
since these would according both to his own standards and to those of his
opponent be unable to determine the identity of the sensorily imagined
objects, leading to the sensory imagination of mere types of objects.80
much like them volitions (see Ginet (1990)), willings (see OShaughnessy
(1980) and McCann (1998)), or tryings (see Pink (1996) and Peacocke (2003)).
See also Hornsby (1980) for a discussion of the subtle differences between the various conceptions of the mental component of actions.
79
Indeed, it is unclear why the thought should be relevant at all if it is not itself the
result of acting on the intention-in-action in question. Thinking (or even judging)
with reference to a given perception that the perceived object is Kings College does
not have any impact on what is actually perceived. Indeed, it is not ensured that the
thought and the perception are concerned with one and the same object. So why
should it be different in the case of sensory imagining? The most natural answer
seems to be that the sensory and intellectual representations concerned are determined by the same factor, namely an intention-in-action to imagine Kings College.
However, more needs to be said to turn these considerations into a proper argument
for the Agency Account of imagining.
80
The same response to Peacockes objection is unavailable in the case of prior intentions, since there is a plausible interfering factor: an intention-in-action that differs
in content from the prior intention, and which is not itself due to a different prior intention (but instead constitutes something like a whim or a spontaneously occurring

Key Features of Sensory Imaginings

105

Of course, the identification of the intellectual element with an intention-in-action presupposes that sensory imagining is an instance of agency.
But, as I argue in Part Four of this book, this assumption should be accepted because it enables us to account best for the unity of imagining. The
Agency Account defended there is, however, also compatible with the alternative proposal put forward by Peacocke: namely to take the intellectual
element to be an imaginative thought (or the imaginative employment of a
concept) which accompanies or even constitutes or penetrates the instance of sensory imagination in question. This compatibility obtains
especially if the imaginative thought is itself assumed to be determined (at
least normally) by the imaginative agency concerned, that is, by the operational intention-in-action.
Whether the intellectual element is understood as being an imaginative
thought or an intention-in-action as well as whether it is understood as a
genuine constituent or a mere accompaniment of the sensory imaginings
concerned also bears on the issue of whether sensory imagining with an
intellectual element may still be episodic in character, or instead constitutes
a (relatively simple) mental project (e.g., one of imagining having a certain
experience; see Section 1.2). The intentions-in-action involved in mental
actions are constituents of the latter, but do not undermine their episodic
character. Actively thinking a thought does not constitute a (non-episodic)
mental project, despite involving both an intellectual representation and an
intention-in-action or a similar element of activity (see Section 13.4).
Hence, equating the intellectual element with an intention-in-action still allows the instance of active imagining concerned to be a mental episode. By
contrast, if the intellectual element is identified with a merely accompanying thought, the sensory representation and the intellectual element form a
complex mental project, rather than a single episode: the thought is simply
added to the episode of sensory imagining (see Dorsch (2010c)).81
intention-in-action or volition). For if it is assumed that what we sensorily imagine is determined by our own mental agency (rather than by our (active or passive)
thoughts), there is no reason to privilege prior intentions over intentions-in-action.
81
This is true even if the thought element depends on the sensory representation in so
far as the former is a de re representation of the latter (see below, as well as Peacocke ((1985): 25f.) on S-imagining, and Dorsch (2012c)). Note that the situation
under discussion is different from the case of a cognitively (or imaginatively) penet-

106

The Nature and Variety of Imagining

(iv) Explanatory Power


Now, my aim here is not to settle the issue of whether the intellectual element involved in imagining is best understood as an imaginative thought or
as an intention-in-action especially since it does not have any bearing on
whether the Agency Account is indeed the best unified account of imagining. But I would like to illustrate that not only the first option, but also the
second option which fits more naturally with the Agency Account and,
indeed, may be taken to imply (i.e., require) the truth of this approach to
imagining enjoy considerable explanatory power. Indeed, while both option can equally well account for the immunity to error through misidentification of sensory imaginings and also for their capacity to refer to
particular and real entities, the view that the intellectual element is
identical with an intention-in-action should perhaps be preferred over its
rival view with respect to their potential to capture the quasi-observationality of sensory imaginings.
Both options assume that, while we are sensorily imagining something,
we are simultaneously having an intellectual representation which identifies and determines what we are currently sensorily imagining. For instance, that we are visualising Kings College (and not something else)
may be due to our intention to visualise Kings College, or to our assumption or thought that the visualised object (or this) is Kings College. Now,
if the content of the intellectual element indeed determines which objects
are sensorily imagined (i.e., are referred to by the episode of sensory imagining), it becomes absurd to question whether we might misidentify them.
For we cannot err about which objects we intend or take our sensory imaginings to represent. We cannot be mistaken, for example, about whether
we intend to visualise Kings College or Hertford College, or about whether we assume or think the visualised object to be identical with one or the
other college.
Of course, we may err about whether our intention or thought and,
hence, also our sensory imagination is in fact concerned with a particular
object. Neither the intention to visualise Ulysses, nor the thought that what
we are visualising is Ulysses, suffices to ensure that we end up with an
rated sensory representation, given that, in the latter case, there is no distinct thought
over and above the episode of sensory imagining (see Dorsch 2012c).

Key Features of Sensory Imaginings

107

episode of visualising that refers to a particular individual. For there is no


specific person in the actual past or present that we denote with our intention or thought. But this is unproblematic for the suggested explanation of
the immunity to error through misidentification, since this example does
not involve any identification in the first place: visualising Ulysses is an instance of visualising a type of man, and not of visualising a particular man.
Indeed, whether sensory imaginings are particular in their representation
is likewise best accounted for in terms of the involvement of an intellectual
element. The idea is that sensory imaginings refer to particular objects
whenever and because they involve some intellectual element which
specifies the particular identity of what is imagined. Accordingly, it is not
surprising that sensory imaginings without particular objects fail to be immune or, for that matter, prone to error through misidentification (see
Section 3.5): both their particularity and their immunity are due to the
same factor, the involvement of some intellectual element.
Similarly, the presence of such an element also explains how sensory
imaginings have the potential to refer to aspects of reality. This time, the
thought is that the intellectual element (i.e., an intention or thought) may
determine the identity of the sensorily imagined objects by means of concepts that refer to real entities (such as genuine names, or definite descriptions that implicitly or explicitly involve the actually operator). Accordingly, sensory imaginings refer to real objects whenever and because
they involve some intellectual element which identifies what is imagined
with some (past or present) part of reality. By contrast, the involvement of
some element of sensory recollection or memory does not suffice to establish or even influence referentiality, as the example of visualising Kings
College illustrates. Even if we know that our episode of visual imagining
portrays the appearance of Hertford College and that it does so because its
content has (willingly or unwillingly) been determined by the content of
our past perception of Hertford College, it still refers to Kings College because this is what we want or take ourselves as visualising. What we do is,
in fact, visualising Kings College as having the visual appearance of Hertford College.
These considerations suggest not only that the strategy of explaining our
immunity to error through misidentification by reference to some intellec-

108

The Nature and Variety of Imagining

tual element may equally be applied to the particularity of sensory imaginings, as well as to their capacity to refer to actual entities; but also that it
does not matter for the success of these further applications whether the intellectual element is assumed to be an imaginative thought or an intentionin-action. But differences in explanatory power appear to surface once the
focus is shifted to the quasi-observationality of sensory imaginings that
is, to the fact that (some of) our knowledge of which kinds of object we are
sensorily imagining from a given perspective is not open to defeat by sensorily imagining the same objects from different perspectives. More specifically, the explanatory difference becomes apparent with respect to cases
where our indefeasible knowledge is concerned with the instantiation of
imperceivable features.
Consider the case in which you are visualising a lemon: you intend or
take yourself to imagine such a kind of object, and your imaginative episode involves the visual representation of the outside of a lemon. The question is now what happens if, after imagining cutting open the visualised object, you end up with a visual representation of the inside of a wax replica
of a lemon. Due to the quasi-observationality of sensory imaginings, your
awareness of this change in visual appearance can neither change the fact,
nor defeat your respective knowledge, that, originally, you started off with
visualising a lemon. It is not as if you suddenly discover that what you
took initially to be a lemon in fact has been a wax replica all along. But
there are still two possibilities with respect to which kind of object you are
visualising after switching from visually imagining the outside to visually
imagining the inside of whatever you are visualising. According to the first
possibility, you continue to imagine the object to be a lemon, albeit one the
inside of which looks like that of a wax replica. According to the second
possibility, you are now visualising a wax replica of a lemon that is,
either you are now visualising a numerically different object, or the object
that you have been imagining all along has, in the mean time, changed in
kind from being a lemon to being a wax replica of one.
Now, in both cases, your knowledge of which kind of object you are
now visualising is still immune to defeat. In the first case in which what
you are imagining changes in appearance, but not in kind you continue to
know that you are still visualising a lemon. This indefeasible knowledge is

Key Features of Sensory Imaginings

109

to be explained by reference to the intellectual element independently of


whether it is an intention-in-action or an imaginative thought that has
already been responsible for the quasi-observationality of your episode of
visualising at the outset of your imaginative engagement. In the second
case in which what you are imagining changes both in appearance and in
kind (and possibly also in numerical identity) you come to know that you
are now visualising a wax replica of a lemon. The question is how to account for this piece of newly acquired indefeasible knowledge. As it turns
out, the best explication privileges the identification of the explanatorily
relevant intellectual element with an intention-in-action, rather than with
an imaginative thought.
Assume, for the sake of argument, that what matters are solely your intellectual imaginings about the visualised object. The idea is that you are
visualising first a genuine lemon and then a replica of a lemon precisely
because you imaginatively take the visualised object first to be a lemon and
then to be a replica. Since you enjoy indefeasible knowledge of what your
imaginative thoughts are about de dicto, you therefore also enjoy indefeasible knowledge of which kind of object you are visualising. What this proposal assumes is that the change in how you visually represent the visualised object as being is accompanied by a change in how you think about.
Moreover, that the two changes occur together is no accident: while the
visual appearance of the outside of a lemon is indicative of there being a
lemon, the visual appearance of the inside of a wax replica of a lemon is
indicative of there being such a replica. This raises the question of what is
responsible for the simultaneity of the two changes.
It cannot be the case that the change in represented appearance determines the change in thought. For this would mean that you form the imaginative thought in response to discovering what the appearance is indicative
of. That is, you would start to think of the visualised object as a wax replica because you realise that it has started to look like one. But this would
render your knowledge of what you are visualising defeasible, since it
would depend on your knowledge of which visual appearances are indicative of which properties. Indeed, you would genuinely discover that what
you are visualising is not a lemon any more, but a replica of one.
In similar fashion, the change in thought should not be understood as

110

The Nature and Variety of Imagining

determining the change in represented appearance. That you are visually


representing the inside of the object as grey, waxen and homogeneous in
texture and not as consisting of different segments of yellow pulp with
seeds, separated by thin white walls is not the result of cognitive penetration. It might be true that visually representing something as waxen requires the possession or employment of the corresponding concept. But
visually representing something as grey rather than yellow, say, does not.
Furthermore, one should be able to visualise an object as having a waxen
and homogeneous inside, without even knowing that wax replicas of lemon
possess such insides.
Association, too, is not involved in the example under consideration.
Your mind may be constituted in such a way that thinking of a wax replica
of a lemon normally gives rise to a particular visual representation of the
inside of a wax replica. But the associatively caused visual representation
would likely be very specific, concerning an object of a certain size, colour, shape, and so on, and involving a certain angle on and distance to this
object. By contrast, there are no restrictions in these or similar respects on
how the inside of the imagined object is visually represented by you in the
given example. Moreover, the change in thought does not happen temporally prior to the change in appearance. For your thought-based knowledge
of what you are visualising is contemporaneous to your episode of visualising. Association, however, typically involves some temporal delay
between the associatively linked representations even though, as Hobbes
notes, associative thought is quick (Hobbes (1651): part 1, sec. 3, p. 9.).82
However, apart from cognitive penetration and causal association (as
well as similar subpersonal processes, such as priming), it is unclear how
thoughts could influence or bring about visual representations at least
without the mediation, say, of mental agency. In fact, our thoughts typically
have no direct impact on how we visually represent things as being. For instance, when seeing or visualising an instance of the Mller-Lyer illusion,
our knowledge that the two lines have the same length does not change our
visual impression of a difference in length. Or our thought that a given
painting of a woman holding a white lily is both intended and widely acknowledged to symbolically represent a innocent woman does not change
82

See Dorsch (2005) for a detailed discussion of the nature of association, and of how
it differs from imagination.

Key Features of Sensory Imaginings

111

the fact that we experience it as visually depicting the woman merely as


having the property of holding a white lily, and not also as having the property of being innocent.83 The example at issue seems to be analogous: how
you are conceiving of the visualised object (e.g., as a lemon, or as a wax
replica of one) has no direct bearing on how you visually represent it (e.g.,
as yellow or, alternatively, grey inside).
Accordingly, the changes in represented appearance and thought during
your imaginative engagement have to be determined by a third element;
and the co-occurrence of the two changes requires that this element is the
same in both cases. The only plausible candidate for this element is an underlying intention-in-action such as the intention to switch from visualising the outside of a lemon to visualising the inside of a wax replica of a
lemon which is responsible for both changes. But this renders any explanatory role of the imaginative thought dependent on its determination
by the intention-in-action. Indeed, the required presence of the intentionin-action is already sufficient to account for our indefeasible knowledge of
which kind of object we are visualising. In short, the needed intellectual
element (in cases like the one described) is ultimately practical rather than
theoretical.
It might be objected that the proposed account cannot capture cases in
which there is no corresponding change in intention-in-action. You might
end up visualising the appearance of a wax replica of a lemon without
(consciously or sub-intentionally) trying to do so, or even contrary to your
occurrent intention to visualise a real lemon. But at least for the Agency
Account of imagining this would not be a case of sensory imagining, given that the sensory representation of the wax replica would not be the result of imaginative activity: there is no continuous engagement in sensory
imagination, but instead a switch from imaginative agency to the
manifestation of some passive mental process (e.g., association or
memory). Hence, the issue of the quasi-observationality of sensory imaginings does not arise. Moreover, in such cases, there is no reason to assume
that what is represented goes beyond the visual representationality and
non-neutrality involved in any type of visual representation. That is, there
83

See Wollheim (1987) and Hopkins (1998) for the distinction between (our experiences of) the pictorial and the symbolic or otherwise non-pictorial content of pictures.

112

The Nature and Variety of Imagining

is no need to postulate the involvement of some intellectual element.


What has been emerging is the following picture of the quasi-observationality of sensory imaginings and, possibly, also of other non-perceptual
sensory representations. Which kinds of object a sensory imagination is
concerned with is, first of all, a matter of which perceivable properties are
sensorily represented as being instantiated (i.e., which states of affairs the
representation is sensorily non-neutral about). Imaginative thoughts may
add to this specification of what is imagined. First, they may determine the
imperceivable nature of the represented objects. Second, given that sensory
imaginings are, as such, neutral about the identity of the imagined objects,
imaginative thoughts are also central to the injection of particularity into
sensory imaginings. Both forms of determination by imaginative thought
are ultimately underwritten by imaginative agency as are the sensory representationality and the quasi-observationality of the imaginative episodes
concerned.84 The outcome is an instance of (voluntary) sensory imagining
that involves a conative and possibly also a thought-like intellectual
element.
Another challenge to the resulting view is Peacockes claim that our
knowledge of what passively occurring sensory episodes represent is
equally immune to error through misidentification, although these images
are not due to and accompanied by any intentions, whether they are prior
or concurrent (see Peacocke (1985): 26). Addressing this example becomes
even more pressing for the Agency Account since Peacocke takes these
passive images to be imaginative. I argue against the classification of passive sensory representations as instances of imagining in Section 14.4. Here,
I just would like to suggest two ways in which the immunity to error of
passive images can be accommodated by a view such as my version of
the Agency Account that takes the immunity to error of sensory imaginings to be due to their active character and their involvement of a respective intention-in-action.
First, it should be questioned whether passive images actually show this
kind of immunity from error at least if we have not yet started to actively
84

What is of importance here is that the agency involved in imagining is direct in the
sense of not relying on epistemic or merely causal means (see Section 13.2 and
Dorsch (2009b)). For we typically do not have the same kind of indefeasible access
to the epistemic or merely causal consequences of what we are directly doing.

Key Features of Sensory Imaginings

113

sustain them in response to their fleeting character, and thus to transform


them into sensory imaginings (see Section 14.4 for a more detailed discussion of this transformation). Lacking any involvement of agency, they
arise completely unexpectedly not unlike certain perceptual experiences
or sensory memories the contents of which we did not foresee. It would
thus be surprising if we would, immediately with their occurrence, come to
know the identity of the objects that they are presenting us with, given that
the same need not happen in the case of unexpected or unclear perceptions
or memories. It seems rather that our lack of expectation or foreknowledge
may leave us in doubt about what the images concerned are actually representing.
Second, even if passive sensory representations were immune to error
through misidentification with respect to the identity of the represented objects, it would still have to be shown that their immunity is of the same
kind and possesses the same origin as that of sensory imaginings. There are
other cases in which we are immune to error through misidentification concerning the represented objects for instance, in cases of first-personal
self-reference which are due to specific features of the thoughts concerned (e.g., that they involve a concept characterised by the fact that it
refers, in all of its applications, to the thinker applying it in that particular
case; see Peacocke (2008): ch. 3). Surely, the alleged immunity to error of
passive images would have to be a matter of very different properties than
that of first-personal thoughts. Hence, there is no good reason to accept
that the immunity of sensory imaginings could not be due to a third set of
specific features such as their active character and their involvement of a
conscious intention.

CHAPTER 4
Imagination and Knowledge

In the last chapter, I argued that one of the two main strict differences
between imaginative and cognitive episodes is that only the latter possess
the epistemic function to exemplify knowledge. This difference is reflected
by the fact that we typically rely on our cognitions when making up our
mind about what to believe and do with respect to the actual world, but
usually do not take our imaginings into account during such deliberations. I
also noted that cognitions can fail to fulfil their epistemic function: not all
perceptual experiences, say, are veridical; and not all judgements or beliefs
amount to knowledge. Relatedly, if we have sufficient doubts about whether a given cognition lives up to its function, we normally stop relying on it
in the formation of beliefs and intentions.
In this chapter, I would like to argue that something similar applies to
sensory imaginings. Although they do not possess any epistemic function,
they may sometimes play a role in cognition and ground or constitute
knowledge in a manner very similar to that of sensory cognitions. It is true
that imagining is not by itself, or intrinsically, constrained by how reality is
like. But it may still be so constrained in an extrinsic manner, namely by
being put to a certain use in the context of our active engagement with a
wider mental project. More specifically, we may actively ensure that certain sensory imaginings are reliable and non-arbitrarily veridical in their
representation of which properties certain objects in the external world instantiate.
The possibility of knowledge-constituting sensory imaginings is central
to my main line of argument and is therefore implicitly or explicitly ad-

116

The Nature and Variety of Imagining

dressed at many different places throughout this book. In particular, I


already discussed some of the preconditions for knowledge by imagination,
notably that instances of sensory imagining are non-neutral about how
things are in the imagined world (see Section 2.1), but may none the less
refer to aspects of reality; and how the second of these precondition is inseparably linked to mental agency (see Section 2.3). Later on, I return to
this important link between imaginative activity and the possibility of
world-directed imaginings, namely when I discuss why active imagining is
normally unconstrained by how the actual world is like, but still leaves
room for the active and reliable implementation of such constraints (see
Sections 13.2f. and 14.1).85
For the time being, I restrict myself in this chapter to presenting two examples for knowledge by imagination, and to illustrating why they should
count as grounding the same kind of knowledge-constituting beliefs as perceptions. One point to be highlighted is that the two examples count as
cognising only because of their being embedded in a suitable cognitive
mental project thus already shedding some light on the nature of imaginative projects and their difference from cognitive ones, which will become
more important later on (see Section 14.2). The possibility of cognising
sensory imaginings enjoys wider importance both with respect to the prospects of the Epistemological Account of imagining, and with respect to
the sometimes defended thesis that imaginings are uninformative with respect to the external world. While I return to the first issue in Part Two (especially Chapter 8), I address the second view and its endorsement by
Sartre and Wittgenstein in the last three sections of this chapter.

4.1. Two Examples of Cognising Imaginings


As already mentioned, the two examples have in common that they concern a situation in which a sensorily imaginative representation fulfils the
same function of grounding knowledge as a comparable perception would
in that situation. That is, the two examples involve sensory imaginings with
85

The different ways in which perceptions and occurrent beliefs are actually restricted
by reality are described in Section 7.1, where I turn to OShaughnessys account of
imagining.

Imagination and Knowledge

117

the potential to figure as bases for knowledge-constituting perceptual


judgements or beliefs.
Consider first the case of wanting to buy a new sofa and being worried
that it might be impossible to get it into the house due to its considerable
size. While in the furniture shop and looking at the sofa, one may try to
visualise the narrow door of ones room at home and mentally rotate the
visual image of the sofa in order to decide whether it might fit through the
door. The resulting image may ground a judgement about the relative sizes
of the two objects in the same way in which a comparable perception of the
two objects together might do. Just as one might be able to see, say, that
the sofa is smaller than the door, one can also recognise this difference in
size by visualising the two objects together. In both cases, the visual representation informs one about which object is larger, and enables one to
make a corresponding judgement. With respect to the case of visualising
the sofa and the door, what is primarily epistemically relevant for the
formation of the judgement is not the whole mental project of trying to correctly and visually compare the sizes of the two objects by mentally rotating one of them, but rather the last visual image of this project: that is, the
image of the piece of furniture as it is spatially related to (e.g., turned upside down and located before or within) the frame of the door in such a
way as to allow for a visual comparison of the sizes of the two objects. The
particular project is instrumental and perhaps even necessary for the
occurrence of the final image and its acquisition of the epistemic standing
required for the provision of knowledge. But only the final image carries
the information necessary for grounding the judgement in question and can
hence occupy the same place as a comparable visual perception or
memory.
The second example is as follows. Assume that one is about to meet
again an old friend whom one has not seen since his youth a long time ago.
One has agreed to an encounter in the busy hall of a station and is not sure
whether one will easily recognise him after all the years. What one might
thus attempt is to extrapolate his current visual appearance from ones
memory of his face and stature in his youth, together with ones general capacity to visually recognise how people change in appearance when getting
older. As a result, one may visualise his likely current appearance by form-

118

The Nature and Variety of Imagining

ing images of him on the basis of ones memory and adjusting them in response to whether one visually recognises the imagined person as being the
same in appearance as the remembered person, now only much older. Once
satisfied that there is such a correlation in content between the visual images and the original memory, one may use the visual image in question to
try to recognise the friend among the real crowd in the station. Perhaps the
imaginative engagement involved constitutes an extended mental project.
But again, it is the final image which is in the same way as a comparable
visual memory might be in other circumstances of primary cognitive importance in the recognition of the friend, or the formation of a corresponding judgement about his appearance. For it is this image which contains the
information required for recognising the friend, or forming the judgement.
What these descriptions of the two examples indicate is that they lead to
visual imaginings which seem to have the potential to ground knowledgeconstituting judgements or beliefs and, in the second case, also the acquisition of certain recognitional capacities. Indeed, the two final images share
important features with each other, as well as with the corresponding perceptions or memories, which should be sufficient to confirm their status as
cognising states and, more specifically, as grounds for intellectual knowledge.
First of all, the sensory imaginings described can non-accidentally and
reliably match reality and lead to practical success. Thus, one may correctly decide on the basis of ones image that the sofa will fit through the
door, and hence buy it; or one may be able to recognise ones friend by
means of ones image of him. The potential veridicality of the visual image
will thereby be a matter of, for instance, the issue of whether one rightly
remembers the size of the door or the appearance of the friend, whether
one does not change the perceived size of the sofa while visually rotating it
in ones mind, or whether ones impression of how the appearance of
people alters with age is adequate (at least with respect to the person in
question). If the imaginative representations turn out to be veridical, it is
therefore to be expected that they typically do so non-accidentally. Because
of their dependency on information provided by perception and memory,
and because of the importance of ones skill to properly visualise the respective scenarios by means of mental rotation or extrapolation, the result-

Imagination and Knowledge

119

ing sensory imaginings will match reality (if at all) not merely by chance,
but in virtue of the reliability both of the underlying cognitive representations and of ones control over the mental capacities involved in visualising.
Not everyone can always or even sometimes act reliably on his intention to sensorily imagine something. But it seems plausible to assume
that this is at least a possibility, in particular with respect to the first example. There is no reason to doubt, for instance, that we can mentally rotate simple objects (or shapes) without changing their sizes relative to other
simple objects (or shapes), at least within a practically permissible margin
of error. Hence, if the imagining person is sufficiently skilled, the mental
processes of actively producing sensory imaginings can be reliable if not
to the same extent as the comparable perceptual mechanisms, then at least
enough to enable practical reliance on them.
Furthermore, one may be aware of the potential reliability and non-arbitrary veridicality pertaining to the imaginative representations, in virtue
of the introspective knowledge of the nature of ones imaginative project
and in virtue of the memory of practical success in the use of visualising of
this kind in the past. For instance, one may be aware that one uses perceptions and memories and tries to come to an appropriate image by not altering the sizes of the represented objects while visualising them together;
and one may be aware that one had success with previous projects of a
comparable nature. Ones awareness of the potential trustworthiness and
non-accidental veridicality of ones imaginative project becomes also apparent in the fact that one may explicitly decide to rely on ones visualising
in order to find out whether the sofa will fit through the door, or whether
the person over there is ones friend.
Finally, the visual imaginings represent and refer to the respective objects in the world (the sofa, the door, and ones friend). Three considerations speak in favour of this idea.
First, visual imaginings may represent real objects if not particular objects (such as the sofa in the shop), then at least particular types or sorts of
objects (e.g., some kind of sofa which has the same size and shape as the
sofa in the shop; see also Section 2.3). Accordingly, it seems reasonable to
say that the two visual imaginings under consideration represent the re-

120

The Nature and Variety of Imagining

spective objects, or at least the corresponding types of object. Depending


on which is the case, the visual imaginings will be said to have provided
new information about the particular objects (e.g., the relative sizes of the
sofa and the door), or about objects of the respective kinds in general (e.g.,
about the relative sizes of objects of the same size as the sofa and the
door). In the latter case, a further step is necessary to be able to apply the
newly gained information to the particular objects in question. For instance, one may have to infer, from the piece of information that sofas with
this specific size are likely to fit through doors with that specific size, that
the sofa in the shop is likely to fit through the door at ones home. This requires, of course, that one knows that the particular sofa and door are instances of those types of sofas and doors. But this can be assumed, since
one set out, and has taken great care, to visualise a sofa and a door with exactly the same sizes and shapes as the real ones.
Second, the imaginings are deliberately designed to refer to the objects
and to correctly represent certain features of them (their relative sizes, their
appearance at a certain age). As a result, if the visualising is done properly,
their contents will covary with the respective facts in the world, and adequately reflect them. In this respect, they resemble perceptions with the
difference that the capacity of perceptions to represent reality is due solely
to factors independent of our intentions (such as evolution), while the capacity of visual imaginings to represent reality is at least partly due to our intentions.86
Third, there may be genuine cases of misrepresentation: if the visualising is not done properly, the visual imaginings need not fail altogether to
refer to the putative objects of knowledge in question. That is, there may be
cases in which the visual imaginings may match reality only partly: for instance, they may represent the size of the door correctly (or at least the relative proportion between its height and width), but not that of the sofa
(whether in comparison to the door, or in respect of the proportions among
its parts); or they may correctly represent only some features of the face of
the friend, but not others. In such cases, the visual imaginings do not ne86

See Dretske (1986) on the difference between natural functions and assigned functions to represent; and Lewis (1980) for the idea that it is essential to the representationality of perceptual experiences that their contents are nomologically dependent
on the respective aspects of the environment.

Imagination and Knowledge

121

cessarily cease to refer to the respective objects as a result of their partial


incorrectness. For the images may still be partly caused by the objects in
question (via the mediation of the respective perceptions and memories,
and by the activity of imagining); and they may still be intended to refer to
those objects. Now, the three factors taken together the partial match
between content and reality, the causal link back to the respective aspects
of reality, and the intention to refer to reality should suffice to guarantee
the referentiality of visual imaginings, even though they are likely not to be
fully correct. Again, visual imaginings may not be very different in this respect from perceptions. For it may be argued that the referentiality of perceptions is the result of the combined presence of a causal link back to the
respective aspects in reality and of an at least partial match with reality. In
any case, the two visual imaginings under consideration allow for genuine
misrepresentation. And this illustrates once more that they intentionally
represent the respective aspects of the world.87
To sum up, the sensory imaginings involved in the two examples can, in
suitable circumstances, represent and refer to certain aspects of the world,
be reliable and non-accidentally veridical with respect to these aspects, be
known and trusted by us to be so, and subsequently help us to successfully
engage with the world by means of action.
All these features, which they have in common with the corresponding
perceptions, contribute to the explanation of why the imaginings can figure
as substitutes for their cognitive equivalents, and how they can make the
same (or at least a comparable) epistemic contribution to the resulting visual judgements or beliefs as the latter. The respective persons engage with
the imaginative activities in the furniture shop or on the way to the meeting
point mainly because they can in this way avoid having to engage with
more complicated or more demanding activities, such as transporting the
sofa home and trying out there whether it will pass through the living room
door, or attempting to get hold of a more recent photograph of the friend.
That is, the imaginative activities are usually meant to be replacements for
certain bodily actions.
In a similar way, the final imaginative representations of the respective
projects may be understood as substitutes in the cognitive process for the
87

The idea that there is a close link between misrepresentation and full intentionality is
further defended in Dretske (1986).

122

The Nature and Variety of Imagining

perceptions or memories which would result from the corresponding forms


of practical engagement with the objects in question. This is reflected in
the great similarity between the respective imaginative and cognitive representations: both may reliably represent and refer to the same state of affairs (possibly in all details); both may lead us to form the same judgements or beliefs; and both may ultimately depend on our (past or present)
perceptual access to the objects in question.

4.2. The Absence of Alternative Grounds


In the light of these considerations, it appears very difficult to deny that the
visual judgements and beliefs formed in response to the final sensory imaginings in the two examples can constitute knowledge. It seems that one
can really come to know, by engaging in the respective imaginative activities, that the sofa is larger than the door, or that there will be particularly
shaped wrinkles on the cheeks of ones friend; and it seems that one really
is able to acquire the capacity to recognise the friend on the basis of ones
visual image. In fact, the whole rationale for the engagement in the imaginative activities concerned may very well be inseparably linked to the possibility of acquiring this kind of judgemental or recognitional knowledge.
But accepting this conclusion about the knowledge-constituting status of
the resulting judgements and beliefs is in principle compatible with rejecting the idea that the sensory imaginings (or their imaginative aspects)
should be taken to be the relevant epistemic grounds. This objection to my
interpretation of the two examples requires, however, the identification of
an alternative foundation for the kind of knowledge in question. The most
promising idea is perhaps that the two cases described involve some cognitive elements whether as a basis for the visual imaginings, or as a part
of them which figure as the actual grounds for the intellectual knowledge
instead. In other words, despite agreeing that the two visual imaginings are
centrally involved in the epistemic process leading to the acquisition of the
respective instances of knowledge, it might be disputed that they are
among the cognising episodes. But the strategy of denying the final sensory imaginings in the two examples the status of epistemic grounds for the
acquired intellectual knowledge faces two difficulties.

Imagination and Knowledge

123

One is that there does not seem to be a plausible candidate for an alternative ground. The two cases described involve certain cognitive elements
such as perceptions or memories of particular objects, and ones general
knowledge or recognitional capacity concerning certain ways in which objects behave (that they remain constant in size, or how they change appearance with the progress of time). But none of these elements carries the information which is contained in conceptualised form in the resulting instances of knowledge or recognition. For instance, neither the perception of
the sofa nor the memory of the door, nor the belief that objects keep their
relative sizes with respect to each other, can give rise to the judgement that
the sofa is larger (or smaller) than the door. Accordingly, the person trying
to find out whether the sofa fits through the door does not remember (or
perceive) that the sofa is larger (or smaller) than the door; nor does he
manifest some previously acquired knowledge of this fact. Equally, neither
being able to visually recognise how the appearance of persons change
with age, nor remembering what ones friend looked like when young, can
ground the acquisition of the capacity to visually recognise the friend, or
the corresponding intellectual knowledge about the current particular appearance of the friend. In particular, the person trying to meet her friend
does not recall his current appearance; all she remembers is his appearance
in the past.
It might, to some extent, be possible to infer the appearance of the friend
on the basis of ones beliefs about how people age, perhaps in conjunction
with ones beliefs about how one remembers him to have looked years ago.
For instance, someone may come to believe on these grounds that her
friend will have some wrinkles, or grey hair. But such beliefs are not involved in the second example. Instead, this example involves the use of
ones capacity to visually recognise age-related changes in the appearance
of people. Moreover, beliefs about these changes will not be rich enough to
determine all the biometrical details (the location, size and shape of the
wrinkles, etc.) with which the visual image may provide us on the basis of
the imaginative modification of the memory of the friend. What becomes
important here is that the example has to involve a visual ground for knowledge, and not merely an intellectual one. For the imagining subject may be
able solely on the basis of this ground to visually recognise the friend. In

124

The Nature and Variety of Imagining

normal circumstances (e.g., when the friend has no distinguishing characteristics, such as a scar on his cheek), a judgement or belief about the appearance of the friends face should not suffice for this task.
Hence, in both examples, what the underlying cognitive states represent
has to be imaginatively combined and modified by using ones ability to
visualise in order to extract the information needed for grounding the respective judgements or recognitional capacities. This is not to deny that
certain aspects of the cognitive elements concerned with the representation
of particular objects may still pertain to the final visual images. For instance, the image of the sofa and the door will presumably represent these
entities with their visually remembered sizes, shapes, colours, and so on.
And the image of the aged friend will presumably represent him as still
having certain of the facial features which one remembers him as having
had in the past (e.g., the distance between the eyes, the shape of the head,
etc.). Similarly, it is still possible that the visual images may be said to include or manifest aspects of ones general knowledge or recognitional capacity concerning how objects behave. That the sofa has not changed in
size, or that the friend is visualised as having wrinkles or grey hair, may
perhaps be taken to be instances of such an inclusion or manifestation. But
what is not true of the two visual imaginings is that they are cognising because they include or manifest aspects of perceptions, memories, beliefs,
or recognitional capacities.
To stress the point again, one does not perceive, remember or bring otherwise back to attention that the sofa is larger than the door, or that ones
friend will have particularly shaped wrinkles on his cheeks. The reason for
this is that ones acquisition of knowledge is based on both the perceptions
or memories of the particular objects and the general knowledge or recognitional capacity concerning how real objects behave. What is thus needed
is an additional process or activity which combines these cognitive elements in such a way as to lead to the extraction of previously unknown
pieces of information. In the two examples, this required additional activity
is imaginative and results in cognising visual imaginings.
In this respect, the two examples differ from the following case. 88 I may
88

See Kosslyn (1980): 1. Pylyshyns case of finding out how many windows ones
house possesses by imagining walking through it thereby visualising each of its
rooms and counting the windows in them is basically of the same kind (Pylyshyn

Imagination and Knowledge

125

wonder whether frogs have lips and, being unable to form a judgement on
this matter by merely thinking about it, may decide to form a visual image
of a frog in order to reach an answer. I may thereby form the image by falling back on my episodic visual memories of a particular frog, or by relying
on my visual memory or knowledge of how frogs generally look linked to
my respective recognitional capacity. This procedure may be veridical and
reliable with respect to the visual representation of the mouth of the
particular frog, or of frogs in general. Hence, it may allow me to form on
the basis of the resulting visual image the knowledge-constituting judgement that, say, frogs indeed have lips. But it can be argued that those elements of my image, which represent the visualised frog as having lips, are
instances or manifestations of my visual memory of the appearance of the
particular frog, or of my general knowledge or recognitional capacity concerning how frogs look. After all, it seems natural to say that, by forming
the image, I come to recall that (some) frogs have lips. 89 Hence, although
the overall visual representation may still be imaginative, its cognising aspects seem to be cognitive. That is, the resulting visual imagining appears
to be cognising only in so far it contains mnemonic representational elements which are cognising: what ultimately grounds the judgement are
mnemonic and not imaginative elements of the visual representation.
In contrast, in the two previous examples, there are no cognitive elements involved (over and above the visual imaginings) which could reasonably be said to ground the respective knowledge-constituting judgements
and, in the second case, the acquisition of the capacity to visually recognise
the friend. The main difference seems to be that, in the case of the frog, we
come to pay attention to previously unnoticed, but none the less consciously experienced aspects (similar to the moment when we suddenly
realise that someone is sad or angry: something which we have suppressed
or ignored before); while in the case with the sofa and the door, or with the
friend, we unearth new information which we yet have not experienced, or
been able to notice, in its entirety, given that it so far has been scattered
89

(2002): especially 164).


In the case of forming an image on the basis of the memories of a particular frog,
the resulting image should count as a memory image, and not an imaginative one, if
its formation does not involve making any imaginative changes either to the appearance of the frog, or to its represented surroundings.

126

The Nature and Variety of Imagining

over several disconnected perceptions, memories or beliefs.


The impossibility to identify an alternative ground for the judgemental
or recognitional knowledge acquired by means of the imaginative activities
is not the only problem for views which reject treating the two visual imaginings as the effective epistemic grounds. The other difficulty is that this
rejection seems to lack any plausible motivation. For the two visual imaginings do not differ from the corresponding visual perceptions or memories in the epistemically crucial aspects; while the actual differences
between the two kinds of visual representation do not undermine the potential of the former to provide sufficient epistemic support for visual judgements in a similar way to the latter.

4.3. Similarities to Perception (and Memory)


Let me begin with the epistemic similarities. I have already argued that the
visual imaginings under consideration are intentional and can, under the
right kind of circumstances, be veridical, reliable and informative in their
representation of the respective aspects of reality. According to traditional
views, this should already suffice for the imaginings to qualify as potential
grounds for knowledge. But the imaginings can (and usually will) also satisfy the further requirement identified, for instance, by OShaughnessy
(see Section 7.1) of rational integration with the subjects current picture
of reality. The person in question will typically form the respective judgement only if he takes the visual imaginings to be veridical, reliable and
more or less consistent with whatever else he believes and, in particular,
with his beliefs concerning the visualised objects, and the epistemic standing of the imaginings. Furthermore, the situation is no different with respect to two other ideas about grounds for knowledge which are present in
the literature: that they have to be constitutively linked to reality (see, e.g.,
McDowell (1994) and Williamson (2000)); and that they ultimately have to
be based on perceptual contact with reality.90
Given that it is plausible to assume that perception is the only way in
which we can immediately access reality (see Section 3.2), one further
90

See, e.g., OShaughnessy ((2000): 1ff.; 68ff.; 113ff.) and the discussion of his relevant ideas in Section 7.4.

Imagination and Knowledge

127

condition on knowledge may be that it has to be ultimately based on perception. But the two visual imaginings fulfil this requirement: they ultimately derive from the respective perceptions of the objects concerned (the
sofa, the door, the friend), as well as from the perceptions underlying the
acquisition of the relevant general knowledge or recognitional capacity
(e.g., perceptions of objects with constant sizes, or of changes over time in
peoples appearances). Of course, the two visual imaginings do not provide
direct access to reality as the corresponding perceptions would do. But
judgements or sensory memories are not immediate in this sense, either,
and too require past or present perceptual mediation, too.But neither prevents them from having the potential to constitute knowledge. Hence, there
is no reason to worry about the indirectness of the imaginings, and their ultimate dependency on perceptions.
Similarly, no particular problem should arise pertaining to the visual
imaginings, if it is assumed that knowledge-constituting states have to be
partly constituted by (a link to) the respective aspects of reality. For the
idea that such imaginings are factive and differ in kind from phenomenologically indistinguishable non-cognising imaginings should not be more (or
less) problematic than the idea that cognising judgements or sensory
memories are factive and differ in kind from phenomenologically indistinguishable non-cognising judgements or sensory memories. In all three
cases, the factivity cannot be realised by means of a direct causal link to, or
immediate awareness of, the respective aspects of the world (as it happens
in the case of cognising perceptions). But while this has the consequence
that the idea of a constitutive link between the cognising visual imaginings
and the world would need further elucidation (which might not be easily
provided), the situation is no different with respect to cognising judgements or sensory memories which are said to be constitutively linked to
reality. Requiring a constitutive link to reality would thus be problematic
with respect to all non-perceptual forms of knowledge; while giving up on
this requirement in the case of memories and judgements would also permit giving up on it in the case of imaginings.
Finally, it is not difficult to identify the cognitive prototype or ideal
which the visual imaginings may be said to realise (if one wants to speak in
this way at all). OShaughnessy is right, it seems, that sensory imaginings

128

The Nature and Variety of Imagining

cannot exemplify the same cognitive prototypes namely successful perception and intellectual knowledge as perceptual experiences or beliefs
(see Chapter 7). But this is also true of sensory memories. And, just like
them, sensory imaginings can have their own cognitive prototype: namely
that of a reliable and veridical representation of reality formed by means of
appropriate imaginative activity. This, too, should count as a prototype,
given that it constitutes the ideal to which the respective imaginative engagement aspires to, and given that it combines the central elements of intentionality, veridicality, reliability, rational integration with ones picture
of reality, and (if also required) a constitutive link to reality.

4.4. Differences from Perception (and Memory)


The situation is very different with respect to other properties of cognitive
episodes, notably their cognitive attitude, their status as providers of prima
facie reasons, and their function or aim of cognising reality. The two visual
imaginings do not possess any of these features and differ in these respects
from the corresponding perceptions, memories or judgements. It is true that
the visual imaginings might be said to have been assigned the function to
lead to knowledge, or to have been aimed at cognition, in so far as they
have been designed to track knowledge. Even so, this assigned function or
aim would not pertain to their nature, but would instead be part of the contingent use to which they have been put (see Dretske (1986)).
Now, that the visual images in the two examples lack the cognitive
properties just listed does not undermine their status as potential grounds
for knowledge. Even if justification is necessary for knowledge, the kind of
justification in question need not amount to prima facie justification. There
is no problem with the assumption that sensory imaginings can justify beliefs, and that we may take them to justify beliefs, if we have appropriate
ancillary beliefs about their reliable and rationally faultless origin (e.g.,
about the nature of the imaginative projects of which they are parts, their
dependency on perceptions or memories, their reliability in the past, or
their coherence with the rest of our beliefs). Indeed, although it is plausible
to assume that perceptions and memories usually do not require additional
beliefs in order to be able to possess and exert justificatory power, they

Imagination and Knowledge

129

sometimes do for instance, when we initially have good reason to doubt


their trustworthiness, only to be ultimately reassured by some further evidence which outweighs our (maybe still remaining) doubts (see Martin
(1992)). Accordingly, the visual imaginings will not be the only mental
episodes the justificatory power of which depends on certain ancillary beliefs.91
Similarly, there is no good reason to assume that the lack of a (say,
evolved) function or aim to cognise reality prevents the two visual imaginings from cognising reality. It should suffice that, in the two examples, the
sensory imaginings are designed by the respective persons to play a certain
role in their cognitive lives and to inform them about how things stand
just as some drawings on a board may be designed to represent the movements of players on a football field (see Dretske (1986)). Finally, the possession of a cognitive attitude that is, first of all, the involvement of a
claim about how things are is equally optional for knowledge, at least as
long as sensory grounds for knowledge are concerned. Making a claim to
truth is not necessary for a mental state to actually be true and cognising
with respect to reality (as it is also not sufficient for being true and cognising). Perhaps, intellectual knowledge implies belief and, hence, a cognitive attitude (see Williamson (2000)). But there is no good reason to assume that something similar applies to non-intellectual grounds of knowledge and perception (or episodic memory). Similarly, the phenomenologically salient attitudinal aspect of episodic cognitions may be said to reflect their epistemic function and also, perhaps, their status as providers of
prima facie reasons (see Dorsch (2009b) and (2013)). But, as I have just
argued, the lack of these features and, therefore, also their phenomenal
manifestation is unnecessary for the general capacity to provide knowledge.
Hence, although visual imaginings possess neither a cognitive attitude,
nor prima facie justificatory power, nor the function to cognise reality, they
can under the right circumstances be used and relied on in cognition.
That is, episodes may serve as grounds for knowledge, not only because of
91

Compare also OShaughnessys view that perceptions can figure as grounds for
knowledge only if they are rationally integrated which is another form in which
the justificatory power of such states may partly depend on the nature of the underlying beliefs.

130

The Nature and Variety of Imagining

their nature, but also because of their contingent use to which we put them.
However, some additional qualifications are requisite here.
First, the use of sensory imagination is presumably often less reliable or
less precise in its provision of visual information than the use of perception; and we will refrain from judgement more often with respect to the
former than to the latter (e.g., if the issue of the sofa fitting through the
door is one of centimetres). But this does not prevent imaginings from being sufficiently reliable and precise as it is reasonable to often rely on
perception, even though the use of calibrated tools of measurement or of
photographs might be more trustworthy and accurate.
Second, the acts of visualising in the examples do not provide us with
new evidence. Instead, they extract information, which has been previously
unknown to us, from the portions of evidence that we already have had
(e.g., in the form of our perception of the sofa, the memories of the door or
the friend, or the knowledge about the ageing process). This is unproblematic since having new evidence (i.e., involving the epistemic intake of new
information) is not a condition on the provision of knowledge as illustrated by the possibility of a priori knowledge. Indeed, this may be taken to
suggest that the imaginative activity fulfils a cognitive role similar to that
of a priori reasoning, given that the latter (at least according to one common understanding) also leads to the gain of information without the acquisition of new evidence. For example, if we had previously measured the
piece of furniture and the door with some kind of tool, instead of our eyes,
we could have just calculated in our heads and on the basis of our
memory of their sizes whether the former would fit through the latter.
And just as such an example of a priori reasoning would make it superfluous to perform the overt action of solving the problem by means of a sliderule or a geometrical drawing, the act of mentally rotating the sofa replaces
the much more demanding task of actually moving it to ones place and determining whether it fits through the door.
Third, it is also important to note that the sensory imaginings can fulfil
their cognitive role only on the basis of information that is gathered, processed and stored by cognitive means, such as perception, memory and belief. This, together with the fact that they do not provide new evidence,
seems to be reflected by the fact that imagining does not constitute an inde-

Imagination and Knowledge

131

pendent source of empirical knowledge.


Fourth, the idea that the visual imaginings in the two examples are capable of grounding knowledge is compatible with their quasi-observationality and their immunity to error through misidentification with respect to
what we sensorily imagine.
As already noted, visual imaginings are quasi-observational in the sense
that, when we are visualising an object from a certain point of view, we do
not appear to be able to acquire knowledge by coming to visualise it from
another point of view about those aspects of the object which are currently visually unrepresented, but which would be visually represented as a
result of adopting an suitably different perspective on the object (see Section 3.4). The quasi-observationality of visual imaginings has the consequence, say, that we cannot acquire any knowledge or make any surprising discoveries about a visualised objects occluded backside by mentally
rotating it. To reuse the example from Section 3.4, it cannot happen that we
think that we visualise a closed cube, only to find out, when visualising it
from a different point of view, that it is merely an open box, or an object
consisting of three planks meeting in one point and attached to each other
at right angles. In contrast, when we see what we take to be a closed cube,
there is always the possibility that it turns out to be an open box or an object consisting of three attached plates and can be visually recognised as
such from a different perspective.
However, the quasi-observationality of visual imaginings does not prevent them from figuring as grounds for knowledge in the manner of the
two examples. First of all, the knowledge in the examples does not concern
currently visually unrepresented aspects of the objects which would be
visualised as the result of an appropriate change of the imaginative point of
view. The discovery that the sofa is larger than the door, or that the friend
has these particular wrinkles, is not (simply) a matter of altering ones imaginative perspective on the visualised objects. Rather, it is a matter of imaginatively combining the images of two objects and visually comparing
their sizes; or of modifying the appearance of an object in conformity with
ones knowledge about how peoples appearances change over time. One
consequence of this is that alterations in perspective are at best instrumental in but never cognitively contributing to the successful pursuit of

132

The Nature and Variety of Imagining

the respective attempts to acquire knowledge by means of imagining. The


person in question may have to change several times his imaginative
perspective on the sofa in order to find out whether it would fit through the
door (as he may in real life have to shift and turn around the sofa several
times before he can be sure about whether it fits through the door). But he
will not get the answer simply by visualising the sofa (or the door, for that
matter) from a different point of view. He also has to merge the two images
of the two objects and use his capacity to visually compare sizes.
In addition, the two examples need not even involve any perspectival
change. For instance, the visualisation of the sofa may begin with a frontal
perception of the side of the sofa, which will allow one to simply spin
around its shape in a two-dimensional plane and compare it with the shape
of the door. And the same is true of the other example: to achieve a cognising image of the friends face, one need not change ones point of view
on the face at all. Third, if the imaginative activity actually involves some
instrumental changes in point of view, its success seems to depend partly
on the quasi-observationality of visual imaginings. If what is visually represented as a sofa turns out, when mentally rotated, to be a longish object
with hardly any depth, the conclusion that it would fit very easily through
the door would not be of much help in deciding whether the real object fits
through the real door. Thus, it seems crucial to the mental rotation of the
sofa and perhaps to mental rotation in general that there are no surprise
discoveries about the previously visually unrepresented aspects of it. The
quasi-observationality of visual imaginings is therefore no threat to their
capacity to cognise reality under suitable circumstances.
The same is true of another fact about visual imaginings which distinguishes them from visual perceptions: namely that we are to a certain extent immune to error through misidentification with respect to what we are
visualising (see Section 3.5). This might be taken to undermine the capacity of the two examples to provide us with knowledge, given that the
piece of knowledge concerned is empirical and fallible. However, the acquisition of knowledge at issue does not concern the identity of the visualised objects. On the contrary, it presupposes that the images indeed refer to
what we take them to refer to. The fallibility involved concerns the ascription of certain features to the imagined objects, and not their identification

Imagination and Knowledge

133

partly because sensory imaginings do not refer to real objects by themselves, but depend for their referential capacity on the wider context, such
as their particular use in some mental project. Consequently, the immunity
to error through misidentification concerning what we are visualising is as
unproblematic for the cognising status of the two examples as the quasiobservationality of visual imaginings. In fact, these two aspects of visual
imagining may stem from the same fact, namely that the referents of our
visual imaginings are ultimately determined by our accompanying desires,
intentions or thoughts about our visual imaginings (see Sections 2.3, 3.6
and 14.1f.).
To conclude, the significant differences between the sensory imaginings
and their perceptual or mnemonic counterparts do not matter for the fact
that the former can, under the right kind of circumstances, cognise the
world in a manner very similar to that of the latter. The final visual imaginings in the two examples should therefore be accepted as potential grounds
for knowledge.

4.5. Imagination and Information


The claim that some instances of sensory imagining realise quasi-perceptual knowledge is in tension with a certain claim that enjoys some prominence in the literature on imaginings:92
(NI) Imaginings are essentially uninformative with respect to the ex92

The N in (NI) stands for negation claim (see the introduction to Part Two). Note
that what is at issue here is the stronger claim that imaginings cannot provide us
with new information; and not merely the weaker and presumably uncontroversial
claim that imagining is not a source of evidence, and that imaginings therefore
cannot provide us with new evidence (see Section 3.4 for further discussion of this
distinction). That imagining is not a source of evidence still leaves room for the imaginative acquisition of new knowledge on the basis of the extraction of information
previously hidden (i.e., unknowingly contained) in already acquired information
(see the two examples discussed above) and is thus incompatible with (NI). Besides,
the uninformativity of sensory imaginings even if being essential to them is
surely not among their fundamental features, but rather derives from one of their
more basic characteristics (e.g., their special origin in the mind or in mental agency).

134

The Nature and Variety of Imagining


ternal world.

It is natural to assume that mental states and, in particular, non-intellectual mental episodes can constitute knowledge of reality only if they
provide us with information about it. 93 Seeing the greenness of a tree, for
instance, contains information about the colour of the leaves of the perceived object, as well as about the fact that it is a tree and has leaves. If it
did not convey this information, then it could not help us to acquire the
corresponding knowledge about the nature and other features of the perceived object. Hence, if imaginings are, by their very nature, uninformative, they cannot constitute knowledge.
This means that either (NI), or my interpretation of the two examples
presented in the last chapter is false. In Section 8.1, I argue that it is (NI)
which should be rejected and partly because of its inability to accommodate the two examples. Here, by contrast, I just aim to undermine different ways of providing support for (NI) that are linked to the features of
sensory imaginings discussed in the last chapter, and which may also at
least to some extent be identified in the writings of Sartre and Wittgenstein, both of whom have endorsed (NI) in one form or another. 94 The conclusion of the remainder of this chapter is that (NI), in its strictness, is
false; and that it can at best be maintained that imaginings typically differ
from cognitions in being uninformative with respect to the external world.
Sartre is not explicitly concerned with the question of whether sensory
imaginings may provide us with information about aspects of the external
world (rather than with information about themselves, such as their content
or origin). But he is very clear about his belief that sensory imaginings differ from perceptions in that [we] can never learn from an image what [we
do] not know already (Sartre (1940): 9). Moreover, Sartre proposes a
Here, I use the term information in a very non-technical sense, mainly because this
is how Wittgenstein seems to employ the expression. In particular, I do not mean to
capture the technical uses introduced by Dretske (1981) and Evans (1982), although
some of the following considerations might apply to their notions as well.
94
OShaughnessy is another philosopher who endorses claims which ultimately entail
(NI). I discuss his reasons for endorsing them in Chapter 7, and object to these reasons in Chapter 8. The discussion here is more restricted in that it focuses on more
direct endorsements of (NI).
93

Imagination and Knowledge

135

straightforward explanation of why this is the case: namely because the information carried by sensory imaginings is put there by us in the process
of producing these imaginative representations, namely in the shape of an
intellectual element which specifies (Sartre (1940): 9).95
Although his concern is primarily with knowledge about the sensory
imaginings themselves and, more specifically, about what they sensorily
represent, there is no obvious reason why Sartres thesis and his explanation could not be extended to knowledge about the external world. If how
our sensory imaginings represent their objects as being is determined by
us, rather than the world, and if, as a result, we cannot learn anything new
about the content of our sensory imaginings, then we should also be unable
to acquire new knowledge about corresponding aspects of external reality
that might be represented by means of the possession of that content. For
example, if I know that what I am visualising is a cube and, moreover, the
cube on the desk of the mathematics teacher next door, precisely because
this is what I am trying or taking myself to do, then I am in no position,
simply on the basis of my episode of visualising, to acquire new knowledge about the real object in question for instance, that it is indeed a
cube that is sitting on a desk, or that it is the cube of the mathematics
teacher next door. I knew this all along, that is, even before engaging in my
imaginative project.96
However, the impact of this line of reasoning is limited. More specifically, it merely rules out the acquisition of three kinds of new knowledge: (i)
knowledge of the (imperceivable) identity of the sensorily imagined object;
(ii) knowledge of the imperceivable nature of that object; and, if applicable, (iii) knowledge of perceivable facts about that object, the imaginative
representation of which is not determined by corresponding intentions-inSee Section 3.4 for a discussion of the longer quotations from which these citations
are borrowed.
96
Information (or knowledge) thereby counts as new for us if we do not already have
access to it by means of other mental representations such as the thoughts or intentions involved in the formation and determinations of the sensory imaginings
which store it in fairly easily recognisable form. Hence, what is claimed is that, for
Sartre, sensory imagining does not lead to a gain in information. That is, all the information, which we can become aware of after the occurrence of sensory imaginings, has already been available to us before this occurrence.
95

136

The Nature and Variety of Imagining

action, or by prior knowledge-constituting representations of those facts


(e.g., some perceptions, memories or beliefs).
To take the same example, visualising the cube does not allow me to
discover that it is identical with the one on the mathematicians desk (or
any other particular cube). I already knew that; and if I had not, I would
have just visualised some cube, but no cube in particular. Similarly, that the
desk, which I am visualising as part of the scene, is in the room next door
is something that I cannot learn from visualising it. Again, either I was
already aware of the desk being next door before starting to visualise it; or
my imaginative project would have stayed neutral on this aspect and not
contained any information to this effect. Something similar is true of the
cubes property of lying on top of the desk (or, if one prefers, of the colour
of the cube) if its visual representation by my imaginative episode is the
result of me having intended to visualise it as lying on top of the desk (or
as having the colour in question): the imaginative representation could not
figure as ground of my knowledge that the cube is (represented as) sitting
on the desk.
The considerations above do not rule out other kinds of knowledge acquisition because they exploit features of sensory imaginings that are themselves limited in scope. The immunity to error through misidentification
concerns only the identity of the sensorily imagined objects, while as argued in Section 3.4 the quasi-observationality of sensory imaginings extends only to changes in perspective. More precisely, sensory imaginings
are quasi-observational in the sense that, despite their perspectivalness, our
basic knowledge of which kinds of object we are visualising is indefeasible
with respect to visual representations of the objects from different points of
view. In particular, we cannot acquire knowledge about the currently hidden visible aspects of visualised objects by changing our perspective onto
them. Both the immunity to error through misidentification and the quasiobservationality therefore come with a substantial restriction on the
amount and nature of new information that we might acquire with respect
to visualised objects. But this restriction is not all-embracing: it just concerns knowledge of type (i) and (ii). Knowledge of type (iii), on the other
hand, is excluded because of the influence of the intellectual element in
sensory imagining that goes beyond the foundation of the immunity to er-

Imagination and Knowledge

137

ror and the quasi-observationality (see Section 3.6, especially the last subsection). But none of these features of sensory imaginings puts any further
restrictions on what we might come to know on the basis of engagement in
sensory imagining.
One consequence is that sensory imaginings may, in some sense, count
as manifestations of knowledge that we were already in possession of before entering into the imaginative projects at issue. My imaginative representation of the cube on the desk is an expression of my knowledge that
there is such a cube on the desk in the room next door. The same happens if
I suppose, as part of a hypothetical argument, that the Earth is flat in response to knowing that the Earth is flat, and to intending to imagine something about this planet as it actually is (e.g., I would like to discover some
metaphysical or scientific truth about the Earth). Or, to take another example, I may use visualising to exhibit my generic knowledge of the visual
appearance of frogs (see Section 4.2). But in none of these cases does the
sensory imagining do any justificatory work, or lead to new knowledge.
Moreover, it is unclear whether the manifestation or expression of knowledge in the way described amounts to the constitution of knowledge (as
defined in the introduction to Part One): that is, it is unclear whether the
(sensory) imaginings concerned should count as instances of knowledge. I
return to this question further below, when discussing Wittgensteins view
on the uninformativity of sensory imaginings.97
In any case, what remains unaffected by the line of reasoning presented
above is the possibility of sensory imaginings providing us with new information about perceivable facts about external objects if their imaginative representation is not determined by prior knowledge of those facts, or
by the prior intention to imagine precisely those facts. This is the case, for
instance, in the two examples discussed in the preceding sections: they
concern perceivable facts namely facts about relative size or visual ap97

Note that this question is different from the question of whether episodes of memory
may constitute knowledge. The answer to the latter should surely be yes, at least in
the sense that memories are ways of preserving and recalling knowledge (see Martin
(2001)). Accordingly, assuming that that the visual representation of a generic frog
should count as an episode of remembering (though not of recalling) rather than an
episode of visualising, this example is to be distinguished relative to its status as
knowledge from the others presented in the main text.

138

The Nature and Variety of Imagining

pearance which the subject concerned is not cognitively or conatively


aware of prior to producing the final episodes of sensory imagining of the
projects in question. As already noted in the previous section, all that
quasi-observationality entails is that imaginatively changing ones perspective on visualised objects cannot lead to the discovery of previously
unknown features of them as it may be the case, for instance, when we
are walking around perceived objects. But this still leaves room for knowledge gained by imaginatively comparing two objects, or by modifying the
appearance of an object in conformity with certain knowledge about how
things change over time. For in neither case does the extraction of the new
information require altering ones perspective on the imagined object in order to discover some previously hidden features of the objects.98

4.6. Spontaneity, Attention, and Filling-In


But let us return to Sartres explanation of why we cannot learn anything
from attending to an instance of sensory imagining: [we] will never find
anything there but what [we] put there (Sartre (1940): 9, as already quoted
in Sections 3.4 and 3.6). The underlying thought is that our sensory imaginings and, especially, which information they carry is determined by
their origin in our active or spontaneous mind, rather than by factors external to it (see Section 5.6 for a more detailed discussion of this idea).
Hence, given that being informative about something requires being determined (i.e., constrained) by it in a suitable way, sensory imaginings cannot inform us about how the world is like.99 This line of thought differs
Indeed, the first example discussed in Section 4.1, which involves the revelation of
previously unnoticed aspects of the objects concerned by means of mental rotation,
essentially exploits the quasi-observationality of sensory imaginings. Further below
in this section and in Section 7.2, I discuss other cases in which attention may disclose previously unnoticed aspects of what is sensorily imagined.
99
See the last part of Section 7.1 for a similar line of thought in OShaughnessys writings. What Sartre says here looks also very much like Wittgensteins observation
that imaginings are uninformative because and to the extent to which they are
actively formed by us (see Section 4.7 below). But Sartres notion of activity is
weaker than Wittgensteins: while the latter is limited to voluntary agency, the
former also includes involuntary spontaneity, such as rational responsiveness (see
98

Imagination and Knowledge

139

from the preceding one in that it focuses on the determinants of sensory


imaginings, rather than on their involvement of an intellectual element and
on the consequences of this involvement (i.e., the immunity to error and
the quasi-observationality). But the present argument still faces similar
problems as the previous one, which is partly due to the fact that the intellectual element is also assumed to be due to the minds spontaneity.
However, there are two different ways in which the mind may be spontaneous and bring about sensory imaginings (see also Section 5.6); and
both are insufficient to establish the uninformativity of sensory imaginings,
though for different reasons. Voluntary spontaneity occurs when a prior
intention to imagine something is acted upon, or when we are simply trying to imagine something. In both cases, an intention-in-action (or volition) occurs that guides and determines our imaginative activity, and
which becomes itself part as an intellectual element of the resulting instance of voluntary imagining (see Sections 3.6 and 14.4). Involuntary
sensory images, by contrast, do not involve anything like an intention-inaction, given that they are not instances of (voluntary) agency.100
Before addressing, in the next section (namely in the context of discussing Wittgensteins thoughts on this issue), the question of why voluntariness is not enough to give rise to uninformativity, I would like to focus on
a specific feature of involuntary images which is at least partly responsible
for the fact that they leave room for discovery. The feature in question is
that, from our subjective perspective, involuntary images arise as much
from nowhere (i.e., unbidden and without the involvement of our own
agency) as perceptions do. This is precisely why the Agency Account in
contrast to Sartres view does not take them to be imaginative. 101 But it
also means that their content is as much open to exploration as the content
Section 5.6).
See, for instance, OShaughnessy (1980) and Pink (1996). It does not matter here
what the precise nature of the intellectual element involved in involuntary sensory
images is, and whether it may still count as conative (see Section 3.6). See Section
5.6 for more on how Sartre and Casey conceive of the spontaneity of the mind.
101
See Chapter 13, especially 13.4. The opposing view is endorsed not only by Sartre
(see Section 5.6), but also by Casey (1976: 34 and 63ff.), Peacocke (1985: 26f.), and
OShaughnessy (2000: 344f. and 349ff.). I discuss the nature of passive images in
Section 14.4.
100

140

The Nature and Variety of Imagining

of perceptions. What is crucial here is the idea that passively occurring


sensory representations as well as passively determined aspects of actively produced sensory representations involve the phenomenon of
filling-in, which again allows for discovery through attention (see above).
When we lie in bed at night and close our eyes, often very vivid streams
of images occur (e.g., hypnagogic images). Although we may guide or develop them further, they initially arise without our doing anything and may
continue to persist and follow their course in this passive manner (see
Dorsch (2010a)). In such a case, their content is not determined by us as
agents, but is instead filled-in by certain associative, mnemonic and other
passive processes in our mind. Because of this, we typically have to find
out what the respective images are about, in the same way as we have to
find out what our perceptions (or even material pictures) are about: namely
by paying attention to what they represent. Of course, this is not always
easy because of the momentary and fleeting character of the involuntary
images. But perceptions may be similar for example, when we look at
near-by trees, buildings or other parts of the landscape while sitting in a
fast-moving train.
There is also no good reason to deny the possibility of sensory imaginings that are voluntarily formed by us, but none the less include some involuntarily determined aspects. To slightly modify an example from Sartre
(who again borrows it from Alain; see Sartre (1940): 38), when I visualise
the frontside of the Pantheon in Paris, I need not attentively visualise each
column of its portico. But the image may still represent a determinate number of columns with a certain size and shape. For instance, my associate
powers, or my specific memory of the church, or my general visual
memory of how temple-like churches and their porticoes look like 102, may
fill in the gaps of the image left undetermined by my imaginative agency.
This filling-in occurs when we deliberately form an image that ends up
being too rich in detail for us to be fully controlled, or with respect to
which we simply do not bother to control every detail. However, given that
these involuntary aspects of sensory imaginings are, from the subjective
What this means is that the visual image does not incorporate the visual memory of
a particular, previously seen building, but instead that the associative or otherwise
involuntary mechanisms filling in the details exploit the general capacity to visually
recognise temple-like churches and their columns.

102

Imagination and Knowledge

141

perspective, given to us in the same passive manner as perceptual experiences, they allow, again, for a perception-like kind of attentive exploration
and discovery. That is, we may find out how many columns we have actually visualised by paying attention to the passively filled-in details of our
image of the Pantheon.
Further examples are deliberate attempts at sensorily imagining something, which are not entirely successful (i.e., involve some form of intentional failure), but still result in the production of some sensory representations (see Peacocke (1985): 26). That is, voluntarily triggered imaginings
may end up representing, not what we have intended them to represent, but
something else. For instance, I may decide to visualise a circle, but end up
visualising a shape which is not absolutely round and which I consequently
do not take to be a circle. Or I may fail in my attempt to daydream about
my visit to Lisbon simply by becoming distracted by thoughts about Rome
and by subsequently starting to daydream about this city instead. Again,
these examples illustrate the possibility of cases in which our will does not
fully determine our sensory imaginings and, hence, of cases in which attention may deliver new information about the passively determined aspects
of what is imagined.
The conclusion so far is that, if sensory representations occur involuntarily or include passively determined aspects of their content, they may
provide us with new information. Moreover, it is in principle possible that
the relevant pieces of information may actually be capable of grounding or
constituting knowledge about aspects of external reality namely if the
passive processes responsible for the determination of sensory content happen to be suitably constrained by how the external world is like. However,
the main candidates for such processes seem to be mnemonic ones (e.g.,
processes that lead to the conscious manifestation of episodic or generic
memory) which means that the resulting (parts of the) sensory
representations should count as instances of memory, rather than imagination. This is the case, for instance, when we succeed in acting on our intention to remember or recall something: we deliberately trigger some passive
mnemonic processes that determine how we sensorily represent the objects
concerned (e.g., consider again the frog example). Hence, while the involuntary origin of sensory representations does not suffice to prevent them

142

The Nature and Variety of Imagining

from being informative, it still supports the claim that there are no sensory
imaginings among the involuntary information-providing episodes. To this
extent, Sartres explanation of (NI), that is, of the uninformativity of sensory imaginings seems to be right.

4.7. Voluntariness and Cognitive Constraints


Indeed, his case appears to be even stronger with respect to voluntary sensory imaginings or, more precisely, with respect to those aspects of the content of sensory imaginings that are determined by our intentions, rather
than by some passive processes in the mind. The underlying thought is,
again, that, while our voluntarily controlled sensory representations may
very well carry information about the external world in so far as we may
choose them to express or manifest some of our knowledge about appearances (e.g., when we decide to visualise a certain friend or frogs just as
they really look like), the knowledge concerned has to be knowledge which
we were already in possession of before embarking on our imaginative
project. That is, voluntary sensory imaginings can never provide us with
new information: they can be informative about reality only to the extent to
which our active control over their content is guided or determined by our
prior knowledge of how the respective objects are like.
Wittgenstein presents a further consideration in favour of this conclusion, which also makes essential reference to the voluntary character of the
sensory imaginings concerned: they are taken by him to be uninformative
to the extent to which they originate in our intentions and other motivational states expressing our will:
Imaginings do not teach us anything about the external world, neither truly,
nor falsely. [...] Imagining subject to the will. (Wittgenstein (1984c): II, 63)
Precisely because imagining is an act of will, it does not inform us about the
external world. (Wittgenstein (1984d): 627)103
Both translations are mine. The originals read as follows: Vorstellungen belehren
uns nicht ber die Aussenwelt, weder richtig noch falsch. [...] Vorstellung dem Willen unterworfen. Weil das Vorstellen eine Willenshandlung ist, unterrichtet es uns
eben nicht ber die Auenwelt. Both passages could have been translated in terms

103

Imagination and Knowledge

143

Despite their incompatible views about which specific methodology


either phenomenology or the study of language games is adequate for the
philosophical investigation of the mind, Wittgenstein adopts a position
strikingly similar to Sartres, even describing it in almost the same words:
namely, that imaginings do not teach us anything about reality. Moreover,
Wittgenstein, too, restricts his observations to sensory instances of the imagination and agrees with Sartre that there is an essential difference between
sensory cognitions and sensory imaginings. In particular, the difference
between the two kinds of representation does not consist in a difference in
degree (e.g., in vivacity), but in a difference in kind (e.g., in attitude),
which makes it impossible for us to mix up the two general kinds of sensory episode (see Wittgenstein (1984c): II, sec. 63; see also Sections 2.1 and
3.5f. above, as well as Section 5.1 below).
But Wittgensteins approach also differs in some important respects
from Sartres. First of all, Wittgenstein endorses the Agency Account of
imagining and therefore rules out the possibility of passive imaginings.
This view is compatible with akratic imaginative episodes which, in some
sense, occur against our will and may be resistant to deliberate attempts to
change them (see Budd (1989): 113). But it denies that sensory representations, which originate in the subjects mind in a completely involuntary
manner, should count as imaginative.104 And then, Wittgenstein defends the
stronger conclusion that sensory imaginings cannot be informative at all
of the verbs to teach us or to instruct us; but only the second passage (the one involving unterrichtet) may also be rendered by using the verb to inform us. Besides, Wittgenstein does not discuss in his sketchy remarks the possibility of voluntarily controlled sensory imaginings that include passive filling-in. Given that
such cases are indeed possible (see above), Wittgensteins claim should be read as
being restricted to the voluntarily determined aspects of sensory imagining.
104
Wittgensteins use of the expression subject to the will might suggest that he favours the Subjection to the Will Account, according to which there can be passive
imaginings, as long as they are in principle open to be actively controlled by us (see
Section 13.5). However, as Budd (1989): 104ff.) has rightly pointed out, what Wittgenstein has had in mind here is the claim that sensory imaginings are essentially
subjected to the will (i.e., under its actual control), and not merely in principle subject to it (i.e., under its possible control). Note that imaginative activity may none
the less be sub-intentional, weak-willed, compulsive or obsessive (see Section 13.3).

144

The Nature and Variety of Imagining

with respect to the external world, in contrast to Sartres weaker claim that
they just cannot provide us with new information about the world. Indeed,
Wittgenstein understands the property of being informative about reality as
the property of carrying correct or incorrect information about it: that is, as
the general property of carrying truth-evaluable information, or having
veridicality conditions.105 Wittgensteins claim is therefore that sensory
imaginings are uninformative in the sense of lacking veridicality conditions because of the voluntary determination of how they represent objects as being. If we act successfully on our intention to visualise a cube as
sitting on the desk next door, it is, for Wittgenstein, inadequate to ask
whether the resulting visual representation is veridical or not in respect of
where it presents the cube as being located.
But Wittgenstein fails to take into account that the two examples of cognising visual imaginings discussed in the preceding sections show that our
voluntary control may actually be responsible for the extraction of new information (e.g., about differences in size, or about likely changes in appearance) on the basis of already stored visual information (e.g., information about the size or appearance of objects, as stored in the content of perceptions, memories and beliefs). The final sensory imaginings of the imaginative projects at issue carry their information precisely because the subject in question very carefully controls their formation and deliberately selects the cognitive contents that figure as informational inputs to the projects (i.e., the perception of the sofa and the memories of the door and the
friend). Moreover, she decides to let her imaginative agency be guided by
her knowledge of how middle-sized objects behave when moved around,
or by her knowledge about how peoples appearances change with growing
age. While she voluntarily determines which states of affairs her sensory
imaginings represent, she sets herself certain constraints (i.e., to respect
The application of the idea of truth to sensory representations is controversial not
the least because it is not unreasonable to take them to be non-propositional instances of object awareness (see Dorsch (2010c)). But since they are representational and non-neutral with respect to the represented situation (and, in the case of
perceptual experiences, come with a cognitive attitude), there is certainly a sense in
which they can (fail to) match whichever situation they represent and, hence, be
(in)correct or (non-)veridical with respect to that situation (i.e., reality in the case of
perceptual experiences; see Siegel (2010)).

105

Imagination and Knowledge

145

how real objects behave) and bases her decision about how to visualise the
sofa, door or friend as being on the relevant knowledge (i.e., concerning
how real objects behave). As a result, the subject is in control when visualising the sofa as fitting through the door, or the friend as having some
wrinkles and greying hair. In particular, her sensory imaginings are not the
passive manifestations of episodic or generic memory, or of some similarly
passive mechanisms, such as association.
The capacity of episodes of sensory imagining to provide us with new
knowledge about reality is therefore due to rather than, indeed, ruled out
by the fact that, in these imaginative projects described, we do not simply
follow our inclinations about how to modify our visual representations, but
try to adhere to certain principles: such as not to alter the size of the objects
when mentally rotating it or changing our perspective on it; or to conform
to our knowledge gained by experiencing the changes over time in the appearance of objects. Indeed, principles like these are constitutive of the
imaginative projects in question: the latter would be different imaginative
projects if they were not aimed, in their purpose to produce specific visual
images, at taking into account how the respective kinds of entity actually
behave. In particular, the imaginative projects could not be embedded in
the wider cognitive project of acquiring knowledge about real objects, given that acting differently in visualising the objects (e.g., by following principles that describe a very different and unrealistic spatio-temporal behaviour of objects) would fail to be conducive to non-arbitrary veridicality.
The underlying thought is, of course, that sensory imaginings possess
the ability to refer to real objects and be non-accidentally correct about
their perceivable features if they are suitably constrained (see Section 7.1).
The point made here is basically that such constraints need not obtain independently of us, but may in fact be established by our own (mental)
agency.106 Note that the sensory imaginings concerned may even count as
object-dependent in the same sense as, say, episodic memories: that is, as
depending for their existence or, at least, for their representationality on
106

Indeed, Budd (1989: 113) understands Wittgenstein as allowing for this possibility
and, hence, for the possibility of knowledge-constituting sensory imaginings in the
way described. But this reading has difficulties to make sense of the strictness of
Wittgensteins remark that imaginings do not teach us anything about the external
world (see the passage quoted above).

146

The Nature and Variety of Imagining

the existence of real objects as their referents. Hence, it cannot be argued


that they lack a truth-evaluable content for semantic reasons (e.g., because
they are like thoughts involving fictional or otherwise empty names that
fail to refer to anything).107 Instead, it should be accepted that the voluntarily controlled sensory imaginings in the two examples are capable of
referring to and matching reality in such a way as to having the potential to
be veridical and to deliver new information (though not new evidence)
about the external world. The underlying fact namely that the voluntariness of sensory imaginings may actually be conducive to their power to
cognise reality is particularly problematic for Wittgensteins endorsement
of (NI), since it is in tension with his central claim that they do not possess
any truth-value and veridicality conditions and, hence, cannot provide us
with any knowledge about the external world because of their actual subjection to the will.
None the less, it is possible to rescue a slightly weaker conclusion from
Wittgensteins considerations: namely that the impact of the will at least
normally prevents our voluntary sensory imaginings from being informative and constituting knowledge. For voluntarily formed sensory imaginings
are typically not constrained by us in such a way that they reliably lead to
the provision of knowledge. And this is, indeed, due to their origin in our
motivational states and, more specifically, our lack of concern for getting
things right when imagining them. Assuming that what we end up visualising is determined by our intentions and desires, rather than by the world,
it is not surprising that the resulting images are usually not sufficiently
curbed by reality in such a way as to be able to lead to quasi-perceptual
knowledge. It therefore stands to reason that what is responsible for the
typical uninformativity of sensory imaginings is their origin in the will (especially assuming the truth of the Agency Account).
This concludes my discussion of the main features of imaginings, which
has been partly motivated by the desire to provide a more detailed
specification of the two desiderata and especially of the second concerning explanatory force for a unified account of imagining. It is now time
For the semantic dependency of imaginings on cognitions, see Sections 2.3 and12.1.
For a discussion of object-dependence, see Martin (2002b). Finally, for the link
between failure of reference and failure to be truth-apt, see McDowell (1977) and
Evans (1985).

107

Imagination and Knowledge

147

to move on the consideration of specific attempts to capture the unity and


nature of the various forms of imagining.

CHAPTER 5
Other Theories of Imagining

The main result so far has been that there are two important differences
between the central cases of imagining and cognition a difference in their
attitude towards what they represent, and a difference in their epistemic
function with respect to the formation and justification of first-order beliefs. Any satisfactory unified theory of imagining should be able to accommodate these differences and, ideally, further elucidate them in terms
of a more fundamental feature present in imaginings and absent in cognitions. The different unified theories vary in their approach to this task. and,
notably, in their choice of the basic feature distinctive of imaginings.
As outlined in the Introduction, most of this book is taken up by the discussion and evaluation of three views that take imaginings to be characterised by the fact that they are not concerned with the cognition of the world
(the Epistemological Account), or by the fact that they constitutively depend on their cognitive counterparts (the Dependency Account), or by the
fact that they are essentially active (the Agency Account). Before discussing each of these three main accounts in detail, I intend to present other approaches to the unity of imagining and the differences between imaginative
and non-imaginative phenomena. Accordingly, I use this chapter to look at
alternative ways of accounting for the differences between (certain kinds
of) imaginings and cognitions, and to assess them both in their own lights
and in their relation to the general task of providing a unified account of
imagining.108
108

One alternative account the Subjection to the Will Account is not to be discussed
before Section 13.5.

150

The Nature and Variety of Imagining

One general problem for almost all of these accounts is that they do not
really apply to forms of imagining which are not (obviously) episodic, notably experiential imaginings and imaginative projects. But each of them
also faces its own more specific difficulties. My conclusion is that they are
all less promising than the Cognitive, the Dependency or the Agency Account.

5.1. The Quantitative Account


One possible option is to endorse the Quantitative Account by taking the
difference between imaginative and non-imaginative phenomena not to be
one in kind, but rather just one in degree. The view of Hume (1739) is
sometimes understood in this way, namely as claiming that the difference
between perceptions (i.e., impressions) and imaginings (i.e., ideas) is
merely a difference in vivacity (see, e.g., Stroud (1977): ch. IV; see also
Section 9.1). According to this specific version of the Quantitative Account, visualising (as well as thinking) is a non-vivid form of seeing:
whether we count as visualising or seeing a tree, say, depends entirely on
the contingent issue of how determinate and intense the visual experience
of the tree is. More generally, the Quantitative Account claims that visualising belongs to the same mental kind as seeing, supposing to the same
mental kind as judging, imaginatively feeling a pain to the same mental
kind as really feeling a pain, and so on. Assuming that sameness in kind
implies sameness in nature, the two members of each pair of imaginative
and non-imaginative phenomena are said to differ, not in their essence, but
in some contingent respect (e.g., their vivacity).
However, it is not easy to see how the Quantitative Account could ensure the unity of imagining. The main problem is that the difference in degree between imaginative and non-imaginative phenomena has to be the
same for all forms of imagining. But it is not clear whether, say, sensory
and intellectual representations could possess one and the same quantitative feature. It is unlikely, for instance, that visual experiences are determinate and intense in precisely the same way as thoughts are. Moreover, the
differences in attitude and epistemic function noted in Chapter 2 are differences in kind, and not in degree. Either a mental representation involves a

Other Theories of Imagining

151

claim about how things really are; or it does not. There is no third option,
and there is no continuum between these two poles. The same is true with
respect to whether a mental episode has the function to ground first-order
knowledge or not: such functions do not come in degrees.109
The claim that the phenomenologically salient difference between imaginative and cognitive episodes is qualitative gets further support by the
observation that we are not subject to a particular kind of error with respect
to our awareness of these episodes. Although we may err with respect to
the nature of our mental episodes, we do not seem to be able to err with respect to the kind of state which we take them (rightly or wrongly) to be at
least not if our conscious and first-personal access to them is concerned.
For instance, we may very well mistake one of our episodes of visualising
for a perceptual experience.110 But it seems absurd to say that we may be
mistaken about the fact that we take the episode of visualising to be a perceptual experience, and not, say, a visual recollection.
This, I think, is at least part of what Sartre and Wittgenstein may have
intended to capture when they stated that we cannot confuse the various
kinds of mental episodes and, in particular, perceptions and sensory imaginings.111 Now, it seems impossible to properly account for this observation if we assume that the phenomenologically salient difference between
imaginings and cognitions comes in degrees. For there would then be episodes (e.g., near the middle of the continuum), concerning which we might
very well be in doubt about whether they are subjectively given to us as
imaginative or cognitive.
In Section 9.1, I note that, for Hume, differences in vivacity come with differences
in attitude and function. Assuming that the latter are differences in quality, Hume
may therefore be able to avoid the two noted problems for the Quantitative Account.
110
Or one of our perceptual experiences for an episode of visualising: which seems to
have happened, for instance, in the Perky experiments at least under the assumption that pictorial experience is a special kind of perceptual experience (see Dorsch
(2012c)). See Hopkins (2011a) for a discussion of what the Perky experiments do
and do not show.
111
See Sartre (1940): 4, Wittgenstein (1984c): II, sec. 85, and Scruton (1974): n. 13, p.
100. See also the comments on the different language games pertaining to seeing
and visualising in Wittgenstein (1984c): II, sec. 63, and Wittgenstein (1984d): sec.
621.
109

152

The Nature and Variety of Imagining

5.2. The Phenomenal Account


According to the Phenomenal Account, the basic difference between imaginative and non-imaginative phenomena is qualitative and essential: it is a
difference in their phenomenal characters. I have already noted that the attitudinal difference between imaginings and cognitions constitutes a phenomenal difference and perhaps also reflects the difference in epistemic
function. But it is to be expected that the characters of the two kinds of
episode differ in other aspects as well. The Phenomenal Account takes
these phenomenal differences to be fundamental: either they are all equally
basic; or only some of them are basic, but ground all others. What this
means is that at least some of the phenomenal differences resist explanation in terms of more basic features that are personal and common to all
imaginings. This is compatible with there being subpersonal explanations,
or different explanations for different forms of imagining. For these additional explanations would not capture the unity of imagining on the personal level.
The phenomenologists such as Husserl, Sartre or Casey tend to focus their energies on providing detailed descriptions of these phenomenal
difference, rather than on inquiring into the possibility of accounting for
them by reference to more fundamental features of imagining. Indeed, as
already noted in the case of Sartre, it is not evident that the phenomenologists have the ambition to formulate a unified theory of imagining, or even
to identify the most fundamental difference between perceptions and sensory imaginings. Casey, for instance, openly expresses the view that there is
no need and perhaps also no possibility to try to discover a more fundamental explanation of the phenomenal difference in attitude, and that it is
better to focus on the descriptive enumeration of the phenomenal traits of
the mental episodes concerned (see Casey (1976): 33, as well as Section
5.6 below). Some passages in Sartre betray a similar outlook for instance,
when he concentrates on which origin imaginings subjectively seem to
have, rather than on which origin they actually do have (see Section 5.5).
Hence, their views come close to an endorsement of the Phenomenal Account.
However, the main objection to this view is that essential phenomenal
difference between two kinds of mental episodes are unlikely to be funda-

Other Theories of Imagining

153

mental. To see this, consider another essential phenomenal difference


among mental episodes. One phenomenal difference between seeing and
recalling something is that only the seen objects, but not the recalled ones,
are phenomenally given as being currently present before ones eyes (see
Martin (2001) and Dorsch (2013)). But this phenomenal difference corresponds to, and can be explained in terms of, an underlying difference in
how the two kinds of episode are linked to ones actual environment: an
episode of seeing, but not an episode of recollection, is causally or constitutively linked (in the right way) to an object in ones current environment (see, e.g., Martin (2002a)).112 It would thus be very surprising if the
phenomenal difference between imaginative and non-imaginative phenomena would not allow for such a more basic explanation especially since
visualising also differs from seeing both in that visualised objects are not
given as being there before our eyes, and in that episodes of visualising are
not causally or constitutively linked (in the right way) to ones current environment.
More generally, it is reasonable to assume that the phenomenal character
of mental episodes is not a primitive feature of them, but is rather determined or constituted by some of their more basic features, such as their representationality, their rational or functional role, their constitutive or causal
relation to the world and to other mental phenomena or, possibly, their
physical or neurofunctional realisation. In short, the phenomenal character
of episodes seems just to be a reflection of their underlying nature (see
Dorsch (2009b), (2010a) and (2012a)). Accordingly, while it is still worthwhile to try to come up with better phenomenological descriptions of the
essential phenomenal differences between imaginings and cognitions, a
unified account of imagining should also aim to explain in more fundamental terms why these differences obtain. This is, indeed, part of the de112

Whether the phenomenal difference between visualising and hallucinating is of the


same kind, and how it should be accounted for, depends partly on whether conjunctivism or (some form of) disjunctivism about perceptual experiences is true (see
Dorsch (2011a) for an overview of the various options). But even if it is assumed
that veridical and non-veridical experiences are of the same kind and differ in the
same way from instances of visualising, this difference can be further elucidated by
reference to underlying causal or representational differences (see, e.g., Dretske
(1995) and Tye (2000)).

154

The Nature and Variety of Imagining

sideratum of explanatory force (see Sections 1.1 and 2.1). Hence, if other
unified theories succeed in providing a satisfactory account of why imaginative and non-imaginative phenomena differ essentially in their phenomenal characters, the Phenomenal Account should not be assumed to capture
the unity of imagining on the most fundamental level possible.

5.3. The Ontological Account


Contrary to some of the observations made in Section 2.3, it might also be
appealing to postulate a strict and basic difference in the ontological status
of the objects that imaginings and cognitions can refer to, and to use this
distinction to account for the less fundamental attitudinal and cognitive differences. More precisely, it might be proposed that there are entities belonging to two different ontological categories: real (or cognised) entities
which are exclusively represented by cognitions; and unreal (or imagined)
entities which are exclusively represented by imaginings.113 The latter may
have some ontological characteristics in common with the former (e.g.,
their being entities and bearers of features). But the unreal entities must
differ categorically from the real ones in some other respects for instance,
in their lack of (the same) spatio-temporality, or in their lack of completeness or full particularity.114 Sartre, for instance, puts forward the idea that
the objects of (sensory) imagining and of cognition differ in their type of
existence (see Sartre (1940): 180) and the kind of space in which they are
The difference in question is primarily one between, say, real and unreal tables or
animals that is, between two categories of concrete entities. This is compatible
with the view that the objects referred to by fictional or similar discourse (such as
Odysseus or unicorns) are to be identified with abstract objects of a certain kind, and
that such abstract objects are part of the same reality or enjoy the same kind of existence as normal concrete objects (see, for instance, Salmon (1998): 293). In other
words, the distinction between real and unreal concrete objects may be reduced to
the distinction between (real) concrete and abstract entities. But this is not the only
option. Another and not necessarily less plausible (see Bodrozic (2009)) approach is to identify the distinction with the distinction between (concrete) existing
and non-existing entities (see, for instance, Meinong (1904) and Routley (1982)).
114
The details of the proposal do not matter for what follows. But the distinction
between cognised and imagined entities parallels that between real and fictional entities or ficta (see, for instance, Voltolini (2006) and Bodrozic (2009)).
113

Other Theories of Imagining

155

located (see Sartre (1940): 52, 86ff., 127ff. and 149); while Casey takes
them to differ in their determinacy (see Casey (1976): 36f. and 104ff.;
and Section 3.3 above). It is tempting to understand these claims in terms
of an ontological difference between two categories of things.
It is important to distinguish the resulting Ontological Account from the
ontological reading of the Phenomenal Account: that is, from the view that
there is a phenomenal difference between (sensory) imaginings and cognitions with respect to whether they present their objects in consciousness as
real (or cognised) or as unreal (or imagined) that is, as being non-neutral
about the actual or some other possible world (see Section 2.1). When
Macbeth becomes aware of the dagger, he does not experience it to be part
of reality and to be literally in front of him: the dagger is not given to him
as a real and perceived dagger, but rather as an unreal and hallucinated (or
imagined) one (see Dorsch (2010a) and (2012a)). The view that there is
such a presentational difference between imaginings and cognitions is
compatible with the idea that imaginings can refer to real objects, or that
there are indeed no unreal objects. The truth of this view may, however,
also support the endorsement of the Ontological Account, given that it
might be thought that the postulation of a categorial difference provides the
best explanation of the phenomenal difference at issue. I return to this issue
shortly.
In some passages of their writings, Sartre and Wittgenstein come close
to accepting the main claim of the Ontological Account. In particular, they
maintain that sensorily imagined objects are part of an entirely different
kind of space (or stand in entirely different spatial relations to each other
and further entities) than perceived ones (Wittgenstein (1984d): sec. 622
and 628; Sartre (1940): 8ff.). McGinn accepts this characterisation of the
distinction between perceived and imagined objects and analyses it as a
difference in the richness of spatial representations: sensorily imagined objects are represented as spatial objects (e.g., as extended), but not as spatially located (i.e., with a specific spatial location; see McGinn (2004):
58f.). However, it is unclear whether any of the three intends this observation to extend to supposed and judged objects; or whether this difference in
space (or spatial relations) implies, or is meant to imply, that we cannot
sensorily imagine the very same objects which we can perceive. Indeed,

156

The Nature and Variety of Imagining

Sartre as well as Casey allows for imaginings to refer to real entities


(see, e.g., Sartre (1940): 18; Casey (1976): 30ff.). Hence, his claims about
the difference in spatiality are perhaps to be understood phenomenologically, rather than ontologically. Wittgenstein, on the other hand, seems to
take the connection of imaginings to our will to be more fundamental than
their seeming spatial detachment from reality (see Section 4.7).115
Now, the first challenge for the Ontological Account is to provide an independent motivation for the assumption of a categorical difference
between real and unreal objects. If no such motivation can be identified,
the postulation of two different categories of entities in response to the task
of distinguishing imaginings from cognitions turns out to be entirely ad
hoc.
It might be a promising strategy to look for semantic considerations in
support of this distinction. It is conceivable to endorse a view about the semantics of statements about fictional or non-existing entities (e.g., sentences like Sherlock Holmes lives in Baker Street 226b or the present
King of France is bald), according to which our engagement with such
statements involves instances of imagining, which again can be expressed
by sentences of the kind at issue. There are furthermore some good reasons
to assume that such fictional or similar statements are best explained in
terms of entities which categorically differ from those referred to in nonfictional discourse (see, for instance, Salmon (1998)). But if such sentences
refer to a special category of objects, the imaginative episodes, that the
sentences can be used to express, are likely to do so as well.
There are general objections to this semantic approach to fictional statements (see Bodrozic (2009)), as well as worries about the possibility of deducing truths about mental and, especially, imaginative reference from
truths about linguistic reference. But even after setting these difficulties
aside, it is still doubtful whether the proposed strategy should be applied to
sensory and other non-intellectual and, possibly, non-propositional
forms of imagining. In particular, as noted in Section 2.3, sensory imagin115

Sartres comments may perhaps also be interpreted as pointing instead to the idea of
a difference in attitude or immediacy (see Sections 2.1 and 3.2, as well as Martin
(2001): 270f.). This understanding may gain further support from Sartres considerations about the quasi-observational character of sensory imaginings (see Sartre
(1940): 8ff., and Section 3.4 above).

Other Theories of Imagining

157

ings may represent mere types of objects, rather than individual objects.
For example, we may visualise some man without thereby visualising any
particular man. Indeed, this is likely to be the default case of visualising,
meaning that any reference to specific individuals has to be due to some
additional intellectual element (e.g., an accompanying thought or underlying intention; see Section 3.6). But non-particular imaginative representations are to be expressed by means of (indefinite) descriptions involving
existential quantification, with respect to which the problem of reference to
fictional or non-existing entities does not arise, and which therefore do not
require the postulation of unreal or non-existing objects (see Russell (1905)
and Bodrozic (2009)).
A second source of support for the idea of two fundamentally different
ontological categories of objects might be phenomenological evidence.
That is, there might be a difference in phenomenal character between imaginings and cognitions, the satisfactory explanation of which requires the
postulation of an ontological difference with respect to the represented entities. As noted in Section 2.1, the phenomenal difference cannot be located
in which properties and state of affairs are represented, given that imaginings and cognitions can at least in this respect have the same content. It
also cannot be located in the phenomenologically salient features pertaining to the mode of representation, given that imaginings and cognitions can
be of the same type (e.g., visual or intellectual). Accordingly, the proposed
phenomenal difference has to be part of the attitudinal difference between
imaginings and cognitions for instance in the shape of the difference in
non-neutrality about the actual or some merely possible world which may
manifest, or be constituted by, the difference in ontological category.
But it is far from clear why the explanation of this phenomenal difference should require or at least strongly suggest the postulation of a categorial difference between real and unreal entities. First of all, there are alternatives explanations available that are at least equally promising. In particular, the phenomenal difference may be due to a difference in immediacy (see Section 3.2 and Chapters 9ff.), or to a difference in voluntary control over what is specifically represented (see Section 14.1). Then, the general metaphysical and epistemological worries about non-real entities (see
Bodrozic (2009) and Reicher (2010)) may in fact render the alternative ex-

158

The Nature and Variety of Imagining

planations better than, and preferable over, the proposed account in terms
of a categorical distinction between real and unreal entities, given that they
do not involve a commitment to some otherwise unneeded and controversial category of objects. Finally, it remains unanswered why other important attitudinal differences for instance, between conative and cognitive
representations, or between perceptions and judgements, on the one hand,
and memories, on the other are not likewise due to ontological differences.
But independently of the possible motives for embracing the Ontological Account, the main problem for this view is that, although many instances of imagining do not refer to real entities and their real features, at
least some do make reference to reality and can even be veridical with respect to it. I may visualise the new car of my friend and thereby happen to
correctly visualise it to blue, without actually possessing any evidence of
its blueness. As already previously noted in Section 2.3, defenders of the
Ontological Account might reply that the imaginative reference to real entities involved in such cases is merely mediated by the more direct imaginative reference to unreal entities similar to the way in which our perceptual experiences of portraits refer indirectly to the portrayed persons by
means of referring directly to the perceived portraits themselves. But this
suggestion presupposes that the unreal objects of the imagination are picture-like in that they refer to aspects of the external world in the same way
as, say, portraits do. Given that imagining does not relate us to pictures external to the mind, unreal entities therefore have to be pictures internal to
the mind.
Sartres view (which is, ultimately, very close to Husserls early view) is
particularly interesting in this context.116 While he denies that there exists
116

See Sartre ((2004): Part II), as well as the excellent and detailed discussion of his
view in Hopkins ((1998): ch. 7). Sartres view is directly derived from Husserls
early account of sensory imagining ((1905/2006): ch. 2 and chs. 4f.).
Both claim that, in visualising a tree, we are not only aware of a tree (i.e., a threedimensional, tall, leafy object that does or could exist as part of a real forest in the
external world), but also of an inner analogon of such a tree (i.e., something
internal that could not form part of a real forest). Moreover, our visual awareness is
completely exhausted by our awareness of the inner analogon (or Bildobjekt, as
Husserl says). By contrast, our awareness of the real tree is mediated by, or a matter
of, our visual awareness of the analogon and the fact that the latter stands in for the

Other Theories of Imagining

159

anything like a mental picture, he still claims that imaginative awareness is


an instance of pictorial experience, namely pictorial experience of internal
rather than external pictures. In other words, Sartre takes imaginings to
intentionally represent entities (i.e., mental pictures) that actually do not
exist. He thus accepts that there is some phenomenological evidence for
endorsing the Ontological Account: the phenomenal difference between
(sensory) imaginings and cognitions consists partly in the fact that the
former, but not the latter, purport to make us aware of pictures. But he rejects the metaphysical conclusion and, hence, the central claim of the Ontological Account that there are indeed (mental) pictures which imaginings succeed in making us aware of. Indeed, Sartres rejection of the claim
that there are mental pictures is well supported (though not all of the following reasons are among his own).
First, it is unclear how the proposal could be applied to imaginative
real thing in virtue of resembling it to some extent (e.g., in outline shape, but not
necessarily in colour or size). Accordingly, visualising does not consist in the fusion
of two distinguishable kinds of awareness, but just in a single visual awareness
which represents one object by representing another.
The pictorial experience of a painting of a tree is even more complex, since it also
involves the perceptual awareness of the depicting material surface. That is, pictorial
experience does merge two kinds of awareness: the perceptual awareness of the
paintings surface, and the visual awareness of the depicted tree, that is, of the
analogon of the real tree.
I do not further discuss the pictorial approach to the nature of the representational
element involved in sensory imagining (and, possibly, also recollection), since the
resulting view is not applicable to intellectual imaginings and, hence, is not of much
relevance for a unified account theory. In the sections below, I discuss other
elements to be found in Sartres and Husserls theories of imagining, which are more
promising as candidates for the principal ingredient of a unified account such as
their observations about the imaginative attitude(s), or about the special subjective
origin of imaginings.
Note also that, later in his life, Husserl changed his mind about the nature of
sensory imagining and replaced the pictorial model with the echo model of the
Dependency Account to be discussed in Part Three. Whether his conception of
sensory imaginings as reproductions of perceptual experiences comes closer to
Humes causal conception of imagining or to the Representational Account (notably
OShaughnessys version) is another question. Note, however, that Husserl agrees
with Hume, OShaughnessy and Martin that episodic memories are also echoes of
perceptions (see Section 9.4, Husserl (2006) and, especially, Marbach (2006)).

160

The Nature and Variety of Imagining

thoughts. The best option is perhaps to treat them as instances of something


like indirect thought, in analogy to indirect speech. But intellectually
imagining some state of affairs need not involve nor presupposes the possession of the conceptual capacities required for the equivalent in thought
of the subjunctive.
Second, when we introspectively reflect, say, on an episode of visualising a tree, what we find is a tree, and not any kind of inner picture or similar intermediary representation (see Section 2.1, Martin (2002a) and
Dorsch (2010c) on the transparency of visualising). Note that the picturelike representations at issue (just like the sense-data postulated in the debate about perceptual awareness and reference) have to be first-personally
accessible in order to count as the direct objects of imaginative awareness
that are assumed to representationally mediate between us and reality.
Hence, whether imaginings involve some subpersonal intermediary representations does not matter here and is, presumably, also unlikely to
distinguish them from their cognitive counterparts.
Third, the inner pictures under consideration would presumably have to
be mind-dependent: that is, depend for their existence on being imaginatively experienced. But the assumption of such mind-dependent mental entities is metaphysically dubious (again, just like in the case of sense-data).
Fourth, even if what we directly imagine were, indeed, some mental pictures, they would likely belong to the same ontological category as perceivable objects. That is, they would enjoy the same kind of reality as trees
not the least because they would be realised by brain cells in a similar
way as trees by their xylem and other cells.
The last reply should also be made to the suggestion that (sensory) cognitions and imaginings differ in their objects in so far as the former make
us aware of external objects, while the latter make us aware of experiences
of such objects. Indeed, in this case, the exemplification of the same ontological category is certain: perceptual and other experiences are as real (or
unreal) as perceived objects. Hence, the suggestion at issue would not
amount to a version of the Ontological Account. Moreover, it would simply
be a consequence of the Representational Account: that is, of the idea that
imaginings are direct representations of cognitions (see Section 9.4).
The proposal that the two kinds of representation differ in that cogni-

Other Theories of Imagining

161

tions are concerned with particular objects (i.e., possess a particular content), while imaginings deal only with types of such objects (i.e., possess a
general content), may be more promising in so far as the difference
between particulars and universals alluded to is perhaps indeed ontologically fundamental. But the resulting view could at best distinguish sensory
imaginings from perceptions, given that thoughts whether judgemental or
imaginative can be particular (e.g., indexical or name-involving) or general (e.g. descriptive).

5.4. The Modal Account


The objections to the Ontological Account do not speak against all theories
of imagining which account for the difference between imaginative and
cognitive phenomena in terms of a basic difference in the status of what is
represented. For it may still be maintained that, while cognitions are fundamentally concerned with actualities, imaginings are fundamentally concerned with possibilities. This is the central thesis of the Modal Account.
Its main challenge is to make sense of the idea of a concern with an actuality or possibility in such a way that it promises to satisfy the two desiderata for a unified account of imagining.
The view that imaginings are concerned with a different modality than
cognitions should not be understood as meaning that imaginings refer to
unreal or otherwise non-actual entities. For this would just lead back to the
Ontological Account and its problems. Instead, the Modal Account should
assume that imaginings represent entities whether they actually exist or
not as possibly being a certain way.
This proposal may, again, be read in two ways. On the one hand, it may
mean that imaginings and cognitions differ in their contents (i.e., possibly
qualifies way in the last sentence): while the former just represent some
objects or states of affairs as instantiating the property of being possible,
the latter do not (or not only). On the other hand, it may mean that imaginings and cognitions differ in their attitudes (i.e., possibly qualifies being
in the last but one sentence): while the former are simply committed to the
presence of the represented entities objects and states of affairs in a possible situation or world, the latter are not (or not only).

162

The Nature and Variety of Imagining

The content version of the Modal Account can be rejected fairly


straightforwardly. First of all, visualising a tree does not involve visualising an instance of the property of being possible, just as imagining that it
rains does not require the possession or employment of the concept of this
property. Then, judgements or beliefs may consist in the cognitive endorsement of modal propositions and, notably, propositions about the possibility
of some object or state of affairs such as the proposition expressed by
Odysseus is a possible object (i.e., a constituent of a possible world), or
the proposition expressed by the flatness of the Earth (i.e., that the Earth is
flat) is possible (see, e.g., Gendler & Hawthorne (2002): 7f.). Finally, it is
reasonable to question whether there is really a property of being possible
(in the same sense, say, that there is a property of being round, or of being
a tree), as well as whether the concept or expression is possible really
functions in a predicative or attributive way (in the same way, say, as is
round or is a tree do).117 In particular, if the standard conception of possibility as existence or obtainment in a possible world and of possible
worlds as specific combinations of objects and (genuine) properties is
largely correct, possibility could not be among the properties, to avoid any
vicious regress. This conclusion about the ontological category of possibility does not necessarily contradict our practice to form judgements or beliefs involving modal propositions, given that they need not be assumed to
be ascriptions of the property of being possible. But it undermines the content version of the Modal Account.
The Modal Account should therefore be embraced if it should be embraced at all in its attitude version, which understands the idea of imaginings as representations of possibilities rather than actualities as qualifying
their attitude or commitment, and not their content. The negative part of
this view is that imaginings and cognitions can be concerned with the same
states of affairs, whether these are modal (i.e., concern the issue of whether
some further object or state of affairs is possible, necessary, and so on) or
not (i.e., concern the issue of whether something instantiates roundness,
tree-hood, and so on). The positive part of the view, on the other hand, is
that imaginings are non-neutral about the obtainment of the represented
117

Compare the similar worries with respect to the phenomenon of truth and the corresponding concept or expression is true noted by Frege (1892): 34f. and (1918): 60,
and Soames (1999): ch. 2, among others.

Other Theories of Imagining

163

states of affairs in some possible world, but neutral about their obtainment
in the actual world. In particular, there are two reasons why the attitude
version of the Modal Account cannot simply identify imaginings negatively as those mental episodes that represent some state of affairs (i.e.,
some objects as having certain features) without involving the claim that
this is really how things are.
The first rationale for a more substantial characterisation is that there are
many representations of states of affairs that lack a cognitive attitude, and
which are not or at least not obviously imaginative. That is, the range
of mental episodes concerned with the non-actual crosses the border
between imaginings and non-imaginings (see Goldie (2002)). One example
are desires or conative feelings: they are precisely concerned with states affairs that they do not yet assume to obtain in the actual world. Similarly,
hopes, expectations or regrets may involve thoughts about features that
they do not take to be actually instantiated. Finally, although wondering
about whether something is the case may lead to the endorsement of a certain proposition as true of the actual world, it need not do so. But it is
doubtful that any of these examples should count as (involving) imaginings
(see Sections 1.2f. and 14.2, as well as the concluding chapter). This does
not mean that the distinction between cognitions and imaginings cannot be
as clearly cut as envisaged in the first two chapters. But it indicates that
there are many kinds of mental representations that are neither cognitive,
nor imaginative.
The second and related reason for requiring a less minimal characterisation of imaginings is that the absence of a cognitive stance towards the actual does not suffice to establish the presence of a concern with possibilities. What is in addition needed is an imaginative stance towards the possible. This is related to the observation that imagining, too, involves a theoretical attitude, that is, some form of non-neutrality towards how things
are (see Section 2.1). Just as cognitions make a claim about how the actual
world is like, imaginings make a claim about some possible world is like.
In other words, imaginings should be characterised, not merely in terms of
the lack of some theoretical attitude, but also in terms of the possession of
another. This is necessary, among other things, to distinguish them from
desires, hopes, wonderings and similar non-imaginative and non-cognitive

164

The Nature and Variety of Imagining

mental phenomena.
One of the few unified theories of imagining, that have been recently
proposed and defended, embraces an instance of the Modal Account which
comes very close to the attitude version just sketched. 118 In his book The
Language of Imagination (1990), Alan White characterises both the sensory and the intellectual imagination in terms of their presentation of possibilities (note that imagining is here understood to be distinct from, and
not to include, visualising; see, e.g., White (1990): 108):
To visualise something is to think of what it does or would look like. [...]
We can, and often do, visualise, when we are imagining, remembering, recalling, expecting, hoping for or fearing something. In visualising we have
an experience which is like that which we have in actually seeing something, though it is an experience which we commonly, though not always,
produce in ourselves. To visualise something is to think of how it does or
would look, it is to picture it to ourselves or to produce a picture of it.
(White (1990): 122f.)
When I am imagining something I am thinking of it as something that could
be happening, whether or not it did, whereas when I am remembering it or
recalling it I am thinking of it as something that actually did happen. (Ibid.:
171)
To imagine something is to think of it as possibly being so. It is usually also
to think of it in a certain way. Thus, even when what we imagine is simply
an object, such a someones face or a logical difficulty, what we are imagining is what it is or would be like. Hence, we can be asked how we imagine
or think of it or what we imagine it to be. This is why what we imagine is
not usually just some object, whether a face or a difficulty, but something
118

Caseys description of the commitment involved in imagining as sheer supposal or


as positing something as purely possible (Casey (1976): 112f.) might also express
an endorsement of the modal version of the Attitude Account. But whether he is indeed a proponent of this view is difficult to say since he does not further expand on
his description; and nor on his motivation for putting it forward. The only thing that
Casey suggests remotely in support of his description is that all intentional states
have to involve an attitude of some sort (see Casey (1976): 112f.). However, although this further claim is plausible in its own rights (at least if restricted to nonneutral representations; see Section 2.1), it does not privilege a modal characterisation of the imaginative attitude over a non-modal one.

Other Theories of Imagining

165

about something. (Ibid.: 184).

First of all, some clarificatory comments on Whites sometimes rather unorthodox terminology are necessary. Imagining, for instance, covers only
intellectual imaginings and imaginative projects with intellectually imaginative components, but not any forms of the sensory imagination (see the
claims and examples in White (1990): 185f.). It is in line with this usage of
imagining that White strictly distinguishes sensory imagining from
visualising: while the former amounts to intellectual imagining (i.e., imaginative thought) about perceivable objects and features (see, e.g., White
(1990): 108), the latter constitutes a form of visual representation. It is true
that he describes visualising sometimes in terms of thinking. But his
other characterisations in terms of experiencing, picturing, and so on,
make clear that he has sensory rather than intellectual representation in
mind (see, e.g., the first passage quoted above). That is, he takes thought
to go beyond the intellectual and to be roughly equivalent with conscious
mental representation.
A more substantial element of Whites view is his identification of
visualising with the visual representational element shared by thinking,
remembering, imagining, and so on, rather than with visually imagining
something (see the first passage quoted above, as well as the examples discussed at White (1990): 122 and 168f.). This means that, strictly speaking,
he endorses the attitude version of the Modal Account solely with respect
to intellectual imagining, which is said to differ from judging and remembering in involving a claim about how some possible world (i.e., what
could be happening) rather than the actual world (i.e., what actually does
or actually did happen; see the second passage quoted above). By contrast, his claims about a comparable difference between seeing and visualising do not distinguish sensory cognitions from sensory imaginings, but
rather perceptions from all other sensory representations (whether they are
mnemonic, imaginative, or something else). However, Whites underlying
assumption seems to be that visually imagining something consists in
visualising it and intellectually thinking (or imagining) of it as possible,
whereas visually recalling something consists in visualising it and intellectually thinking of it as actual and past (see White (1990): ch. 18, especially pp. 168f.). Hence, his restricted Modal Account may still count as a

166

The Nature and Variety of Imagining

unified account at least with respect to sensory and intellectual imaginings since it distinguishes sensory imaginings from sensory cognitions in
terms of the difference between intellectual imaginings and intellectual
cognitions.
Independently of whether proponents of the attitude version of the Modal Account follow White in his views on sensory imagining, their theory
needs to be further qualified. First of all, as already noted by White in the
quoted passages, what we imagine may coincide with what is actually the
case. For instance, how I visualise the car of my friend as being may happen to be identical with how it actually looks like. But this is unproblematic since it does not (wrongly) entail that my visualisation of the car or
any other instance of imagining involves a cognitive commitment towards how the actual world is like.119 In addition, since the actual world is,
or corresponds to, one of the possible worlds, making a claim about how
things are in the actual world implies making a claim about how they are in
some possible world. Accordingly, the Modal Account might have to assume that the (stronger) cognitive attitude includes a (weaker) imaginative
attitude. Hence, what is distinctive of imaginings might not be the possession of an imaginative commitment, but rather the possession of nothing
but an imaginative commitment.120
Therefore, if Whites claim that imagining something (in my sense) consists in
[thinking] of what it does or would look like, or of what it is or would be like (see
the first and third of the passages quoted in the main text) should instead be read
as meaning that imagining may at least sometimes involve a commitment to how
things actually are (i.e., a cognitive attitude), it would clearly turn out to be false
(see Section 2.1) and also undermine his strict distinction between imaginings and
cognitions. But his additional statement that imagining something means thinking
of it as something that could be happening, whether or not it did (my emphasis; see
the second passage quoted in the main text) indicates that he does not take imagining
to involve a cognitive attitude and merely allows for the case that what we are imagining may happen to be actually the case.
120
This qualification of the attitude version of the Modal Account seems even more
plausible if it is accepted that possible worlds do not really exist, or at least not in
the same substantial sense as the actual world. According to this view, possible
worlds are entities of a completely different ontological category than the actual
world: while the latter enjoys substantial material existence, the former are most
likely sets of propositions or statements that describe things as being a certain way
(see Divers (2002): ch. 2). Hence, cognitions may have in common with imaginings
119

Other Theories of Imagining

167

But the Modal Account gets even more complicated. The difference
between judgementally and imaginatively entertaining a modal proposition
cannot simply be identified with a difference in attitude towards the same
non-modal proposition, since this would be unable to deal with iterated
cases. For instance, there would be no difference between imagining that it
rains, imagining that it is possible that it rains and imagining that it is possible that it is possible that it rains. It is therefore more promising to relativise possibility to particular worlds of evaluation (see, e.g., Priest
(2008)). Accordingly, there is a (formal) difference between something being possible in (or from the perspective of) one world and the same thing
being possible in (or from the perspective of) some other world, given that
which alternative possible worlds are relevant for the possible truths in one
world is assumed to be independent of (though not necessarily different
from) which alternative possible worlds are relevant for the possible truths
in the other world. Imagining that it rains can then be differentiated from
imagining that it is possible that it rains in the following way: while the
former endorses the proposition it rains as being actually true in some
possible world, the latter endorses the same proposition as being possibly
true in some (e.g., the same) possible world. Further iterations allow for a
similar treatment: imagining that it is possible that it is possible that it
rains, for instance, endorses the proposition it rains as being possibly possibly true in some possible world (i.e., as being possibly true in some world
that is possible from the perspective of the (distinct) imagined possible
world).
Finally, since the best version of the Modal Account (i.e., its attitude
version) identifies the attitudinal difference between imaginings and cognitions as their most basic difference, it in fact constitutes an instance of the
more encompassing Attitude Account. I evaluate the general prospects of
this broad approach to the unity of imagining and, hence, also of its explicitly modal version in the subsequent section. But, before that, I would
that they are concerned with a description of how things are, but differ from imaginings in that they take this description to be true of or, more neutrally, to apply to
the actual world. This does not turn the attitude version of the Modal Account into a
version of the Ontological Account, though, since imaginings and cognitions would
still represent and refer to objects of the same ontological category (i.e., trees or cars
that may exist in the material actual world).

168

The Nature and Variety of Imagining

like to note a problem that is specific to the Modal Account in so far as it


concerns the element of modality, rather than the attitudinal element.
The problem in question is that the Modal Account cannot easily accommodate cases in which we imagine impossible states of affairs, such as 5
plus 7 equalling 13, or water being identical with CO 2 (see Gendler (2006)
and Weinberg and Meskin (2006)), or somethings being a Devils tuning
fork or an impossible (or infinite) staircase. 121 Sensorily or intellectually
imagining an impossible state of affairs (e.g., the falsity of a true logical or
mathematical theorem or of a true identity claim, or the visual appearance
of an impossible figure) is to be distinguished from sensorily or intellectually imagining several possible states of affairs that cannot all obtain at the
same time (e.g., that something is square and, independently, that it is
round). The second case is unproblematic for the modal version of the Attitude Account since it leads just to a rational tension between different representations that make contradictory claims about one and the same possible world. By contrast, the first case is problematic since it concerns a
single representation (i.e., a single state of affairs) and, hence, requires the
introduction of a stance towards (something like) an impossible world
(e.g., a world in which the theorem that 5 plus 7 equals 13 is true, or water
is not H2O). In particular, it would not help to insist that imagining something impossible is always an instance of taking an impossible state of affairs to be part of a possible world. For although cases of the kind alluded
to are possible (e.g., if we are unaware of the impossibility of the state of
affairs concerned), the proposal would not capture cases in which imagining something impossible clearly involves being aware of it as impossible,
rather than possible.

5.5. The Attitude Account


The Attitude Account and, therefore, also the most promising interpretaIndeed, there are many metaphysical impossibilities that we can perceptually experience or depict in a visual mode (see Macpherson (2010) or the pictures of M. C. Escher) and should therefore also be able (at least in principle) to visualise them. It is
an interesting question whether visualising also allows for illusions like the Waterfall illusion (see Crane (1988)).

121

Other Theories of Imagining

169

tion of the Modal Account constitutes a specific version of the Phenomenal Account and therefore also faces the general objection raised above
against this approach to imagining (see Section 5.2). The worry was that
essential phenomenal differences such as the one in attitude are unlikely to be basic. Instead, it should be expected that they can be further
elucidated in terms of some underlying features in this case, perhaps, by
reference to the difference in epistemic function. Indeed, the phenomenologists, who have identified the strict difference in commitment between
imaginings and cognitions, need not necessarily be opposed to such an explanatory approach, despite their focus on phenomenological description.122
The main thesis of the Attitude Account is that it is distinctive of, and
fundamental to, imaginings that they possess an imaginative attitude (as
well as a cognitive one). For the same reasons as in the case of the Modal
Account (see the preceding section), the Attitude Account has to specify
the imaginativeness of representations in terms of the presence of an imaginative attitude, rather than merely in terms of the absence of a cognitive
attitude. For, otherwise, imaginings would not be distinguished from mental episodes such as conative feelings or episodes of wondering that are
neither cognitive, nor imaginative. Similarly, at least some versions of the
Attitude Account have to characterise imaginings also by reference to their
lack of a cognitive attitude namely if the possession of an imaginative attitude is weaker than, and included in, the possession of a cognitive attitude. This qualification applies, for instance, to the modal version of the
Attitude Account (see the preceding section).123
It is interesting to ask whether phenomenology understood as a set of methodologies is purely descriptive, or also explanatory. I tend to favour the second option,
which assumes that phenomenological observations should also aim at making sense
of phenomenal aspects in terms of non-phenomenal features (which still count as
phenomenologically salient in so far as they are reflected by phenomenal aspects;
see Dorsch (2012a) and Soldati & Dorsch (2011)). The debate in question is also
closely related to the issue to which extent phenomenologists like Husserl were realists about the external world (see Soldati (1994)).
123
Whether a mental episode may simultaneously possess a cognitive and an imaginative attitude depends on what the postulated imaginative attitude specifically amounts
to. The Modal Account is a case in point: since it characterises the imaginative attitude in terms of a commitment about how things are in some possible world, and
since the actual world is or corresponds to such a world, the view has to maintain
122

170

The Nature and Variety of Imagining

Different versions of the Attitude Account diverge from each other in


how they characterise the distinctively imaginative commitment (as well as
perhaps in how many imaginative attitudes they thereby distinguish). The
attitude version of the Modal Account formulated in the last section understands this commitment as a stance towards some possible world (or situation): imaginings are committed to things being a certain way (i.e., how
they represent them as being) in some possible world (but neutral with respect to the actual world). But, as noted at the end of the same section, the
Modal Account faces the specific problem of accommodating cases in
which we imagine impossibilities. Other characterisations of an imaginative attitude can be found, for instance, in the writings of the earlier advocates of the phenomenological tradition, notably Husserl and Sartre.
Husserl describes the (single) attitude distinctive of sensory imaginings
both in negative and in positive terms (see Husserl (1901): 443f.). His negative claim is that imaginings lack a cognitive attitude: they do not involve
belief and do not present things as realities, that is, as really being a certain way: in sensory imagination, the imagined objects are not themselves
before our mind. Husserls positive characterisation understands the imaginative attitude as a commitment to fiction, that is, to things as fictionally being a certain way, or as being something non-existent. 124 He further elucidates this imaginative attitude as being concerned with a commitment to something as being possible, while staying completely neutral on
what is the case in the actual world (see Husserl (1918/2006)). So his view
comes close to the attitude version of the Modal Account. Besides, in his
early writings, Husserl is adamant that having an imaginative attitude is
compatible with sensory immediacy (see Husserl (1901): 444; see also
Section 3.2).125 But later on, his switch from a pictorial model of imaginatthat cognitions show an imaginative attitude by showing a cognitive one (see the last
section). But other ways of drawing the contrast between the two kinds of attitude
rule out such a possibility. Due to the nature of the commitments involved, no state
of affairs can be phenomenally given both as actual and as non-actual, for instance.
In any case, all unified accounts should assume that imaginings lack a cognitive attitude.
124
Indeed, in the second edition of his Logical Investigations, Husserl substituted fiction (Fiktion) for imagination (Einbildung), and the fictitious (das Fingierte)
for the imagined (das Eingebildete) (see Husserl (1901)).
125
His rather controversial example is as follows. While walking around in a cab-

Other Theories of Imagining

171

ive awareness to an echo model (see Note 9 above) is accompanied and


also partly motivated by a recognition that sensory imaginings lack
immediacy, after all (see the texts in the second half of Husserl (2006), as
well as Marbach (2006)).
Sartre, on the other hand, claims that all sensory imaginings involve an
attitude, but that they may involve one of four different imaginative commitments (which all characterise, in one way or another, the lack of immediacy essential to sensory imagining):
inet of curiosities, we seem to see an unknown woman waving at us. However, as we
come closer, it turns out that it is just a mechanical puppet looking like a woman.
According to Husserl, our state of mind changes in so far as we stop having a (nonveridical) perceptual experience and start to have a complex visual experience which
involves two conflicting attitudes towards the same visual content (accompanied by
two conflicting interpretations of that content): that there is something real (i.e., a
puppet), or that there is something imagined or non-real (i.e., a woman). Husserls
thought is now that the visual content and the imaginative attitude (plus the respective interpretation) form an instance of visualising albeit one which gives us the impression that the imagined woman is bodily present and directly before our eyes,
given that it involves the same representational element as the underlying perception
of the puppet.
The main problem with this example is that it is not a clear case of imagining. An
alternative suggestion is to treat our visual awareness as of the presence of a woman
as pictorial (or, if that is different, as the same kind of awareness involved in the
visual recognition of what a sculpture represents). Although Husserl explicitly
denies the plausibility of this option, his only reason for doing so seems to be the
fact that the puppet was not meant to be a depiction or representation of a woman,
but rather merely a moving automat (see Husserl (1901): 443). However, even if
such an underlying intention is necessary for something to count as a picture (or
sculpture), it is not necessary for the experience of something to count as pictorial:
we can have pictorial experiences of marks on a wall, or see something in
formations of clouds (see Wollheim (1987): Part I). Besides, the claim that our
visual awareness of the woman should not count as pictorial does not imply that it
should count as an instance of visualising.
Another alternative is that it may be an instance of seeing something under or
noticing an aspect in the light of an imaginatively entertained concept; and a third
option, that it may be a visually based recognition of a similarity (i.e., we are
visually aware of the fact that the puppet looks, to some extent, like a woman). In
Dorsch (2012c), I argue that all apparent cases of seeing something under an
imaginatively employed concept should in fact be analysed as such experiences of
resemblance.

172

The Nature and Variety of Imagining


Every consciousness [i.e., intentional mental episode FD] posits its object,
but each in its own way. Perception, for example, posits its object as existing. The image [i.e., a sensory imagining FD] also includes an act of belief or positional act. The act can take four and only four forms: it can posit
the object as nonexistent, or as absent, or as existing elsewhere; it can also
neutralize itself, which is to say not posit its object as existent. (Sartre
(1940): 12)
This suspension of belief remains a positional act. (Sartre (1940): 12; n. 10)
Two of these acts are negations; the fourth corresponds to a suspension or
neutralisation of the thesis. The third, which is positive, assumes an implicit
negation of the natural and present existence of the object. These positional
acts this remark is crucial are not superimposed on the image after it is
constituted: the positional act is constitutive of the image consciousness.
(Sartre (1940): 12f.)126

The four identified imaginative attitudes (or imaginative positional acts)


have in common that they oppose the cognitive attitude involved in perception in such a way that the two kinds of attitude are incompatible with each
other. Taking the represented entities to be non-existing, or to be absent
from ones actual environment, or to be existing elsewhere in the actual
world (e.g., in a distant location), is clearly in tension with taking them to
be existing and present before ones eyes: a given sensory representation
can either do the one, or the other, but not both, without becoming incoherent in its commitments. Similarly, neutralising a cognitive attitude is incompatible with staying committal to how the actual world is like. So, the
four imaginative attitudes proposed by Sartre are unified in their opposition
to, and incompatibility with, the cognitive attitude of perceptions.
In his description of the attitudinal difference between imaginings and
cognitions, Sartre assumes that the cognitive stance towards the actual is
something like the default case for sensory representation. This is reflected,
for instance, in his characterisation of the fourth type of imaginative attitude as the suspension or cancellation of a commitment about how things
are in reality. A commitment can be nullified only if it has been somehow
present before. Furthermore, the idea that the default stance of sensory repFor this last point, see also Casey (1976): 112.

126

Other Theories of Imagining

173

resentation is cognitive in nature is a specific instance of the idea already


discussed as part of Caseys view that all (sensory) mental representations involve an attitude towards their content.
Furthermore, it is important to note that Sartre believes that the cancellation of the default cognitive commitment does not (merely) lead to the
absence of that commitment, but (also) involves the presence of an attitude
of neutralisation. It is not absolutely clear how he intends this thought to be
spelled out in detail. But he understands the difference between perception
and neutralising imagination to be of the same kind as the difference
between suspending belief in something and not believing anything in the
first place.127 Without this analogy with suspended belief, it would be difficult for Sartre to avoid the possibility that the nullification of a cognitive
attitude is also involved in other non-cognitive phenomena (such as
expectations or wonderings) and, hence, not distinctive of imaginings.
In any case, Sartres proposal is problematic for several reasons. Two of
them are specific to his view, while a third generalises to all versions of the
Attitude Account, including Husserls. Indeed, Husserls characterisation of
the (single) imaginative attitude seems to be identical with Sartres specification of the first imaginative commitment, according to which imaginings take their objects to be non-actual (and stay neutral about their actual
existence). But let me now turn to the three objections.
First, it postulates four distinct and independent kinds of sensory imagining which are not really related to each other, except by the idea that
they are all incompatible with a cognitive commitment to reality. Being
committed to the absence of an object from ones actual environment is
perhaps more basic than, and a common ingredient of, being committed to
the objects non-existence or existence elsewhere in actual space. But this
does not help to overcome the difficulty that these three imaginative attitudes oppose their cognitive counterpart in a different way than the fourth:
while the first three serve as a substantial replacement of a cognitive attitude, the fourth imaginative attitude consists just in the cancellation of a
127

Note that the quoted passage about the suspension of belief applies to sensory representation and, hence, is not literally concerned with intellectual belief and its suspension. Rather, what Sartre seems to have in mind here is the cancellation of the
cognitive attitude (or thesis) that is common to both conscious perceptions and occurrent beliefs (i.e., judgemental thoughts).

174

The Nature and Variety of Imagining

cognitive attitude. So, Sartres view does not provide a fully unified
elucidation of what renders an attitude imaginative, rather than cognitive;
and it is unclear how his characterisation could be completed. 128 Husserls
postulation of a single imaginative attitude obviously does not face a similar challenge.
Second, the unity of imagining is also under threat by the fact that it is
unclear how his account could be extended to non-sensory imaginings
(something which Sartre himself does not aspire to, though). Intellectual
imaginings seem to share the same imaginative attitude with each other.
But none of the four listed by Sartre is a natural candidate for it. Imagining
that the Earth is flat does not involve a commitment to the absence of that
state of affairs from our actual environment. In fact, the representation is
neutral on what is really the case. This is reflected by the fact that imagining that the Earth is flat is in rational tension neither with believing that the
Earth is flat, nor with believing that it is not flat. Accordingly, the imaginative representation under consideration also does not make the committal
claim that the represented state of affairs does not exist or obtain. Similarly, imagining that the Earth is flat does not purport to be about a state of
affairs that exists elsewhere in the actual world (e.g., on the other side of
the hemisphere, or beyond the Moon). Last, switching attention to the neutralising attitude does not help since suspension of belief involves the neutralisation of judgemental endorsement and, hence, would be wrongly classified as an instance of imagining. Again, Husserls proposal is immune to
this problem. But with respect to the following two objections, Sartres and
Husserls views are on equally bad footing.
Third, episodic memories and other non-imaginative representations
may involve exactly the same commitments. When I remember the appearance of a friend, who now lives on the other side of the planet, my memory
is likely to present him or her as absent or existing elsewhere. Similarly,
memories of people who have died may very well involve an impression of
their non-existence. This is also reflected in the fact that such episodic
It does not matter here whether Sartre also thinks that each sensory imagining constitutively involves a thought (see Section 3.6), which may furthermore be responsible for its exemplification of one of the four types of imaginative representation.
For such additional thoughts could at best be taken to constitute the various commitments, but not to unify them in one general kind.

128

Other Theories of Imagining

175

memories just like sensory imaginings do not move us to the same kind
of actions towards the represented people, which perceptions of them
would incline us to. The natural reply is, of course, to say that memories
but not imaginings are still non-neutral about the existence of the represented entities in the actual past. So, the first of the four attitudes seems to
fare best because of its ability to accommodate episodic memories. But
there are still other mental representations such as our visual experiences
of the pictorial content of pictures, hallucinations that are recognised by us
as such, or spontaneously occurring images and thoughts that involve
some or all of the proposed imaginative attitudes, despite failing to be instances of imagining (see, for instance, Dorsch (2010a) and (2013), as well
as Section 14.4). Accordingly, Husserls and Sartres views do not succeed
in singling out what is distinctive about imaginings.129
The last problem (and possibly also the second) persists even if only one
of the four attitudes identified by Sartre is chosen to be definitional of imaginings. Not all of them manage to distinguish imaginings from memories;
and none of them succeeds in distinguishing imaginings from the representational element involved in pictorial experience that is concerned with
the depicted. This points to a general problem for the Attitude Account,
which arises independently of how the attitudinal difference between imaginings and cognitions is characterised: namely the problem to accommodate non-imaginative mental representations that are, none the less, committal in a very similar (non-cognitive) way as imaginings. Hence, at least in
its current state, the Attitude Account is untenable as a unified theory of
imagining.
Even the modal version of the Attitude Account has difficulties to distinguish sensory imaginings from pictorial experiences or spontaneous images, without wrongly identifying them as being imaginative themselves.
129

To be fair, Husserl acknowledges as much in his statement that sensory imagination


and pictorial experience are both instances of fantasy in virtue of their involvement
of what I have called an imaginative attitude and differ in other phenomenologically
salient aspects, most notably whether they involve also a perceptual awareness of a
depicting material surface. The resulting account, however, comes very close to the
problematic view that our awareness of what is depicted is imaginative in nature
(see Hopkins ((1998): ch. 1), Dorsch (2005) and Dorsch (2012c) for reasons why
this view is untenable).

176

The Nature and Variety of Imagining

For instance, when we look at a painting of Pegasus, our resulting visual


awareness of a winged horse is both neutral about its reality and non-neutral about its possibility. Hence, our awareness of something depicted
should count as imaginative according to the Modal Account. But, as has
been argued elsewhere (see Hopkins (1998) and Dorsch (2012c)), this implication should be rejected. Something similar applies to spontaneous images and their incorrect identification as imaginings (see Section 14.4).
Note that, in the case of pictorial experience, it does not matter that the
awareness of the winged horse is inseparably linked to a perceptual awareness of the depicting surface. The Modal Account does not distinguish
between representations that may occur on their own and representations
that are constitutively dependent on perceptions or other kinds of mental
state. Moreover, it should not draw this distinction for instance, by restricting its theory to self-contained representations in order to rule out
pictorial experience. For there are both dependent and independent instances of imagining: visualising the instantiation of a surface colour requires also visualising the instantiation of some shape property, whereas
imagining the instantiation of a shape property in a tactile manner does not
presuppose any further imaginative activity.130

5.6. The Spontaneity Account


A final important idea, prominent in the writings of the phenomenologists,
is that imaginings are special because of their apparent origin in our minds
and, more specifically, in its spontaneity. The Spontaneity Account adopts this idea and takes it to be most central to its account of the nature and
unity of imagining. Accordingly, imaginings are said to be fundamentally
characterised by their spontaneous origin.131 In particular, the view claims
I describe and argue for the dependent nature both of our experiences of surface colours and shapes, and of our experiences what paintings depict, in more detail in
Dorsch (2012c).
131
Both the Epistemological Account and the Agency Account stress the importance of
the fact that imaginings originate in the mind, rather than in the world. But the
former allows for two independent ways in which they may arise from the mind and
does not attempt to unify them in a single kind of origin; while the latter takes the
origin in mental agency to be essential to all instances of imagining. At the end of
130

Other Theories of Imagining

177

that, while perceptions originate in the perceived external objects, imaginings originate in the subjects spontaneity. The Spontaneity Account
maintains that this fact constitutes a basic and essential difference between
imaginative and cognitive episodes.
The main challenge that this view faces is, however, to come up with a
satisfactory characterisation of this special way of originating in the subjects mind. Sartre starts off with contrasting the spontaneity of sensory
imaginings with the passivity of perceptual experiences (note that Sartre
uses consciousness in roughly the same sense in which I use episode):
A perceptual consciousness appears to itself as passive. On the other hand,
an imaging consciousness gives itself to itself as an imaging consciousness,
which is to say as a spontaneity that produces and conserves the object as
imagined. It is a kind of indefinable counterpart to the fact that the object
gives itself as a nothingness. The consciousness appears to itself as creative,
but without positing as object this creative character. (Sartre (1940): 14)

In this passage, Sartre observes a difference in how perceptual experiences


and sensory imaginings are given to us in phenomenal consciousness: that
is, a difference in their phenomenal character. For Sartre, the phenomenologically salient difference at issue concerns the determination of how perceptual and imaginative episodes represent their objects. In the case of perceptual experiences, this determination seems to be a matter of passivity:
that we start to perceptually experience a green tree, say, and also that we
continue to do so appears to be due to the causal impact of some factor external to the mind, most notably the greenness of the perceived tree. By
contrast, that we sensorily imagine something as being a certain way (i.e.,
the object as imagined) seems to be spontaneously determined: our
episode of visualising a green tree, say, appears to be brought about and
sustained (i.e., produce[d] and conserve[d]) by our spontaneous mind,
rather than by the some (causal) factors external to it.132
this section, I will suggest that both theories can be understood as improvements on
the Spontaneity Account.
132
Sartre also employs the terms activity or creativity to describe the apparently subjective origin of our sensory imaginings. I use quotation marks to make clear that
my own terminology is different: following Casey (1976), I reserve spontaneity for
the involuntary and, respectively, non-perceptual or non-rational occurrence of con-

178

The Nature and Variety of Imagining

However, this noted phenomenal difference does not by itself suffice to


establish the desired difference in origin. That sensory imaginings present
themselves in phenomenal consciousness as arising from the mind, rather
than the world, is not enough to ensure that they in fact originate in the
mind. And something similar applies to the apparent, but not necessarily
true provenance of perceptual experiences from reality. To get the desired
contrast, we need to secure a difference in actual origin, and not merely
one in seeming origin. Moreover, it does not help much simply to assume
that how mental episodes phenomenally appear to come into being always
reflects how they really do so. Hallucinatory perceptual experiences do not
originate in the world, despite phenomenally appearing to do so.133
The best solution to this problems is perhaps to assume that Sartres
phenomenal distinction is meant to track, and render phenomenologically
scious images and thoughts. I further discuss and compare Sartres and Caseys notions of spontaneity in Section 5.6, where I also take a critical look at the prospects
of formulating a unified theory of imagining in terms of what Sartre calls spontaneity. Spontaneous representations (in Caseys and my sense) and their connection to imagining are then the subject of Section 14.4. The distinction between spontaneous and voluntary imaginings that is, between involuntary and voluntary
spontaneity (in Sartres sense) is also closely related to OShaughnessys distinction between will-impervious and will-susceptible imaginings (see Section 7.1).
133
See Dorsch (2012a) and Soldati & Dorsch (2011), as well as Section 1.2. The
presented considerations are compatible with Sartres independent claim that we
have certain and infallible first-personal access to the phenomenal character of our
mental episodes (see Sartre (1940): 4f.). For what is at issue here in the main text is
not whether we can err introspectively with respect to the phenomenal aspects of
mental episodes, but instead whether the phenomenal character of those episodes
can mislead us about their non-phenomenal features (such as their causal origin).
In the two works just cited, I argue that both kinds of error can occur as
evidenced, say, by the case of perfect hallucinations which present themselves as
veridical and relational experiences, but are in fact neither; and the case of our
introspective judgements about such hallucinations which may fail to recognise their
hallucinatory status and, instead, take them to be veridical perceptions.
Besides, in the two texts I also discuss the main differences phenomenal and
perceptual awareness. In particular, our phenomenal (and self-reflexive) awareness
of the non-phenomenal features of episodes is not object-directed, or at least not in
the same sense as (perceptual, imaginative, etc.) awareness of external objects, like
tables or landscapes: the creative character is not [posited] as object (see the
passage by Sartre quoted above in the main text).

Other Theories of Imagining

179

salient, a more substantial difference between internal and external origin: while perceptual experiences are due to factors external to the subject
or his or her conscious mind (whether these factors are perceived external
objects, processes in the brain or something else), sensory imaginings arise
from within the subject or mind (such as mental agency, rationality, association, and so on).134 This assumption would further stress the presumably
intended resemblance between Sartres distinction between passivity
and spontaneity and the Kantian distinction between receptivity and
spontaneity: that is, the distinction between the causal impressions that
we suffer and our own subjective contributions in the forms of mental acts,
responses, and so on.135 But the assumption of an underlying difference in
real origin fits also well with Sartres identification of two kinds of spontaneity (with image, he refers here to an instance of sensory imagining):
We may be reproached for loading the dice by choosing a voluntarily produced representation as an example of a mental image. In most cases, no
doubt, the image springs from a deep spontaneity that cannot be assimilated
to the will. It seems that the involuntary image appears to consciousness as
my friend Pierre might appear to me from around the corner in a street.
(Sartre (1940): 18)
[I]nvoluntary and voluntary images represent two very closely related types
of consciousness, of which one is produced by a voluntary spontaneity and
the other by a spontaneity without will. (Sartre (1940): 19)

It is not clear how best to describe this contrast. Among the options are mental/nonmental, subjective/objective, or personal/impersonal. There is also the problem
of where to place mental dispositions, unconscious mental states, and so on. See also
the objection below (and in Section 14.4) to the effect that unbidden (but non-perceptual or non-judgemental) images and thoughts resemble perceptions more than
mental actions in their origin.
135
I return to this issue in Section 5.6, in the context of discussion the Spontaneity Account of imagining. The distinction presented in the main text is not really Kants
own distinction (see his (1993)), since he does not take receptivity to be causal
(i.e., causality is a relation between empirical entities, and not between the subject
and the transcendent entities impinging on it), and since he assumes the subjective
contribution to be transcendental and pertaining exclusively to reason and understanding.
134

180

The Nature and Variety of Imagining

Accordingly, sensory imaginings may either occur voluntarily, as the result


of our voluntary mental agency and the underlying motivational states; or
involuntarily, in virtue of some other subject-involving process or event in
the mind. In both cases, they appear to arise from within the mind, either
due to mental agency or, alternatively, to some involuntary mental act or
subjective response.136 Perceptual experiences, sensations and similar phenomena, on the other hand, are simply caused and passively determined by
the perceived external objects and similar factors and do not originate in
our spontaneous minds (see also Sartre (1940): 14f.; 33). This is why they
phenomenally present themselves as impressions on the mind caused by
such factors external to the mind.
However, while we have a fairly good understanding of mental agency
(see, e.g., OBrien & Soteriou (2009), as well as Dorsch (2005) on mental
projects) and its imaginative employment (see Part Four), it is still necessary to provide a more detailed specification of what it means for a mental
representation to occur spontaneously in an involuntary manner and, in
particular, to be the result of a subjective act or response that is not present
in perception.137 Casey another proponent of the Spontaneity Account
has indeed tried to characterise this phenomenologically salient involuntary
spontaneity in more substantial terms. He accepts Sartres thesis that all aspects of our imaginative episodes are either controlled by the will or due to
More may have to be said to get clear about the precise nature of the difference in
mental origin that Sartre and others assume to be phenomenologically salient. For
instance, perceptual experiences, too, might be caused by other mental states. But
the causation concerned is presumably rather like the causation by external objects,
and not like a subjective response. Mental agency, association or the manifestation
of mental dispositions, on the other hand, certainly involve causation as well. But
they presumably also involve the subjects mind in a non-causal (or not merely causal) manner for instance, by involving rationality or similar forms of causation in
virtue of content.
137
One particular problem, that I ignore here, is that perception may already involve
elements of conceptualisation or interpretation, not the least in cases of aspect perception or seeing something under a concept (see, for instance, Scruton (1974): ch.
8, and Budd (1989): ch. 4). Indeed, already Kant believed that perception involves
more than receptivity, and his contemporary followers who uphold his distinction between receptivity and spontaneity as central to our understanding of the
mind typically agree with him on this point (see McDowell (1994)).
136

Other Theories of Imagining

181

the involuntary spontaneity of our minds. But, in contrast to Sartre, Casey


assumes that these two kinds of mental origin do not have much of significance in common. As a consequence, he reserves the term spontaneous
for the involuntary form of origin and discusses voluntary agency and involuntary spontaneity separately from each other.138
More specifically, Casey has identified three basic characteristics of
(involuntary) spontaneous representations (see Casey (1976): 68; see also
34f. and 67ff.). The first is that they arise in an unsolicited manner, that is,
appear unbidden or on their own accord: they do not involve any agency or
effort on behalf of the subject and are therefore not intended or otherwise
practically motivated (and not to be evaluated in terms of such motivation).
Their second property is that partly as a result of their unbidden nature
they can, and often do, occur unexpected and, hence, may surprise us. A
third attribute of spontaneous representations is that they typically arise instantaneously, or at once, and do not involve the kind of temporal dynamics or development that deliberate imaginings may include (e.g., when we
imagine something very complex by imagining one aspect after another).
138

See Casey (1976): 33-35, and ch. 3. Caseys choice of terminology seems to be
partly motivated by the thought that voluntary and involuntary formations of mental
representations do not have anything significant in common. OShaughnessy, too,
takes these two possible kinds of origin to differ essentially from each other. But he
is fully aware of the consequences of this essential difference for a unified account
of imagining that allows for involuntary imaginings: namely that voluntary and involuntary imaginings do not belong to the same mental kind (see the last part of Section 7.1).
In what follows, I generally accept Caseys characterisation of spontaneity and,
especially, the contrast that he draws between involuntary spontaneity and voluntary
mental agency (see Section 13.4), as well as the observation that perceptions are as
spontaneous as unbidden images or thoughts (see the discussion below in this
section). This continuity in understanding is the main reason why I have decided to
signal the distinct and independent notion of spontaneity to be found in Sartres
writings by using single quotation marks.
Casey also suggests there that all imaginings are in principle controllable, that is,
subject to the will. But it is not clear to which extent spontaneous images or thoughts
which he takes to be imaginative are indeed subject to the will (see Section
14.4). Besides, describing imaginings as being subject to the will is neither very
clear, nor very helpful for the formulation of a theory of imagining (see Section
13.5).

182

The Nature and Variety of Imagining

To these three features identified by Casey, a fourth may be added: spontaneous representations are usually fleeting in character and disappear as
quickly as they have occurred (see Section 14.4). This is closely related to
the fact that they serve no specific function in our mental lives (see Dorsch
(2010a)).
However, perceptual experiences, too, may show these marks of spontaneity. They are not actively formed and therefore arise without effort and,
typically, also instantaneously and unexpectedly.139 Furthermore, at least
some of them are rather transient because they are concerned with rapidly
changing or moving objects (e.g., cars passing by) or involve rapid changes
in perspective (e.g., because the subject is falling, running, or sitting in a
train). In the light of these observations, it is not surprising to see that Casey tries to say a bit more about the nature of (involuntary) spontaneity and
how it differs both from perception and from imaginative agency:
[A] truly spontaneous phenomenon initiates itself rather than being initiated
by other phenomena: it is autogenous. [...] [A] spontaneous imaginative experience does not present itself as embedded in any [well-ordered, causally
concatenated series of entities or events]. Instead of emerging from a nexus
of causally efficacious factors, it arises suddenly, in a psychical vacuum. To
be sure, upon reflection the imaginer may succeed in locating the spontaneous imaginative appearance within a certain causal context, but this context does not as such enter into the specific content of his experience. What
is experienced is rather an imaginative act-cum-presentation, which presents
itself as unconnected with preceding or surrounding circumstances, including even those circumstances provided by former acts of imagining. This
act-presentation appears of and by itself, as quite independent of the imaginers immediate practical situation (a situation that may even include particular plans and projects for imagining). It is as if this situation, though undeniably present, were irrelevant to the spontaneity of imagining. (Casey
(1976): 68)

This is not to say that perception is completely passive. Perceptual experiences may
involve the activity of attention and may also form the core of certain kinds of mental action, such as watching something or looking for something to happen (see the
work of Crowther, such as his (2010)). But even in the latter cases, the perceptual
experiences concerned occur unbidden, and without any voluntary determination of
what is represented.

139

Other Theories of Imagining

183

The distinction that Casey draws in this passage concerns, first of all, the
phenomenal character of the kinds of mental episode concerned. His first
observation about the character of spontaneous episodes is negative: they
are not subjectively given as being causally (or otherwise) related to other
mental phenomena or entities external to the mind (see the second sentence
in the quoted passage). In addition, Casey puts forward a second, more
positive description: spontaneous representations are phenomenally presented as being causally (and otherwise) unconnected from everything else
(see the fifth and the sixth sentences). Not appearing to stand in any causal
connection is clearly different from appearing to stand in no causal connection: while the former corresponds to the absence of a certain phenomenal
aspect (i.e., the aspect present in perception, judgement, and so on), the latter pertains to the presence of a different phenomenal aspect (i.e., one that
would be distinctive of imaginings). Accordingly, Casey takes the phenomenal character of spontaneous representations to include an aspect which
positively marks them as lacking any causal (or other) links to other elements of the mind and the world.
But, in contrast to Sartre, Casey also makes some assumptions about the
underlying actual origin of imaginings. First of all, he accepts the natural
thought that spontaneous images and thoughts may have causes that are external to, and independent from, their spontaneous effects. These causes are
just not phenomenologically salient. More important, Casey notes that
spontaneous representations occur of and by [themselves] and [initiate
themselves]. This cannot mean that spontaneous images and thoughts are
their own causes, given that this would not only involve a potentially incoherent and very strong notion of self-causation, but also because it would
contradict the accepted possibility of underlying external causation. Rather,
what Casey seems to have in mind is that spontaneous representations do
not occur in virtue of the impact of other phenomena that are accessible
from our subjective, personal perspective. Both mental agency and perceived external objects are such phenomena: we are consciously aware of
them, either through phenomenal awareness and introspection, or through
perceptual awareness (see Dorsch (2012a) for more on these three distinct
kinds of awareness).
Accordingly, what is said to be distinctive of spontaneous representa-

184

The Nature and Variety of Imagining

tions is their phenomenologically salient absence of an origin in some conscious mental episode or some consciously represented external object. Together with the specification of voluntary representations in terms of mental agency, the presented characterisation of the involuntary form of spontaneity (in Sartres sense) enables us to provide a more precise formulation
of the main claim of the Spontaneity Account: imaginings are differentiated from all other mental phenomena in their origin either in mental
agency or in nothing that we are consciously aware of.
The resulting view promises to be able to distinguish spontaneous representations from most other mental phenomena. For instance, while perceptions are determined by salient aspects of the external world, judgemental thoughts and conscious decisions are rational responses to consciously represented facts or considerations, and voluntary imaginings and
other mental actions originate in some conscious tryings or intentions-inactions (see Dorsch (2013)). Moreover, the Spontaneity Account is compatible with the basic contrast that Sartre appears to track with his phenomenological distinction discussed above between spontaneity and perceptual passivity: namely the contrast between what merely happens to us
and what we are ourselves doing, or between what we suffer as mere objects (e.g., in the shape of impressions or reflexes) and what counts as our
own response as subjects (e.g., in the shape of rational responses, associations, manifestations of mental dispositions, and the like). 140 The idea is
that spontaneous representations do not come into existence due to the impact of external objects, but originate in some associative, dispositional or
similar aspects of the mind.
In addition, that we experience sensory imaginings as creative (and as
arising from within the mind), while we experience perceptions as passive
(and as arising from reality), may be used to establish an intimate connection between the attitudinal and the origination difference. The thought is
that perceptual experiences involve a commitment to reality because they
The Kantian tradition understands this distinction largely in rational terms: namely
by differentiating the rational in the minimal sense of a responsiveness to epistemic, practical or other kinds of reason from the non-rational or merely causal
(see McDowell (1994) for a good example). A comparable though far less strict
and divisive Humean contrast is perhaps that between impressions and ideas (see
Chapter 9).

140

Other Theories of Imagining

185

are brought about by (or nomologically linked to) parts of the world; while
sensory imaginings lack such a commitment because they are brought
about by (or otherwise closely linked to) elements in the mind. Hence, the
phenomenal difference in commitment may be constituted partly by the underlying, phenomenologically salient difference in origin. The Spontaneity
Account therefore promises to explain at least some of the principal features of imagining.
But one major objection to the Spontaneity Account is that it cannot distinguish (sensory) imaginings from certain non-imaginative phenomena,
notably hallucinations and episodic memories. In particular, neither hallucinations, nor involuntary recollections occur due to the influence of external objects or conscious manifestations of theoretical or practical rationality (i.e., in response to noticed evidence, motivational states, and so
on).141 Indeed, the subpersonal processes, which are responsible for the occurrence of hallucinations and unbidden memories, may not be very dissimilar from those involved in (non-perceptual and non-mnemonic) spontaneous representations.
The other main objection to the Spontaneity Account is that it wrongly
assumes two basic forms of imagining namely voluntary and involuntary
ones rather than one. This threatens the unity of imagining in so far as
spontaneity (in Caseys sense) and mental agency do not have anything
significant in common, which they also do not share with, say, the rational
origin of judgemental thoughts in perceptions or conscious inferences.
Judgements originate as much in the mind rather than in the world as
voluntary imaginings or spontaneous representations. That is, all three
kinds of mental episode are not impressions on the mind, but belong to our
subjective doings or responses. Hence, Sartres notion of a spontaneous
origin is to wide to single out only imaginings; and it is unclear how it
could be narrowed down to exclude rationally formed judgemental
thoughts without becoming simply a disjunction of two kinds of origin
(i.e., mental agency and Caseys spontaneity).
141

See Dorsch (2010a). Voluntary memories are no problem since they do not allow for
the same kind of direct agency that is characteristic of voluntary imaginings (see
Section 13.2 and Dorsch (2009b)). In Section 14.2, I distinguish the different mental
origins of (voluntary) imaginings and voluntary memories with respect to whether
they involve a specifically imaginative kind of mental agency.

186

The Nature and Variety of Imagining

This second objection is closely related to the phenomenological observation that the way, in which spontaneous representations occur, seems to
be subjectively much closer in nature to that of perceptions and sensations
than to that of judgemental thoughts or mental actions. Talk of spontaneity should not cloud the fact that unbidden images or thoughts are not
subjectively experienced as coming from ourselves, or as subjective doings or responses. Indeed, as Casey has already noted in the quoted passage, the only thing that we can come to know from our first-personal perspective is where spontaneous representations do not come from: namely
neither from the world, nor from our will or other conscious mental phenomena. But we do not become subjectively aware of where they do come
from: they are simply given as unrelated to all entities that we are consciously aware of. Not surprisingly, we take them to come from nowhere,
to occur out of the blue or, indeed, to appear by and of [themselves].
Phenomenological evidence therefore indicates that spontaneous representations are not among our (involuntary) doings or responses, but enjoy
rather a similarly passive origin as perceptions and sensations. 142 But it also
suggests that spontaneous representations in contrast to perceptions and
sensations do not originate in the world (including our body). Hence,
they seem to differ significantly, not only from voluntary imaginings and
judgemental thoughts, but also from perceptions and sensations. It is therefore perhaps best to treat them neither as being imaginative, nor as being
cognitive, but instead as belonging to an entirely different, third category
of representation (see Section 14.4).
It might be rightly insisted that the four characteristics of spontaneity
listed above constitute aspects of the phenomenal character of the respective spontaneous representations. But, as already mentioned, these aspects
are not sufficient to distinguish spontaneous representations from (certain)
perceptions. Hence, they also do not suffice to phenomenally mark the difference in origin between the two kinds of mental representation. PercepInterestingly, Sartre himself seems to make a similar observation when he states in
the passage quoted above that the involuntary image appears to consciousness as
my friend Pierre might appear to me from around the corner in a street (Sartre
(1940): 18). But, as also illustrated above, he still thinks that involuntary images differ significantly from perceptions due to their spontaneous origin (see Sartre
(1940): 14 and 33).

142

Other Theories of Imagining

187

tual experiences do possess a further phenomenal aspect, that identifies


them as relations to, and determined by, some external objects (see Martin
(2002a) and Dorsch (2010a)). Hence, their distinctive origin in the world is
reflected by their phenomenal character. Similarly, voluntary imaginings
and judgemental thoughts are phenomenally marked as being motivated
and determined by, respectively, practical or theoretical reasons, thus making us first-personally aware of their distinctive origin in imaginative mental agency or rational endorsement.143 By contrast, spontaneous representations lack such an additional phenomenal aspect that informs us introspectively about how they have come into existence. Thus, our conclusion that
they must somehow originate in the mind turns out to be the result of an informed guess or inference to the best explanation, rather than a matter of
phenomenal awareness or introspection.
Now, that their spontaneity is not distinctive of imaginings does not
mean, however, that we should give up on the idea that their specific origin
in the mind, rather than in the world, is of significance for a unified theory
of imagining. It is perhaps already enough to change the way, in which the
mental origin unique to imaginings is specified. Two of the three main theories of imagining may, indeed, be understood as modifying and improving on the Spontaneity Account along this line.
The Epistemological Account claims that the difference in how imaginings and (some) cognitions arise from the mind is linked to how they are
constrained by reality; and that this difference in their relation to reality is
most fundamental and ensures the differences in attitude and epistemic
function. The Epistemological Account avoids the problems of the Spontaneity Account by focusing primarily on the idea that imaginings originate
in a way incompatible with cognition, rather than on the idea that they
either spontaneously emerge or are actively formed. As OShaughnessy,
the main proponent of the Epistemological Account, argues, it is still very
143

See Dorsch (2009b) and Dorsch (2013). Note that the kind of first-personal awareness concerned is sometimes erroneous. Perfect hallucinations are not relational perceptions; but we still experience them as such (see Dorsch (2012a) and Soldati &
Dorsch (2011)). And judgemental thoughts are given to us in consciousness as responses to epistemic reasons, even if they have been irrationally caused by some
other mental phenomena (e.g., an emotion or preference), or by some drug or pathology (see Dorsch (2009b)).

188

The Nature and Variety of Imagining

important for their imaginativeness that imaginings originate in the mind in


one of the two ways distinguished by Casey and Sartre but only because,
and to the extent to which, their mental origin has the consequence that
they fail to lead to the acquisition of knowledge.
Another option is to accept the conclusion that spontaneous representations are not imaginative, and to locate the distinctive feature of imaginings
in their active character, that is, in their origin in mental agency. To establish an essential difference with respect to cognitions, the involvement of
agency should be taken to be constitutive of imaginative phenomena, and
to be direct with respect to the determination of what specifically is represented. This difference in direct agency is, furthermore, phenomenologically
salient because imaginative mental agency like any other form of agency
involves an awareness of agency (see Dorsch (2009b)). Before defending
the Agency Account in Part Four, I first turn, in the following two parts of
the book, to the discussion of the Epistemological and the Dependency Account of imagining.

Part Two
The Epistemological Account

Introduction to Part Two

The three chapters of this part of the book are devoted to the presentation
and assessment of the Epistemological Account of imagining and, primarily, of its exemplary version, the unified account of imagining developed
by Brian OShaughnessy. In the first chapter, I start with providing a succinct summary of OShaughnessys position, before expanding in more detail on his main claims and overall argumentative strategy. The middle
chapter is then dedicated to a thorough study of OShaughnessys complex
considerations and arguments in favour of endorsing his theory of imagining. The evaluation of the Epistemological Account, and of OShaughnessys account in particular, will have to wait until the last of the three
chapters.
The insertion of a short synopsis in addition to an extended discussion is
motivated by the wish to enable the reader to follow the main argument
against the Epistemological Account without having to acquaint him- or
herself with many of the details of OShaughnessys particular version of
it, or with the textual exegesis that the presentation of these details require.
Reading the middle chapter and much of the first chapter is therefore not
required to engage with the objections to the Epistemological Account
presented in the last chapter of this part of the book.
The main idea of the Epistemological Account is to characterise imaginings as non-cognitive in one form or another. This does not mean that all
theories of imagining which assume episodes of imagining to be different
from cognitive phenomena are versions of the Epistemological Account.
For the Epistemological Account goes beyond the thesis that imaginative

192

The Epistemological Account

episodes are not cognitive episodes, and imaginative projects not cognitive
projects. What it also claims is that it is distinctive of imaginings that they
are non-cognitive, and that their non-cognitivity is among the basic features in virtue of which they count as imaginative. Hence, any version of
the Epistemological Account of imaginings makes essential reference to
their non-cognitivity: it assigns their non-cognitivity a central place in the
specification of their nature (and not merely, say, in the specification of the
implications of their nature).
That the Epistemological Account takes the non-cognitivity of imaginings not only to be essential to them, but also to be among their fundamental essential features which cannot be explained by pointing to even
more basic characteristics, distinguishes it from other unified theories that
also accept the essential non-cognitivity of imaginings. One example is the
Agency Account of imagining, or at least some versions of it. It differs
from the Epistemological Account partly because it does not take the noncognitivity of imaginings to be basic, but characterises and explains it as a
consequence of their status as instances of mental agency (see Sections 4.7
and 14.1). Similarly, proponents of the Dependency Account may very well
accept the necessary non-cognitivity of imaginative phenomena, while deriving it from their more fundamental feature of being representations or
echoes of cognitive phenomena (see Part Three). None the less, the Epistemological Account may be combined with another theory about the fundamental essence of imagining, as long as the two accounts are concerned
with different and independent aspects of their basic nature. In other words
is, the negative and epistemic conception of imaginings as essentially noncognitive may be supplemented with a more positive and non-epistemic
characterisation of the other part of their nature.
Indeed, such a supplementation may be deemed necessary because of
the fact that the Epistemological Account characterises imaginings primarily in terms of what they are not, rather than in terms of what they are. In
Section 8.1, I return to the issue of whether some versions of the Epistemological Account have the resources to provide an exhaustive unified account of imagining, or whether they need to be supplemented by some other fundamental theory about the nature of imaginings.
The worry about the comprehensiveness of the Epistemological Account

Introduction to Part Two

193

may also help to explain why there is no actual defender of its exhaustiveness. In the preceding chapters, I highlighted the fact that, while Sartre and
Wittgenstein accept the essentiality of the non-cognitivity of imaginings,
they deny the fundamentality of their non-cognitivity and try instead to explain it in terms of more basic features such as a specific origin in the
mind or the will (see especially Sections 4.7 and 5.6). OShaughnessy, the
chief proponent of the Epistemological Account, on the other hand, combines the thesis about the fundamental non-cognitivity of imaginings with a
claim about their equally fundamental representational connection to cognitions. That is, he also endorses the following echo claim about imaginings (see Part Three for more on such claims):
(ER)

Imaginings are essentially representations of their cognitive


counterparts.

OShaughnessy supplements the Epistemological Account with a representational version of the Dependency Account partly because of the difficulty of the Epistemological Account to characterise the nature of imaginings in more positive terms (see Section 8.1). 1 Therefore, in what follows, I
treat the Epistemological Account primarily as one key element in a more
complex unified theory of imagining, rather than as constituting a self-sufficient unified account in its own right. Besides, my discussion is concerned as much with the fundamentality of the claim that imaginings are
essentially non-cognitive as with its truth.
As already suggested, the Epistemological Account can come in various
forms, depending on what is understood by the claim that imaginings are
non-cognitive. But the central claims of these different versions of the Epistemological Account share a structure, because of which they may be la1

OShaughnessys writings do not completely rule out the possibility that the noncognitive aspect of the nature of imaginings may be traced back to, and explained by
reference to, the representational one. It is therefore perhaps also possible to read
him as defending a pure version of the Dependency Account, which does not require
the simultaneous endorsement of an Epistemological Account. But it would then be
difficult to make sense of his argumentative strategy of supporting the negation
claims endorsed by him without any reference to the echo claim equally accepted by
him (see Section 6.3).

194

The Epistemological Account

belled negation claims. What is distinctive of negation claims is that they


characterise imaginings in purely negative terms: they say merely what
imaginings are not, or cannot be. All theses about imaginings having this
structural feature are negation claims. Here, I focus exclusively on negation claims that describe imaginings as being non-cognitive in one form or
another. Only such negation claims are relevant for the Epistemological
Account.2
Different versions of the Epistemological Account endorse different
negation claims as fundamental to the characterisation of the nature of imaginings. Moreover, these claims may differ in strength notably in that
some of them may imply others. We already came across a specific negation claim, namely the one endorsed by Sartre and Wittgenstein which focused on the relation between imagining and informativity (see Chapter
4):3
(NI) Imaginings are essentially uninformative with respect to the external world.
Two different negation claims are central to OShaughnessys account of
imagining (see Section 6.2). They are concerned with the cognitive role of
imaginings and differ primarily in their strength:4
2

There might be other interesting ways of characterising imaginings in negative


terms. But I take it that the corresponding non-cognitive negation claims are likely
to face the same difficulties as the cognitive negation claims introduced here and
critically discussed in Chapter 8.
See Sartre (1940): 8ff., and Wittgenstein (1984d): sec. 627. To simplify the discussion, I ignore from now on that Sartre and Wittgenstein conceive of informativity in
slightly different terms, and also that they identify slightly different sources of the
uninformativity of imaginings. See Chapter 4 for more on these differences.
The first and hence also the second claim is endorsed by Sartre (1940): 8ff., Wittgenstein (1984d): sec. 627, Ryle (1949/1963): ch. 8, OShaughnessy (2000): 345
and 359ff., and, at least to some extent, McGinn (2005): sec. 1.2. Given that, for
Hume, the ideas of the imagination are not impressions (i.e., perceptions, sensations, feelings, etc.), do not adequately reflect the order of impressions (as memories do) and are also not directly linked to an impression (as beliefs are), he seems
to endorse something like (NC) or (NC*) as well (see Hume (1739): 1.1.1.1, 1.1.3,
1.3.5, 1.3.7.7, and 1.3.7.15). But only OShaughnessy assumes that such claims
identify a fundamental feature of imaginings (see Section 6.2).

Introduction to Part Two

195

(NC) Imaginings are essentially unable to cognise reality.


(NC*) Imaginings are essentially unable to cognise reality in the same
way as their cognitive counterparts (or other cognitive
states).
The last two negation claims are both concerned with the inability of imaginings to constitute (or ground) knowledge about reality. But they differ in
whether they focus on merely some or on all possible ways of constituting
knowledge. For instance, while (NC*) just states that imaginings cannot
cognise reality in virtue of satisfying precisely those epistemic constraints
that govern perceptual experiences, beliefs, memories, and so on (see Section 7.1), (NC) maintains that they cannot cognise reality in virtue of satisfying any potential epistemic constraints.
It is worthwhile to note some of the logical relations between the three
different negation claims. (NC) is stronger than (NC*) in so far as it entails
the latter: if an episode cannot at all constitute knowledge, it obviously
cannot do so in the same manner as, say, perceptual experiences or beliefs.
But there is no similar implication in the other direction, given that (NC*)
leaves it open whether imaginings can cognise reality in a way that is completely different from perceptual, mnemonic or intellectual cognition.
Then, as already suggested in Chapter 4, Wittgensteins reading of (NI)
seems to entail (NC). Arguably, knowledge implies aptness for truth or
veridicality. Hence, if imaginings indeed turn out to be uninformative in
the sense of failing to be truth-evaluable, it follows that they cannot constitute knowledge. Perhaps the same can be said of Sartres interpretation of
(NI) in terms of the provision of new information, given that it comes very
close to Wittgensteins. Finally, there are many other plausible negation
claims that entail (NC*), as will become clear in Section 8.1.
Now, there are at least three reasons why OShaughnessys complex
and in parts unorthodox theory of imagining deserves special attention.
First of all, his theory of imagining is both rich and sophisticated. Furthermore, it is one of the few clear and sufficiently developed examples of
a unified account of imagining to be found in the philosophical literature.

196

The Epistemological Account

Lastly, and perhaps of most interest in the current context, it constitutes the
main and possibly sole manifestation of the Epistemological Account.
For these reasons, it is apposite to devote two whole chapters (Chapters 6
and 7) to the presentation of the view of OShaughnessy and his arguments
for it. However, since many of the details of his view are not crucial to the
assessment of the Epistemological Account, I start the first of these
chapters with a brief summary of OShaughnessys principal claims and his
foremost reasons for endorsing them, thus rendering engagement with the
rest of that chapter and the subsequent one entirely optional. The last
chapter (number 8) is reserved for my objections to the Epistemological
Account and, notably, OShaughnessys version of it.

CHAPTER 6
OShaughnessys View

The most developed unified account of imagining presented recently is that


found in Brian OShaughnessys book Consciousness and the World.5 It incorporates the Epistemological Account of imagining by endorsing specific
versions of the negation claims (NC*) and (NC) as fundamental truths
about imaginings, and supplements it with a representational instance of
the Dependency Account by also including a representational echo claim.
Since OShaughnessys theory has by no means received the same degree
of attention as other theories of imaginings, but well deserves the same
kind of consideration, I have opted to devote a whole chapter to the discussion of his claims and arguments. This is also justified by the fact that both
OShaughnessys views and his writings are less straightforward and penetrable than those of others, which in part explains the general neglect that
they have been suffering in the recent debates on the imagination.
One particular difficulty largely presentational in nature stems from
the complex, and not always absolutely clear, dialectical structure of
OShaughnessys argumentation, which I try to reconstruct in the last section of this chapter. Another specific problem in dealing with, and presenting, OShaughnessys account of imagining is that it forms just one part of
a much wider project, namely to account for the nature of waking consciousness (i.e., the state of being awake) and for the nature of perception
5

See OShaughnessy (2000): chs. 11f.. As already mentioned in Section 1.1,


OShaughnessys theory is weaker than other unified accounts because it denies that
the different forms of imagining share a common intrinsic or relational essence. Instead, it merely aims to identify necessary and sufficient conditions for being an instance of imagining.

198

The Epistemological Account

as its most important aspect.6 As a result, many of OShaughnessys claims


seem to be inseparably linked to his claims about consciousness and, especially, perception. As a result, his discussion of imaginings continually
relies on ideas and arguments presented elsewhere in his book, in relation
to other and often wider issues pertaining to conscious and perceptual
awareness. I none the less intend to present his theory as neutrally as possible with respect to his other often rather controversial views. In particular, I aim to steer away from his belief in, and subtle conception of, a
sense-data theory of visual perception, and to present his view on imagining as non-committal as possible in this respect.7
Readers with time on their hands may skip the first section and move
straight to the second, in which I present the main ideas of OShaughnessys unified account of imagining, notably the three specific negation
and echo claims alluded to above. The third section clarifies some important aspects of his strategy in arguing for his theory and, in particular, for
the three claims. The key issue is how the nature of the two cognitive prototypes mentioned by his version of the negation thesis (NC*) should be
understood. After this clarification, it will be possible to move on in
Chapter 7 to the presentation of OShaughnessys two crucial arguments
for his Epistemological Account.
Readers pressed for time, on the other hand, may just concern themselves with the first section of this chapter. It provides a brief synopsis of
OShaughnessys main points and arguments and permits continuing directly with Chapter 8, while passing over everything else that comes in
between.
6

Waking consciousness has to be distinguished from (what I have called) phenomenal


consciousness, and which OShaughnessy calls contemporaneous experience (or
simply experience), or a stream of consciousness (OShaughnessy (2000): 5;
37ff.). We are concerned with consciousness in OShaughnessys sense of being
wakeful when we say that a sleeping person wakes up, or that a patient regained
waking consciousness, or came to himself (see ibid.: 68ff.). The following discussion should make clear why OShaughnessy takes waking consciousness to be more
fundamental than, and independent of, phenomenal consciousness
However, whenever this may clarify his position on imaginings, or more generally
may contribute to a better understanding of his picture of how the mind works, I elucidate relevant aspects of his underlying ideas about perception in the notes.

OShaughnessys View

199

6.1. A Brief Synopsis


The main idea in OShaughnessys unified account is to define imaginings
in terms of their relation to, and contrast with, perception and intellectual
knowledge. He labels the latter two cognitive prototypes in the sense that
they constitute the epistemic ideal that is, the forms of knowledge that
perceptual experiences and beliefs (as well as judgemental thoughts) aspire
to. Cognitive episodes have a chance to reach that ideal: they may indeed,
under the right circumstances, constitute knowledge. By contrast, imaginings are, for OShaughnessy, bound to fail to realise either of these two
cognitive prototypes. The resulting negation claim that imaginings cannot constitute successful perception or intellectual knowledge is in fact a
version of the negation thesis (NC*), albeit one restricted to the two kinds
of cognition mentioned.8 For OShaughnessy argues that imaginings fail to
realise the cognitive prototypes because they fail to satisfy the cognitive
constraints that define those prototypes, and which need to be met in order
to lead to perceptual or intellectual knowledge. In other words, what he
maintains is that imaginings do not succeed in constituting knowledge in
the same manner in which perceptual experiences or beliefs may do so.
OShaughnessys explanation of this fact is that, while sensory imaginings are not causally linked to the world in the same way as successful perceptions are, intellectual imaginings are not rationally supported by other
cognitive states (whether they are perceptions, memories or beliefs) in the
same way as knowledge-constituting beliefs are. As a result, both kinds of
imagining are unlikely to be veridical and, even if they are, fall short of being sufficiently non-accidental or reliable in their veridicality in order to
count as cognising. That imaginings do not stand in the required causal and
rational links, on the other hand, is due to the fact that they originate either
in the will (i.e., in the case of voluntary imaginings) or in other factors in
the mind which lead to a breakdown of the normal mental processes involved in the occurrence of perceptual experiences and judgemental
thoughts (i.e., in the case of involuntary imaginings). Because of this reference to how imaginings come into being, the resulting argument may be
8

Again, I ignore here that OShaughnessy formulates his view in terms of necessary
and sufficient conditions for being an imagining, rather than in terms of an essence
common to all imaginings (see Sections 1.1 and 6.2).

200

The Epistemological Account

called the Argument from Origin (though this terminology is not


OShaughnessys): imaginings cannot realise one of the two cognitive prototypes because of their special origin in the mind.
In both cases, what matters is that imaginings are not sufficiently constrained by reality. OShaughnessys reasoning with respect to involuntary
imaginings is roughly as follows: given that the formation of passive imaginings is causally or rationally deviant, it cannot ensure that they end up
being non-accidentally veridical with respect to how the world actually is.
The argumentation concerning voluntary imaginings, on the other hand, is
similar to Wittgensteins (see Section 4.7): since what we imagine is determined by what we want or intend, it remains largely unconstrained by
the facts. But, just like Wittgensteins consideration, this line of reasoning
faces the problem of accommodating cases such as the two examples
from Chapter 4 in which we acquire knowledge about the world not despite, but because of our voluntary control over what we imagine (see Section 8.2).
As mentioned before, OShaughnessy also endorses a version of the
stronger negation claim (NC): namely that imaginings cannot cognise reality, irrespective of how they are formed. He does not, however, take this
second negation claim to be fundamental. Instead, he insists that it can be
derived it from (NC*) and from the assumption that there are no other cognitive prototypes over and above successful perception and intellectual
knowledge for imaginings to exemplify. In particular, he does not mention the cognitive ideal of successful episodic memory, indicating that he
does not assume it to be relevant for the question of whether imaginings
can constitute knowledge.9 His main motivation for this omission is presumably his additional endorsement of a representational version of the
Dependency Account (see Section 9.4). For, given that he takes imaginative episodes to be representations of their cognitive counterparts, and given
that the counterparts of sensory imaginings are perceptual experiences, and
not episodic memories (see Chapter 10), it makes sense to concentrate on
perception rather than memory as the relevant cognitive prototype for
9

Indeed, OShaughnessy does not mention memory at all in his discussion of imagining. Moreover, he talks about memory only very briefly in his book, and only in the
context of its role in the continuity of conscious experience and agency (see
OShaughnessy (2000): 55ff.).

OShaughnessys View

201

sensory imaginings.
One thing to note about this omission is that it is not entirely warranted.
Sensory imaginings seem to be much closer to sensory memories than to
perceptual experiences both in their phenomenal character and in their
causal history (see Martin (2001) and Hopkins (2011a)). So, it seems
reasonable to assume that (if at all) sensory imaginings are more likely to
satisfy the cognitive constraints on successful memory rather than those on
successful perception. This fits well with the fact that it seems more
promising to treat sensory imaginings and episodic memories as belonging
to one and the same mental kind (i.e., imagining) than to do the same with
respect to sensory imaginings and perceptual experiences.10 But the
omission is none the less understandable in so far as it may be reasonably
argued that the special origin of imaginings in the will or in a breakdown
of normal informational processes also prevents them from satisfying the
requirements pertaining to knowledge-constituting memories. That is, there
is no reason to question the extension of (NC*) as well as
OShaughnessys Argument from Origin in favour of it to memory as a
third cognitive prototype. The open formulation of (NC*) and the focus
on this negation claim rather than on OShaughnessys more restricted one
is meant to reflect the possibility of this extension and to allow my
discussion to stay neutral on the issue of how many cognitive ideals there
are, and what their specific nature is.
As already indicated, OShaughnessy supplements his versions of the
negation claims (NC*) and (NC) with a more positive thesis that postulates
a particular dependency relation between imaginative and cognitive episodes: the former are said to be essentially representations of the latter. This
claim which is just the thesis (ER) is at the heart of the representational
10

Hume might be read as endorsing the second view (see Chapter 9, also for objections to his underlying picture of the mind). Hopkins (2011a) defends the first view
by arguing that episodic memories are contextually constrained sensory imaginings:
namely sensory imaginings that are veridical and, in the right way, determined by
the past. OShaughnessys reply to Hopkins could be that imaginings and memories
belong to essentially different mental kinds because of their differences in functional
role, phenomenal character, and determination by the past (see Martin (2001)).
Hence, for him, (non-perceptual) visual experiences involving a claim about the past
and being constrained by the past should not count as imaginative.

202

The Epistemological Account

version of the Dependency Account of imagining and is one of the main


examples of what I call echo theses (see the introduction to Part Three).
Again, OShaughnessys actual view is more specific than (ER). One reason for this is that it restricts the relevant cognitive phenomena to the two
cognitive prototypes mentioned (and also drops the reference to the essence of imagining). This means that sensory imaginings are said to be representations of successful perceptions, while intellectual imaginings are
said to be representations of knowledge-constituting beliefs. In addition,
OShaughnessy has a particular take on the nature of the representational
link involved most notably, that it is non-intentional and thus similar to
way in which perceptual experiences are taken by him to represent sensedata (see Section 9.4).
OShaughnessy takes (NC*) and (ER) to be fundamental and to be individually necessary and jointly sufficient for defining imaginings (or, in my
preferred way of speaking, for describing the nature of imaginings); while
he treats (NC) merely as a less basic, direct consequence of (NC*), plus the
already assumption that there are no other forms of knowledge over and
above perception and intellectual knowledge which imaginings might realise. OShaughnessys unified account of imagining therefore consists in a
specific combination of the Epistemological Account and the Dependency
Account. In the remainder of this chapter, I spell out OShaughnessys
mixed view, while I use the next chapter to present his two main arguments
for the Epistemological Account. I then move on, in the closing chapter of
this part of the book, to the evaluation of the Epistemological Account,
primarily in relation to its prospects as a unified theory of imagining. If
you have read this summary of OShaughnessys position, you may very
well proceed straightaway to that final chapter (i.e., Chapter 8).

6.2. The Three Main Claims


The main part of OShaughnessys theory of imagining consists in a definitional project aimed at specifying necessary and sufficient conditions for
being an imagining. The core idea of his definition is that imaginings are
necessarily similar to, and modelled upon, perception and knowledge, but
necessarily never amount to them: they are, in his words, only quasi the

OShaughnessys View

203

types of cognising state in question:


[I]maginings are imaginings [...] through satisfying the definition. That is,
through instantiating the following character: being of necessity merely
quasi some cognitive phenomenon. (OShaughnessy (2000): 361)

Hence, whether something counts as an imagining is for him a matter of


whether it conforms to this definition. OShaughnessy elaborates on how
he conceives of the definition in the following passage:
Imagining is quasi some cognitive prototype which represents Reality as
endowed with a certain character, it is not that prototype and is thus merely
quasi, indeed it is of necessity merely quasi, and it is as such directed to
unreals. These are all necessary conditions of being an imagining. (Ibid.:
359)

He therefore defines imaginings as mental representations which are quasi


an instance of a cognitive prototype or ideal, but necessarily never really
instantiate that prototype. In addition to quasi exemplifying a cognitive
prototype, OShaughnessy often speaks of them being as if that prototype,
or being a seeming instance of it (ibid.: 362ff.). Furthermore, the quoted
passage (as well as other parts of his text) may be interpreted as suggesting
that he supplements his definition with two further conditions on imaginings, namely that they have to be merely quasi cognitive items, and that
they have to be directed to unreals. But in fact, these conditions are
already contained in his definition. That imaginings are merely quasi a cognitive prototype means that they only quasi, but not actually, exemplify
such a prototype which is already a consequence of the definition. Similarly, that imaginings are concerned with unreals means that they do not
really, but at best only seemingly, represent the world, because they are not
cognitively linked to it which, again, follows from the definition (ibid.:
359 and 363f.).11 In short, imaginings are for OShaughnessy therefore ne11

In fact, OShaughnessy thinks that cognition is the only means by which our mind
can really be linked in contrast to merely seeming, or merely presenting itself, to
be linked to reality (see the introduction to OShaughnessy (2000): for instance 6f.
and 12f.). Correspondingly, he also defines the common cognitive significance of
knowledge and perception in terms of the fact that they [represent] Reality (ibid.:

204

The Epistemological Account

cessarily only quasi some cognitive prototype.


As the last quote has shown, OShaughnessy takes the satisfaction of
this definition to be necessary for something to be an instance of imagining. Before addressing the issue of why he takes it also to be sufficient, it is
indispensable to come to a better understanding of the definition by clarifying which cognitive prototypes OShaughnessy has in mind, and what it
means for a mental state to exemplify one of them merely quasi.
The cognitive prototypes to which he refers are perception and knowledge (ibid.: 361, 363 and 366). I say more about his view on these prototypes in the following sections. For the time being, it suffices to note that
OShaughnessy maintains that there are no other cognitive prototypes involved in imagining. Here is how he describes the fact that imaginings
have to be defined in terms of either one or the other prototype:
[T]he imagining phenomenon is (at the very least) a function of that mental
item, insofar as its essential description makes reference to that item. [...]
[I]maginings are necessarily functions uniquely of mental phenomena.
More, they are of necessity functions uniquely and only of the fundamental
mental cognitives knowledge and perception. (Ibid.: 363)

This quote makes clear that, for OShaughnessy, all imaginings are to be
elucidated by reference to either perception (and its modal variants) or
knowledge (and its variants, such as self-knowledge, etc.). However, it also
suggests that he thinks that there are no other cognitive prototypes than
these two: after all, they are the fundamental mental cognitives. This is in
line with the fact that OShaughnessy does not mention in his book the existence of any other cognitive prototype. But even if there were other prototypes, they would not according to what OShaughnessy claims be
relevant for imaginings: the latter pertain uniquely and only to perception
and knowledge. His definition of imaginings thus expresses the view that
all instances of imagining are related to one or other of the two prototypes.
The relation in question is that of quasi or seeming instantiation (ibid.:
363). For OShaughnessy, there are two aspects to this relation of quasi exemplification. The first is that, although imaginings resemble their cognitive counterparts in many respects, they cannot really instantiate perception
363).

OShaughnessys View

205

or knowledge.
As OShaughnessy notes, there are certain similarities (or a certain resemblance, as he calls it) between instances of the relevant two kinds of
mental state (ibid.: 363-6). While sensory imaginings are like perception,
intellectual imaginings are like knowledge.12 For not only are both the imaginings and the cognitive prototypes representational, but the imaginings
can have the same intentional objects and be of the same types of representation as the respective prototypes (ibid.: 166f.). Thus, both sensory imaginings and perceptions represent perceivable objects and features relative
to a certain modality; and both intellectual imaginings and states of knowledge involve concepts and take propositions as their objects (ibid.: 365f.).
Imaginings imitate and are like their cognitive prototypes and may
hence also sometimes be mistaken for them (ibid.: 365).
None the less, OShaughnessy maintains that imaginings necessarily
differ from, and do not fully amount to, their cognitive prototypes (ibid.:
359ff.). Given that the similarity claim does not play a very significant role
in his account,13 it is this negation claim about imaginings which constitutes the first important part of what he means when he says that imaginings are only quasi the cognitive prototypes, but never the real thing
(ibid.: 361). More specifically, OShaughnessy endorses the weak negation
claim that no imagining can cognise reality by instantiating one of the two
Here, I largely ignore a third kind of imagining that OShaughnessy identifies,
namely the imaginative perception supposedly involved in pictorial experience (but
see Note 13 below, Note 7 in Chapter 8, Note 13 in Chapter 11, as well as Dorsch
(2005): sec. 6.5., and Dorsch (2012c), for some further discussion). OShaughnessy
also does not speak of sensory and intellectual imaginings, but instead of directobject and propositional imaginings. His main reason for this is that he believes
that all sensory imaginings are literally imaginings of experiences and are therefore
always directed at objects (i.e., the experiences), while only intellectual imaginings
can have propositions as their objects (see OShaughnessy (2000): 342f.). However,
since this extra claim about the nature of sensory imaginings is controversial (see
Dorsch (2010c)), I use the more neutral terminology.
13
Apart from elucidating the representational nature of imaginings, the similarity
claims primary significance is that it may give some support for the endorsement of
OShaughnessys version of (ER) to be introduced below. The idea is that only this
echo thesis the claim that imaginings have cognitive prototypes as their directly
represented objects can explain why the similarities hold. I return in Part Three to
the issue of how strong a support this provides for (ER) or similar echo claims.
12

206

The Epistemological Account

cognitive prototypes, successful perception and intellectual knowledge.


Given that perceptual experiences and beliefs (or judgements) constitute
knowledge by exemplifying the relevant cognitive prototype, this claim is
a version of (NC*), limited to the two specific forms of knowledge
mentioned.14
The other aspect of the relation of quasi exemplification is that imaginings constitutively depend on the cognitive prototypes by taking them as
their immediate objects. More precisely, imaginings have types of perception or knowledge as their immediate, non-intentional and necessary objects (not unlike sense-data in the case of perception). This does not, however, deprive imaginings of their less immediate intentional objects (see, in
general, OShaughnessy (2000): 166f. and 363ff.):
[T]he intentional object of imagination must be distinguished from the (socalled) immediate object of imagination, which is to say from the cognitive prototype or that which is being imagined for example, in the case
(say) of visual imagining from the visual perceiving of an object. [...] [T]he
distinction [is] between intentional object content and immediate object
content of imagining [...]. [...] [T]he event of imagining is filled by a perceptual experience which is posited as being intentionally directed to the
very same object as the imagining itself. (Ibid.: 363)

This passage reveals two interesting aspects of OShaughnessys picture of


imaginings as second-order functions, shadows or replicas of the cognitive prototypes (ibid.: 363-5). First, he takes the content of imaginings to
involve two objects (or, if one prefers, to have two contents involving different objects). The first object is an imagined type of mental episode (e.g.,
a visual perception), while the second object is an imagined visible entity
(e.g., a tree or a car), namely the intentional object of the imagined type of
14

The only notable difference is, again, that OShaughnessys account is concerned
with providing a definition of imagining, while my concern is with characterising its
essence. This difference in aim is reflected in the concrete formulation of OShaughnessys specific version of (NC*), as well as in the formulation of his specific versions of (ER) and (NC) to be introduced further below. The issue of contention is not
whether each instance of imagining possesses an essence: it surely does, as do all
other entities. Rather, we disagree on whether all imaginings share one and the same
essence (see Section 1.1).

OShaughnessys View

207

episode. Accordingly, OShaughnessy assumes that all sensory and intellectual forms of imagining are instances of experiential imagining: they
consist in imaginings of types of mental episodes (ibid.: 359). What he has
in mind is that we imagine something by imagining our relevant cognitive
access to it: that is, we visualise a face by imagining seeing (or seem seeing) one, or we make-believe that it rains by imagining believing (or
seem believing) the corresponding proposition (ibid.: 363f.). 15 Second,
the imaginings and the imagined cognitive prototypes share the same inten15

In fact, the situation is (again) a bit more complicated. As already noted, OShaughnessy insists that there is no such thing as imagining per se: different forms of imagining possess different essences (see OShaughnessy (2000): sec. 11.6 and ch. 12).
We can visually, auditorily or propositionally imagine something; but we cannot
simply imagine it. This is why he writes that there is no such thing as merely imagining an object, and that imagining a seeing acquires its identity not under imagining, but under visual imagining (ibid.: 364; see also 366). Indeed, it is structurally misleading to treat imagining as a fundamental mode in which we may be
consciously related to entities (e.g., types of experiences and their external objects);
rather, the fundamental modes of awareness include imagining seeing, imagining
hearing, and so on, which are intentionally directed at external objects, and not at
types of experiences (ibid.: 364).
Noordhof (in a talk presented at the Perceptual Memory and Perceptual Imagination conference in Glasgow 2011) takes this to mean that OShaughnessy rejects the
Dependency Account: that visual imagining is just another kind of visual experience,
the content of which is exhausted by the external object visualised. But OShaughnessys insistence that that which is being imagined [...] in the case (say) of visual
imagining is identical with the visual perceiving of an object (ibid.: 363; this passage is also quoted above) undermines this interpretation. Instead, OShaughnessys
point here is, again, that the different forms of imaginings do not share a common
essence (and, especially, a common mode of awareness).
More specifically, the imaginative element in question that imaginings constitutively depend on, and non-intentionally represent, types of cognition (in a way similar to the representation of sense-data) does not correspond to a detachable basic
mode of awareness precisely because the non-intentional nature of the representation
ensures that different forms of imagining are constitutively linked to different types
of cognition.
This also fits well OShaughnessys observation that perceptual and imaginative
awareness are not two independent first-order forms of awareness that may have the
same objects of awareness, but instead a pair of independent first-order and dependent second-order awareness: [t]here are not two distinct ways of relating to an
identical object-content, the perceptual way and the imaginative way: there is one

208

The Epistemological Account

tional objects and, presumably, represent them as having the same features
(e.g., as being a green apple).16
In addition, OShaughnessy understands the relation between imaginings and the imagined prototypes as non-intentional (ibid.: 363 and 371).
He also maintains that imaginings of a certain kind (e.g., auditory imaginings) could not exist if the corresponding kind of prototype (in this case,
auditory perception) did not exist (364; 377). Consequently, imaginings are
said to be functions of the cognitive prototypes in two senses: the former
have to be described in terms of the latter (as indicated in the quote above);
and the former are constitutively related to the latter. In my terminology,
this implies that imaginings are echoes of the cognitive prototypes (see the
introduction to Part Three). It is precisely this specific echoing relationship
between imaginings and the respective prototypes, which the second aspect
of the relation of quasi exemplification introduced above consists in. It can
way, and a shadow of that way (ibid.: 365).
There is, however, still the problem to make sense of OShaughnessys idea of a
non-intentional, relational form of representation that has, as its second relatum, not
a particular mental episodes (or other object), but just a type thereof. Indeed, proponents of an intentionalist understanding of the representational relation between
imaginings and cognitions may point out, for instance, that their view also implies
that imaginings are relations to types of cognitions, given that the general content of
intentional representations constitutively depends on the represented properties (e.g.,
we cannot have an experience as of something red, if the universal redness does not
exist). This may very well constitute one motivation of preferring the intentionalist
reading of (ER) over the relationalist interpretation.
16
Note, however, that the first aspect of OShaughnessys claim that imagining is
merely a [second-order] function only and uniquely of its prototype (OShaughnessy (2000): 366) implies that imaginings do not directly (or on their own) relate to
the intentional objects, but only via the imagined cognitive representations. For the
immediate objects of imagining (i.e., that which is being imagined (ibid.: 363))
are possible instances of one or the other cognitive prototype. As a consequence,
there are not two distinct ways of relating to an identical object-content, the perceptual way and the imaginative way: there is one way, and a shadow of that way
(ibid.: 365). It is in this sense that imaginings possess a second-order nature and are
like replicas or shadows of the corresponding cognitive states (ibid.: 363ff.; see
also Section 3.2 on immediacy, and Chapters 9f. on Humes view and on the Representational Account). Moreover, this explains, according to OShaughnessy, why
imaginings can be and sometimes are confused with the respective prototypes
(ibid.: 365).

OShaughnessys View

209

be captured by the echo thesis that all imaginings are non-intentional representations of one of the two cognitive prototypes, successful perception
and intellectual knowledge.
This echo claim constitutes a specific version of (ER) which characterises the postulated representational link as non-intentional. As illustrated
above, OShaughnessy takes it to be necessary for imaginings to be necessarily only quasi instances of the two cognitive prototypes. This means, in
particular, that he takes the echo thesis and the weak negation claim introduced above to be necessary for imaginings, since they specify what it
means to merely quasi exemplify the prototypes.
But OShaughnessy intends to provide a full definition, that is, one
which also specifies sufficient conditions for imaginativeness: he is looking [...] for the defining marks of being an imagining (ibid.: 342). Although OShaughnessy is not explicit about when sufficiency is reached,
he thinks that his definitional project is completed once the relation of
quasi exemplification is understood in the manner just described: adding
the conception of imaginings as imaginings of cognitive prototypes to the
necessary conditions mentioned earlier [concludes] the analytical characterisation of imagining (ibid.: 366). Hence, it is reasonable to assume that,
for OShaughnessy, the satisfaction of his definition of imagining is both
necessary and sufficient for something to be an imagining if the relation of
quasi exemplification between imaginings and the respective cognitive
prototypes is understood in terms of the weak negation and the echo claim.
This interpretation is in line with the fact that all of OShaughnessys
comments, which emphasise the fact that the conditions so far specified by
him are only necessary, precede his introduction of the non-intentional representational link between imaginings and the cognitive prototypes (ibid.:
359 and 362). Furthermore, this reading gets additional support from the
fact that OShaughnessy argues that imaginings are unique in being constitutively linked to other mental phenomena by having them as their nonintentional objects: such a measure of dependence of one psychological
phenomenon upon another is without precedent in the mind (ibid.: 371). In
contrast, higher-order thoughts, intentions or desires at best take mental
phenomena as their intentional objects; and their identity as distinctive
types of mental phenomena does not depend constitutively on the mental

210

The Epistemological Account

phenomena they may represent (ibid.: 363 and 371f.).


As a result, OShaughnessy aims to satisfy the first desideratum for a
unified theory of imagining, which requires full extensional adequacy with
respect to all central cases of imagining as well as all paradigm instances
of non-imaginative phenomena. But he also takes his theory of imagining
to satisfy the second desideratum demanding explanatory power and, in
particular, explanatory fundamentality. His envisaged theory is meant to
provide not only a definition of the concept of imagining, but also an explanation of how the different types of imagining do indeed satisfy this
definition in virtue of certain properties that they instantiate (ibid.: 339ff.,
357ff. and 362). Accordingly, he is also concerned with the issue of whether their conforming to the definition may be traced back to certain intrinsic
or relational features of the respective mental representations.
As will become clear, OShaughnessy argues at length that one reason
why all imaginings (but not, say, perceptions or judgements) satisfy his
version of (NC*) is that they arise entirely from the mind, without any influence from how thing are in reality (ibid.: sec. 11.2 and 11.4). However,
he also tries to show that different forms of imagining do so in virtue of
different specific origins: some because they are voluntarily produced, and
others because they are the result of some breakdown in the usual causal
pathways leading to perceptions or beliefs. Hence, there is for him no unified way in which imaginings come into being. In addition, OShaughnessy
maintains that there is a further reason for why (NC*) applies to sensory
imaginings in particular: namely their specific constitutive differences
from perceptions (ibid.: 354ff.).17
17

OShaughnessy adopts the same strategy to pictorial experience, which he takes to


involve an imaginative experience of what is depicted. For him, pictorial experience
(or at least its imaginative element) satisfies his definition because of both its origins
and its constitution (OShaughnessy (2000): 346ff.). In what follows, I do not discuss his treatment of pictorial experience, mainly because it parallels that of sensory
imaginings. I take it that he succeeds in showing that his weak negation claim (i.e.,
his version of (NC*) is true of pictorial experience. But, as I argue in the next
chapter, I do not think that this is sufficient to show that they are really imaginative,
or that they conform to (ER) (i.e., involve the imagining of perceptions of the depicted; see ibid.: 349). Besides, I am not sure whether (NC*) or a similar negation
claim has much plausibility for pictorial experiences, given that they manage to
ground knowledge about what is depicted (see Note 7 in Chapter 8, as well as

OShaughnessys View

211

I discuss his arguments for his version of (NC*) in more detail in the
subsequent sections. At the moment, it suffices to stress that this weak
negation claim is an essential part of OShaughnessys definition of imagining; and that he therefore concludes that different kinds of imagining
conform to the definition for different reasons: some because of their origin
in the will, others because of their origin in some mere mental causes, and
some in addition because of their constitution (ibid.: 359ff. and 367). Accordingly:
[T]he property of being an imagining is neither a relational nor constitutive
property, even though relational and/or constitutive properties determine its
existence. What matters is that a certain necessity is binding, namely that
imaginings cannot exemplify their cognitive prototype. (Ibid.: 361)

This means that, for OShaughnessy, being an imagining that is, satisfying the definition of imagining cannot be reduced to, or explained in
terms of, a more fundamental feature shared by all imaginings: there is no
feature common to all imaginings, over and above their satisfaction of the
definition in virtue of which they count as imaginings. In particular, there
is no proper imaginative essence: the class of imaginings does not constitute a natural mental kind (ibid.: 361 and 368ff.). Moreover, his conclusion
that there is no underlying intrinsic or relational feature common to all
imaginings, which can account for their imaginativeness, has the consequence that OShaughnessy takes the property of satisfying the definition
of imaginings to be explanatorily fundamental. Given that it is also explanatorily illuminating because it tells us in substantial terms what imaginativeness consists in, his theory of imagining is designed to satisfy the
second desideratum for a unified account of imagining as well.
However, OShaughnessy does not stop here. For him, his analysis of
what it means for something to be an imagining entails that imaginings
cannot cognise the world even though they otherwise resemble cognitive
states in many respects. His main idea is that the imaginings cannot cognise reality because they cannot really exemplify their respective cognitive
prototypes. Indeed, he links the failure of imaginings to exemplify the two
cognitive prototypes with their failure to be able to cognise reality by
Dorsch (2012c)).

212

The Epistemological Account

identifying the same cause for both, namely, the specific origins of imaginings in the mind (ibid.: 345 and 359). In addition, OShaughnessy also
does not seem to think (as already illustrated above) that there are any other cognitive prototypes which imaginings might instantiate. For him, there
do not seem to be more than two fundamental mental cognitives knowledge and perception (ibid.: 363). Accordingly, he concludes with respect
to all forms of imagining:
The fact that the mind acts [...], not as a representative of Reality but in direct opposition, guarantees that imaginings must be cognitively void. [...]
[F]ailure to realise the [cognitive] prototype emerged as a universally necessary feature of imaginings. (Ibid.: 359)

And concerning intellectual imaginings (which he calls propositional imaginings), he writes:


No propositional imagining can be a knowing. [...] Propositional imagining
is as such out of the cognitive circuit. (Ibid.: 345)

Hence, he does not take imaginings to possess the capacity to cognise reality. This aspect of his theory, which is presented not as part of his definition of imagining, but rather as a consequence of it, can be expressed by
the strong negation claim that no imagining can cognise reality a claim which is
in fact identical with (NC) (with the qualification mentioned above in Note
10 in mind).
To conclude, his versions of (NC*), (ER) and (NC) adequately reflect
OShaughnessys account of imagining: while the weak negation claim and
the echo thesis constitute the two main ingredients of his definition of imagining, he takes the strong negation claim to be one particular consequence
entailed by this definition. For him, these three claims apply to all imaginings; and there are no non-imaginative mental phenomena of which all
three claims are true. Accordingly, he takes imaginings to be precisely
those mental phenomena of which these three claims are true. Since
OShaughnessy thinks that certain entailment relations hold between the
three claims, his theory can presumably be presented in a simpler form
(e.g., by reference to only the first two theses, or to his own statement of

OShaughnessys View

213

his definition). But it becomes clear further below that partly due to some
apparent ambiguities in OShaughnessys notion of the cognitive prototypes it is better to keep these three aspects of his account apart.
With respect to the project of providing a unified account of imagining,
OShaughnessys theory is meant by him to satisfy the two desiderata, notably the demands for extensional adequacy and explanatory fundamentality. As it stands, the account can also reasonably hope in particular due to
his version of (ER) to elucidate the nature of imaginings in positive
terms. Moreover, due to the nature of the three main claims involved,
OShaughnessys theory is a version both of the Epistemological Account
and of the Dependency Account: its key idea is to characterise imaginings
as non-cognitive in various ways, as well as representationally dependent
on cognitions.

6.3. The Argumentative Strategy


Let me now turn to the issue of how OShaughnessy aims to argue for two
of his claims, namely the weak and the strong negation thesis. I postpone
the discussion of his arguments for his echo claim until the next section. As
already indicated, OShaughnessy tries to derive the strong negation thesis
the thesis that imaginings cannot cognise reality from two premisses.
The first consists in the weak negation claim, which claims that imaginings
cannot exemplify either of the two cognitive prototypes. The second
premiss is that there are no further cognitive prototypes which imaginings
might be able to instantiate. Taken together, these two premisses entail the
intermediate conclusion that no imagining can be an instance of any cognitive prototype (of whichever sort). Since OShaughnessy plausibly takes
this intermediate conclusion to imply if not even to amount to his
strong negation claim, the two premisses appear to establish the desired
conclusion. The main problem with this derivation is, however, that
OShaughnessy does not provide any explicit support for the second
premiss in his argument. While he tries to show in detail why imaginings
cannot realise the cognitive ideals of successful perception or intellectual
knowledge, and hence why his weak negation thesis is true, he does not
concern himself with the issue of the possibility of other forms of know-

214

The Epistemological Account

ledge which imaginings might be able to instantiate.


That this issue is clearly pressing is illustrated by the possibility of cognising sensory memories. These cognitive episodes should count as exemplifying a cognitive prototype especially if there is indeed the assumed
entailment relation between the intermediate conclusion and the strong
negation claim. But it is difficult to see how they could be said to instantiate the prototype for perceptual experiences or the prototype for beliefs.
Perhaps it is still possible to show that successful episodic memories satisfy the same cognitive constraints as perceptions or beliefs and, therefore,
are not linked to a third cognitive prototype. This might then enable
OShaughnessy to rule out the existence of additional cognitive prototypes,
and to establish the second premiss in his argument.
However, if this strategy turns out to be unfeasible, he would have to
provide a different argument for the second premiss. The idea here might
be that his Argument from Origin in favour of the weak negation claim also
supports or at least can be easily modified to support the idea that there
are no further cognitive prototypes that imaginings could exemplify. In other words, it might be hoped that the specific origins of imaginings prevent
them from constituting, not only perceptual and intellectual knowledge, but
also any other possible form of knowledge. It remains to be seen, however,
which of the two alternatives (if any) OShaughnessy might be able to
choose. I return to this issue in the last part of Section 7.1.
For the time being, I turn to OShaughnessys argumentative support for
the weak negation claim: that is, for the thesis that no imagining can be an
instance of successful perception or intellectual knowledge. As noted
above, OShaughnessy thinks that imaginings conform to this claim either
because of their constitution, or because of their specific origins, or because of both. In relation to this conviction, he presents two arguments:
one arguing for the idea that all imaginings cannot instantiate the two cognitive prototypes because of their roots in the mind (what I call the Argument from Origin); and the other identifying a constitutive difference
between sensory imaginings and perceptions which likewise prevents them
from exemplifying the respective cognitive prototype (what I call the Argument from Attention).
For a better understanding of the two arguments, it is necessary to get

OShaughnessys View

215

clearer about how OShaughnessy conceives of the precise nature of the


two cognitive prototypes that he has in mind. The main thing to be noted is
that OShaughnessy identifies the cognitive prototypes with forms of
knowledge. That is, they constitute the cognitive ideals for which perceptual experiences and beliefs (or judgements) always strive, and which the
latter can and sometimes do realise. This parallel treatment of perceptual and intellectual knowledge can be illustrated by the following two
quotes, in which OShaughnessy specifically singles out the cognitive prototypes as something that cognitive states can and should instantiate:
[V]isual experience is aspirational in allowing for the possibility of its coinciding with its own visual prototype [...]. (Ibid.: 366)
Belief is the essence of knowledge, it aspires to the condition of knowledge, and constitutes a core phenomenon which can find itself redescribed
as knowledge. (Ibid.: 368)

Accordingly, both perceptual experiences and beliefs are meant to constitute knowledge; and they may or may not succeed with respect to this ambition.18 Both fulfil this function if they instantiate their respective prototypes. This indicates that OShaughnessy has two forms of knowledge in
mind when he speaks of perception and knowledge as the two cognitive
prototypes: what he refers to are perceptual and intellectual knowledge.
Moreover, he seems to embrace the traditional picture of intellectual knowledge, according to which belief is a constitutive part of this form of knowledge.19
Strictly speaking, OShaughnessy thinks of successful perception, not as a form (or
mode) of knowledge, but rather as a way of acquiring knowledge (OShaughnessy
(2000): 551). That is, he restricts the constitution of knowledge to beliefs. Here, I ignore this aspect of his position and continue with my general practice in this book to
treat what others take to be merely grounds of knowledge such as successful perception or memory as instances of knowledge as well (see the introduction to Part
One).
19
This appears to be in tension with OShaughnessys idea that there are certain (imaginative) beliefs which cannot constitute knowledge: Propositional imaginings can
be real beliefs [...]. And yet it is clear that their truth must be accidental. No propositional imagining can be a knowing (OShaughnessy (2000): 345). But what he
should have in mind is simply that (non-imaginative) beliefs cannot constitute
18

216

The Epistemological Account

However, OShaughnessys conception of the two cognitive prototypes


raises an immediate puzzle. His version of (ER) states that imaginings are
non-intentional representations of the two cognitive prototypes. According
to the considerations above, this means that they are representations of successful perception or intellectual knowledge. However, this already entails
that imaginings differ constitutively from those forms of knowledge. For if
imaginings really represent types of perceptual or intellectual knowledge,
they cannot themselves exemplify those types. Instances of successful perception do not have other successful perceptions as their objects. And instances of intellectual knowledge can have other instances of intellectual
knowledge at best as their intentional objects, namely in the shape of
higher-order representation. Hence, assuming that imaginings are representations of forms of knowledge already implies that they cannot be instances of those forms: OShaughnessys version of the echo thesis entails
his weak negation claim. This raises a pressing question about OShaughnessys argumentative strategy: namely why he does not make any use of
this straightforward argument for the weak negation thesis, but instead
spends a lot of time on developing the two alternative arguments already
mentioned notably the Argument from Origin.20
Given the simplicity of the argument just sketched, it is unlikely that
knowledge under certain circumstances: namely when they are based either on imaginative visual hallucinations which are phenomenologically indistinguishable from
perceptions, or on dream experiences.
First of all, OShaughnessys prime example of such a delusional belief (ibid.:
344) Macbeths belief that Banquo [is] at the table (ibid.: 345) could constitute
knowledge, given that Macbeth could form exactly the same belief on the basis of
veridically perceiving Banquo at the table. Then, treating this and similar beliefs as
imaginative would be incompatible with (ER). Finally, OShaughnessy does not
have any good reason to take these beliefs to be imaginative. It is at best
controversial (and at worst false) that the underlying experiences should count as
imaginative (see Sections 1.3 and 14.4). And the nature of beliefs does not seem to
be affected by the nature of the experiences on the basis of which they are formed.
In particular, I have argued in Chapter 4 that beliefs, that are conceptualisations or
endorsements of visual imaginings, can sometimes constitute knowledge.
20
The details of the Argument from Origin will not be discussed until the end of Section 7.1, mainly because this discussion presupposes that we have a good grasp of
OShaughnessys view on how imaginings are supposed to differ in origin from their
cognitive counterparts.

OShaughnessys View

217

OShaughnessy just overlooked it. Instead, the best explanation seems to


be that he intended to endorse a weaker echo thesis, namely that all imaginings are non-intentional representations of perceptual experiences or
(occurrent) beliefs. This echo claim is weaker than the one introduced
above in so far as it takes imaginings to be representations of types of perceptual experience or belief (including episodes of illusion or hallucination, as well as states of ignorance), rather than of types of successful perception or intellectual knowledge. Despite making no reference to the two
cognitive prototypes, the weak echo claim is still in line with OShaughnessys claims about the similarities and differences between the imaginative episodes and the imagined types of representation. Moreover, although
the resulting definition of imagining becomes slightly more complicated
since the two crucial claims the weak negation thesis and the weak echo
claim are concerned with different types of cognitive phenomena, its key
idea is still preserved: that imaginings cannot exemplify the cognitive prototypes; and that they have cognitive phenomena as their direct objects.
However, there is still a further problem. In the last passage quoted
above, OShaughnessy endorses the orthodox picture of intellectual knowledge, according to which the latter implies belief. Similarly, he seems to
think that it is part of the nature of the sensory cognitive ideal that its realisation requires perceptual experience. What this precisely means is that
the cognitive constraints, which define the two ideals or forms of knowledge and govern all mental representations aiming to realise them, include
not only requirements on veridicality, reliability, and so on, but also a requirement on, respectively, being a perceptual experience or being a belief.
Two pieces of textual evidence speak in favour of this interpretation of
OShaughnessys view on the sensory cognitive prototype. The first pointer
is simply that he uses the term perception to describe this sensory prototype (e.g., ibid.: 361, 363 and 366). And it is very compelling to conclude
that something has to be a perceptual experience in order to be able to
amount to successful perception. The second indication is that OShaughnessy actually provides an argument to the effect that perceptions and sensory imaginings differ constitutively; and that he takes this constitutive difference to imply that sensory imaginings cannot instantiate the perceptual
prototype (i.e., that the weak negation claim is true in this respect).

218

The Epistemological Account

The argument in question is in fact what I have labelled the Argument


from Attention. Its main idea is that perceptual experiences but not sensory imaginings involve attention and allow for the distinction between
what is present to consciousness in the perceptual field (though perhaps
unnoticed) and what is actually noticed by the perceiver (ibid.: 354ff.). The
details or the soundness of the argument need not interest us here (I discuss
both in Section 7.2). But what is of importance is that OShaughnessy uses
it to conclude that there is an essential difference between sensory imaginings and perceptual experiences which prevents the former from becoming
instances of the respective cognitive prototype:
[P]erceptual imaginings generally are not merely causally guaranteed not to
be their prototype, they are in addition constitutively incapable of being the
prototype, since they are [...] essentially different experiences from the prototype experience. (Ibid.: 360)

That there is an essential difference between sensory imaginings and perceptual experiences does not on its own suffice to prevent the former from
instantiating the same cognitive prototype as the latter. What is in addition
needed is the assumption that the nature of the sensory cognitive prototype
is such that only perceptual experiences can exemplify it. Hence,
OShaughnessys Argument from Attention relies on a conception of this
prototype which already excludes sensory imaginings and other non-perceptual experiences as candidates for its realisation. This is reflected,
again, by his identification of the sensory cognitive ideal with successful
perception.
But, now, more or less the same question about OShaughnessys argumentative strategy arises as above. For the weak echo thesis entails the
weak negation claim on the assumption that only perceptual experiences
and beliefs can, by definition, constitute instances of perceptual or intellectual knowledge. The truth of the weak echo thesis ensures that imaginings
differ essentially from perceptual experiences and beliefs. The definitional
link between the cognitive episodes and their respective forms of knowledge then suffices to establish the weak negation claim. Hence, there is no
need to look for any further arguments in favour of this claim. In particular,
the issue of how imaginings are actually produced becomes completely ir-

OShaughnessys View

219

relevant for the issue of whether they can exemplify the cognitive prototypes. This explains why what I have called the Argument from Origin
which traces back the truth of the weak negation claim to the specific ways
in which imaginings arise from our minds would turn out to be completely superfluous. It thus becomes, again, puzzling why OShaughnessy
spends to much time on developing and defending this argument, while not
explicitly mentioning the short and simple line of argument just outlined.
As spelled out in Chapter 2, it is not very promising to give up on the
idea that there is an essential difference between imaginings and cognitions. Instead, the best solution to this puzzle seems to drop the assumption
that the cognitive constraints, which define the sensory and intellectual
forms of knowledge mentioned in the weak negation claim and govern all
attempts at their realisation, include the requirement that the mental representations concerned have to be perceptual experiences or beliefs. This
means that the cognitive constraints should be understood as just being
concerned with more centrally epistemic conditions like veridicality, reliability, and so on. As a consequence, nothing about the nature of the two
forms of knowledge under consideration prevents mental representations
other than perceptual experiences and beliefs from realising them. In accordance with this, the truth of the weak negation claim that is, the failure of imaginings to constitute perceptual or intellectual knowledge has
to be due to some other reasons, such as the inability of imaginings to be
veridical or reliable representations of the external world. This is exactly
what the Argument from Origin is meant to show: that imaginings cannot
exemplify the two cognitive prototypes because they occur in ways which
do not involve their causal or rational determination by reality (see Section
7.1).
Whether this would really be OShaughnessys preferred reply to the interpretative problem raised above is, to some extent, a mute question. His
argumentative strategy and, in particular, his insistence on the significance of the Argument from Origin in the derivation of his weak and the
strong negation claim makes sense only if he is regarded as endorsing the
weak rather than the strong echo thesis, and as characterising the cognitive
ideals primarily in epistemic terms, rather than also in terms of specific
kinds of mental episode.

220

The Epistemological Account

Before I turn to the discussion of this argument, it is important to highlight an important consequence of the preceding considerations. If
OShaughnessys two negation claims would in fact be derivable from his
echo thesis in the way sketched above, they would stop singling out a basic
aspect of the nature of imaginings. As a consequence, all that would fundamentally matter would be the echo claim; and OShaughnessys theory of
imagining would just be a version of the Dependency Account, but not also
one of the Epistemological Account. But his actual argumentative strategy
suggests, rather, that the two negation claims need to be established independently of the representational dependency of imaginings on cognitions.
Hence, the echo claim should better not be read as implying the weak and
the strong negation claim; and OShaughnessys unified account of imagining should indeed better be understood as a combination of the Epistemological and the Dependency Account.

CHAPTER 7
OShaughnessys Arguments

As already discussed in the last chapter, OShaughnessy puts forward two


arguments in favour of his version of (NC*): the Argument from Origin
and the Argument from Attention. Each section of this chapter is devoted to
the presentation of one of these arguments.
The main idea of the Argument from Origin is that imaginings are prevented by their specific origins in the mind from satisfying certain cognitive constraints, and therefore from exemplifying the cognitive prototypes
defined by constraints. The discussion of the Argument from Origin proceeds in two steps. The first step is concerned with the clarification of the
proposed link between the prototypes and the constraints. OShaughnessys
does not only endorse certain traditionally accepted epistemic requirements
governing perceptions and beliefs (e.g., veridicality and reliability), but
also introduces some additional and unorthodox epistemic demands, the
postulation of which is closely linked to his belief that only waking subjects can cognise reality. In a second step, I expound on OShaughnessys
idea that imaginings cannot satisfy these constraints because of the various
ways in which they arise from the mind.
While I address the main weakness of the Argument from Origin (i.e.,
that it cannot be extended to support also (NC), and not only (NC*)) not
before Section 8.2, the second section of this chapter combines the summary presentation of the Argument from Attention with a critical discussion of OShaughnessys main reason for accepting its soundness. The key
thought of the argument is that sensory imaginings do not allow for the
same kind of attention to what is sensorily represented as perceptions do,

222

The Epistemological Account

and that this difference between the two kinds of sensory episode marks an
essential difference, given that the object-directed attention in question is
constitutive of perception. My aim is to show that the relevant examples do
not support the conclusion that sensory imaginings do not involve perceptual attention in the way envisaged by OShaughnessy.

7.1. The Argument from Origin


At one place, OShaughnessy summarises his Argument from Origin as
follows:
Thus, we saw in the case both of propositional [i.e., intellectual FD] and
direct-object [i.e., sensory FD] imaginings generally, that causes alone ensured they could not realise their prototypes. All imaginings arise from the
subjects mind in such a way that the constraint of Reality is necessarily inoperative, whether through substituting ones will for Reality or through
confusing subjective Reality with Reality itself. (OShaughnessy (2000):
359)

This argument consists of two chief claims. The first links the capacity to
exemplify one of the two forms of knowledge to the capacity to satisfy certain cognitive constraints (or Reality constraints, in OShaughnessys terminology). The satisfaction of these constraints is necessary and perhaps
also sufficient for the exemplification of the respective cognitive prototypes. The second thesis of the argument states that imaginings cannot satisfy the respective cognitive constraints because of their special origin in
the mind. Cognitions, by contrast, possess this capacity since they can and
typically do originate in the world in the way required for that potential
satisfaction. His overall strategy is thus to derive the difference in cognitive capacity from a difference in origin, thereby making essential use of the
idea of cognitive constraints. Before presenting OShaughnessys defense
of the second thesis concerning the specific origins of imagining, I begin
with his first claim that cognising representations have to satisfy certain
cognitive requirements.

OShaughnessys Arguments

223

(i) Standard Constraints on Cognitions


From what OShaughnessy says, one can distil the idea that the constraints
on perceptual experiences and beliefs, which aspire to exemplify their respective prototypes, consist minimally in three demands: (i) that the state
in question represents reality; (ii) that it represents reality correctly; and
(iii) that it does so in a reliable way. The first demand requires that the candidate states are intentionally linked to objects. This presupposes that they
can be correct or incorrect in respect of how they represent objects as being, and that they can genuinely misrepresented objects (i.e., misrepresent
them without ceasing to represent and refer to them: see Dretske (1986)).
The third demand, on the other hand, requires a suitable link to or origin in
reality, or evidence about it. Among the significant features of the cognitive
ideals (and their instances) are thus intentionality, veridicality and reliability. Accordingly, OShaughnessys picture of knowledge and its acquisition
appears to be fairly standard.
Now, to illustrate that he really endorses a view like the one just
sketched, I turn to what he has to say about visual perceptions and beliefs
as possible instances of knowledge. In particular, I focus in this section on
the differences in the third demand on each type of state, which are due to
the different kinds of representation in question (and, ultimately, their different links to reality). It is here that certain aspects of OShaughnessys
analysis of the nature of perceptions, beliefs and the respective forms of
knowledge become relevant. I then, in the following section, address the issue of whether his account of perceptual and intellectual knowledge also
involves some non-traditional elements which reach beyond the three elements of the standard view just sketched.
Let me begin with the sensory form of knowledge (or with the sensory
form of epistemically sound representation if one disputes the use of the
term knowledge in this context). OShaughnessys general analysis of the
concept of perception is extensive, stretching over most of his book. It is
also controversial regarding some of its main aspects, such as its embracing a version of a sense-data theory (OShaughnessy (2000): chs. 17f.),
the identification of phenomenal consciousness with attention (ibid.: ch. 7),
the interpretative activity of attention involved in perception (ibid.: chs. 8f.,
14 and 20), and the related conception of perceptions as essentially non-

224

The Epistemological Account

propositional (ibid.: ch. 10). For the current purposes, however, it is possible to set aside these parts of his account, as well as the general issues
concerned. Instead, I would like to focus on several features of perceptions
in general, and of successful perceptions in particular, which seem to be
less controversial. The first is that perception is intentionally directed to
mediated physical objects (people, sky, trees, and the public occurrence of
secondary qualities) (ibid.: 16, 19 and 166f.). Furthermore:
A material object can come to visual attention only if one of its qualities
does for no physical item can be just seen. (Ibid.: 537)

Consequently, perceptions always represent material objects as being certain ways. They can do so either correctly or incorrectly: the perceptions
can agree or disagree with reality (328). 21 OShaughnessy thereby dis21

According to OShaughnessys sense-data theory, perception has two kinds of object


(OShaughnessy (2000): 16ff.). The immediate objects of perception are sensuous
and mental sense-data (or sensations), to which perceptual experiences are essentially extensionally rather than intentionally directed (ibid.: 16). This means
that the experiences could not occur without the existence of these objects and do
not show indifference to (in)existence towards the latter (i.e., are committed to, and
require, their existence; ibid.: 16).
In addition, perceptual experiences are also intentionally directed at another kind
of object, namely physical objects constituting the external world. This second representational link is mediated by the first, that is, by the immediate mental objects of
perception, which figure as the unique potential representative of the physical in the
psychological realm (OShaughnessy (2000): 18). Moreover, the second link is intentional since misrepresentation can occur: the represented physical object need not
exist, and it need not possess the properties that it is represented as having.
For the current purposes, however, it is possible to ignore this aspect of
OShaughnessys account of perception. What is relevant here is solely that: (i)
physical objects in the world are typically causally responsible for the occurrence of
perceptions; (ii) the resulting perceptions are intentionally directed at these external
objects; and (iii) they can be and, under normal circumstances, are veridical.
These three elements of OShaughnessys view are independent of his commitment
to sense-data and their mediating role for intentional representation.
OShaughnessys prime example of a non-veridical visual experience or seemseeing, as he often says is the projection of a light-image of pink elephants onto
ones retina (ibid.: 350 and 479). According to his sense-data theory, this experience
is necessarily related to a particular mental object that is, a sensation or

OShaughnessys Arguments

225

tinguishes between internal and external objects of perception. The


internal object is what is part of the content of the perception: it is the object which appears to us in perception and which is the result of the interpretation of the given sense impressions (i.e., the configuration of colours
and shapes in the two-dimensional visual field) by the activity of attention
(ibid.: 557f.). Accordingly, the internal object is the intentional object as
it appears in perception (e.g., as a mountain). The external object, on the
other hand, is the perceived object, such as the relevant physical object in
the environment (e.g., the real mountain). If it exists, it is the object that is
perceived. Accordingly, the external object is whatever the perception
refers to (if anything). And if it exists, it is the intentional object as it really
is.22 Then, OShaughnessy thinks that it is essential to perceptual experiphenomenal reality (9; 16ff.). But it does not veridically represent ones real
environment, in particular because it is not reliably caused by pink elephants.
22
For OShaughnessys complex account of perception and, in particular, the relationship between internal and external objects of perception, see OShaughnessy
(2000): chapters 16 and 19f. (especially pp. 454ff., 466f., 479, 536ff., 549f., 559 and
569). It is important to note that he thinks that perceptual experiences may have
more than one external object, and that all perceptual experiences have at least one,
namely a sensation or sense-datum (ibid.: 454ff. and 479). Indeed, OShaughnessy
concludes that there are no perceptual hallucinations, but only perceptual illusions
(ibid.: 350f. and 479). And he believes that, in normal perception, there is a series of
causally connected perceived external objects: for example, I see Mt Blanc
through seeing its south side, its south side through seeing its south surface, its south
surface through seeing a patch of snow thereon, and so on (ibid.: 549). The list of
perceived items includes also the light emanating from the objects and the resulting
mental sense-datum (or sensation) (ibid.: 444, 457 and 569).
Furthermore, with respect to each of these external objects, distinct internal
objects are part of the content of the respective visual experience (OShaughnessy
(2000): 559), namely as different interpretations of the given visual field (ibid.:
457f.). This means that there is a single visual perception by means of which we
perceive simultaneously many different, but none the less closely linked items: we
perceive the mountain by perceiving its surface and, ultimately, by perceiving the
respective sensation, but we enjoy only one perceptual experience (ibid.: 466,
536f. and 549f.). This requires, for OShaughnessy, that the perceived objects have
to share the features which we perceive them to have. The only plausible candidates
for such features are secondary qualities (ibid.: ch. 19, especially p. 537).
However, we do not consciously perceive or notice all perceived objects: although
we perceive and notice the light, we do not experience our perception as of being of

226

The Epistemological Account

ences that they can be an instance of successful perception (and perhaps


also that it is their cognitive function to cognise reality; see also ibid.: 362):
[V]isual experience is aspirational in allowing for the possibility of its coinciding with its own visual prototype [...]. (Ibid.: 366)

OShaughnessy describes the natural veridicality of perceptions (in contrast to the at best accidental veridicality of hallucinations) as the natural
correspondence between the internal [object] and the outer object (ibid.:
566), or as a general matching correspondence between the internal object
[...] and the outer [...] material object (ibid.: 567), or as the fact that we
harbour internal objects which naturally so match up with the physical environment (ibid.: 568). This means in the light of what OShaughnessy
takes internal and external objects to be that he treats perceptions as
veridical just in case how they represent their intentional object to be reflects how it really is which presupposes, of course, that there is an external object at which the perception is intentionally directed. In the case
of veridical perception, there is a match between the content of the perceptual experience and the represented reality. Furthermore, the natural
veridicality of perceptions is the result of its reliable causation by the perceived object in the world (ibid.: 567):
Perception has the function [...] of generating knowledge of the physical
here-and-now. Now perception can discharge that function only if the internal objects of the perceptual experience tend in general to match the outer objects. (Ibid.: 565)
[I]n seeing the colour and contour of the light and thereby also of the object,
and in that way experiencing the visual appearance of that object, we are
dependent upon a reliability that is situated, not just within our perceptual
apparatus, but externally to our bodies in the environment. (Ibid.: 450)
light (ibid.: 454 and 457). We come to know that we perceive light only by means of
(philosophical) reflection (ibid.: 457). This might also explain why the perception of
light does not seem to involve a respective internal object (ibid.: 569). In any case,
it is important to note that the sense-datum (or sensation) is among the perceived
objects; and that its perception involves a content including a corresponding
internal object. This means that internal objects and sense-data should not be
mixed up with each other (ibid.: 467 and 569).

OShaughnessys Arguments

227

For OShaughnessy, the reliable link between the perceived object and our
visual experience which represents it as being a certain way is realised by
suitable causal links between the two items (ibid.: 451). What he presents
is thus a causal theory of perception which promises to tackle an important
issue for any such theory, namely to take due note of deviant and nondeviant physical causal chains both within and without the perceiver
(ibid.: 450).23 Thus, OShaughnessys general picture of perception (though
23

For OShaughnessys causal theory of perception and its reliability, see OShaughnessy (2000): chapters 16 and 20 (especially pp. 451ff., 537ff., 542ff., 549ff. and
565ff.). Whether his theory succeeds is another matter, which need not interest us
here. But an uncommented sketch of his thoughts may be helpful.
His general idea is that the reliable causation happens in four stages (ibid.: 542ff.
and, especially, 569). The first stage comprises the causal chain from the perceived
object and its respective perceivable parts (e.g., its surface) via light to the
stimulation of the retina. The second causal chain bridges the gap between the
physical and the mental by linking this stimulation to the resulting sensation (or
sense-datum). The third and fourth stages involve only mental causation: one leads
from the sensation to the visual experience in its most minimal form (i.e., as a
representation of a two-dimensional ordered coloured expanse), while the other
occurs within the visual experience and establishes the other internal objects (e.g.,
the surface and the side of the mountain, and finally the mountain itself).
Now, the first two causal chains are reliable because of their subjection to
respective laws. The suitability of the causal links in question is due to their
conformity to laws, or law-like regularities, which govern the interaction of light
with objects and the retina, the processes in the visual system and its connections to
the mental realm (ibid.: 451, 453, 537, 544 and 549ff.). In contrast, the other two
causal chains are reliable, not because of some nomic regularity (ibid.: 544), but (as
it seems to me) because the interpretative steps from one internal object to the
next are small enough to typically adequately reflect the corresponding links
between the perceived external objects (ibid.: 565ff.). The transition from the first
interpretation (concerned with a two-dimensional configuration of colours) to the
last (concerned with the perceived object: e.g., a mountain) is thereby a matter of
trial-and-error, always open to correction by experience or, as OShaughnessy puts
it, we see from experience (ibid.: 567).
That the nomic regularity holds only between the external object and the relevant
sensations is thereby a consequence of the minds ability (e.g., in the form of
associations, desires, emotions, etc.) to interfere with the causal link between the
sensations and the resulting experience (ibid.: 544 and 552). However, if such an
interference occurs, the resulting experience is not a perception any more (and thus

228

The Epistemological Account

not its details) seems to be a fairly standard causal theory (see, e.g., Lewis
(1980) for a similar view). In accordance with these considerations, the
cognitive prototype of sensory knowledge has to be characterised minimally in terms of the complex property of being a representation which is intentionally directed at an object, represents it correctly (and thereby refers
to it), and does so in a reliable manner.
One aspect of this cognitive constraint on perceptual states is that, if
something is to count as an exemplification of the perceptual form of
knowledge, it has to be due to an appropriate external causal determination. Perceptions that do not satisfy the condition of being constrained by
reality in this way cannot play the cognitive role of perceptions (e.g., of
constituting knowledge, or at least of grounding judgements or beliefs that
count as knowledge). More generally, perceptual experiences are subject to
the cognitive constraint of being suitably causally linked to external objects
and of correctly reflecting the nature of these objects (if this is not already
also not an experience in OShaughnessys terminology), but an imagining. This is
due to the fact that the mental episode in question is not caused by the external
object (via the sensations) in the normal way distinctive of perception (ibid.:
352f.): that is, by means of the influence of attention and sometimes also of
understanding which structures the sensations in conformity to their received
nature (ibid.: ch. 20). In other words, cases, in which the causal determination breaks
down completely, fail to be perceptual, given that OShaughnessy takes sensory
representations that lack the normal causes of perceptions to be perceptual
imaginings (ibid.: 352).
Consider the example of vision, in which sensations are identical with noticeable
aspects of the structured visual field (ibid.: 500 and 554). Abnormal causation of the
visual experience by a particular determination of the visual field would occur, if
this specific determination (say, involving a small red colour expanse) would bring
about that other mental phenomena (such as the thought about ones having cruelly
shed blood, and ones related feeling of guilt) let the whole visual field in ones
awareness to become red while in fact only parts of it really register redness (for
similar examples, see ibid.: 352f. and 478f.). That the perceptual character of an
experience depends, for OShaughnessy, on its causal origin in this way might none
the less suggest, however, that the nomic regularity holds also between the external
object and a veridical perception (given that it is indeed appropriately caused and a
perception) in particular since the different ways in which the visual field can be
organised or interpreted by the attention and the understanding are restricted by its
particular determination.

OShaughnessys Arguments

229

ensured by the intentional and the suitable causal link). They exemplify
their cognitive prototype (i.e., constitute successful perception) only if they
satisfy this requirement which means, the three demands of representationality, veridicality and reliability.
Now, moving on to intellectual knowledge as the cognitive prototype for
beliefs, it is first important to note that OShaughnessy takes belief to be
constitutive of intellectual knowledge. His picture is that, while some beliefs do not count as constituting knowledge, others do. Moreover, he takes
it to be the epistemological function of beliefs to constitute knowledge:
Belief is the essence of knowledge, it aspires to the condition of knowledge, and constitutes a core phenomenon which can find itself redescribed
as knowledge. (OShaughnessy (2000): 368)

It seems that whether a candidate belief constitutes knowledge or not is for


OShaughnessy primarily a matter of whether it is non-accidentally true
and reliably formed (or sustained). The first feature is necessary to distinguish beliefs that amount to knowledge from intellectual imaginings (or
propositional ones in OShaughnessys terminology) which are true by
mere chance. For instance, I may imagine that my friend is sitting in the
Opera Bastille and listening to a performance of Bergs Wozzeck because I
know that my friend is in Paris and that the opera is on there; while unbeknown to be, my supposition happens to be true (indeed, I might even believe that he hates operas and would never listen to one). In a similar vein,
OShaughnessy observes:
[P]ropositional imaginings can be true. [...] And yet it is clear that their truth
must be accidental. No propositional imagining can be a knowing. (Ibid.:
345)

Accordingly, beliefs (or other intellectual representations) have to be nonaccidentally true in order to count as knowledge. However, OShaughnessy
also believes that the ways in which we form (and sustain) beliefs typically
lead to knowledge: that is, they are reliable. And although he is not explicit
about this, it seems reasonable to assume that this reliability ensures that
most true beliefs are non-arbitrarily true. Interestingly, OShaughnessy sees

230

The Epistemological Account

the reliability of our ways of forming beliefs to be an essential part of waking consciousness (i.e., the state of wakefulness). According to him, what
distinguishes people who are awake from those who are asleep, in a trance,
or unconscious (i.e., knocked out or anaesthetised) is precisely that only
the former are in direct contact with reality (see, in general, ibid.: 1ff., 68ff.
and 113ff.):
To be in the state consciousness is to be in the experiential condition of being aware of the World. As we say of the conscious, they are in touch with
Reality as those lost in a trance or dream are not. (Ibid.: 1)
[C]onsciousness is not a mode of epistemological success as if the world
was an object. Rather, consciousness is correct epistemological posture on
the part of an experiencing subject. (Ibid.: 117)

For OShaughnessy, being awake is thus necessary for coming into epistemic contact with reality and acquiring knowledge about it by means of
perceptual or similar experiences (see also ibid.: 119). Moreover, wakefulness involves the possession of the usually successfully employed capacity
to form beliefs in a reliable way:
[I]f consciousness is to exist [...] a cognitive sensitivity to perceptual
experience must inhere, and of the right kind. This cognitive sensitivity, and
the mode of belief-formation concerning the environmental realities, are I
suggest intrinsic to consciousness, which requires in general that beliefs be
well-formed [...] indeed, should normally lead their owner to Reality. Thus,
in the conscious the mode of belief-formation out of veridical perception
should be such as normally to make knowledge of that belief. (Ibid.: 85)

Hence, he assumes (given that we are usually awake) that our ways of
forming beliefs reliably lead to knowledge; and, furthermore, that typically
most of our beliefs constitute knowledge (ibid.: 157).24 For OShaughnessy,
the feature of wakefulness responsible for the reliability of our belief
formations is rationality. Thus, waking people are said by him to differ
24

OShaughnessy allows for the possibility of wide-spread error, but assumes that a
system of mostly or entirely false beliefs would be open to correction by experience
(OShaughnessy (2000): 157).

OShaughnessys Arguments

231

from dreaming or hypnotised ones with respect to the presence of a rational


integration of their occurrent mental episodes with their underlying picture
of the world. And among waking subjects, normal and sane people differ
again from people with unusual psychological conditions with respect to
the degree of rational coherence among their cognitions (ibid.: 86; see also
sec. 3.4). For OShaughnessy, the rational consistency with other beliefs
and with the occurrent perceptions and judgements is therefore part of
what turns beliefs into knowledge (ibid.: 142ff. and 157f.). Here is how he
describes the way in which we may come to acquire new knowledge about
the world on the basis of our actual veridical perceptions:
[A belief-system will] unlock the veridical incoming data, whereupon the
Physical World will stand forth in all its glory [...] when the [belief-system]
is for the most part a veridical [knowledge-system] which is internally rationally consistent, when sense-intake is veridical, and the experiencing
subject in a state such that the principle of admission for novel members of
the [belief-system] which is operative at the moment is that the putative
novel member is rationally consistent with incoming data and the prevailing
[belief-system]. [...] [R]eason plays a decisive part on the fulfilment of epistemological function [...]. [...] [R]eason is the unique guide to [truth].
(Ibid.: 157)

According to this quote, when we veridically perceive something, the respective belief will enter our net of beliefs about the world only if it rationally coheres both with the already existing and largely true beliefs and with
the veridical perceptions on the basis of which it is formed. But the quote
also specifies that, if the new belief satisfies these conditions and becomes
a part of our picture of the world, it is likely to amount to knowledge itself
(to unlock the veridical information contained in the perceptions). Hence,
OShaughnessy presents the following minimal condition for beliefs to
constitute knowledge (whether they are newly formed or already existing):
they have to be rationally consistent both with other beliefs and judgements (most of which should be true) and with any occurrent perceptions
(again, most or all of which should be veridical). In addition, OShaughnessy endorses the view that beliefs are intentionally directed at their objects and can be either true or false; and he requires knowledge-constituting beliefs to be true themselves (ibid.: 19, 84ff. and 328). Therefore,

232

The Epistemological Account

OShaughnessy seems to accept a fairly orthodox view of intellectual


knowledge: namely that intentional beliefs constitute knowledge only if
they are true and rationally linked in the right way to other (largely veridical) beliefs and perceptions. For him, any belief has to satisfy this cognitive
constraint in order to exemplify its cognitive ideal (i.e., in order to be intellectual knowledge).
To sum up, both perceptions and beliefs are subject to cognitive constraints that are structurally similar in that both demand representationality,
correctness and reliability. But since the two kinds of representation are
linked to reality in different ways, their reliability is established in different
ways: for perceptions, by the appropriate causal origin in the world; for beliefs, by the appropriate rational connection to other mental representations. None the less, the satisfaction of the three demands and hence of
the cognitive constraints are necessary for the respective representations
to count as cognising the world.

(ii) Wakefulness as a Further Constraint


This raises the question of whether OShaughnessy thinks that intentionality, veridicality and reliability (or rational consistency in the case of judgements and beliefs) are also enough for the provision of knowledge. 25 In
25

With the general qualification that, say, Gettier-style examples and similar cases still
have to be satisfactorily dealt with. Besides, it should be clear that reliability in conjunction with some form of correct indication would not be sufficient for knowledge,
at least not in the full sense intended by OShaughnessy. If one would by chance be
inclined to visualise a pink elephant whenever one would begin to suffer from an ear
infection, and if one would come to know about this correlation, one could use the
occurrence of such a visual image as an indicator for a beginning ear infection. But
the image of the elephant is not intentionally directed at the ear. In particular, it does
not represent it as being a certain way (i.e., as infected), with respect to which it
could be either correct or incorrect. For the hypothesised correlation between the
image and the infection does not allow for misrepresentation: if there is an infection,
the resulting image indicates that this is the case; but if there is no infection, the
image does not indicate anything at all (except, of course, if it is correlated to some
other fact).
The underlying reason for this is, presumably, that the correlation between the
image and the infection is not linked to some kind of function of the respective

OShaughnessys Arguments

233

particular, it is still left open whether his account of epistemically sound


perceptions and knowledge-constituting judgements or beliefs is really limited to the traditional conception of sensory and intellectual forms of
knowledge mentioned. In this section, I aim to illustrate that OShaughnessys claims about perception and knowledge reach beyond the traditional conception, according to which knowledge can be analysed in terms of
intentionality, veridicality and reliability (or justification) alone. For, as I
have already mentioned, he maintains that knowledge of the external world
requires the establishment of a contact with reality; and that the latter requires, again, wakefulness. But, for OShaughnessy, wakefulness is irreducible to the features traditionally employed in the analysis of knowledge,
notably rational consistency among ones mental states. Accordingly, he
does not take the satisfaction of the three requirements of representationality, correctness and reliability to be sufficient for knowledge; what is also
needed is wakefulness.
It is not necessary to engage here with all details of OShaughnessys
account of what it means for someone to be awake. 26 Instead, it should suffice to identify two aspects of wakefulness which are said by him to be necessary for the establishment of a contact with reality, but which cannot, it
appears, be traced back to the traditional ingredients in the analysis of
knowledge.
images to represent ear infections. Given that the images are not normally or ideally
meant to track ear infections, they cannot be said to fail to do so if they occur
without any infection present (see Dretske 1994). In OShaughnessys words, there
would be no internal object corresponding to the external ear infection, and hence
no general matching correspondence between the [two] (OShaughnessy (2000):
567), and no possibility of disagreement with Reality in stark contrast to the
cases of perception and belief (ibid.: 328). Perhaps, such an indicatory correlation
may lead to some other kind of knowledge (see Dretske (1986) on natural signs).
But it would not enjoy the rich nature of the two cognitive prototypes which
OShaughnessy has presented; and it would not be subject to similarly demanding
constraints.
26
At some points, he suggests that the presence of a belief system, of some conscious
episodes and of rationality are both sufficient and necessary for wakefulness (e.g.,
OShaughnessy (2000): 152 and 167). At others, he stresses also the importance of
self-knowledge (or insight) and practical self-determination (e.g., ibid.: 107f., 121
and 142).

234

The Epistemological Account

The first is that wakefulness requires the presence of a stream of consciousness (or of experience; see OShaughnessy (2000): 5, 82f., 123 and
142). It seems plausible enough to maintain that a person not enjoying any
mental episodes should not count as being awake. As OShaughnessy
points out, it seems that only completely unconscious people (such as those
who are anaesthetised or in a coma) lack phenomenal consciousness altogether; and they are clearly not awake. On the other hand, he illustrates the
fact that the presence of phenomenal consciousness is not sufficient for
wakefulness by reference to dreaming and hypnotised people who experience phenomenally conscious states as well (ibid.: 119 and 123).
OShaughnessys idea that epistemic contact with reality presupposes
wakefulness and therefore phenomenal consciousness should not be taken
to imply that unconscious people cannot have any (knowledge-constituting) memories or beliefs. This would be an absurd view, given that we usually do not lose our memories and beliefs as well as our knowledge
during periods of unconsciousness. Rather, it should be understood as
meaning that unconscious people have no conscious access to their cognising memories or beliefs and hence cannot actually use them (e.g., by
manifesting them in consciousness) to get into contact with reality.27
The second relevant aspect of wakefulness is that it involves what
OShaughnessy calls awareness of the outer world. 28 He distinguishes
this form of awareness from perceptual awareness of reality. In particular,
he takes it to be present even when a waking subject has no actual perceptions or similar episodic representations of his environment. He takes it
thus to be prior to, and independent of, any actual perceptual link to reality
(see also ibid.: 117 and 152):
Strictly speaking, this seems to put a limitation on OShaughnessys account of
knowledge. For wakefulness seems to be a necessary condition only on available
knowledge (i.e., knowledge that can be used to get into contact with reality), but not
on knowledge in general at least not if one wants to allow that temporarily inaccessible memories or beliefs may none the less continue to constitute instances of
knowledge. But it is not clear whether OShaughnessy really has such a limitation in
mind.
28
He also speaks of realistic, linked or connective awareness (OShaughnessy
(2000): 119, 122 and 155f.); of w-consciousness (ibid.: 117); and of being in
touch or contact with Reality (ibid.: 1 and 124).
27

OShaughnessys Arguments

235

[W]aking consciousness is consistent with the perceptual awareness of absolutely nothing in the (outer sector of) (the real) world. None the less,
such a perceptually empty consciousness, since it is awareness of the world,
continually orients to the world: it carries the real world with it all the time,
as the framework within which to site anything it happens perceptually to
encounter. (Ibid.: 119)
Now awareness of the outer world does not mean awareness of the present
facts of the environment. (Ibid.: 152)
[Awareness of the outer world] is not the perception of anything, and in particular is not the perception of the contents of [reality]. Rather, it precedes,
and enables where possible, the perceptual and cognitive reading of the
data indicative of those contents. (Ibid.: 155)

None the less, as these passages suggest, OShaughnessy assumes that


there is an important link between awareness of the outer world and perception. His idea is that the former facilitates perceptual contact with concrete aspects of the world (awareness of the present state of the world;
see ibid.: 86, 118 and 121) and any subsequent acquisition of knowledge
(see ibid.: 84ff., 122, 142ff. and 156f.). As he writes, it is this aspect of
wakefulness that enables us to be in a state [of mind] apt-for closing the
epistemological gap between mind and world (ibid.: 152). OShaughnessys view seems to be accordingly that wakefulness puts us into a suitable epistemic position thereby enabling us to get cognitively in touch
with the world primarily because it involves awareness of the outer
world (see also ibid.: 86).
His primary motivation for introducing an additional element in the establishment of a perceptual link to reality is that he thinks that there is a
particular epistemological gap between mind and world which cannot be
bridged even by non-arbitrarily (or reliably) veridical perception:
The perception of present physical realities as what they are [...] that nonaccidentally identifies its object and apprises one of its existence [...] and
manages veridically to site its object in a continuous sector of spatio-temporal reality which in turn it also perceives and identifies, is still insufficient. [...] [S]omething more [is required] if we are to be credited with contact with the outer world, and with perceptual experience which reflects that

236

The Epistemological Account


contact. (Ibid.: 123f.)

The main reason why OShaughnessy assumes that non-arbitrarily veridical perception is not sufficient for epistemic contact with reality is that
somnambulists or hypnotised people and, to a lesser extent, dreaming
subjects may enjoy such perceptual experiences, or at least very similar
ones, without thereby being in a position to acquire knowledge about reality (see ibid.: 118ff., especially 123f.). His view is that such people may reliably and veridically perceive the objects in their environment (e.g., where
they are located, or which shape and size they have), but none the less are
bound to fail to form knowledge-constituting judgements about them. This
failure may manifest itself in various ways: in the fact that they do not correctly interpret what they perceive (e.g., they may take the table in front of
them to be an automobile because the hypnotist says so (ibid.: 122f.)); in
the fact that they do not take the perceived objects to be part of reality (but
instead, say, of some unreal world which is dreamt, or imagined in response to the instructions of the hypnotist (ibid.: 120 and 123)); or in the
fact that they do not properly connect the perceived objects to the rest of
reality (e.g., in the case of a somnambulist or a hypnotised person who can
cognise only small and disconnected parts or regions of reality (ibid.: 120
and 122)).
In contrast, waking people can as long as they are not in a psychologically unusual or abnormal state of mind (e.g., when intoxicated or mentally disturbed) avoid these failures and subsequently acquire knowledge
about reality by means of perception. They then enjoy what OShaughnessy calls realistic perception or awareness: the perceptually based contact with reality which locates its particular objects in reality (ibid.: 119,
122, 124 and 155), links them to other parts and objects of reality (ibid.:
122, 124 and 156), and enables us to acquire knowledge about them under
many different interpretations or conceptualisations (ibid.: 124 and 156).
It is hence not surprising that OShaughnessy concludes that there has to
be an intimate link between wakefulness and perceptually grounded knowledge about reality. He observes that what is missing in all the cases of
dreaming or trance and what is present in the case of wakefulness is the
establishment of a (sufficient) rational consistency between how things are
perceived to be and how they are believed to be. Accordingly, he takes the

OShaughnessys Arguments

237

location of perceived objects in reality, the interpretation of them in the


light of what one believes and the discovery of their relationships to other
parts of reality to be a matter of rationality: they occur due to the rational
integration of ones perceptions into ones already existing picture of the
world (ibid.: 119, 124 and 152). Here is how he describes the necessary
and sufficient conditions for a full and proper awareness of the outer
world that is, for a realistic and hence successful epistemic contact
with the world on the basis of perception:
Then a full and proper awareness of the outer world will typically be dependent firstly upon a [belief-system] [...] pervaded by the property of rational consistency.
And secondly [...] occurrently the subject finds himself in such a state
that his occurrent cognitive attitudes are rationally determined by the relevant parts of the above belief-system [...]. If all of this is fulfilled then a full
and proper awareness of the outer world, and so also a wakeful consciousness, must obtain [...].
Here we have a statement of the conditions needed for a full and proper
awareness of the outer world. (Ibid.: 152)

Now, the specific aspect of wakefulness which is responsible for the difference in the epistemic position between waking and non-waking people is,
of course, identified by OShaughnessy as the involvement of awareness
of the outer world: it is this aspect which equips occurrent perceptions
with the potential to lead to knowledge (ibid.: 119 and 155f.).
In accordance with the preceding considerations, OShaughnessy understands the property of being aware of the outer world as involving three
different things. First, it involves having a sufficiently rich and rational
system of background beliefs about the world which we can readily apply
to what we perceive. It is in this sense that the form of awareness under
discussion is awareness of the world [and] carries the real world with it all
the time, as the framework within which to site anything it happens perceptually to encounter (ibid.: 119; see also 156). Consequently, if we perceive
something while being awake, we perceive it as part of the reality and are
disposed to describe the perceived entity in accordance with our picture of
the world. As OShaughnessy writes, the perceived item brings an entire

238

The Epistemological Account

world with it, as mental back-drop (ibid.: 156). 29 Second, to be aware of


the outer world also means to be rational in certain ways: namely to be rationally responsive to newly occurring perceptions or similar experiences
of the world; and to possess the capacity to establish rational consistency
between them and ones underlying picture of the world (ibid.: 119 and
152). Third, being aware of the outer world likewise involves striving for
contact with reality and, hence, being inclined to actually use the capacity
for rational integration in the concrete case of an occurring perception. At
least this is what OShaughnessy seems to have in mind when he describes
a waking person whose stream of consciousness is devoid of all perceptual
and similar representation of his environment, but whose mind is none the
less aware of the outer world and strives to establish epistemic contact
with it:
A man could suddenly surface from deep unconsciousness to an alert wideawakeness in which all is black and silent and devoid of feel of any kind,
and his mind at that instant be straining beyond itself in an effort to epistemologically make contact with its surroundings. Is he not conscious at
that moment of the presence of the reality lying beyond his mind? (As a
dreamer is not.) (Ibid.: 152)

If, however, this awareness of the outer world which pertains to the waking mind as a whole (i.e., this set of underlying beliefs, capacities, inclinations and enabling conditions) is indeed accompanied by an actual perceptual experience, it may and usually will come to pertain to that particular instance of perception as well. According to OShaughnessy, the result
will be full and proper awareness of the outer world or, in other words,
29

What seems to be assumed here by OShaughnessy is that perceptions involve not


only a phenomenologically salient claim about how things are (i.e., a cognitive attitude), but also a phenomenologically salient claim about how things are in reality. In
other words, he seems to presuppose that perceptions present their intentional objects as real (see also OShaughnessy (2000): 155). Moreover, this aspect of the conscious perceptual contact with reality appears for OShaughnessy to be due to the rational integration of the respective perceptions with our picture of the world. That
perceptions locate their objects in reality thus seems to mean that we take these objects to be part of reality in the light of our beliefs about the world (ibid.: 124 and
155f.).

OShaughnessys Arguments

239

perceptually grounded knowledge of reality. By contrast, the capacity and


tendency to access ones beliefs in order to rationally integrate ones perceptions with them is missing in people who are not awake. Hence, they
fail with respect to the acquisition of perceptually based knowledge. It
should now be clear why OShaughnessy highlights the importance of
wakefulness in general and awareness of the outer world in particular
for the establishment of epistemic contact with reality. For him, wakefulness involves awareness of the outer world, which again makes possible
and is typically accompanied by a sufficient rational integration of what
one perceives and believes. And this rational coherence enables us to
bridge the epistemological gap between mind and world and acquire knowledge on the basis of perception.
However, OShaughnessys emphasis of the link between knowledge
and wakefulness has important consequences for the cognitive constraints
on perceptions and beliefs. The general upshot is, roughly, that, for him,
the two kinds of state can provide knowledge only in close connection with
each other: while beliefs appear to constitute (available) knowledge only if
they are part of a waking and sufficiently rationally consistent mind which
includes (or at least can include) conscious perceptual or similar experiences of the world, perceptions seem to constitute or ground knowledge
only if they are rationally integrated with the underlying system of beliefs.
But that perceptions and beliefs can cognise reality only in cooperation entails further conditions on the epistemological soundness of perceptions
and beliefs.
With respect to perceptual grounds of knowledge, consider again his
quote above on the insufficiency of reliability to bridge the epistemological
gap between reality and perception indicates (ibid.: 123f.). It indicates that
OShaughnessy intends the requirement that perceptions have to be rationally integrated with ones picture of the world to be a condition on the epistemic appropriateness of perceptions and not merely a condition on the
epistemic appropriateness of beliefs acquired by means of perceptions.
Something more than non-arbitrary veridicality and reliability is required
if we are to be credited with with contact with the outer world, and with
perceptual experience which reflects that contact (ibid.: 124). Moreover,
as this passage also indicates, this further requirement on cognising percep-

240

The Epistemological Account

tions resists reduction to the requirement that they have to be reliable and
non-arbitrarily veridical at least as long as one is convinced by
OShaughnessys examples of somnambulists and hypnotised persons
which are meant to illustrate the respective difference in epistemic standing. For him, the subjects in these examples do not really perceive reality,
that is, are aware of objects as part of the real world, even if their perceptual experiences are reliably formed and non-accidentally veridical. Consequently, OShaughnessy takes rational consistency to be an additional aspect of the cognitive constraint linked to the cognitive prototype for perceptions.
His account thus reaches beyond the traditional picture, according to
which a perception is already epistemically sound if it is non-accidentally
veridical and reliable. The traditional view may very well include the claim
that our perceptual beliefs are epistemically appropriate only if they or
the respective perceptions rationally cohere to a sufficient degree with
our other beliefs; and that we endorse our perceptions in belief only if we
do not recognise them as not being rationally consistent with what we otherwise believe (i.e., only if there are no respective defeaters). But it is not
part of the orthodox position that the epistemic soundness of perceptions
depends also on their rational relations to our beliefs, and not merely on
how they are brought about. Indeed, it is traditionally controversial whether perceptions can stand in rational relations. 30 Therefore, OShaughnessys
theory of perceptual grounds of knowledge makes a substantial addition to
the customary view of the epistemic soundness of perceptions.
His position on the cognitive demands on beliefs is also more complex
than the orthodox approach, mainly because the latter usually does not address the issue of the importance of wakefulness (e.g., in connection with
Cartesian dream arguments). The traditional picture requires cognising be30

Coherentists, for instance, have responded to traditional foundationalism with skepticism about the idea that perceptions can enter into rational relations and hence
provide reasons for beliefs (see, for instance, Davidson (1989)). In recent years,
however, it has become more common to endorse some kind of modest foundationalism and to assume that perceptions possess rational status as reason-givers (see
Martin (1992); Pollock & Cruz (1999): chs. 4f.; Pryor (2001)) although sometimes
only in conjunction with the claim that the content of perceptions is conceptual (see
McDowell (1994)).

OShaughnessys Arguments

241

liefs to be true and justified (plus perhaps to satisfy some further conditions
in order to rule out Gettier-style cases). This may involve that they have to
be rationally consistent with other beliefs and ones current perceptions and
be rationally responsive to ones potential perceptions or judgements. But
OShaughnessy demands in addition that the believing subject is awake,
that is, in a position to have perceptual experiences of the right kind
namely, perceptual experiences which become rationally integrated with
ones background beliefs and which hence indeed present objects as being
located in the real world. This is one aspect of what he means by claiming
that beliefs (and, presumably, also memories) can provide knowledge about
the external world only in conjunction with, and because of their link to,
actual or potential perceptions. The other aspect is the empiricist claim that
all kinds of external knowledge have to be ultimately based on perceptual
access to reality a claim which, again, is not necessarily part of the orthodox view on knowledge. Accordingly, OShaughnessy understands the
cognitive constraint on beliefs as involving two requirements over and
above the traditional demand for their rational integration with ones other
cognitive states: they have to occur in a waking subject; and they have to
be ultimately based on perceptual access to reality. In fact, the two requirements appear to be closely related, given that OShaughnessy assumes that
only waking people can get into perceptual contact with the world.
None the less, his view on knowledge-constituting beliefs is not inconsistent with the traditional analysis. OShaughnessys requirement that cognising beliefs about the world have to be generally linked to wakefulness
and, hence, to the presence of phenomenal consciousness and perceptual
experience can be easily accommodated, given that it is satisfied by the beliefs of normal cognising subjects (with the exception of the states of
knowledge or memory which are part of temporarily unconscious or nonwaking minds and which constitute available knowledge during periods of
wakefulness). The orthodox picture is also to compatible with the empiricist idea that all knowledge about the external world is ultimately grounded
in perception. Indeed, it does not seem to be unreasonable to assume that
proponents of the traditional view sometimes implicitly assume that cognising subjects have to satisfy these or very similar conditions, and that
normal waking people typically do satisfy them.

242

The Epistemological Account

To conclude, OShaughnessy adds two further requirements to the traditional account of the cognitive constraints on perceptions and beliefs in
terms of intentionality, veridicality and reliability (as well as rational consistency in the case of beliefs). The first is that both perceptions and beliefs
and, presumably, also other kinds of cognitive states, such as memories
have to be rationally integrated with ones picture of the world. The second
additional demand is that all kinds of knowledge are ultimately based on
perceptual contact with reality. Both of OShaughnessys further requirements have in common that they can be satisfied only by cognitions which
occur in, or are part of, a waking mind.

(iii) The Origin of Imaginings


After clarifying OShaughnessys view on how the realisation of the two
forms of knowledge is inseparably linked to the satisfaction of certain cognitive constraints and the conditions pertaining to wakefulness, I move on
to the second and main part of the Argument from Origin: OShaughnessys defense of the claim that imaginative representations cannot satisfy
these cognitive constraints. As already suggested, he takes this fact about
imaginings to be the consequence of their specific origins in the mind:
All imaginings arise from the subjects mind in such a way that the constraint of Reality [i.e., the relevant cognitive constraint FD] is necessarily
inoperative, whether through substituting ones will for Reality or through
confusing subjective Reality with Reality itself. The fact that the mind acts
here, not as a representative of Reality but in direct opposition, guarantees
that imaginings must be cognitively void. By contrast, while a mind reasoning on empirical matters is a causal force in the engendering of its own cognitive attitudes, the rationality of the process ensures that mind and Reality
act here in consort. It is precisely not so in imagining. Here the mind operates genetically in such a way that the mental products are guaranteed not to
realise their cognitive prototype, and Reality is simply short-circuited out of
the causal transaction. (Ibid.: 359)

What seems most important in this quote is that OShaughnessy emphasises the causal role of reality: the suitable causal determination of mental
representations by the relevant aspects of reality either in a direct way (as

OShaughnessys Arguments

243

in the case of successful perception), or mediated by evidential representations and rational mechanisms (as in the case of intellectual knowledge)
is crucial to the acquisition of knowledge. The problem with imaginings is
thus not that they are causally determined by the mind (as OShaughnessy
notes, the mind is causally involved in the formation of cognitions as well);
but rather that they lack any suitable causal link to reality. In other words,
all significant causal impact on imaginings (e.g., determining what they
represent) comes from the mind.
Now, the quote also suggests that this may happen in two ways: either
the will takes over the role of reality in determining the representations; or
the representations occur passively due to some other causal factors in the
mind and are (perhaps because of sharing this passivity with some cognitions) to some extent mistaken for their cognitive counterparts. But let me
discuss the two ways in which imaginings can arise from the subjects
mind in a bit more detail.
On the one hand, the formation of an imagining may be due to activities
of the will, which is directed (e.g., by means of an intention) at producing
an imaginative representation. In other words, the will is the active starting-point of the process of bringing about the imagining, and the result is
the substitution of ones will for Reality (ibid.: 359). The products of this
process are, at least to some extent, subject to the will. Even though we
sometimes cannot influence their occurrence, we have at least control over
their persistence. Furthermore, the resulting imaginings are selfconscious
in the sense that we are aware of, or experience, them as imaginings and as
presenting their objects not as real (i.e., as not making a claim about how
things are; see ibid.: 358f.), and consequently do not trust them or let them
have an impact on our view of the real world. OShaughnessys prime examples of will-susceptible imaginings are engagement in make-believe and
active visualising (ibid.: 343f., 346 and 351ff.).
On the other hand, the formation of an imagining may involve a weakening of ones sense of reality (ibid.: 352). What he seems to have in mind
here is that we come to be unsure or disturbed (ibid.: 352) about what is
real; and that we may even begin to mistake things, which are not part of
reality and normally recognisable as such (e.g., imagined or hallucinated
entities), for parts of reality hence the weakening of our sensitivity (see

244

The Epistemological Account

Dorsch (2010a)). Although OShaughnessy is not very explicit about the


origins of imaginings of this second kind, he seems to think that they are
not due to the will, but at least in the case of sensory hallucinations to
an event in the mind which involves, or effects, the absence or breakdown
of the suitable causal processes which, if undisturbed, would lead to cognitive states (see OShaughnessy (2000): 352f.). This interpretation is specifically in line with his further claim that the resulting imaginings (including intellectual ones) are impervious to the will (see ibid.: 340f.). Hence,
the imaginings involving a weakening of ones sense of reality seem in
general to be due to the causally induced absence or breakdown of the suitable causal chains coming from reality which typically result in the formation of cognising representations. The will has no role to play in bringing
them about or sustaining them.
The involuntariness of these imaginings and the involved weakening
of ones sensitivity concerning what is real are closely related to the fact
that such imaginings are at least to some extent unselfconscious: they do
not fully involve the awareness of what is imagined as unreal (ibid.: 352f.
and 358f.).31 As a consequence, the subject in question can distinguish her
imaginative representation from its cognitive counterpart less clearly than
in the case of actively formed imaginings. This unclarity comes in degrees,
as his example of three different kinds of passive visual hallucinations illustrates:
[V]isual hallucination [...] can be conveniently grouped into those
hallucinations experienced with belief (such as Macbeths hallucination of
Banquo), those which leave one in doubt (like Macbeths hallucination of a
dagger), and those one knows to be illusory (say, in the first stages of mescalin [sic] intoxication). (Ibid.: 349)
[T]o the extent to which self-awareness is absent, to that same extent imaginings are experienced, not as imaginings, but as what is being imagined,
namely as the cognitive prototype, given as directed to reals. (Ibid.: 359)

Hence, although in many cases, the subject still notices a phenomenal dif31

It is not always clear what is presented as unreal. The passages referred here seem
to concern the represented external entities, while other passages seem to concern
the directly imagined cognitive prototypes (see OShaughnessy (2000): 349).

OShaughnessys Arguments

245

ference between her imaginings and her perceptions, beliefs or judgements,


there are extreme cases in which she cannot any more distinguish her imaginings from her cognitions. His main examples for imaginings of this extreme kind are certain visual hallucinations (such as Macbeths vision of
Banquo; see ibid.: 341, 345. and 349) and, perhaps, certain dream experiences (ibid.: 344f.).32 OShaughnessys view on these examples implies, in
particular, that it must be possible that we sometimes experience our imaginings as possessing a cognitive attitude (i.e., as making a claim about the
truth or reality of the imagined content). Otherwise, we (or Macbeth) could
not fail to subjectively distinguish them from perceptions. And we (or
Macbeth) would not as OShaughnessy claims endorse those imaginative hallucinations in belief and action, without the endorsed representations
thereby losing their imaginative character (ibid.: 349 and 352f.).
Of course, this last element of OShaughnessys account is in contradiction with my initial assumption that there is always a phenomenologically
salient difference in attitude between imaginings and cognitions. But the
examples he has provided visual hallucination indistinguishable from
perception and, perhaps, similar dream experiences are not central cases
of imagining: their classification as imaginative is highly controversial (see
Sections 1.3, 12.4 and 14.4). In particular, it is natural to treat visual hallucinations, that are subjectively indistinguishable from perceptions, instead
as perceptual experiences though perhaps as perceptual experiences of a
different kind than successful perceptions (see Martin (1994) and (2002);
as well as Dorsch (2010c) and (2012a), and Soldati & Dorsch (2011) for
discussion). Moreover, as I argue in the next chapter, OShaughnessys
main reason for treating these examples as imaginative their lack of the
capacity to cognise reality (due to their specific origin) is not a good
reason: it does not capture the nature of imaginings. Hence, it is reasonable
to take these extreme examples to be genuine cases of cognition, rather
than imagination.
Now, OShaughnessy is not very explicit on why exactly the origin of
imaginings in the mind in either of the two ways specified implies that
their formation and representationality cannot be cognitively constrained
by how the real world is. But it is plausible to extrapolate the following
32

His intellectual examples are the imaginative beliefs discussed in Note 15 in


Chapter 6 above.

246

The Epistemological Account

ideas from his premisses and conclusions and, notably, his comments on
the nature of the cognitive constraints on perceptions and beliefs.
According to OShaughnessy, these constraints consist at least partly in
the requirement that certain epistemically sound processes of formation
as mediated by suitable causal (and perhaps also rational) links precede
the occurrence of the representations in question. Will-impervious imaginings cannot satisfy the respective cognitive constraints because they emanate from the breakdown of exactly these processes. In the case of passive
sensory imaginings (such as hallucinations), the normal causal determination by external entities does not obtain; and in the case of passive intellectual imaginings (such as dream-beliefs), the same is true of the rational impact of our cognitive representations of reality. Consequently, neither kind
of will-impervious imagining can realise a cognitive prototype.
Similar considerations apply to imaginings originating in the will.
OShaughnessys idea is here that the causal determination of the imaginative representations by the will, which is already sufficient for their occurrence or persistence, prevents them from being causally (and rationally)
determined in the appropriate ways by either aspects of the actual world or
evidential representations about it. That is, representations cannot derive
both from the will and from reality in the fashion required for the provision
of knowledge: with respect to will-induced imaginings, the epistemic processes involved in the normal formation of perceptions or beliefs are
simply not involved (see also Dorsch (2009b)).
Considerations like these let OShaughnessy form the more general conclusion that both ways, in which imaginings may occur, rule out the possibility of the satisfaction of the respective cognitive constraints. The issue is
thereby not that imaginings cannot be representational or veridical. As
already mentioned, OShaughnessy assumes that they can be both
(OShaughnessy (2000): 166f., 345 and 363). Similarly, the worry is not
that imaginings cannot occur in a waking mind; they obviously can. Instead, the problem is that imaginings are formed in ways which cannot ensure that they are reliable, rationally integrated with our picture of the
world, and ultimately based on our past or current perceptual experiences.
But since it is essential to instances of the two forms of knowledge that
they do satisfy those constraints, OShaughnessy can infer his weak nega-

OShaughnessys Arguments

247

tion claim: imaginings cannot be such instances.33


Although OShaughnessy puts so much emphasis on the ways in which
imaginings can be produced, he still does not take their origin to be constitutive of them. For him, imaginings are imaginings because they satisfy
his definition (i.e., his versions of (NC*) and (ER)), and not because they
arise from the mind in specific ways. His two reasons for this choice of
view are, first, the impossibility to unify the various origins which imaginings can have and, in particular, to bridge the division between voluntary
and involuntary cases (ibid.: sec. 11.5f.); and, second, the fact that some
imaginings are imaginative (i.e., satisfy his definition), not only in virtue
of their origin, but also in virtue of their constitution (ibid.: 359f.). None
the less, the specific origins of imaginings explain, in OShaughnessys
view, why his weak negation claim is true of imaginative episodes: that is,
why they are bound to fail to exemplify the two cognitive prototypes.

7.2. The Argument from Attention


OShaughnessys other argument for his weak negation claim which applies solely to sensory imaginings is, in comparison, short and simple
(see OShaughnessy (2000): 355f.). It supports the conclusion that there is
a difference in constitution or kind between perceptions and sensory imaginings by arguing for the claim that only perceptions involve attention and
therefore allow for the distinction between what is present to consciousness
in the perceptual field, on the one hand, and what is actually noticed by the
perceiver, on the other hand. One of his examples is the visual perception
of a host of golden daffodils (e.g., on a field): while we typically do not
33

To ensure that his version of (NC*) can make a substantial contribution to a unified
account of imagining, OShaughnessy has to presuppose that there are no entirely
voluntary cognitions and, especially, no purely actively formed beliefs or judgements. If cognitions could be determined exclusively by the will, instead of at least
partly by reality or evidence about it, they would not differ in this respect from willinduced imaginings and hence would be subject to OShaughnessys Argument from
Origin as well. In this respect, OShaughnessys version of the Epistemological Account is in agreement with the Agency Account to be defended in Part Four. I
provide an independent argument for the involuntariness of judgements and beliefs
in Dorsch (2009b).

248

The Epistemological Account

notice each of the daffodils as such, but only their totality, they are all none
the less part of the visual content. In such a case, we are aware of each
daffodil in one sense, but not in another: we see them, but we do not notice
them. The difference consists in the fact that we notice things that is, see
them only if we specifically attend to them. Noticing thus requires more
than mere seeing: it necessarily involves attention (see also Dretske (1979)
and Martin (1994)).
But why should we think that we are perceptually aware of all the daffodils, although we are not attentively aware of (most of) them? OShaughnessy presents two reasons. The first is that what we notice is determined
by everything that is part of the content of the visual impression (see
OShaughnessy (2000): 501). In particular, we have to be aware of the
single daffodils in order to be able to notice their totality: we could not see
a host or field of daffodils without seeing each of them (ibid.: 356). The
second reason is that there are physical features, that determine which area
of the content of the visual impression shows which colour-value and
hence that enable us to test whether our visual impression in fact includes
information about each of the daffodils (ibid.: 355f.). Although he does not
say much more about the issue, the underlying idea seems to be that, in the
case of perception, the visual content is completely determined by nonpsychological, physical factors (e.g., the stimulation of the receptors in the
eyes and, presumably, certain areas of the visual cortex), that are part of the
normal causal link between external world and experience, and hence can
be investigated by scientific means. Hence, there is a distinction between
what we merely perceive and what we actually notice in the spotlight of
our attention.
A similar distinction, as OShaughnessy argues, cannot be drawn with
respect to sensory imaginings. His main reason for this claim is that the
presence of unnoticed, but none the less sensorily presented elements in the
content of the imaginative representations cannot be shown by means of a
scientific investigation (of, say, the visual system in the brain). For, due to
the absence or breakdown of the mechanisms that typically link perceptions to the external world, there are no determining physical causes to be
investigated. But if the distinction between being mere sensory awareness
and additional attentive awareness cannot be demonstrated, it should not be

OShaughnessys Arguments

249

assumed that sensory imagination involves attention. Hence, for


OShaughnessy, perceptions and sensory imaginings differ essentially:
while perceptual consciousness admits of attention and the distinction
between what we are merely aware of and what we notice, imaginative
consciousness does not. However, his argument in favour of this conclusion is flawed in two respects.
First, it is not clear whether OShaughnessys idea of a physical or scientific test for the colour-values of the content of visual impressions is
plausible in the light of neurophysiological evidence (let alone whether it
can be generalised to cover all kinds of perception and sensation).
Moreover, even if such a test turns out to be possible, there is no reason to
rule out the possibility that there are, after all, causal factors in the brain
that determine, in a similar way, the content of visual imaginings, and that
can be scientifically investigated as well. Of course, such causal factors
would presumably differ from those involved in perception, or at least
would concern a later stage in the general process leading to the generation
of visual representations. But it none the less ensures that OShaughnessys
argument that it is impossible to demonstrate that there are unnoticed, but
presented elements in the contents of sensory imaginings because certain
causal factors involved in the generation of perception play no role in determining sensory imaginings looses all its force.
Second, the idea that sensory imaginings involves attention, after all, is
supported by the fact that the other reason in favour of attentive perception
can be applied to the imagination as well. Just consider again the example
of me visualising the frontside of the Pantheon in Paris (see Section 4.6). 34
It is possible that I visualise its facade with a determinate number of
columns, without consciously specifying this number. The number may
thereby be determined by some associative, mnemonic or other passive
processes, or alternatively by some sub-intentional form of imaginative
mental agency (see Section 13.3). In both cases, my conscious control over
what I am visualising is limited, and I am unaware of the exact number of
columns that my imaginative episode visually represents the Pantheon as
34

Another good example are hypnagogic images. While falling asleep, visual images
may arise without our (conscious) active influence and surprise us with what they
represent, or allow us to attend to previously unnoticed details of the represented objects.

250

The Epistemological Account

having. But I may discover this number by focusing my attention to the respective part of the visual field, or to the imagined scene that it presents to
me. Accordingly, the sensory imagination is open to attention and discovery: we may come to notice previously unnoticed details of a visualised
scene by attending to them.
It might be insisted that, rather than discovering the number of columns,
we determine it by means of imaginative activity when trying to find
out how many columns we are visualising the Pantheon as having. That is,
it might be argued that we shift our attention from one imaginative project
(i.e., visualising the Pantheon with no determinate number of columns) to
another (i.e., visualising it with this determinate number), rather than from
one aspect of the visual field of our episode of visual imagination (e.g., the
stairs) to another (e.g., the columns and their number). As a consequence,
visualising would not, after all, involve the kind of attention that is characteristic of perception, and which enables us to notice previously unnoticed
things. It would just be able to occupy our attention in the same way as all
other mental episodes, including thoughts and mental actions.35
However, two things speak against this reply. First, the proposed alternative explanation of the role of attention in visualising cannot explain the
fact that our recognition of how many columns we are visualising may be
accompanied by a genuine feeling of surprise, or discovery. In such cases,
we find ourselves visualising a certain number of columns, without being
aware of actively determining this number. This is true even if the number
of imagined columns is determined by sub-intentional mental agency: just
as we can discover that we are actively scratching our back, we may find
ourselves visualising a particular number of columns. Second, visualising
shares with seeing the property of being transparent to the represented objects (see Section 2.1). But this just means that, if we try to shift our attention inwards to our episodes of visualising, we end up noticing the external
35

Hopkins (2011b) defends a view on visualising and attending that comes very close
to this reply. The distinction between the two kinds of attention pertaining to perceptual awareness and conscious thought and, notably, the fact that attending to perceptual experiences means, at least in part, attending outward (i.e., to some external objects), while attending to conscious thoughts does not is introduced in
Peacocke (1998), and discussed in more detail in Martin (1998), OBrien (2007): ch.
6, and Dorsch (2012a).

OShaughnessys Arguments

251

objects and features that we are visualising. In other words, visualising involves the same kind of attention as seeing and not the same kind of attention as, say, thinking. It is true that visualising lacks the immediacy of
seeing in the sense of not presenting those objects and features as part of
our actual environment (see Section 3.2). But the kind of attention involved in perception is linked to transparency, and not to immediacy.
These considerations do not show that perceptions and sensory imaginings could not still differ in kind with respect to the involvement of attention. But they demonstrate that OShaughnessy has not presented us with a
good reason to believe the opposite. We should therefore take the Argument from Origin to be his main motivation for endorsing the weak negation claim and, hence, (NC*) also because that argument is not limited to
any particular form of imagining. This concludes the expository part of my
discussion of OShaughnessys theory of the imagination. I now move on
to the assessment of this and other instances of the Epistemological Account.

CHAPTER 8
Critical Assessment

In this final chapter of Part Two, I investigate whether the Epistemological


Account and, in particular, OShaughnessys version of it can make a
substantial contribution to a unified account of imagining. More specifically, what I shall try to do is to undermine the plausibility of the idea of
providing a unified account of imagining in terms of negation theses that
characterise imaginings by reference to their non-cognitivity. My primary
aim is not to discredit these claims per se, but rather to question their relevance for the study of what is common and distinctive to instances of imagining. But I also intend to assess the truth and scope of each of the main
negation claims notably Sartres and Wittgensteins thesis (NI), as well as
the claims (NC) and (NC*) which OShaughnessys weak and strong negation claims are variants of in independence of their inclusion or function
in theories of imagining.
In accordance with the two desiderata outlined in Section 1.1, negation
claims may fail to make a substantial contribution to a unified account of
imagining for two reasons: first, they may be extensionally inadequate; and
second, they may be explanatorily unilluminating. I start with the explanatory weaknesses of different negation claims and move then on to query
their extensional adequacy, before closing with some diagnostic remarks.

254

The Epistemological Account

8.1. The Explanatory Power of Negation Claims


(i) General Limitations
The first thing to note is that negation theses are not of the right kind to be
able to establish a unified account of imagining on their own, that is,
without the help of more positive claims about the nature of imagining.
This is due precisely to their negative character: the fact that they tell us
what imaginings are not, instead of informing us about what they positively are.
The underlying assumption is that, at best, only very special mental phenomena possess a nature which does not allow for a positive description.
Perfect hallucinations that is, experiences which are subjectively indistinguishable from (veridical) perceptions without being such perceptions
are arguably examples of mental episodes that can be specified only in
terms of what they are not: namely in terms of the negative facts that they
are not perceptions, and that we cannot tell them apart from perceptions
(see Martin (2006) and Dorsch (2010a)). As a consequence, perfect hallucinations do not form a natural (mental) kind, but merely an epistemic kind
similar, say, to the class of all lemon-looking entities, which include not
only real lemons, but also perfect fakes made out of wax or other materials.
But there is no reason to assume that imaginings are like perfect hallucinations in this respect. In particular, they are not to be characterised in
terms of the subjective indistinguishability from some cognitive episodes.
Although there are important phenomenologically salient similarities
between imaginings and cognitions, there are also significant differences,
not the least the attitudinal one. Moreover, the phenomenal character of
imaginings cannot just be characterised by the lack of aspects which cognitions show. If there is something at all that it is like to imagine something,
it should be describable in positive terms if not even explainable by reference to some more fundamental feature distinctive of imaginings.
As a consequence, there is a limit to the explanatory power of negation
theses: none of them is able to reveal (fully) the distinctive nature of imaginative phenomena and, especially, their phenomenal character. Hence, if
one or more of the negation claims are meant to be at the core of a unified
account of imagining, they need to be supplemented by some further posit-

Critical Assessment

255

ive claims in such a way that the resulting theory discloses fully what imaginings essentially are. The preceding considerations also show that the
positive claim to be added has to be more substantial than, say, the statement that imaginings are episodic mental representations. For the view that
what is distinctive of imaginings is that they are merely mental, episodic
and representational provides, again, primarily a negative characterisation
of their nature, given that the positively mentioned features are too ubiquitous among mental phenomena. It is much more promising to combine a
negation claim with one of the echo theses that are central to the different
versions of the Dependency Account of imagining. This is, for instance,
what OShaughnessy is proposing, partly in response to the noted negativity of negation claims (see Section 9.4). There is then, however, the worry
that the chosen echo thesis may describe the nature of imaginings in more
fundamental terms than the respective negation claim that, for instance,
the lack of a certain cognitive feature is due to the fact that imaginings are,
in one way or another, dependent on cognitions.
In addition to their negative character and the resulting general limitation of their explanatory power, particular negation theses may fail to be illuminating with respect to the common nature of imaginings. Consequently, they may fail to add anything to a unified account of imagining,
even if they are true of all central cases of imagining and supplemented by
some more positive claims. Three cases are of particular significance here.
First of all, negation claims may lack explanatory power to the extent to
which they focus on features the absence of which is not particularly distinctive of imaginings. The thesis that imaginings are not moods tells us
not much of interest about what imaginings are or are not, even though it is
presumably true of all imaginings. The main reason for this is that many
non-imaginative phenomena lack the specified features as well: most mental states are not moods. Accordingly, the lack of the respective features is
not very specific to imaginings. Even less informative is the claim that
imaginings are not tables, given that no mental phenomenon is a table (and
neither most non-mental phenomena). Negation claims that apply only to
imaginings or at least just to imaginings and to a few other mental phenomena promise to be much more illuminating.
Then, negation claims may be unilluminating with respect to the pur-

256

The Epistemological Account

pose of providing a unified theory of imagining because they describe different kinds of imagining in terms of different kinds of features (or, rather,
lack thereof). To say that visual imaginings are not seeings, intellectual
imaginings not judgements, affective imaginings not real feelings, and so
on, has precisely this result: it does not tell us about the nature of imagining since it does not identify a single feature the lack of which characterises all kinds of imagining. As a consequence, such a negation claim may
be true and illuminating with respect to particular forms of imagining; but
it will not be able to say anything about what imaginings have in common
and what renders them imaginative.
Finally, negation claims may fail to have proper explanatory power because they do not pick out the most fundamental feature(s) the lack of
which is distinctive of imaginings. The idea that imaginings are not tables,
say, is unilluminating in this way, given that it is merely a consequence of
the fact that imaginings are not entities in the external world (which again
may perhaps be traced back to another negation claim, or the positive idea
that imaginings are mental phenomena).

(ii) The Limitations of (NC*)


Let us start with the explanatory power of (NC*). What renders it immediately problematic is the impression that it does not seem to describe the
nature of the various forms of imagining by reference to the absence of one
and the same feature (or set of features). As OShaughnessy and many others have noted, perceptions, beliefs and judgemental thoughts are subject to
different epistemic constraints.36 That is, they count as knowledge-constituting in virtue of different properties. In particular, the reliability of their
formation is due to different factors (e.g., while reliable perceptions are
nomologically depend on aspects of the environment, reliable beliefs are
sufficiently supported by the available evidence); and they are rationally
integrated with our picture of reality in different ways (e.g., while beliefs
are subject to rational revision, perceptions are open to be rationally disowned). Accordingly, the claim that sensory imaginings are not reliable
36

See Chapter 7 and, for instance, Martin (1992), Pollock & Cruz (1999), or Williamson (2000).

Critical Assessment

257

and rationally integrated in the same way as perceptions is concerned with


the lack of very different properties than the claim that intellectual imaginings are not reliable and rationally integrated in the same way as judgements or beliefs.
However, (NC*) seems precisely to consist in the conjunction of these
two claims, as well as similar claims concerned with the relationship
between other forms of imagining and cognising. Hence, endorsing (NC*)
appears to amount to nothing more than taking different types of imagining
to lack different properties. But such a conjunctive negation claim cannot
be of much relevance for a unified theory of imagining, given that the latter
aims, by definition, to provide an account which captures the nature common to all instances of imagining, instead of merely highlighting the particular characteristics distinctive of the various forms of imagining. As a
consequence, although (NC*) may very well be true and informative about
the specific nature of particular kinds of imagining (e.g., sensory imaginings, or intellectual imaginings), it does not appear to reveal anything of interest about the nature of imaginings in general.
Someone, who like OShaughnessy takes (NC*) to be vital to the
provision of a unified account of imagining, might reply that the thesis
should be understood differently not as a conjunction of more specific
negation claims, but rather as identifying a general aspect lacked by all
forms of imagining (which is, moreover, fundamental in the sense that it is
not due to a more basic property common to all imaginings). In other
words, the idea is to read (NC*) as claiming that imaginings are bereft of a
certain single feature (or set of features) that is necessary for the realisation
of each of the cognitive prototypes that is, which is necessary for the
realisation of a cognitive prototype, independently of whether we are concerned with perception, intellectual knowledge, successful memory, and so
on.37 The property in question has to satisfy two conditions: it has to be ne37

OShaughnessy seems to have been aware of the need for such an interpretation of
his claim that no imaginative episode can realise the particular prototype linked to
its specific cognitive counterpart for instance, when he stresses that imaginings are
bound to remain cognitively void, whatever may happen to them (OShaughnessy
(2000): 359). However, OShaughnessys discussion of imaginings and, in particular, his Argument from Origin does not address the issue of whether there are or
can be any further prototypes, over and above perception and intellectual know-

258

The Epistemological Account

cessary for the occurrence of knowledge in all of its forms; and it has to be
absent in the case of imagining. Only then can it account for the assumed
fact that no imagining can cognise reality in the same single way as cognitive episodes. The result is an interpretation of (NC*) which understands it
in the light of a shift in focus from the particularities of the related cognitive constraints to their commonalities and, indeed, identifies (NC*) with
another, more schematic negation claim.

(iii) The Limitations of (NI) and Similar Claims


There are indeed many initially plausible candidates for the property just
mentioned. Three of them play a central role in OShaughnessys understanding of the nature of cognitive prototypes and of the related cognitive
constraints (see Section 7.1): (i) being epistemically reliable; (ii) being rationally integrated with our theoretical picture of the world; and (iii) being
ultimately based on perceptual contact to reality. As already mentioned,
Sartre and Wittgenstein concentrate on a different cognitive feature: (iv)
being informative with respect to the external world. In addition, three other relevant features have already been discussed in Chapter 2: (v) having a
cognitive attitude; (vi) providing prima facie reasons for belief; and (vii)
having the function or aim of cognising reality (or at least of being true).
Besides, it has been argued that it is constitutive of cognising representations that they are relationally (and not merely intentionally) linked to the
aspects of the world which they represent; while it is not part of the nature
of imaginings or other non-cognising representations, such as nonveridical or epistemically unsound cognitions to stand in such a relation
to reality.38 So, another candidate might be: (viii) being constitutively
38

ledge, that imaginings might realise.


The claim that cognising perceptual experiences, memories or beliefs differ in nature
(though not necessarily in phenomal character) from non-cognising perceptual
experiences, memories or beliefs (e.g., non-veridical, illusory or hallucinatory ones)
in that only the former are constitutively linked to reality is central to Williamsons
non-conjunctive account of knowledge (see Williamson (1995): see especially
47f.), as well as to the view known as disjunctivism a position that has been
defended, with respect to different kinds of cognitive states, in, for instance, Evans
(1982), McDowell (1994), Martin (2002a) and (2004), or Dorsch (2010c) and

Critical Assessment

259

linked to reality. What we thus get is a negation claim of the following


schematic form (with y as the variable standing for the respective
cognitive property):
(Ny)

Imaginings essentially lack the cognitively relevant feature


y.

The property in question is cognitively relevant in so far as its possession


is necessary for cognising reality, at least in the same way as cognitive
states. Applying this schema to one or more of the features (i)-(viii) results
in more specific negation claims. For instance, if this schema is brought to
bear on property (iv), we get (NI) or something very close to it.
Now, each of the respective negation claims is, at least to some extent,
illuminating and may serve as a plausible interpretation of what (NC*) is
really saying. For each of them promises a plausible and substantial explanation of why imaginative episodes are unable to figure in cognition in
the same way as cognitive episodes. Accordingly, what (NC*) may in fact
be all about is the general idea that imaginings differ in their (in)capacity to
cognise reality because they lack one or more of the features such as features (i)-(viii) that are necessary for the cognition of reality. However, the
various specifications of (Ny) are seriously limited in their explanatory
power because, in some central cases, they apply equally well to imaginative and to non-imaginative episodes. Among the five main forms of imagining identified in Section 1.2, there are at least two, to which this limitation applies: affective imaginings and imaginative projects.
With respect to many affective imaginings, the negation claims remain
(2012a). The idea that cognising states are constitutively linked to reality implies the
idea that they are factive (i.e., that, if they represent something to be the case, it is
really the case); though not necessarily vice versa (see Dorsch and Soldati (2011)).
It is, however, important to note that OShaughnessys account of perceptions is
not of this kind. Although he assumes that perceptions are necessarily related to their
most immediate external object (i.e., the respective sensation or sense-datum; see
OShaughnessy (2000): 350f., 454ff. and 479), they are only intentionally related to
the represented objects in the external world (see, for instance, ibid.: 124, 166f. and
363). In this respect, perceptions resemble imaginings who also have two objects:
types of perceptions as immediate objects, and parts of reality as intentional
objects (see ibid.: 363, as well as Section 6.2 above).

260

The Epistemological Account

largely unilluminating because the corresponding non-imaginative affective states lack the cognitive properties in question as well. 39 The affective
imaginings concerned correspond either to non-imaginative episodes
which are non-representational (e.g., certain moods or bodily sensations,
such as anxiety or orgasm), or to non-imaginative episodes the representationality of which is not concerned with cognition, but rather with conation
or affection (e.g., feelings of desire or hope, or object-directed emotional
feelings, such as love). It may be denied that there are non-imaginative affective states of the first kind (see Crane (2001)). But the existence of
emotional states of the second kind should be beyond serious doubt (see
Goldie (2000)).
Now, none of these non-imaginative episodes seem to possess any of the
cognitive properties listed above. For instance, episodes of feeling anxious
or of longing for a cigarette cannot really be said to cognise reality, be reliable in their representation of it, be integrated in our epistemic picture of
the world, be informative about reality, be constitutively linked to it, possess a cognitive attitude, provide prima facie reasons for first-order belief,
or have the function of cognising reality. The explanation of this fact is
precisely that the non-representational, conative or otherwise affective
episodes concerned lack the kind of representationality linked to cognition
and the instantiation of properties related to cognition: they are not putatively cognitive representations, and their representationality (if they possess any at all) is not concerned with cognition. 40 As a result, the discovery
that the corresponding affective imaginings such as imagined feelings of
anxiety, orgasm, longing, hope, love, and so on lack these cognitive properties as well is not very informative. Given that longing for a cigarette
cannot cognise aspects of the world, it is no surprise that imaginatively
In addition, the negation claims remain unilluminating with respect to the non-cognitive affective aspects of affective imaginings that have belief-involving counterparts (such as conscious emotions that involve a belief about an actual situation).
40
See Section 2.1; as well as Velleman (2000): 182 and 248f.. The episodes under discussion may, however, be integrated with our mental life in other ways for instance, as reasons or motives for actions, or for higher-order beliefs. In particular, all
conscious mental phenomena (including imaginings) may be said to be rationally integrated in so far as they can ground introspective judgements concerning their presence (see Dorsch (2013)). But this form of rational integration surely does not render
them cognitive or cognising in any plausible sense.
39

Critical Assessment

261

longing, or imagining the feeling of longing, for a cigarette cannot play a


role in the cognition of reality either.
Hence, although the negation claims formulated in terms of the properties (i)-(viii) may be true of the affective imaginings under consideration,
they cannot provide much insight about the distinctive nature of these imaginings, given that they also capture the corresponding aspect of the nature
of the non-imaginative counterparts of these imaginings. The underlying
reason for this is that the affective states at issue whether imaginative or
not are simply not of the right kind to allow for the instantiation of the
cognitive properties listed above. The issue of whether they play a role in
cognition and if not, why not does not arise because of their lack of the
kind of representationality linked to cognition.41
In a similar fashion, most of the negation claims of form (N y) are unable
to elucidate the nature of imaginative projects. Again, the main problem is
that this form of imagining does not belong to the right kind of mental
category, which would allow for the exemplification of the specified
cognitive properties. For most of these properties pertain to representational episodes or states, and not to mental projects. In particular, it is not a
cognitive project as a whole which provides knowledge, but only certain
parts of it, that is, certain of its episodic constituents (e.g., its final conclusion in the shape of a judgement). When we try to calculate a sum in our
heads, find the best next move in a game of chess, or judge a certain philosophical issue, it is not the whole process or project of calculating, of imagining the different possible positions, or of considering and weighing the
evidence, which cognises reality. Instead, only part of the process some
of the episodes involved (e.g., some perceptions, memories or judgements)
may ground or constitute knowledge. In a similar vein, other cognitive
features such as being reliably formed and rationally integrated with
ones view of the world, being informative about and constitutively linked
to reality, being ultimately based on perception, possessing a cognitive atti41

Indeed, this may mean that, strictly, speaking, the negation claims formulated in
terms of the properties (i)-(vii) fail to contribute to a unified account of imagining in
this way not (merely) because they are unilluminating with respect to affective imaginings, but (also) because they do not even apply to them and, hence, cannot be true
(or false) of them. The problem with these negation claims would then pertain to
their extensional adequacy rather than to their explanatory power.

262

The Epistemological Account

tude, or providing prima facie reasons for first-order belief may pertain
to the episodes making up a cognitive project, but not to the project or its
pursuit as a whole.
One reason for this is that mental projects do not themselves constitute
representations. Rather, they are activities which partly consist of representations. And since cognition and knowledge are exclusively a matter of
representation, it is only the episodic representations involved in projects
that can exemplify the cognitive properties (i)-(viii), but not the complex
projects comprising them. The other reason for the fact that cognitive or
other mental projects cannot reasonably be said to instantiate the cognitive
features under discussion is that many of them may lead to the manifestation or acquisition of knowledge, while none the less containing imaginative elements or non-cognising cognitions. When aiming to acquire knowledge about a certain issue, we may make extensive use of imagined hypothetical cases; and many successful cognitive projects inadvertently involve the formation and later revision of false judgements or beliefs.
But it would be strange if the project as a whole instantiated certain cognitive properties lacked by many of its elements. Again, saying that imaginative projects lack the features (i)-(viii) reveals hardly anything about their
distinctive nature, given that cognitive and other projects are due to their
very nature as mental projects bereft of them as well.
The only exception is perhaps the negation claim that imaginings do not
aim at the cognition of the world. This property related to cognition can
perhaps be instantiated, not only by episodes, but also by projects. Indeed,
it is reasonable to maintain that cognitive projects are to be characterised
and to be distinguished from imaginative projects as having the acquisition of knowledge as their aim. In addition, adopting an Agency Account of
imaginings and identifying them as instances of mental agency with a particular non-cognitive purpose would explain this teleological difference
between the two kinds of mental project (see Section 14.2).
The negation claim that imaginings are not reliably formed, on the other
hand, does not constitute any exception. For the occurrence of a particular
mental project that is, the occurrence of several, suitably linked mental
episodes is reliable only in so far as its episodic constituents are reliably
formed and connected to each other. This is the sense, for instance, in

Critical Assessment

263

which the two projects discussed in Chapter 4 are reliable: they are so because they involve the reliable formation of certain episodes of the sensory
imagination.
It might be thought that imaginative projects can be defined in terms of
imaginative episodes: namely as precisely those mental projects which
contain, aim at, or lead to, episodic imaginings (of a certain kind). And the
hope might then be that the negation claims are able to illuminate the
nature of imaginative projects by illuminating the nature of imaginative
episodes and characterising imaginative projects in terms of those episodes. However, this proposal should be rejected for four reasons.
First, it could still not ensure that the negation claims apply to all kinds
of imaginative projects. For it is possible that there are imaginative projects
which are to be characterised by reference to affective imaginings of the
kind discussed above: that is, to imaginings whose imaginative nature the
negation claims would still have nothing to say about.
Second, the resulting account would not be illuminating with respect to
the ambition of providing a unified theory, since it would account for the
imaginativeness of episodes and projects in terms of two distinct features:
while episodes would be said to be imaginative because of their lack of a
certain cognitive property, projects would be said to be imaginative because they involve imaginative episodes. Hence, the theory would be disjunctive: it would not identify a common feature of all imaginings, but instead postulate two very different (though related) ways of being imaginative.
Third, the idea that imaginative projects can be defined by reference to
their involving imaginative episodes is untenable. Although it seems true
that all imaginative projects have to involve at least some imaginative episodes while cognitive projects need not involve any imaginings, the latter
may involve and even be deliberately directed at the occurrence of imaginative episodes just like the former. Above, I already mentioned cognitive
projects that involve imagining in the shape of hypothetical reasoning. But
an even better illustration of the possibility of cognitive projects, that lead
to and, in fact, are aimed by us at the generation of imaginative episodes of
a certain kind, are the two examples presented in Chapter 4.
Fourth, specifying imaginative projects in terms of the aim or con-

264

The Epistemological Account

sequence of actively bringing about the occurrence of imaginative episodes


does not describe them in most fundamental terms. It will turn out (see
Chapter 13) that it is distinctive of actively formed imaginative episodes
that we exert direct control over which specific states of affairs they represent. Accordingly, the mental project of actively producing imaginative
episodes can also be identified as the mental project of actively producing
episodic representations the specific content of which is directly determined by our mental agency. But the second identification is more fundamental than the first, given that being an actively produced representation
with a directly determined content is more fundamental than being an actively produced imaginative representation: that the concerned episodes are
imaginative is a consequence of their direct determination by mental
agency, and not the other way round (I return to this issue of fundamentality in Section 14.1).

(iv) The Limitations of (NC)


The most abstract of the negation claims to be discussed namely (NC)
identifies a single feature which all imaginings are said to lack: the capacity to provide (i.e., ground or constitute) knowledge. Moreover, the possession or lack of this feature demarcates an important divide among our
mental episodes and states. There are also quite a few of mental phenomena among them all cognitive phenomena which possess this capacity
and therefore differ essentially from imaginings. As a result, (NC) promises to be relatively illuminating (with the general limitation mentioned at
the beginning of this chapter in mind): if true, the idea that imaginings cannot cognise reality would inform us about an important aspect of the nature
of these mental states.
One serious worry about the explanatory power of (NC) is, however,
that there may be a more fundamental feature in terms of which the lack of
the capacity to provide knowledge can be elucidated. This more basic feature may be either a property which all imaginings lack and which is necessary for the provision of knowledge, or a property which all imaginings
possess and which prevents them from providing knowledge. Wittgenstein
has suggested the presence of a more fundamental feature of the second

Critical Assessment

265

kind (i.e., the essential link of imaginings to the will; see Section 4.7). In
Part Four, I argue that it is advisable to follow his suggestion at least with
respect to imaginings of which (NC) is indeed true and trace back the absence of the capacity of cognise to the presence of another feature, namely
their active character. The cognitive properties (i)-(viii) discussed in the
last section constitute plausible candidates for a more basic feature of the
first kind that is, of a feature that is responsible for the cognising power
of all cognitions, but which all imaginings lack. Indeed, (NC) may perhaps
be understood as the specification of (N y) in terms of the most general
cognitive property, the power to cognise reality, which is derivable from a
more basic specification in terms of one or more of the features (i)-(viii).
But even if (NC) is assumed to be the most basic negation claim available, which cannot be elucidated by reference to a more fundamental thesis
about the nature of imaginings, it is still limited in its explanatory power.
For just like the (other) specifications of the schema (Ny), it is incapable of
differentiating affective imaginings and imaginative projects from their
non-imaginative counterparts, and for the same reasons. Again, the issue of
whether they can cognise reality does not arise with respect to many affective episodes irrespective of whether they are imaginative or not because of their non-representational, conative or otherwise non-cognitive
nature. And, again, mental projects are in contrast to their episodic parts
not of the right kind of entity for the realisation of cognising power, given that the constitution of knowledge is a matter of representational episodes or states. Hence, (NC) does not fare better with respect to explanatory
power than the other negation claims.
To sum up, the various explanatory limitations of the negation theses
discussed should already be enough to disqualify them as substantial components of a unified account of imagining. They may be true with respect
to affective imaginings and imaginative projects, but not distinctively
that is, not in virtue of their imaginativeness since the claims are true of
all affective episodes and mental projects, and not only imaginative ones.
As a consequence, negation claims are severely restricted in their contribution to the elucidation of the common imaginativeness of the different
forms of imagining. Hence, they should not be part of a theory of imagining, or at least not if the aim is to provide a unified account of imagining

266

The Epistemological Account

capturing the imaginative character common to all central cases of imagining.

8.2. The Extensional Adequacy of Negation Claims


But the negation claims under consideration also fail to be extensionally
adequate with respect to the ambitions of a unified account of imagining.
This means, primarily, that they may fail to be true of all central cases of
imagining. As a result, any theory of imagining which aspires to be unified
and which contains them will simply be false. An example of an extensionally inadequate negation claim would be the thesis that imaginings are nonaffective: after all, we can have emotional imaginings, or imagine experiences of feelings (see Section 1.2 and Chapter 11). Whether negation
claims are true of central cases of non-imaginative phenomena matters
only if they are meant to establish a unified account of imagining on their
own. Only if they are put forward with this particular ambition might they
fail to be extensionally adequate by also being true of non-imaginative
phenomena. For example, the idea that imaginings are not moods seems to
be true of all instances of imagining. But it cannot by itself provide a theory of imaginings, given that it is equally true of many cognitions. At best,
it can fulfil the less demanding ambition to contribute, together with other
claims, to a true theory of imagining.
One already mentioned difficulty for the extensional adequacy of negation claims is that they are usually also true of certain central cases of nonimaginative phenomena, such as emotional or conative states, moods, nonrepresentational bodily sensations, cognitive projects, and so on. Consequently, the negation claims have to be supplemented by further claims
capable of distinguishing imaginings from such non-imaginative phenomena.
A more serious challenge, however, is that the two examples discussed
in Chapter 4 constitute counterexamples to (NC), (NI) and any other negation claim closely associated with them.42 For, as I have argued there dur42

From OShaughnessys own perspective, there may actually be another counterexample: pictorial experience. For him, such experiences involve an imaginative experience of what is depicted, that is, an imagined perception of it (OShaughnessy

Critical Assessment

267

ing my extended discussion of the examples, their final imaginative


episodes do not only cognise reality, but also may possess many of the cognitive features (i)-(viii) introduced above. In particular, they exemplify a
cognitive prototype, are reliable in their representation of reality and rationally integrated with our picture of the world, are ultimately based on
perceptions and informative with respect to the external world. Besides, assuming that knowledge in its various forms is relational (see, e.g., McDowell (1994), Williamson (2000), and Martin (2002a) and (2006)), the visual
imaginings in the examples may arguably even be constitutively linked to
their objects in so far as they may inherit the respective connection to the
world from the perceptions, memories and beliefs on which they are imaginatively based perhaps in the same way in which episodic memories
and perceptual beliefs are understood as inheriting this link from the underlying perceptions. In short, the two imaginings instantiate the properties
(i)-(iv) and (viii). The respective specifications of (N y) are therefore not
true of all central cases of imagining and should hence be given up.
By contrast, (NC*) and negation claims formulated in terms of the features (v)-(vii) turn out to be extensionally adequate. For no instance of
imagining including the visual imaginings involved in the two examples
can cognise reality in the same way as cognitions, that is, by satisfying
the same cognitive constraints. And none of them possesses a cognitive attitude, provides us with prima facie reasons for belief, or has the function
or aim to cognise reality (see Sections 2.1f.). It is true that the sensory imaginings of the sofa and the friend are formed by us with the aim in mind to
acquire knowledge about reality. But this is not enough for them to possess
the respective function. I may use a key to open a bottle of beer, without
the key acquiring the function to be used for bottle-opening. The designated instrument for this task is a bottle-opener, while the key has the function to open one or more doors. The same is true of the externally bestowed
justificatory power of the two episodes of visual imagining.
However, the fact that these episodes lack the cognitive properties (v)(2000): 349). However, it seems plausible to maintain that the relevant aspect of
pictorial experience informs us precisely about what the picture represents, and that
hence this aspect of the experience can ground knowledge. But I do not develop or
press this point here, mainly because it does not seem to be necessary for
OShaughnessy to take pictorial experience to be imaginative.

268

The Epistemological Account

(vii) means also that they cannot be necessary for cognising reality. In other words, mental episodes may realise a (non-intellectual) form of knowledge without possessing a cognitive attitude or function, and without
providing us with prima face reasons for belief. 43 Besides, (NC*) and the
specifications of (Ny) in terms of the features (v)-(vii) are still subject to
the severe explanatory limitations noted above and, hence, cannot make a
substantial contribution to a unified account of imagining.
It seems that the mistake in OShaughnessys argumentation does not
pertain to his defense of (NC*) by means of the Argument from Origin. Indeed, it is reasonable to assume that he is right in arguing that imaginings
cannot satisfy the particular cognitive constraints on perceptual experiences and beliefs; and that they therefore cannot count as cognising reality
in the same manner as these cognitive states. The problematic aspect of his
argumentation is, rather, his derivation of (NC) from (NC*). The crucial
assumption in this derivation is that there are no other cognitive prototypes
which imaginings could exemplify.
His reason for making this assumption seems to be his belief that there
are only two cognitive prototypes: the fundamental mental cognitives
knowledge and perception (363). Accordingly, he appears to have overlooked the possibility of other cognitive prototypes, the exemplification of
which also requires intentionality, veridicality, reliability and possibly other cognitive properties, but which differ from the perceptual and the intellectual one in that they demand a different kind of reliability. Perceptions
have to be reliably caused by the perceived aspects in the world. Beliefs (or
judgements) have to be rationally supported by other relevant cognitive
states. Sensory memories have to reliably depend on our past perceptions
of things. The two visual imaginings in the examples have to be based on
reliable (and veridical) perceptions or sensory memories; and they have to
be the result of imaginative activity which is reliable at least with respect
to the extraction of information already contained in the underlying cognitive states. Hence, there seem to be at least four different forms of knowledge, each with its particular type of reliability, its specific cognitive constraints and its particular cognitive prototype or ideal.
This raises the question of why the examples do not seem to have
43

This may explain why these features do not figure at all in OShaughnessys arguments for the negation theses (NC)*) and (NC).

Critical Assessment

269

crossed OShaughnessys mind or the mind of other philosophers, who


have defended (NC) or similarly strong negation claims. An answer might
be found in the general tendency to neglect imaginative projects in discussions of imagining. The two examples can occur only as parts of imaginative projects (which are, again, part of even more comprehensive cognitive
projects (see Section 14.2)), the nature of which is essential for ensuring
the epistemically sound status of the resulting imaginative representation.
Without being embedded in these projects, the visualisings of the sofa in
the door and of the appearance of the friend would not be reliably formed,
nor be able to fulfil the cognitive role assigned to them by us. If their production did not originate in our perceptions or memories of the respective
objects and did not involve the controlled modification of the information
contained in these, they would at best randomly match reality, and we
should not epistemically trust them when forming beliefs. Something very
similar applies to the other example concerned with the visual recognition
of a familiar person after a long period without any (visual) contact.
Now, OShaughnessy, for instance, discusses imaginative projects
such as daydreams only very briefly (mainly at OShaughnessy (2000):
216ff.). Moreover, he does not link his discussion to his self-assigned task
of providing a unified account of imagining, despite daydreaming clearly
counting as a central instance of imaginative activity. In similar fashion,
many important philosophical works about the nature of imaginings such
as those by Hume (1739), Sartre (1940), Wittgenstein (1984a, 1984b), Casey (1976), White (1990), or McGinn (2004) do not really engage with
imaginative projects and focus instead more or less exclusively on imaginative episodes as instances of imagining. On the other hand, texts that do
address the issue of the nature of imaginative projects such as Wollheim
(see his (1973), and (1984): chs. 3 and 4), Peacocke (1985), Hopkins
((1998): ch. 7), or Noordhof (2002) are usually not concerned with the issue of why they have their imaginativeness in common with imaginative
episodes, but instead with other issues. It is thus perhaps unsurprising that
OShaughnessy and other philosophers do not seem to have noticed, or
taken into account, examples like the ones presented in Chapter 4.
There might also be another reason why such counterexamples to (NC)
are not as obvious as one might think when confronted with them. For al-

270

The Epistemological Account

though there are many cognitive projects which (have to) involve imagining, few of them involve cognising imaginative episodes. Or, to speak in
Dancys terms, the imaginative episodes involved in those projects do not
play a role in the justification of the resulting knowledge-constituting beliefs, but merely enable us to acquire them (see Dancy (2000): 127).
Consider the following three examples: (a) we can find out about the
thoughts and feelings of another person by imagining being in her position
and state of mind; (b) we can discover the logical consequences of a certain proposition by engaging in hypothetical reasoning on the basis of imaginatively entertaining it; and (c) we can find out about the aesthetic qualities of a certain configuration of objects (e.g., whether hanging a specific
picture on a certain wall would look nice; or whether adding a yellow
square to a Mondrian painting would destroy its balance and harmony) by
visualising them and observing our emotional reaction towards them. The
last category may also include emotional reactions to imagined objects
which are not necessarily or clearly aesthetic (e.g., feeling disgust when
imagining the taste and consistency of worms in ones mouth).
In each of these cases, it seems reasonable to assume that we can gain
some knowledge about reality (understood in a wide sense, so as to include
the recognition of response-dependent features): we may learn something
about the minds of other people, about the logical properties of propositions, or the aesthetic qualities of certain combinations of objects. But a
proponent of (NC) has the plausible option to deny that the respective cognising representations are imaginative.
In the example from aesthetics, the relevant ground of knowledge is the
(aesthetic or other) emotion felt towards the imagined object: we judge the
configuration to be beautiful or balanced on the basis of our real (i.e., nonimaginative) feeling towards it (see, e.g., Budd (1995); though see also
Dorsch (2007)). With respect to imaginatively gained knowledge of what
can be inferred from a given proposition, on the other hand, it is not the final imaginative intellectual representation of the inferential train of suppositions which grounds the knowledge, but the inference as a whole. For
the last supposition does not contain any information about its inferential
origin: only (higher-order) reflection on the whole inference can establish
the recognition of the inferential link between the original and the inferred

Critical Assessment

271

propositions. If I infer, on the basis of assuming that determinism is true,


that there is no free will, I do not gain knowledge about whether there is
free will. At best, I can gain knowledge about whether determinism implies
free will. But the respective piece of information is contained, not in the final supposition of the inferential chain, but in the chain as whole.
Finally, a very similar strategy may be adopted with respect to the case
of empathy, given that ones judgemental ascription of certain mental states
to another person cannot be based simply on ones experience or introspection of ones own (simulated or off-line) mental representations. In addition, we need to make the transfer from us to the other person: we have to
recognise the resemblance or type-identity between our experienced or introspected mental states and the mental states of the other person. But the
representation of this resemblance or type-identity cannot be imaginative,
since it needs to ground the belief that the other person feels and thinks just
like we do in our simulation of her. Hence, none of the examples (a)-(c)
constitute a counterexample to (NC) or related negation claims. However,
this does not mean that there are no such counterexamples, or that they
have to be particularly rare.
The main conclusion of this chapter is that the Epistemological Account
of imagining as put forward by OShaughnessy should be rejected.
That is, negation claims should not figure as fundamental truths in a unified account of imagining. This is compatible with some negation theses
still being strictly true, and most others at least typically. After all, imaginings do lack a cognitive attitude and function and normally do not constitute knowledge. Hence, the negation claim (NC*) and OShaughnessys
Argument from Origin in favour of it may be retained just like the
stronger thesis (NC), as long as it is restricted to affective and intellectual
imaginings. But the conclusion also means that the remaining candidates
for a unified theory of imagining should be able to explain why or in virtue of what imaginings are non-cognitive and usually do not cognise reality.
The Agency Account, which identifies imaginings with a special kind of
mental actions, promises to account for both aspects of imaginative episodes (see 14.1). They are said to lack the function to cognise reality as
well as its phenomenal reflection in a cognitive attitude because having

272

The Epistemological Account

this phenomenologically salient function is incompatible with having a


phenomenologically salient practical purpose (see also Dorsch (2009b)).44
And they are taken to typically fail to be reliable and, hence, to provide
knowledge because the imaginative activity involved is usually not sufficiently constrained by how reality is and works. However, the Agency
Account has difficulties to accommodate examples of what seem to be involuntary or will-impervious imaginings, like those identified by Sartre
(1940: 19) Collingwood (1938: 179 and 195), OShaughnessy (2000: ch.
11), and others (see Section 14.4). These non-cognitive episodes, too, fail
to cognise reality but obviously not because of the involvement of a certain kind of mental agency. Indeed, the Agency Account appears to lack the
resources to account for both the imaginativeness and the non-cognitivity
of these examples.
So, it seems that we face a dilemma: if we embrace the world-related
Epistemological Account, we appear to be unable to accommodate cases in
which our controlled imaginative activity leads to the acquisition of knowledge; while if we endorse the will-related Agency Account, we seem to be
unable to handle what are apparently will-impervious imaginings. In Part
Four, I argue that the prospects of choosing the latter horn are far less
daunting than the prospects of choosing the former. There, I develop the
details of the Agency Account of imagining sketched in this paragraph and
return to the significance and implications of the worry about the possibility of passive imaginings. In the meantime, I aim to explore a third view
the Dependency Account which avoids both horns of the dilemma and
promises a different explanation of the main features of imaginings,
namely in terms of their constitutive dependency on their cognitive counterparts.

44

Note that the practical purpose of cognitive projects differs from the epistemic function of cognitive episodes. For the general goal of cognitive projects is not to cognise reality, but to bring about episodes which cognise reality (see Section 14.2).

Part Three
The Dependency Account

Introduction to Part Three

The central idea of the Dependency Account of imaginings is that, fundamentally, imaginings are constitutively dependent on cognitions (and similar phenomena, such as emotional feelings). This constitutive claim is often accompanied by a claim about the concept of imagining and its necessary elucidation in cognitive terms. Here, I focus mainly on the metaphysical claim, given that unified accounts of imagining are primarily concerned with the nature of imaginings, and not with how we conceive of
them.
The various versions of the Dependency Account differ, first of all, in
their notion of the assumed fundamental dependency relation. In particular,
imaginings may be said to have cognitions either as their necessary causes,
or as a necessary part of their representational contents, or as the necessary
origin of their representational capacity, or as the necessary target of their
imitative or simulative employment. Depending on the kind of dependency
concerned, there are also differences which cognitive phenomena serve as
the dependency base. While representational and imitative links concern
the whole set of cognitive counterparts of imaginings (e.g., judgements or
beliefs as well as perceptions and sensations), causal and semantic dependency bases consist primarily of cognitions that link us directly to reality
(e.g., perceptions and sensations). The resulting theories may be labelled
the Causal, the Representational, the Semantic and the Imitation Account
of imagining.1
1

Hume endorses both the Causal and the Representational Account, given that he
takes imaginings to be ideas that causally and representationally depend on particular impressions (see Hume (1739): 1.1.1.6ff. for the double echo claim; and 1.3.7.7

276

The Dependency Account

OShaughnessys echo claim, introduced in Section 6.2, is a more specific version of the Representational Account in that it takes the representational relation to be non-intentional, and the relevant cognitions to be either
perceptions or knowledge-constituting judgements or beliefs. Indeed, his
claim illustrates that versions of the Dependency Account may also differ
in their characterisation of the cognitive phenomena that serve as the dependency base for instance, whether they include only knowledge-constituting episodes or, more generally, all cognitive episodes. But they all
have in common that they specify the nature of imaginings by reference to
perceptual experiences, judgemental thoughts (or occurrent beliefs) and
similar mental episodes (e.g., sensations or feelings).2
Just as in the case of the Epistemological Account, the various endorsements of the Dependency Account show the same structural form. More
specifically, their central claims such as the theses listed above model
imaginings on their cognitive counterparts or (as I will say) understand
them as echoes of the latter. Claims of this kind which I call echo claims
involve at least three components: the idea of dependency, the idea of inheritance, and the idea of counterparts.
The idea of dependency is, of course, that imaginings are dependent in
some way or another on cognitions which explains why they have to be

for the characterisation of imaginings). OShaughnessy (2000) endorses the Representational Account with respect to all kinds of imaginings (see the discussion in Section 6.2), while Peacocke (1985): 22, Martin (2001): 273, (2002): 404, and Dorsch
(2010c) endorse it with respect to (central cases of) sensory imaginings (see also
Note 10 in Chapter 1). Finally, Currie & Ravenscroft (2002): 49, explicitly defend a
semantic dependency (or echo) claim and may also together with Ryle
(1949/1963): 250ff. be read as embracing the Imitation Account. See the Chapters
9f. and 12 for detailed discussions of each of these endorsements.
It would in principle be possible to account for imaginings in terms of a dependency
on non-cognitive phenomena. But no one has, to my knowledge, defended such a
view and for a good reason, given that imaginings are by far most similar to cognitions and, hence, should be modelled upon them (if on anything at all). Note that entertaining a proposition understood as a necessary ingredient to all instances of
thought and possibly also to all sensory representations, if the latter turn out to be
propositional as well does not count as an exception, despite being non-cognitive
and essential to imagining. For entertaining a proposition in this sense does not constitute a full-blown mental episode, but merely a necessary part of one (see Section
14.2).

Introduction to Part Three

277

elucidated in terms of cognitions. Echo claims may differ in whether they


take this dependency thesis to apply to all or only to certain forms of imagining (e.g., just to sensory imaginings). And, as already mentioned, echo
claims may differ in whether they postulate a causal, a representational, a
semantic or an imitational dependency. Depending on which dependency
relation is assumed, imaginings may be said to be dependent, not merely
on specific types of cognitive episode, but perhaps also on particular
tokens (see Section 9.4 for more on this point).
The idea of inheritance, on the other hand, consists in the idea that imaginings inherit some of the main features of the cognitions on which they
depend. This idea need not be an explicit part of particular echo claims, but
is always present in their background. The thought is that there are certain
striking similarities between imaginings and cognitions notably similarities in content (e.g., which entities and features can in principle be represented) and in type of representation (e.g., whether they are visual, intellectual, perspectival, etc.) which cannot be explained otherwise than by reference to an assumed dependency relation between the two. For instance,
that what we can visualise are visible entities, and that we visualise them in
a visual and perspectival manner, is traced back to the dependency of visualising on seeing (see Chapter 10, as well as Hopkins ((1998): ch. 7) and
Martin (2002a)). Indeed, one major reason for accepting an echo thesis is
that it promises to explain why these similarities hold, without having to
give up on the idea that there is still an essential difference between imaginings and cognitions: while the former are said to be similar to the latter
in so far as they echo them, only the latter are still taken to be the real
thing.3
Finally, the idea of counterparts is identical with the idea of assigning to
each kind of imagining a particular type of cognitive state on which it is
modelled (see Budd (1989): 100; Currie & Ravenscroft (2002): 11). Thus,
sensory imaginings are usually modelled on perceptions, intellectual ones
on beliefs or judgements, affective ones on feelings or similar states, and
so on (depending on the intended scope of the particular echo thesis to be
3

Not dissimilar to the way in which fake lemons are made in order to resemble real
lemons (by recreating their visual or otherwise perceivable appearance), without
amounting to real lemons (i.e., without recreating their DNA; see Dorsch (2010c)
and (2012a)).

278

The Dependency Account

formulated). What prompts the specific assignments are, again, the similarities holding between the respective relata. The counterpart of visualising,
say, is seeing because both kinds of episode represent the same kind of entities and features (i.e., visible ones), and do so in the same manner (i.e.,
visually). This explains why cognitions are the best candidates for the dependency base for imaginings: the two kinds of episode resemble each other (almost) perfectly in what they are capable of representing, and in which
type of representation they may instantiate (see Section 2.1).
The Dependency Account of imagining may be taken to provide an exhaustive unified account of imagining in the sense that it suffices to completely characterise the distinctive nature of imaginings (as well as in the
most fundamental terms). But, as the example of OShaughnessys view
has illustrated, it is possible to supplement the Dependency Account with
another view. One particular reason for doing this is the existence of sensory memories. The thought is that sensory imaginings are even more similar to sensory memories than to perceptions, notably in respect of whether
they represent their objects as currently present in our actual environment
that is, in respect of whether they represent their objects in an immediate
way, as specified in Section 3.2. To account for this fact about the three
kinds of sensory episode, and to avoid taking sensory imaginings to be dependent on episodic memories, it is perhaps necessary to treat sensory imaginings as echoes of perceptions as well (see Martin (2001)). But, then,
maintaining that imaginings are dependent on cognitions may not suffice
any more to distinguish them from episodic memories. Hence, some further fundamental claim about the nature of (sensory) imaginings may have
to be needed to be added to the echo thesis in question.
I return to this issue in Section 9.4, where I suggest that the purity of
the Dependency Account may be saved by taking imaginings to be dependent on types of cognitions rather than on tokens, as in the case of episodic
memories which serve as recollections of particular past experiences. None
the less, the following discussion and assessment of the Dependency Account should be understood as staying neutral on whether the acceptance of
an echo claim may suffice to provide a complete unified account of imagining, or whether it may just make a substantial contribution to such a theory.

Introduction to Part Three

279

The chapters and sections of this part of the book are structured around
the discussion of the various versions of the Dependency Account, as well
as the corresponding echo claims listed above. Three of the chapters
(Chapters 9 to 11) deal with the Representational Account, while one
chapter (Chapter 12) concentrates on other versions of the Dependency Account. The long discussion of the Representational Account is warranted by
the fact that this theory of imagining is the most promising rival to the
Agency Account and, moreover, reveals a lot about the nature of two particular forms of imagining, namely sensory and affective imaginings.
Chapter 9 is devoted to the prospects of the Representational Account as
a unified theory of imagining, as well as to the Humean origins of this account. In the first three sections, I critically discuss Humes conception of
imaginings. His view of imaginings as representations (or copies) of cognitions differs from contemporary versions of the Representational Account by endorsing a specifically causal rather than intentional or relational understanding of the representational link. As a result, Hume ends
up endorsing both the causal and the representational version of the Dependency Account: imaginings are causally and representationally dependent on their cognitive counterparts in so far as they represent precisely
those particular cognitions on which they causally depend. 4 Accordingly,
the discussion of Humes conception includes also a discussion of the causal version of the Dependency Account that is, what might be called the
Causal Account of imagining. My conclusion is that his conception does
not provide us with a satisfactory account of the distinctive nature of imaginings which, arguably, also was not Humes ambition.
The fourth section of Chapter 9 introduces the main intentional and relational variants of the Representational Account and highlights their continuity with the Humean conception spelled out before. In the remaining
two sections, I discuss two forms of imagining namely intellectual imagining and imaginative projects the nature of which cannot be properly
captured by means of a representational echo claim. If my arguments are
4

OShaughnessy is therefore not the only philosopher who combines an echo claim
about imagining with another thesis about their nature. But Humes approach is different from OShaughnessys in that it combines two echo claims rather than an
echo claim and a negation claim and thus still provides a pure Dependency Account.

280

The Dependency Account

correct, the Representational Account does not generalise to all paradigm


instances of imagining and is therefore bound to fail as a unified account of
imagining. None the less, as I will argue in the Chapters 10 and 11, representational echo claims are true of sensory and emotional imaginings. Although the Representational Account fails to capture the nature of their
imaginative character, it manages to capture the nature of their representationality (which they, incidentally, share with sensory memories). Besides,
in the final chapter of this part (Chapter 12), I briefly criticise attempts to
account for the common nature of imaginings in terms of semantic dependency or imitation, with a special focus on the ideas of simulation and pretense.

CHAPTER 9
The Representational Account

The Representational Account has its origin in the empiricist idea that all
mental representation starts with, and derives from, sense experience
(where this may include sensation as well as perception). One particular respect, in which Hume stands out from the other British Empiricists, is that
he assigned a very prominent place in his philosophy of mind to the imagination. While this is primarily true of the role of the imagination (understood as something like a mental faculty or capacity) in the distinction,
combination or association of ideas, it also applies to the imaginative complex ideas (understood as mental episodes) that are the particular results of
imagining. Indeed, Hume was one of the first philosophers to discuss episodes of imagining in any detail and, moreover, had a great influence
though mainly by serving as a target of objections on the subsequent discussions of imaginings, notably in the writings of Husserl, Sartre, Ryle and
Wittgenstein.
In addition, he inspired together with his empiricist predecessors the
recent endorsement of the idea that imaginative (and other) episodes are
representations of perceptions, which constitutes a version of the Representational Account of imagining. The first two sections of this chapter are
concerned with, respectively, the presentation and criticism of Humes conception of imaginings (and other non-perceptual episodes) in terms of the
copy principle, which understands representation in essentially causal
terms. In the third section, I briefly argue that the causal element of
Humes view has no role to play in a unified account of imagining and
should therefore be dropped from such an account. The last section illus-

282

The Dependency Account

trates different contemporary ways of spelling out by reference to intentional or relational conceptions of representation the resulting purely representational version of the Dependency Account.

9.1. Humes Copy Principle


For Hume, episodes of imagining belong to the class of ideas. And, as
such, they are taken to be dependent on the corresponding impressions in
the sense that they are copies of these impressions. To understand this
claim, it is necessary to take a closer look at some of the details of Humes
theory of the mind. He divides the class of mental episodes (or perceptions in his terminology) into impressions and ideas or, as he also says,
into feelings and thoughts. The former comprise sensations, passions
and emotions, as they make their first appearance in the soul (Hume
(1739): 1.1.1.1) that is, perceptual experiences, bodily sensations, and basic feelings of desire and emotion. By contrast, the latter include the faint
images of [impressions] in thinking and reasoning, such as imaginings and
occurrent beliefs or judgemental thoughts.5
Hume characterises the difference between both kinds of mental episode
in two different, though related ways. According to the first, impressions
and ideas differ in vivacity: the former are said to be generally more vivid
than the latter. Vivacity comes in degrees, however. And some impressions
and ideas may be of almost equal vividness, so that we may sometimes be
unable to distinguish them.
Those perceptions, which enter with most force and violence, we may name
impressions [...]. [...] The common degrees of [impressions and ideas] are
easily distinguishd; tho it is not impossible but in particular instances they
may very nearly approach to each other. Thus in sleep, in a fever, in madness, or in any very violent emotions of soul, our ideas may approach to our
impressions: As on the other hand it sometimes happens, that our impressions are so faint and low, that we cannot distinguish them from our ideas.
But notwithstanding this near resemblance in a few instances, they are in
general so very different, that no one can make a scruple to rank them under
5

See Hume (1739): 1.1.1.1, and also some of the passages quoted further below.
Memories are a special case and will be discussed separately further below.

The Representational Account

283

distinct heads, and assign to each a peculiar name to mark the difference.
(Ibid.: 1.1.1.1)

The near resemblance of instances of the two types of mental episode and
our subsequent problems to tell them apart are still not meant to imply that
some ideas might be more vivid than some impressions they only approach each other very closely.6
Another point is that Hume takes the vivacity of our mental episodes to
be an aspect of their subjective characters which enables us to tell apart,
from the inside, instances of the various kinds of mental episode. For he
acknowledges that the vividness of mental episodes is part of how they appear to us in consciousness and of what lets us distinguish them from our
first-personal perspective. This becomes apparent in the passages where
Hume describe the subjectively accessible differences among ideas of three
kinds, namely those of judgement, memory and imagination.
For tho it be a peculiar property of the memory to preserve the original order and position of its ideas, while the imagination transposes and changes
them, as it pleases; yet this difference is not sufficient to distinguish them in
their operation, or make us know the one from the other; it being impossible
to recall the past impressions, in order to compare them with our present
ideas, and see whether their arrangement be exactly similar. Since therefore
the memory is known, neither by the order of its complex ideas, nor the
nature of its simple ones; it follows, that the difference betwixt it and the
imagination lies in its superior force and vivacity. (Ibid.: 1.3.5.3)
An idea assented to feels different from a fictitious idea, that the fancy
alone presents to us: And this different feeling I endeavour to explain by
calling it a superior force, or vivacity, or solidity, or firmness, or steadiness.
[...] [I]t is something felt by the mind, which distinguishes the ideas of the
judgement from the fictions of the imagination. (Ibid.: 1.3.7.7)

In Hume (1739): 1.3.5.7, the focus is also on the effect a de- or increase in vivacity
has on what we take a given episode to be. Hume also notes there the possibility that
an imaginative episode may change into a cognitive one that an often enough repeated idea of the imagination may become an idea of judgement or memory. But
again, this makes clear that a sufficient de- or increase in vivacity leads to a different
kind of episode.

284

The Dependency Account

One may wonder whether the enumeration of different terms really already
helps to explain what vivacity amounts to. But Hume offers more, namely
the claim that a higher degree of vivacity comes with two other important
aspects: (i) an increased sense of presence or reality with respect to the objects and features presented7 to us by the episodes at issue; and (ii) a higher
motivational (or rational, as one would feel inclined to say today) impact
on our beliefs, emotions and actions.
This variety of terms, which may seem so unphilosophical, is intended only
to express that act of mind, which renders realities more present to us than
fictions, causes them to weigh more in the thought, and gives them a superior influence on the passions and imagination. (Ibid.: 1.3.7.7)

In the case of judgemental thoughts, their high degree of vivacity also ensures that they lead to the formation of a more stable and enduring belief.
It gives them more force and influence; makes them appear of greater importance; infixes them in the mind; and renders them the governing principles of all our actions. (Ibid.: 1.3.7.7)

Accordingly, the subjective vivacity of a mental episodes reflects the closeness of its connection to perception and its impact on belief which is, of
course, in line with Humes thought that perceptions are the most vivid
episodes that we enjoy, and that imaginings are characterised by the least
degree of vivacity.
His second and less explicit characterisation of the difference between
impressions and ideas introduces both the idea of a resemblance between
the two and the notion of a causal dependency of the latter on the former.

My use of the term presentation is meant to be neutral enough to allow for both intentional or relational forms of presentation of objects, as well as for the sensory or
intellectual presentations of objects that may be given as past, present, actual, nonactual, and so on. The presentation of an object is, however, always taken to be a
conscious presentation. The expression representation, on the other hand, is intended to highlight the fact that the represented perceptions or judgements are thereby
not themselves present in the stream of consciousness, but instead merely re-presented by the respective episodes.

The Representational Account

285

Thus we find, that all simple ideas and impressions resemble each other;
and as the complex are formd from them, we may affirm in general, that
these two species of perception are exactly correspondent. [...] Let us consider how they stand with regard to their existence, and which of the impressions and ideas are causes, and which effects.
The full examination of this question is the subject of the present treatise;
and therefore we shall here content ourselves with establishing one general
proposition, that all our simple ideas in their first appearance are derivd
from simple impressions, which are correspondent to them, and which they
exactly represent. (Ibid.: 1.1.1.6f.)

Since Hume describes this multi-faceted relationship between impressions


and ideas also in terms of the latter being copies of the former (see, e.g.,
ibid.: 1.1.1.5, 1.1.3.4, 1.3.7.5 and 1.3.14.15), it has come to be known as
his copy principle. Simple perceptions are thereby understood as perceptions which cannot be further divided or analysed into smaller perceptions (ibid.: 1.1.1.2). And since complex ideas are composed of simple
ones, the former inherit their resemblance with and dependency on simple
impressions from the latter. Hume thus maintains that we cannot think of,
imagine or remember something, the various parts and aspects of which
(e.g., their colours, shapes, etc.) we have not previously perceived. This
does not require that complex ideas have to be actually caused by corresponding complex impressions. It is sufficient if they are constructed out of
simpler ideas which are causally dependent on precedent simple impressions (ibid.: 1.1.1.4f.). This entails that complex ideas despite being possibly caused by corresponding complex impressions causally depend for
their occurrence only on the respective simple impressions.
Now, the causal derivation of ideas from impressions is not the only aspect of their relationship highlighted by the copy principle. One further aspect is that Hume also understands ideas as corresponding to the respective
impressions by resembling them in all respects but their degree of vivacity
(ibid.: 1.1.1.3). In particular, they resemble each other in which objects and
features they present us with, albeit presenting them in differently vivid
manners (ibid.: 1.3.7.5). Moreover, there is a third aspect which, like the
causal dependency but unlike the resemblance, is asymmetric in nature:
ideas are images or representations of impressions (ibid.: 1.1.1.1,
1.1.1.4, 1.1.1.7 and 1.1.3.4).

286

The Dependency Account

That a given complex idea may be caused merely by several simple impressions, and not also by a corresponding complex one, raises the issue of
which impression(s) it is said to resemble. It seems plausible to maintain
that the complex idea resembles each of the simple impressions in so far as
it possesses parts (i.e., simple ideas) which resemble the latter. On the other
hand, and as already noted, the resemblance at issue here concerns primarily the (non-mental) objects and features presented by the ideas and impressions compared. And this similarity is unlikely to hold between complex ideas and simple impressions, at least with respect to the presentation
of objects and of higher-level properties. For Hume understands simple impressions as presentations of basic features, such as colours, tastes or
smells (ibid.: 1.1.1.2). Hence, although complex ideas may resemble each
of the respective simple impressions, this is not the kind of similarity referred to in the copy principle. Instead, what is meant is the resemblance of
ideas on those impressions that present the same objects and features as the
former in this case, the resemblance of complex ideas on similarly complex impressions. However, it remains unclear which complex impression
could be relevant in cases where a complex idea is not preceded by any
corresponding complex impression. It is true that such ideas would resemble a complex combination of the relevant simple impressions, if any
would actually come into existence. But this is not the same as saying that
the complex idea resembles an already given impression.
The same set of issues does not arise with respect to the proposed representational link between ideas and impressions. On the one hand, in contrast with resemblance, representation need not be genuinely relational in
the sense of requiring an actually given second relatum. And, on the other
hand, a complex idea represents not simply individual simple impressions,
but also their complex structure. Imagining a blue book on a brown table
corresponds to seeing a blue book on a brown table, and not to seeing a
brown book on a blue table, despite both complex impressions involving
the same simple ones. Accordingly, what complex ideas represent are complex impressions, though not necessarily particular ones (e.g., the one I had
yesterday when looking at my friends house).8
8

The preceding considerations liken Humes complex ideas to pictures, given that
both allow for the presentation of some kind of object, without the need to present a
particular and actual instance of that kind; and given that both may perhaps still give

The Representational Account

287

This suggests perhaps also a solution to the resemblance issue. The key
thought is that ideas present certain (non-mental) objects and features precisely because they are representations of impressions which present those
objects and features. This has the result that impressions and ideas differ in
immediacy. But it also means that ideas resemble the impressions that they
represent in so far as they present the same (non-mental) objects and features as the latter albeit typically in a less vivid way. For they represent
an impression with a certain property such as a certain content or phenomenal character and thereby acquire themselves this property, or at
least a property very similar to it (see the discussion of immediacy in Section 3.2). A complex idea therefore resembles the complex impression that
it represents. And although the latter need not enjoy actual existence in the
past or present, it is clear which complex impression is relevant. Moreover,
we do not generally find claims about the resemblance between actual and
merely represented entities problematic. We may recognise similarities
between friends of ours and characters in a film; or between a real person
and our mental image of her which we have formed prior to meeting her or
knowing anything about her appearance.
These considerations allow now for a more precise reading of Humes
copy principle. Since the symmetric resemblance between ideas and impressions is probably best understood as a consequence of their asymmetric
representational link, we need not any more mention the former separately.9 According to the resulting interpretation, the principle maintains that
particular ideas are copies of particular impressions in that they (i) causally
depend on the later, and (ii) are representations of the latter in such a way

rise to an experience of resemblance. A painting may depict a type of man with a


certain appearance, without depicting a particular man (e.g., Socrates or Napoleon);
and it may still look like such a type of man (see Hopkins (1998) and Sections
10.3f.). This analogy may provide further support for the suggested interpretation of
copies as reproductions similar to photocopies or photographs.
At one point, Hume writes that impressions and ideas [...] are exact copies of each
other (Hume (1739): 1.1.1.5). This seems to equate the relation of copying with the
relation of resembling. My different usage follows instead that to be found in the
contemporary literature on Hume, as well as presumably in other passages in
Humes text (e.g., ibid.: 1.3.7.5). My aim is, in any case, a reconstruction, not of
Humes use of the word copy, but of his conception of the relationship between
ideas and impressions.

288

The Dependency Account

as to end up presenting the same (non-mental) objects and features. Hume


thus endorses the conjunction of the causal echo thesis (EC) and the representational echo thesis (ER).
(EC) Imaginings are essentially causally dependent on perceptions
(and other cognitions).
(ER) Imaginings are essentially representations of their cognitive
counterparts.
In the case of simple ideas, both the causal and the representational relation
hold between the ideas and simple impressions. And a specific simple idea
always represents a particular simple impression. Complex ideas, on the
other hand, causally depend on each member of a certain set of particular
simple impressions, while representing and resembling a complex impression though not necessarily a particular one.
Humes conception of the relationship between simple ideas and simple
impressions seems thus to be that the former are causal reproductions of
the latter perhaps not dissimilar to photocopies or photographs. 10 For
such reproductions combine the same three elements of causal dependency,
representation and resemblance. It is constitutive of photocopies of sheets
of paper that they are causally dependent on the respective original sheets,
and also that they are photographic representations of the latter and resemble them in respect of what is written or drawn on them. Complex
ideas deriving from complex impressions may equally count as causal reproductions of the latter with the exception that they are not causally dependent on them, given that they might have come into existence without
the involvement of the complex impressions. For those cases, the analogy
of a collage of photocopies seems more fitting. If we glue together several
photocopies, the resulting collage is causally dependent on the initial
sheets of paper, while representing and resembling a corresponding
potential collage of the originals, though no particular one.
Drawing the analogy with photocopies may perhaps also offer an explanation of why ideas are less vivid than impressions and therefore also
10

I follow here a general line of interpretation proposed by Mike Martin in a research


seminar on the Treatise at University College London in the academic year 2002/03.

The Representational Account

289

of how the two distinct characterisations of the difference between the two
types of mental episode fit together. In the case of photocopies, the contrast
and saturation of the marks on their surface tend to be less than those of the
marks on the original sheets of paper. Similarly, the vivacity of episodes
might be understood as an aspect of their subjective character which is
bound to decrease when mentally reproduced. That is, this reproduction
might be of such a nature that it results not only in the episodes inheritance of the presentation of certain objects and features, but also in a diminishing of their sense of reality or presence and of their impact on beliefs,
emotions and actions.
We have finally reached the point where we are in a position to become
more concrete about Humes conception of imaginings. Both his examples
of imagining and his discussion of the difference between imagining and
remembering something suggest that he takes imaginings to be complex
ideas. Imagining the New Jerusalem, whose pavement is gold and walls
are rubies (ibid.: 1.1.1.4), or winged horses, fiery dragons, and monstrous
giants (ibid.: 1.1.3.4), means combining simpler ideas whether the latter
are still somewhat complex (such as the ideas of horses, wings, and so on)
or indeed among the simplest (such as the ideas of colours, tastes, smells,
and so on). Correspondingly, ideas of the imagination differ from ideas of
memory in whether they preserve the order in which the relevant impressions occurred before (ibid.: 1.1.3.2).11 And this, again, presupposes that
both are complex ideas possessing an internal structure. According to the
proposed reading of the copy principle, Hume therefore maintains that it is
constitutive of imaginings that they represent complex perceptions and
causally depend on the prior perceptual occurrence of the simple aspects of
those perceptions.12
Note that this fact is not directly subjectively accessible (see the passage from Hume
(1739): 1.3.5.3 quoted above). It is interesting to ask whether the also postulated and
subjectively salient difference in vivacity between memories and imaginings might
be said to indirectly reflect this difference of how the two kinds of episodes are
taken to relate back to the original perceptions.
12
I assume here that our perceptions are simple that is, for instance, of a single colour and no other sensible quality only in rare and artificial circumstances. Perhaps
there may also be simple instances of imagining, in which case Humes conception
would have to be slightly modified as to allow for simple ideas of the imagination as
well.
11

290

The Dependency Account

9.2. Objections to Humes View


Humes theory of the mind faces many challenges, some of which concern
his theory of mental episodes in general. One of these has already been acknowledged by Hume himself, namely that it seems possible to have
simple ideas (e.g., the one of the missing shade of blue) without having
had before the corresponding simple impression (see Hume (1739):
1.1.1.10). The universality of his copy principle becomes therefore questionable, even in its restriction to simple ideas and impressions.
Another objection targets the fact that the differences in vivacity and
hence the resulting differences between mental episodes are taken by
Hume to be quantitative, and not qualitative. This contradicts the observation that perceptions, judgements, memories, imaginings, and so on, differ
in kind, and not merely in degree. They differ, for instance, in whether they
are sensory or intellectual, in whether they involve a cognitive attitude
(i.e., whether they involve the claim that things are as they present them to
be), in whether they possess an epistemic function, or in whether they
provide us with reasons for belief, or are responsive to them (see Note 6 in
Chapter 1, as well as Dorsch (2009b) and (2013)). These differences are
not only subjectively salient; they also distinguish the episodes concerned
qualitatively, and not merely quantitatively (see Sections 2.1f.).
Furthermore, it should be explained why some ideas (i.e., judgemental
thoughts and conscious memories) are more vivid than others (i.e., imaginings) to such an extent that the former, but not the latter, have an impact on
what we believe about the world in a way very similar to that of impressions. Especially Humes discussion of episodic memories reveals that he
himself struggled more or less explicitly with this problem. While generally assuming them to be ideas (see, e.g., Hume (1739): 1.1.1.4 and
1.1.3.1), he then moves them sometimes closer to impressions to account
for the perception-like impact on belief. Thus, when trying to distinguish
them from imaginings, Hume locates memories betwixt an impression and
an idea, to reflect the fact that their vivacity is in between those of
perceptions and imaginings (ibid.: 1.1.3.1). And when trying to specify
why memories have the same power as perceptions to give rise to beliefs,
Hume even speaks of them as impression[s] of the memory (ibid.:
1.3.5.1) though not without seemingly relativising this statement shortly

The Representational Account

291

afterwards by apparently reintroducing the contrast between perceptions


and memories:
To believe is [...] to feel an immediate impression of the senses, or a repetition of that impression in the memory. (Ibid.: 1.3.5.8)

If repetition of an impression is understood here as meaning the literal reoccurrence of the original impression, the explanation of the memories impact on beliefs has the price of rendering them indistinguishable from perceptions. But if it is instead taken to denote, in accordance to the copy principle, the less vivid representation of a perception, the initial problem of
accounting for the memories perception-like link to belief reoccurs. This
illustrates that Hume has difficulties to accommodate the fact that memories are very similar to perceptions in their epistemic function, while also
holding on to their differences in vivacity and immediacy that is, their
differences in how they present us with objects.
Finally, the first characterisation of the difference between impressions
and ideas is in tension with the second one precisely because of this aspect
of Humes conception of vivacity (and despite the explanatory link
between the two mentioned above). For treating ideas as copies of impressions in the sense just specified means treating the two as being different in
kind. Hence, it clashes with the claim that the only difference between the
copies and what they are copies of is one of vivacity (ibid.: 1.1.1.3). In
fact, this inconsistency seems already to be inherent to the copy principle
itself, given that its second clause postulates a resemblance in all respects
except vivacity (i.e., a difference in degree), while the third clause puts forward the idea of a representational link (i.e., a difference in kind).
Now, Humes theory of the mind is subject not only to general challenges to his theory of mental episodes. Some objections are also more
specifically related to the particular incorporation of imaginings in his
overall view of the mind. Hume claims that episodes of imagining are
causal reproductions and therefore representations of their cognitive counterparts. In addition to the difficulties already mentioned, this thesis is
problematic for at least two reasons.
The first is that it is unclear how to avoid the conclusion that all imaginings (just like all thoughts or judgements) involve some sensory or affect-

292

The Dependency Account

ive element, given that they are or include copies of perceptions with sensory or affective qualities. This idea makes sense in the case of affectively
imagining a pain or of sensorily imagining something red: the character of
the first episode contains some element of painfulness, and the second
some quality of reddishness.13 But we can suppose (or, more generally,
think) that an object is a certain way without any sensory or affective element involved. The underlying problem is, again, that Humes theory cannot properly accommodate the distinction between sensory, affective and
intellectual episodes, especially given that this distinction is one in kind.
The second reason for being uneasy about the proposed claim about
imaginings is that it can at best play only a minor role in an account of
imagining. That imaginings are dependent on perceptions in the way described is not distinctive of them within Humes theory. For the same is
said to be true of episodes of memory, thought and judgement. This leads
back to the worry that, ultimately, reference to degrees in vivacity is his
only means to establish differences among mental episodes, while the more
fundamental or significant ones of these are in fact differences in kind.
Moreover, even if Humes copy principle is read in such a way as to describe a qualitative difference between impressions and ideas, it does not
suffice to strictly distinguish imaginings from other kinds of ideas, such as
judgemental or mnemonic ones. Humes conception of imaginings therefore does not and also was probably not intended by him to provide the
resources to formulate a proper theory of imaginings, given that it does not
really treat them as a distinct kind of mental phenomena.

9.3. Causal Echo Claims


Before addressing the possibility of improving on Humes conception of
imaginings in such a way as to use it as the basis for a unified account, it is
worthwhile to discuss the prospects of the causal echo claim (EC) in its
own right that is, especially, independently of its representational analog
(ER) and of Humes theory of the mind. One interesting thing to note about
13

Though neither suffices to qualify the episode as a genuine experience of pain or


redness, given that they do not present these qualities as actually being instantiated
(see Sections 10.1, 11.5 and 14.3, as well as Dorsch (2010a) and (2013)).

The Representational Account

293

Humes characterisation of imaginative episodes as representationally and


causally depending on cognitive ones is that the two postulated dependency relations concern different cognitive relata. According to Humes
copy principle, while the (complex) imaginings are said to represent corresponding (complex) perceptions, they are taken to causally depend on the
prior occurrence of simple elements of those perceptions. An episode of
visualising a red cube, for example, is assumed to reproduce an episode of
seeing a red cube, without having to be caused by such a perceptual episode. Instead, the occurrence of this episode of visualising is understood as
requiring only the prior occurrence of an episode of seeing something red
and an episode of seeing something cubic (or with even simpler shapes).
Now, the assumption of such a difference in cognitive relata is well justified, at least with respect to the causal echo claim. For no version of (EC)
should make the stronger claim that each instance of imagining has to have
a causal cognitive predecessor with the same content and the same type of
representation. We can visualise aliens without ever having seen any, we
can imagine how a papaya fruit might taste without ever having eaten one,
and we can suppose that Timbuktu is the capital of Mali without ever having judged or even thought about it. It seems that, at best, it may be argued
that, in intellectually imagining, we cannot use concepts which we did not
already acquire during certain past processes of belief-formation; or that, in
sensorily or affectively imagining, we cannot combine sensory or affective
representational elements concerned with basic features (simple colours,
tastes, etc.) if we did not already experience them as part of some prior perceptions, sensations, feelings, and so on (see Section 12.1). In short, imaginings should at best be said to causally depend, not on cognitions with
the same content, but on cognitions which ensure the availability of the relevant representational means or materials. As a consequence, the defensible versions of (EC) turn out to be rather weak and, in particular, not to
hold between imaginings and their cognitive counterparts.
This entails, however, that they cannot contribute much to a unified account of imagining. It is arguably true that our power to imagine something
depends on our sensory and conceptual representational capacities; and
that we can acquire the latter only by perceiving and cognising the world.
But the same applies to our power to desire or remember something, or to

294

The Dependency Account

present it in an emotionally loaded manner, say. If we cannot intellectually


imagine water to smell of roses or visualise a red cube without having believed something about water or seen something red, then the same should
be true of desiring water to smell of roses or visually recalling something
red. At least, the standard externalist arguments do not explicitly or implicitly assume any relevant distinction between imaginings, on the one hand,
and non-imaginative states, on the other (see, e.g., Putnam (1975) and
Burge (1979)). Hence, the fact that the acquisition and employment of representational capacities presupposes having had suitable prior cognitions is
not distinctive to imaginings, but concerns many non-imaginative forms of
representation as well. If a related version of (EC) turns out to be true, it
can therefore constitute at best a small part of a unified account of
imagining.

9.4. Representational Echo Claims


Representational echo claims of the form (ER) promise to be much more
illuminating about the nature of imaginings than their causal complements
of the form (EC). In particular, the assumption of a representational dependency is more substantial than the assumption of a causal dependency
in being more resourceful in the explanation of central features of imaginings such as their lack of a cognitive attitude, or their diminished vivacity
(see Sections 3.2 and 10.4). Indeed, this difference in substantiality becomes manifest in the fact that the application of (ER) to imaginative states
is far more problematic and controversial than the application of (EC) to
them. In consequence, it should be expected that (ER) is more likely than
(EC) to single out what is distinctive of imaginings.
Indeed, considerations like these may provide enough motivation to formulate an account of imagining solely in terms of (ER), and to argue that
the postulated representational dependency is limited to imaginings (and,
possibly, memories). Dropping (EC) in this way from the Dependency Account (without necessarily denying the truth of the causal dependency
claim) makes it possible to treat imaginings as distinctive mental phenomena, and thus to avoid one specific difficulty for Humes conception,
namely that it cannot ground a proper theory of the nature and unity of

The Representational Account

295

imagining. In accordance with these thoughts, the Representational Account of imagining (i.e., the representational version of the Dependency
Account) aims to elucidate the nature of imaginings primarily if not exclusively in terms of the following fundamental echo claim:
(ER)

Imaginings are essentially representations of their cognitive


counterparts.

Endorsing (ER) should in fact mean endorsing the view that all instances
of imagining are instances of experiential imagining (see also Sections 1.2,
10.4 and 11.5, as well as Walton (1990): ch. 1, and Walton (1997)). The alternative option of treating imaginings as intellectual representations of
(i.e., thoughts about) cognitions is evidently untenable because sensory and
affective imaginings cannot be reduced to purely intellectual episodes (see
Section 11.3). Moreover, even intellectually imagining that the Earth is
flat, say, is not the same as entertaining the (higher-order) thought that one
judges or believes that the Earth is flat. In particular, supposing that the
Earth is flat does not require the possession or application of the concept of
judgement or belief.
But how does the idea that the objects of imaginative representation
that is, what is imagined are cognitive episodes square with the fact that
we may visualise landscapes, or suppose that the Earth is flat? The idea is
to distinguish between direct and indirect objects of representation. What
imaginings are said to represent directly are cognitions. That is, the representational content of the imaginative episodes consists just of cognitive
episodes. But cognitions are themselves representational: they represent
landscapes, the flatness of the Earth, and so on. Hence, what proponents of
the Representational Account assume is that, by (directly) representing
cognitions, imaginings also (indirectly) represent what the cognitions represent. The imaginative episodes are understood as inheriting the representationality of the cognitive episodes in so far as the latter are together
with their content part of the content of the former. 14 Of course, (ER)
14

See Section 10.4 for a more detailed discussion of how this is supposed to work in
the case of visualising. In the case of intellectual imagining, this form of inheritance
is likely to be similar to the sort of embedding that may occur in the case of higherorder thought. Thinking that one is thinking that it rains literally includes thinking

296

The Dependency Account

could be reformulated in such a way as to make this assumption more explicit. But this would render it unnecessarily complicated. It seems sufficient to say that imaginings are representations of cognitions to indicate
that they represent external objects not in the same (direct) way, but instead
(indirectly) by representing cognitive representations of them.
Now, simply dropping (EC) and shifting the focus exclusively to (ER) is
not enough to address all the problems for the Humean approach that have
been noted in the previous sections of this chapter. But this general approach to imaginative episodes is open to further improvements, without
having to give up on the fundamental idea that imaginative episodes depend on cognitive ones by being representations of them. Two additional
important modifications are, first, the introduction of the qualitative distinction between sensory and intellectual presentations; and second, the
substitution of the qualitative differences in attitude and epistemic function
for the quantitative difference in vivacity. Humes contrast between impressions and ideas is thus replaced by the opposition of cognitions and
imaginings, as well as the orthogonal opposition between sensory and intellectual episodes. As a result, not only perceptions, but also episodic
memories and judgemental thoughts are taken to be episodes that do not
represent others in the relevant way, but instead can figure as the objects of
such a representation. And only imaginings plus possibly memories (see
below) and spontaneous images continue to count as representations of
cognitions.
These two further modifications suffice to solve several of the other difficulties facing Humes theory of the mind. The resulting version of the
Representational Account can accommodate the fact that the various kinds
of cognition and imagining differ in kind from each other, and along the dimensions outlined above. It also avoids any of the problems linked to the
introduction of vivacity as an important element in an account of imaginings. Intellectual imaginings need not involve any (more or less vivid)
sensory elements any more, since they can now be construed as representations of intellectual cognitions, such as judgemental thoughts or occurrent
that it rains. That is, there are two thoughts and two contents which are such that
one of them contains the other. A similar form of inclusion or overlap can be expected to be present in the case of imagining thinking that it rains, which also entails
thinking that it rains (see Section 9.5).

The Representational Account

297

beliefs. Finally, the modified theory comes closer than the original to the
identification of a distinctive feature of imaginings, which separates them
strictly from cognitions and other non-imaginative mental episodes. For
while it is said to be constitutive of imaginings that they are representations
of cognitions, the same is not true of perceptions, judgemental thoughts,
bodily sensations, feelings of emotion or desire, and so on.
The application to episodic memories may appear to remain problematic, however. They still seem to fall in between perceptions and instances
of sensory imagining. While they share their cognitive attitude and epistemic function with the former, they do not present their objects as being
there before us in our environment and, in this respect at least, resemble
the latter. This raises again the issue of how they can actually share their attitude and impact on belief with perceptions, despite their lack of the latters direct connection to reality. And it also generates the question of
which set of features is distinctive of imaginings, if it turns out that episodic memories, too, are best treated as representations of perceptions.
But the Representational Account might have the resources to satisfactorily address these two issues, and thus to properly deal with episodic
memories (see Martin (2001)). If episodic memories are indeed representations of past perceptions, they may inherit the particular content of the latter. That is, they may also be (indirect) representation of the specific objects and features then perceived and, moreover, represent these objects
and features as they were once represented by ones past perception. In this
way, episodic memories may provide us with access to particular aspects of
the past. And this fact may very well explain why they involve a subjectively salient commitment to how things actually were, and why they influence our beliefs in roughly the same way as the original or other perceptions. They would still differ from the latter in that they do not represent
their objects as being there before us, but locate them in the past.
The contrast with sensory imaginings might then be established by arguing that the imaginative episodes although representing some type of
perceptions do not represent particular perceptions (see Martin (2001)).
This would ensure that they do not bring us into contact with the (present
or past) actual world something which is reflected by the fact that their
objects are not given to us as actual, and that they do not show the cognit-

298

The Dependency Account

ive attitude and impact on belief of cognitions. The Representational Account may therefore identify the representation of non-particular cognitions
as the distinctive feature of imaginings. And this would not only suffice to
distinguish them from episodic memories and other non-imaginative episodes, but would also promise an account of their lack of epistemic features.
Indeed, it should become clear now that separating the representational
echo claim from the causal one and understanding solely the former as being central to the Representational Account does not only promise the reward of a unified account of imagining, but also has the further advantage
of allowing for non-mechanistic interpretations of the postulated
representational link between imaginative and cognitive episodes. While
Humes empiricist picture of the mind leads him to take imaginings to be
causal reproductions of particular cognitions, it is now possible to conceive
of them also as intentional or relational representations of types of cognitions. That is, instead of stressing the causal dependency between imaginative and cognitive token episodes, it may be argued that imaginings constitutively depend on types of cognition in the sense that the latter are necessarily part of the representational content of the former (see Peacocke
(1985), OShaughnessy (2000): 365f., or Martin (2002a): 404-407).15
The result are endorsements of (ER) that maintain, for instance, that
visualising amounts to imagining seeing, or that supposing is nothing but
imagining judging or believing. In other words, visualising and supposing
are assumed to be nothing else than instances of experiential imagining. 16
As already mentioned above, the content of the cognition is thereby taken
to be embedded in some form or another in the content of the imagining. With respect to intellectual imaginings, the conceptual content of the
judgement or belief simply constitutes part of the content of the supposiThe idea of representing types, in contrast to representing tokens, certainly needs
further investigation (see also Section 10.4). But that this distinction obtains can be
independently illustrated by reference to pictures: while some paintings (e.g., portraits) are of particular people, others are of no particular people at all, although they
depict people or, rather, types of people (e.g., in genre paintings; see Wollheim
(2003)).
16
Note that the claim here is not the (relatively uncontroversial claim) that there are instances of experiential imagining, but rather the claim that the basic episodic forms
of imagining (i.e., sensory, intellectual and, presumably, also affective imagining)
can be reduced to another form of imagining, namely experiential imagining.
15

The Representational Account

299

tion. In the case of sensory imaginations, the situation is a bit more complex since the sensory content of perceptions may loose some of its aspects
(e.g., its immediacy, or its vivacity) when imaginatively represented (see
Sections 3.2 and 10.4, as well as Martin (2002a)).
Taking imaginings to be dependent on types of cognitions, rather than
on tokens, has the consequence that imagining something does not require
any more the prior occurrence, in the same subject, of instances of the cognitive types concerned. Hence, the proposal can accommodate the possibility (if it is a genuine one) of, say, the visualisation of a previously unperceived shade of blue: we visualise this colour, not in virtue of having previously seen it, but by imagining having a certain type of perceptual experience, namely one of that colour. In addition, the representational view under discussion does not need to restrict (ER) to simple cases or aspects of
imaginative representation. For the view can explain why we are able to
visualise a unicorn without previously having seen one by reference to the
existence of the corresponding type of perception which, after all, could be
instantiated if unicorns would exist. Besides, the view may also be able to
account for the dependency of sensory memories on perceptions in a way
that does not render them imaginative (i.e., that does not render their dependency indistinguishable from that of imaginings). For, as suggested
above, it may be argued that, while episodic memories represent and depend on previously existing tokens of experience, sensory imaginings are
so related to types of experience (see Martin (2001): 279). What these considerations show is that formulating the Representational Account in terms
of a type-based representational echo claim avoids most of the remaining
problems facing Humes causation-centred conception of imaginings.
It might seem most natural to understand the postulated representational
link between imaginings and cognitions in intentional terms. After all, we
are capable of imagining cognitive episodes which no one ever has actually
had (e.g., when imagining the perceptions of the first man on Mars, or the
memories of the pyramids). But that this is not the only option is illustrated
by OShaughnessys account, the core idea of which is much closer to
Humes copy principle than any intentionalist echo thesis.17 In OShaugh17

See Section 6.2 for a more detailed discussion of OShaughnessys view. He hardly
ever makes explicit references to the philosophers who have had a (positive or negative) impact on the presentation and content of his own arguments. But he does so

300

The Dependency Account

nessys theory, imaginings have similar contents and involve similar types
of representation as cognitions because they are shadows or replicas of
the cognitive prototypes (see OShaughnessy (2000): 363ff.). What he
has here in mind is the idea that imaginings have types of veridical perception or knowledge as their immediate, non-intentional and necessary objects (not unlike sense data in the case of perception) and borrow their
world-directed intentionality from the latter. This is actually what he means
when he speaks of cognitive prototypes for imaginings: the imaginings
must not only be described in terms of the prototypes, but are also necessarily modelled upon them, that is, constitutively related to them in the special way described. It is in this sense that we are said by him to imagine
something by representing the relevant cognitive prototype: that is, visualise a tree by imagining seeing (or seem seeing) one, or make-believe that
the Earth is flat by imagining believing (or seem believing) the respective
proposition (see OShaughnessy (2000): 363f.).18
The closeness to Humes original account manifest in the postulation
of a genuinely relational, and not merely intentional, link between imaginings and their cognitive counterparts has the consequence that, for
OShaughnessy, imaginings of a certain kind (e.g., with a certain content or
of a certain type) could not exist if the corresponding kind of perception
would not exist (see OShaughnessy (2000): 377). It is not clear whether
in this context and, of course, to Hume which suggests that even OShaughnessy
himself sees himself as standing in the tradition of Hume (see OShaughnessy
(2000): 365).
18
As already mentioned in Note 11 in Section 6.2, there is at least one good reason to
prefer an intentionalist understanding of the representational link between token
imaginings and types of cognitions over a relationalist conception: namely the difficulty of rendering intelligible the idea of a genuinely relational form of representation, which has, as one of its constituents or relata, a type of episode, rather than a
particular episode. Since there cannot be genuinely relational forms of awareness
with purely general contents, the solution has to be something like the proposal to
treat the represented prototypes themselves as particulars, rather than as universals
(or their nominalist equivalents). But it is unclear how this might be supposed to
work. Note, however, that OShaughnessy clearly speaks of imaginings as representations of types of cognition although he acknowledges that the imagining
subject is not aware of the meta-representational aspect of imagining, but instead enjoys a mental episode that is, so to speak, transparent to the imagined external objects and their features (see OShaughnessy (2000): 364f.).

The Representational Account

301

the intentional variant has to embrace this strong constitutive dependency


as well although it might generally be plausible to endorse it in the light
of the apparent impossibility of imagining things which we could not perceive or think of. In any case, both the intentional and the relational reading of the claim that imaginings are representations of cognitions differ
from Humes causal conception in that they do not postulate a relation to
tokens of cognition.
What remains to be seen is whether the Representational Account is
true. In the remaining two sections of this chapter and the following two
chapters, I discuss the application of (ER) to, respectively, intellectual imaginings, imaginative projects, sensory imaginings and affective imaginings.
My conclusion is that the Representational Account captures an important
aspect of the nature of the last two forms of imagining, but not of the first
two. This result fits well with the fact that proponents of the Representational Account with the notable exception of Hume (see above) and
OShaughnessy (see below) tend to endorse its main echo claim only
with respect to visual or, more generally, sensory imaginings (see Chapter
10).
In the next two chapters, I argue more specifically for the thesis that instances of visualising and emotionally imagining something amount to instances of representing corresponding experiences of seeing or feeling. Although I do not defend this further claim here, I also think that as already
suggested above the respective imaginings have this feature in common
with other sensory or affective presentations, such as episodic memories or
spontaneous images. More needs therefore to be said about what distinguishes the imaginative from the non-imaginative cases such as its involvement of imaginative activity, say (see Part Four).
The remaining two sections of the current chapter, on the other hand, are
reserved for the discussion of intellectual imaginings and imaginative projects, and their resistance to the application of (ER). Hence, although this
echo thesis is relevant for a unified account of imagining in so far as its
truth with respect to sensory and affective imaginings needs to be accommodated and accounted for, it cannot form a proper part of a unified theory
since it fails to capture what is common to all instances of imagining. In
short, while the Representational Account is right about the representation-

302

The Dependency Account

al nature of the sensory and affective elements involved in the respective


forms of imagining, it is wrong about what renders them or other forms
of imagining imaginative.

9.5. Intellectual Imagining


What the different versions of the representational echo thesis (ER) have in
common is that they take all imaginative episodes to be experiential imaginings, that is, non-propositional representations of cognitive episodes
(such as perceptual experiences or occurrent beliefs). But they differ, as
already noted, in respect of how they conceive of the representational connection. Hume took it to be a form of causal reproduction (see Sections
9.1). Contemporary philosophers defending (ER) with respect to sensory
imaginings typically assume an intentional link (see Section 9.4). 19
OShaughnessy, the main proponent of the application of the Representational Account to intellectual imaginings, on the other hand, understands
the representationality involved in non-intentional terms, perhaps to be
modelled on the relation of awareness holding between perceptions and
perceived sense-data (see the relevant notes in Section 7.1, as well as
OShaughnessy (2000): 363ff.).20 In what follows, I evaluate the general
prospects of (ER), independently of whether causal indication, intentionality or another form of representationality is taken to be constitutive of imagining.
With respect to intellectual imaginings, the proposal is to treat them as
representations of types of occurrent beliefs (or judgemental thoughts): intellectually imagining that p means, essentially and fundamentally, imagining consciously believing that p. Accordingly, what is maintained is that
imagining that something is the case is literally imagining believing (or
See Peacocke (1985) and Martin (2001: sec. 3; 2002: sec. 3), as well as the discussion in Section 9.4.
20
Just like in the case of sensory imaginings, OShaughnessy assumes not only a nonintentional link between intellectual imaginings and intellectual cognitions, but also
an intentional link between the imaginings and the propositional objects of the nonintentionally represented cognitions: a make-believe quasi-belief is intentionally
directed to the very same propositional object as its cognitive model (OShaughnessy (2000): 363).
19

The Representational Account

303

judging) that it is the case. It is important to understand this as a substantial


point, and not merely as a manner of speaking: it implies that all intellectual imaginings involve a complex content (or perhaps two kinds of content) including not only the respective proposition but also a type of belief.
It will be helpful to contrast this echo thesis with a plausible alternative
picture not involving the endorsement of (ER) with respect to intellectual
imaginings. The core idea of this alternative view is to assume that the content of intellectual imaginings is exhausted by the entertained proposition.
Accordingly, imaginative thoughts have a type of belief as their (intentional) object only in case they are higher-order thoughts, in which we explicitly imagine that someone has a certain belief. With respect to their
content, intellectual imaginings are therefore said to be no different from
mere thoughts, judgements, wonderings, (conscious) desires, and the like.
The alternative view may be supplemented by the idea that intellectual
imaginings involve some further element which distinguishes them from
(other) propositional attitudes: their active formation, say, or an imaginative attitude towards their proposition. But what counts most here is that the
two proposals differ in what kind of content they ascribe to the imaginings.
The high implausibility of (ER) in the intellectual case derives from two
sources: (i) there seem to be no good arguments in favour of it; and (ii)
some considerations actually speak strongly against the application of (ER)
to intellectual imaginings. One might also be skeptical about this application because it assumes that occurrent beliefs (or judgemental thoughts)
possess a phenomenal character, which is moreover fine-grained enough to
distinguish imagining the instantiation of the character of consciously believing that p from imagining the instantiation of the character of consciously believing that q. But I take this worry to be unfounded, as I argue
at length in Soldati & Dorsch (2005) and, more briefly, in Dorsch (2009b)
(see also Strawson (1994) and Siewert (2000)). Once it is accepted that
conscious beliefs possess a sufficiently specific phenomenal character, the
postulated representational link between intellectual imaginings and types
of occurrent beliefs becomes unproblematic: it just amounts to object
awareness directed at a mental episode and its phenomenal features (see
Sections 1.2, 10.4 and 11.5).
There seem to be two plausible motivations for endorsing (ER) with re-

304

The Dependency Account

spect to intellectual imaginings. The first is that this endorsement can fulfil
a certain explanatory role: namely to explain the similarities between intellectual imaginings and beliefs, and yet to elucidate why and perhaps even
to entail that the former cannot instantiate types of the latter (e.g., the respective cognitive prototype). More specifically, one component of the idea
is that intellectual imaginings involve the same contents as beliefs simply
because they are representations of types of beliefs, and hence include the
contents of the latter as part of their own contents. And the other component of the idea is that this representational link to types of beliefs also illustrates how they differ from beliefs, and perhaps why they cannot provide
knowledge in the same way as beliefs.
The second plausible motivation for the endorsement of (ER) in the intellectual case is that it promises to shed more light on the relationship
between these imaginings and the corresponding judgements or beliefs.
That there is an intimate link between the two kinds of phenomena has
been expressed by means of the idea that the elucidation of the nature of
imaginings has to make essential and substantial reference to beliefs (see
Scruton (1974): 100; Currie & Ravenscroft (2002): 32; see also Section
12.1). For OShaughnessy, this is part of the truth of his version of (NC*):
that imaginings are, essentially and fundamentally, incapable to exemplify
the two cognitive prototypes of successful perception and intellectual
knowledge. He concludes from this that any account of imaginings has to
elucidate them in terms of the cognitive prototypes and, in particular, intellectual imaginings in terms of knowledge-constituting beliefs. But
simply saying that imaginings cannot realise the two forms of knowledge
does not say much about the relationship between intellectual imaginings.
OShaughnessy may therefore have felt the need to specify the connection
between imaginings and, especially, intellectual imaginings and the respective cognitive phenomena in more positive terms: namely by means of
(ER) or, more precisely, his particular version of (ER) (see Section 6.2). A
similar desire may have moved other philosophers to endorse their versions
of the representational echo thesis.
However, these suggestions cannot settle the issue in favour of the application of (ER) to intellectual imagining. On the one hand, endorsing this
representational echo thesis is not the only way to account for the similarit-

The Representational Account

305

ies and differences between intellectual imaginings and beliefs. On the other hand, (ER) is not the exclusive option of providing a positive account of
what it means to be an intellectual imagining. Let me discuss each point in
turn.
First, the introduction of (ER) is not the only way to explain how intellectual imaginings are both similar to and different from beliefs (or judgements).21 For instance, the alternative picture of intellectual imaginings described above can account for the similarities by reference to the possibility that intellectual imaginings and judgements can involve the entertainment of the same propositions. And it can trace back their differences to
the presence or absence of a cognitive attitude to this content. In particular,
one consequence of this lack of cognitive attitude seems to be that intellectual imaginings cannot stand in the required rational relations to other beliefs (and also to perceptions). Imaginatively (or merely) entertaining propositions cannot provide rational support for the judgemental endorsement
of a proposition, even if the latter can be inferred from the former. If two
propositions, which we endorse, entail a third proposition, we will not imaginatively (or merely) entertain the implied proposition, but endorse it as
well. In accordance with this, the form of normativity or rationality pertaining to intellectual knowledge concerns reasons for endorsement, and
not reasons for entertainment.
This may, indeed, be a satisfactory explanation of why intellectual imaginings cannot constitute knowledge. It may also imply that being an intellectual state of knowledge requires the involvement of a cognitive attitude
which may explain why there do not seem to be any intellectual counterexamples to (NC). It does not, however, necessarily entail that states of
knowledge have to be of the same kind of mental state as false beliefs: disjunctivism may still be an option (see Dorsch (2011a), and Note 3 in
Chapter 8). Nor does it imply that sensory grounds of knowledge have to
involve a cognitive attitude. For such grounds do not have themselves to be
rationally supported. The idea here is just that reasons for belief are always
reasons for endorsement and not that reasons for belief can be provided
only by rational endorsements (see, for instance, Pollock & Cruz (1999)).
Second, a similar strategy can be adopted towards the claim that what
21

For a similar argumentative strategy with respect to sensory imaginings, see Hopkins (1998): ch. 7.

306

The Dependency Account

speaks in favour of the introduction of (ER) is that it can specify the nature
of intellectual imaginings in positive terms. For the alternative model may
identify other, equally suitable candidates for the feature which marks intellectual imaginings as imaginative (e.g., their involvement of an imaginative attitude; or their being active). In particular, even if it is true that a
satisfactory account of imaginings has to make reference to corresponding
types of cognitions, there is no need to assume that this reference has to
part of the positive side of such an account. Instead, it may pertain to its
negative side such as when (NC*) or a similar negation thesis is taken to
be central to any successful theory of imagining. Besides, it is not certain
that an account of imaginings has to refer to the respective kinds of cognitions. As the discussion above of OShaughnessys motivation to assume
the necessity of such a reference illustrates, this assumption may depend
largely on the controversial idea that (NC*) applies to all kinds of imagining, thus establishing an intimate link between imaginings and cognitions
which is in need of further elucidation. But OShaughnessys main motivation for endorsing his version of (NC*) namely his belief that there is no
intrinsic or relational feature common to all imaginings, in virtue of which
they count as imaginative, because there is too great a variety of origins
and constitutions among imaginings (see Sections 6.3 and 7.1) need not
be true, or at least is not obviously so. Indeed, I will argue in Chapter 13
that it is false.
It might be insisted that there are other differences between intellectual
imaginings and beliefs which can be accounted for solely in terms of (ER).
But the first worry that (ER) itself is unilluminating with respect to intellectual imaginings would still apply. Moreover, it is not obvious which
further differences could demand such an explanation. With respect to
sensory or affective imaginings, it may be argued that (ER) is necessary to
account for some of their phenomenologically salient features (e.g., how an
imagined itch can somehow feel itchy without really being an itch; or how
visual imaginings can possess a perspectival content; see Martin (2002a),
as well as Chapters 10f. and Section 14.3); or for the fact that we can be
immune to error through misidentification with respect to what is imagined
(see Peacocke (1985), and also Section 3.5). But in the case of intellectual
imaginings, there do not seem to be similar aspects in need of explanation.

The Representational Account

307

Their phenomenal character is not very complex and comprises not much
more than their lack of an attitude (or the involvement of an imaginative
one) and, presumably, also their entertaining of a specific proposition (see
Soldati & Dorsch (2005) and Dorsch (2009b)). Similarly, our knowledge of
what intellectual imaginings represent can be accounted for in roughly the
same terms in which we explain our knowledge of what our beliefs (or
judgements) represent: namely by reference to the concepts and semantic
mechanisms which the first- and the higher-order representations involve
(see Peacocke (1985)). Hence, intellectual imaginings do not seem to involve any features whose explanation depends essentially on the acceptance of (ER).
However, what needs to be elucidated might not be some feature of intellectual imaginings, but rather their distinctness from other (non-sensory)
forms of imagining. The introduction of an echo claim with respect to sensory imaginings may be motivated by the idea that it can explain the differences between visual, auditory and other modes of sensory imagining (see
Peacocke (1985) for an endorsement, and Hopkins ((1998): ch. 7) for a rejection, of this motivation). In a similar vein, it might be thought that the
representational echo thesis with respect to intellectual imaginings has the
power to explain the differences between, say, imaginatively believing
something and imaginatively desiring it. The idea is that the two kinds of
imagining differ because they involve different types of propositional attitude as part of their contents: the first echoes belief, while the second
echoes desire. But at least two considerations speak against this view.
First, there seems to be an asymmetry between the two phenomena:
while it does not come natural to us to speak of imaginatively desiring
something (there is no obvious desire-like equivalent to supposing, assuming or imagining-that), it is commonly accepted that there are cases of
imaginatively believing something (suppositions, etc.). But this suggests
that the two phenomena should not be treated on a par. It is much more
straightforward to say that, while intellectually imagining something denotes a sui generis way of entertaining a proposition, imaginatively desiring something does not. Otherwise, people would have noticed the desirelike way of imaginatively entertaining a proposition and would have invented a name for it (see Soldati (2006)). Hence, this alternative picture seems

308

The Dependency Account

to provide a much more plausible explanation of how intellectual imaginings differ from cases of imaginatively desiring something (in whichever
other way they have to be characterised) than the idea that the two phenomena merely differ in what they are imaginings of.
Second, it is not easy to come up with cases of imaginatively desiring
something which cannot be analysed either as cases of intellectually imagining that one desires something (plus perhaps having certain closely related affective imaginings, such as imagining a feeling of wanting or longing), or as cases of experientially imagining (the feeling of) desiring something. Imaginatively desiring something if there is such a thing in the first
place (see Note 16 in Chapter 1) is therefore best understood as an instance of intellectual or experiential imagining. Consequently, the application of (ER) to intellectual imaginings lacks argumentative support.22
Moreover, as noted above, the fact that there is no satisfactory argument
in support of the claim that intellectual imaginings are imaginative representations of types of beliefs is not the only difficulty. There is also a consideration that speaks strongly against the application of (ER) to intellectual imaginings. This consideration concerns the fact that we can, and often
do, entertain propositions without either endorsing them or imagining (or
otherwise representing) them to be endorsed. We may consciously desire
that it will rain; we may hope that it will be sunny; we may worry that we
will be too late to reach the shop before it closes; or we may simply wonder whether Quito is the capital of Ecuador. None of these cases need (and
some perhaps cannot) involve a judgemental endorsement of the enter22

It is interesting to note that OShaughnessy himself does not explicitly argue, or


mention his motivations, for endorsing his (or some) version of (ER). Up to the
point where he properly introduces this echo thesis (see OShaughnessy (2000):
365), he has only argued for the acceptance of his weak negation claim (i.e., his version of (NC*)): that imaginings cannot exemplify the two cognitive prototypes. And
in the context of discussing his echo thesis and completing his definition of imagining, OShaughnessy adds to this only his plausible considerations about the fact that
imaginings share many features with cognitive states, such as the scope of representable entities and features, or their types of representation (see OShaughnessy
(2000): 364f.). But clearly neither the weak negation claim, nor the similarity claim,
nor the two together, are sufficient to entail (ER): they do not imply that imaginings
should be understood as having cognitive prototypes (or other types of cognitive
states) as their immediate objects of representation.

The Representational Account

309

tained proposition.23 But it is also false to require that they have to involve
imagining an endorsement of the entertained proposition. At best, it may be
argued that they presuppose having some prior beliefs about the possibilities available (e.g., that it could rain or be sunny in the future). However, if
we can entertain propositions without either endorsing them or imagining
them to be endorsed, it seems arbitrary and unlikely that this should none
the less be impossible in the case of intellectual imaginings. In particular, it
is reasonable to assume that the simple active entertainment of a proposition without endorsing it is already an instance of imagining (see Section
14.2). As long as possibilities like this are not ruled out, (ER) should be rejected if intellectual imaginings are concerned.
The situation appears, again, to be different in the sensory case. The instances of the main non-imaginative kind of non-perceptual sensory representation namely sensory memory possess a cognitive attitude and may
very well turn out to have perceptions as part of their contents (see Section
9.4). Hence, there may be no visual representations that lack a cognitive attitude, and of which (ER) is not true. After-images seem to involve a cognitive attitude and, in this respect, to be perception-like: they involve a
claim about the presence of an experience-dependent patch of colour.
Spontaneously occurring or otherwise unbidden images, on the other hand,
are likely to be covered by (ER) at least if imaginative and mnemonic
images are. This just leaves the visual awareness of the depicted that forms
a necessary part of pictorial experiences which, too, should be taken to
be perceptual and to involve a cognitive attitude, given that it is not imaginative (see Hopkins (1998) Dorsch (2012c).

9.6. Imaginative Projects


The representational echo claim fares even worse with respect to imaginative projects. Clearly, imaginative projects are not representations of cognitive ones. In particular, they are not experiential imaginings of cognitive
23

See Gordon (1987) on epistemic emotions (in contrast to factive ones), which
need or even can not involve any cognitive attitude. Merely entertained thoughts
may be another example, at least if they are not taken to be imaginative themselves
(see Section 14.2).

310

The Dependency Account

projects, given that only the episodic parts of projects but not the project
as a whole possess by themselves a phenomenal character. At best, imaginative projects may be said to consist of episodes (i.e., imaginative episodes) that represent types of cognitive episodes. But two things speak
against this proposal. First, it has to assume the truth of (ER) with respect
to intellectual imaginings which, as just argued, is highly problematic.
And, second, it has to presuppose either that imaginative projects consist
solely of imaginative episodes, or that cognitive projects do not involve
any imaginative episodes. For, otherwise, imaginative projects would not
differ from cognitive ones in respect of whether or in respect of the extent to which they involve the representation of cognitive episodes. But,
as already noted, imaginative projects may also include non-imaginative
episodes, while cognitive projects may also involve imaginative episodes.
For instance, when we daydream about being wealthy, we make use of our
beliefs about the world of the rich; and we may use imaginative thought
experiments in order to solve a theoretical problem (see also Sections 2.3
and 14.2, as well as Chapter 4).
Very similar considerations undermine the different suggestion that it is
distinctive of imaginative projects that they aim at, and lead to, the formation of episodes which are imaginative representations of types of perceptions, beliefs, and so on. This strategy has already been discussed as a way
of trying to avoid roughly the same problems in the case of negation
claims; and its application to echo claims faces substantially the same objections (see Section 8.1). First of all, given that it is unlikely that (ER) applies to intellectual imaginings, the suggestion leaves out imaginative projects that have the purpose to bring about solely intellectual imaginings
(e.g., the project of imaginatively drawing out the consequences of a certain counterfactual situation in a philosophical context). Then, imaginative
episodes and imaginative projects are assumed to count as imaginative for
very different reasons, hence undermining the unity of imagining not the
least because the imaginativeness of projects is said to be derivative of that
of episodes. Finally, the proposed characterisation of imaginative projects
remains extensionally inadequate. In particular, there are cognitive projects
which specifically aim at, and lead to, the formation of imaginative episodes as well. The two examples discussed in Chapter 4 are cognitive pro-

The Representational Account

311

jects of this kind: they successfully aim at the formation of knowledgeconstituting visual imaginings. Accordingly, the suggestion under
consideration would have to give up the ambition to provide an account of
the imaginativeness common to all forms of imagining. (ER) should therefore not be assumed to form part of a unified theory of imagining.

CHAPTER 10
Visual Imagining As Experiential Imagining

That the Representational Account should not be accepted for some forms
of imagining notably intellectual imaginings and imaginative projects
does not mean that it should not be accepted for others. Indeed, in the following two chapters, I aim to argue in favour of the truth of (ER) when applied to sensory and emotional imaginings. Hence, even though the Representational Account does not single out the common nature shared by
episodic and non-episodic, as well as intellectual and non-intellectual, instances of imagining, it still succeeds in elucidating some important aspect
of the nature of sensory and emotional imaginings namely their distinctive representationality. This is, indeed, part of the reason why the Representational Account has been endorsed in the first place even though typically only with respect to visual imaginings (see, for instance, Peacocke
(1985), Martin (2001) and Martin (2002)). Now, while Chapter 11 is focused on emotional imaginings, I intend to use this chapter to show that
(ER) is true of at least some instances of visualising.
Although this does not suffice to establish the conclusion that (ER) is
true of all instances of sensory imagining (or even all instances of visualising), it constitutes an important step towards that conclusion. 24 My argument is, to a considerable extend, a defense of the argument to the same effect that is presented in M. G. F. Martins paper The Transparency of Ex24

Some philosophers notably Peacocke (1985) and OShaughnessy (2000): ch. 12


have put forward stronger versions of the claim, extending it to all kinds of sensory
imagining. Martin is clearly sympathetic with this more general conclusion, as can
be witnessed in Martin (2002a, 404f.), as well as Martin (2001).

314

The Dependency Account

perience.25 This argument has been often misunderstood (or ignored), and
it is worthwhile to discuss it in detail and to illustrate how it is immune to
the objections standardly raised against it. In addition, I present a second
and independent argument for the claim that visualising is imagining seeing, which is not part of Martins paper (and might not be endorsed by
him), but which none the less makes use of roughly the same example as
Martins argument. Both arguments show that it should better be accepted
that visualising is, at least sometimes, essentially imagining seeing.

10.1. Martins Argument: Perspectivalness


If we limit (ER) to sensory imaginings, what we get is the claim that sensorily imagining something means, basically, representing a perception of
it. This is, more or less, Martins Dependency Thesis:26
[T]o imagine sensorily a is to imagine experiencing a [.] (Martin
(2002a): 404)

The thesis to be defended in this chapter is, again, a more specific version
of both (ER) and the Dependency Thesis, this time restricted just to one
particular kind of sensory imagining, namely visualising.
Now, what we are aware of when imagining a perceptual experience is
just that, some experience. More specifically, we are aware of the first-personal side of an experience, that is, of its phenomenal character. We imagine some experience by imagining some instantiation of its character. Its
third-personal side (if experiences have any) is, so to speak, invisible to
object-directed imagining. Of course, we can have additional thoughts
about it. But it is not presented as part of a case of imagining with an exSee his (2002a). Here, I ignore the fact that Martins argument for the claim that
some instances of visualising are instances of imagining seeing forms part of a wider
argument against intentionalist (and for certain disjunctivist) theories of perception. I
discuss and reject this further application in Dorsch (2010c), from the middle
sections of which most of the material in this chapter is taken.
26
Note that Martin uses experience here as short for sense experience, such as perception or bodily sensation. My own use is less narrow in also including, say, visualising or other imaginative instances of object awareness.
25

Visual Imagining As Experiential Imagining

315

perience as its object. The reason for this is that this form of imagining is
experiential, in the same sense in which seeing and visualising are visual.
Just as the latter are limited to the presentation of visible entities, experiential imagining is restricted to the presentation of phenomenal aspects of
mental episodes. The non-phenomenal (or structural) features of episodes
lack an experiential appearance, so to speak. Again, experiential awareness does not differ in this respect from, say, visual awareness. When we
see or visualise a lemon, we see or visualise its visual appearance, but not
its biological nature, for example.
Martin uses the Dependency Thesis to formulate an argument against intentionalism about perceptual experiences the view that veridical and hallucinatory experiences share the same intentional nature. Indeed, intentionalists typically select the Dependency Thesis as the main target of their criticism of Martins argument. This is partly explained by the fact that intentionalists treat imaginative experience as involving the same general kind
of intentional representationality as perceptual experience. For them, it is
therefore difficult to understand why, say, seeing and visualising should not
make us aware of the same objects, namely external things. If perception
does not involve an awareness of an experience, why should imagination
do so, if both are assumed to involve the same kind of intentional awareness? This doubt should be taken serious not the least because it simply
confirms that there is in fact some tension between intentionalism and
(ER). While Martin draws the conclusion that the former should be given
up, his opponents question the plausibility of the latter.
Some of the intentionalist objections, however, seem to have misunderstood the nature of Martins argument. What is crucial here is that it concentrates on, and exploits the special features of, cases in which our episodes of visualising involve certain subjective properties. Subjective properties are characterised by the fact that they are experience-dependent: their
instantiation is dependent on the occurrence of a specific experience. Martins focus is on instances of visualising which involve a certain kind of
perspectivalness. Accordingly, the conclusion of his argument should be
understood as being restricted to those cases (or to similar cases, such as
imaginative experiences involving aspects of painfulness or itchiness).
What it therefore claims is that visualising is identical with imagining per-

316

The Dependency Account

ceiving when it involves the subjective perspectival element at issue. Perhaps sensory imaginings are generally perspectival in this way. But, if not,
the argument fails to establish (ER) for all of them.
So, what is the kind of perspectivalness at issue? By perceiving an object, we may acquire knowledge about the latters specific spatial location.
But our perception does not thereby place the object in objective space.
When we look at a building that is located roughly to the South-East of the
bench on which we are currently sitting, we do not see it as being to the
South-East of that bench. In particular, we do not perceive objects as being
oriented in accordance with the cardinal directions. Instead, we see them as
being oriented towards ourselves for instance, we see the building as being to our left. What this means is that we perceive objects as part of
egocentric space, and not as part of objective (or absolute) space. 27 One
manifestation of this fact is that our perception of the building inclines and
entitles us to judge that it is to our actual left, but not that it is to the SouthEast of the bench. Coming to know the latter requires additional information notably about our own location and orientation in objective space
(see Campbell (1994): especially ch. 4).
None the less, our perceptions of egocentric locations are still as much
concerned with actual space as is our knowledge of objective locations. We
see the building as being to our actual left, as part of our actual environment. If this were not so, our experience would not be able to provide us
with all the information necessary to properly interact with what we see
for example, to succeed in walking over to the actual building. But that
perception does provide us with this information is illustrated by the fact
that such interaction does not require inferring the presence of the building
to our actual left from perceiving it as being to our left and believing that
27

It does not really matter for Martins main argument whether we are concerned here
with two different sets of spatial properties of objects (e.g., one objective, and the
other subjective), or instead with two modes of presentation of one and the same set.
What is relevant here is primarily the fact that our perceptual access to spatially located objects is perspectival and, in particular, presents them as oriented towards us,
rather than in more objective terms. But many of the points involved in the argument
can be described more easily by reference to egocentric properties. Besides, the postulation of subjective orientations is not much different from the postulation of subjective modes of presentation (see the similar issue raised below with respect to the
aspect of leftishness and similar phenomenal aspects).

Visual Imagining As Experiential Imagining

317

such experiences are (typically) concerned with the actual world. The issue
of which world our perceptions are concerned with simply does not arise
it is our world, the world in which we perceive. Something similar is true
of the temporal relevance of our perceptions: they concern our present environment. We see the building as presently being to our left, and not as
having been there in the past, or as going to be there in the future.28
Part of our perception of the building as being to our actual left is implicit, however. We do not explicitly experience ourselves and our spatial relation to the building when perceiving the latter. We are not among the entities presented to us by our experience. Of course, we can see other perceivers and even ourselves, say, by utilising a mirror or some similar apparatus which turns us into the object of our own perceptions. But normally, when we are simply subjects of perception and perceive the orientation of objects relative to us, we do not see us, but only the objects. Our
own perspective is only implicitly reflected in our perceptual experiences,
namely as the point of view oriented to which objects are presented to us.
As a consequence, what figures explicitly in our experience is not the relational property of being to the left of us, but the monadic quality of being
leftish.
It seems that such a quality can figure in perception in two different
ways. The perceptual experience may instantiate the quality; or it may instead present an external object as having that quality. In both cases, this
has consequences for the phenomenal character of the experience concerned. In the first case, the quality constitutes one of the non-presentational aspects of that character. In the second case, it is a constituent of one
of the characters presentational aspects.29 That the quality of leftishness
figures in our perception of the building therefore means that the latter instantiates a certain character aspect either a non-presentational aspect, or
the presentational aspect of presenting the building as being the monadic
property of being to the left.30
In what follows, I concentrate on the fact that perceptions present objects in actual
space and mention the temporal dimension only when it becomes relevant.
29
It should become clear very shortly that there is a third possibility: the experience
may represent another experience as instantiating or presenting the quality.
30
Using the expression being to the left to denote a monadic property is not ideal,
since this expression clearly has some connotations of relationality. But it is not easy
28

318

The Dependency Account

Which view is to be preferred in the end does not matter here. Indeed, it
is not so clear whether they actually differ in any substantial way which
might explain why Martin appears to switch between both views in some
of his formulations (e.g., when talking about the quality of itchiness). The
step from acknowledging the presence of a non-presentational aspect of the
character of a perception to projecting this aspect onto the perceived object
is indeed small as discussions about blur or similar phenomena illustrate
(see, for instance, Peacocke (1983), as well as Soldati & Dorsch (2011)).
Moreover, any presentation of something as being leftish would lack the
status and force of the presentation of it as being to our actual and present
left. In particular, we do not see the building as having the monadic property of being to the left; and we are not inclined or entitled to believe it to
genuinely instantiate this property. Of course, we may say the building is
to the left. But when prompted, we will happily clarify that what we really
meant was that it is to the left of us.
In any case, that the character of our perception of the building include
this phenomenal aspect let us call it the aspect of leftishness should not
be doubted. We can attend to it; and we can exploit it when drawing a picture of how the building looks like when seen from our current point of
view. That is, we can depict an object as being to our actual left by drawing
it on the left side of the canvas instead of, say, by drawing both ourselves
and the object.31
But how is the instantiation of the aspect of leftishness linked to the perception of the property of something as being at some specific location to
our actual left? More generally, how does the perspectivalness of an
to come up with another formulation, without altogether loosing the connection to
the perceived property of being to our actual left. I am grateful to one of the referees
for making me aware of this issue.
31
Very similar issues arise, for instance, with respect to the status of the quality of
ovalness another perspectival aspect of perception which figures in our experience when we are looking at objects from an angle and perceive them as round.
Again, we typically draw round objects by tracing elliptical shapes on the canvas.
But it is debatable whether our experiences present round objects as elliptical in addition to presenting them as round (see, for instance, Peacockes discussion (1983)
of what he calls sensational properties). One significant difference from egocentric
orientation is, however, that, while roundness is an objective property, being to the
left of us is not.

Visual Imagining As Experiential Imagining

319

experience relate to the determination of what is experienced? Martins insightful observation is that the former suffices for the latter (see Martin
(2002a): 410). If an experience of an object exemplifies leftishness that
is, shows a respective non-presentational phenomenal aspect or, alternatively, presents the object as being to the left then it is an experience of
the object as being to our actual left. More specifically, the presence of the
perspectival aspect of leftishness is sufficient to ensure, first, that the experience concerned is an experience of something as being to our left
(rather than to our right) and, second, that it is an experience of something
as being to our actual left (rather than to our left in a merely possible situation). Indeed, Martin claims even more, namely that it also suffices for
having a perceptual experience of something as being to our actual left.
This makes sense since the other two kinds of visual experience, which
may involve the aspect of leftishness, are not or at least not in their
simplest forms concerned with our current environment. Episodes of
visualising present objects as part of imagined situations (see below), while
episodes of visual memory present objects as part of past situations. I return below to the issue of how important this additional claim is for Martins argument.
That the instantiation of the aspect of leftishness turns the respective experience into an experience of something to our actual left is a direct consequence of the implicitness involved in our perception of the spatial relations that objects bear to us in egocentric space. As noted above, we see
objects as being to our actual left (and not, say, as being at an egocentric
location in some merely possible space). But this relational property is typically not explicitly given to us. Instead, what figures in our experience is
solely the monadic quality of leftishness. Hence, we perceive the instantiation of the property of being to our actual left simply by being aware of
the quality of leftishness. When we see the building as being to our actual
left, no aspect of our perception but its aspect of leftishness plays a role in
determining that we experience the building at that specific location in our
actual environment. If the aspect of leftishness is taken to be presentational, this thought becomes even more straightforward: our perception
presents the building as being to our actual left just by presenting it as being to the left; no other presentational element is needed or involved. What

320

The Dependency Account

we are confronted with here is the particular subjectivity of the aspect of


leftishness. Its actual instantiation is both necessary and sufficient for the
experience of something as being to our actual left.32
However, as Martin notes, these considerations about perception give
rise to a puzzle in the case of visualising (see Martin (2002a): 410). On the
one hand, our episodes of visualising involve the same kind of perspectivalness as our episodes of seeing (see Hopkins (1998): ch. 7). We
visualise buildings as being to the left of certain subjective points of view.
And we normally do so without explicitly presenting those points of view
or any subjects occupying them. What figures in our respective imaginative
experiences is therefore, again, the monadic quality of leftishness, and not
the relational property of being to the left of some subject in the subjects
environment. But this means that our episodes of visualising may involve
the same phenomenal aspect of leftishness as our episodes of seeing. Indeed, this is partly due to the relative simplicity of our visual presentation
of the egocentric orientations of objects. It is devoid of any explicit reference to the subject of experience and, therefore, allows us to visualise
something as being to the left without thereby visualising it as being to the
left of any particular subject.
On the other hand, when visualising buildings as being to the left of
subjective points of views, we need not and typically do not imagine
them as being to our actual left. At least in the simplest cases, our episodes
of visualising do not locate their objects in our actual environment, but instead in some imagined space (see Sartre (1940): 8ff., and Wittgenstein
(1984d): sec. 622 and 628).33 Of course, we can project our image onto our
actual environment by taking what we imagine to be part of actual space.
Of course, Martin cannot assume in his argument that an experiences instantiation
of the aspect of leftishness is also sufficient for the existence of something to our actual left. This would follow only if the experience is a perception, and if perceptions
are always factive something that intentionalists deny.
33
There is also the issue of whether visualising always locates objects relative to us
(understood as part of the imagined situation), or rather to some other imagined subject. The default case seems to be that we visualise objects as oriented towards
ourselves (i.e., our point of view within the imagined world), and that imaginative
projects involving imagined subjects different from us require the additional identification of our imagined point of view with that of those other subjects (see Wollheim (1984) and Martin (2002a): 411).
32

Visual Imagining As Experiential Imagining

321

But even then, there is no real competition between what we see and what
we visualise. For example, when looking at a certain picture on our kitchen
wall, we may visualise with open eyes how things would look if there were
a different picture at the same spot on the wall. But such a complex and
mixed presentation does not amount to a presentation of the impossible
state of affairs of two pictures occupying the same part of space.
So, episodes of visualising may involve the aspect of leftishness without
presenting something as being to our actual left. But due to the subjectivity
of the aspect of leftishness noted above, its instantiation is inseparably
linked to the presentation of something as possessing the relational property of being to our actual left. Hence, the instances of visualising concerned cannot exemplify the aspect of leftishness. This raises the question
of how it is involved in visualising instead. Martins proposal is that, in
visualising, we imagine an experience as instantiating the aspect of leftishness that is, we imagine a perspectival experience of something as being
to the left in the imagined situation. When we visualise a building as being
to the left, our imaginative episode does not instantiate the aspect of leftishness. But it still involves the latter by representing another experience as
instantiating it.
The proposal captures the specific subjectivity of the aspect of leftishness. For it takes the instantiation of that aspect in a certain world to be
sufficient for the occurrence of an experience of something as being located to the experiencing subjects left in that very same world. Actual perspectival experiences concern actual space, while imagined perspectival
experiences concern imagined space. Moreover, what needs to be imagined
is a perceptual experience. As noted above, other perspectival experiences
are not concerned with the current state of the world in which they themselves occur. Instead, they are concerned with the past of that world (as in
the case of visual recall), or with an entirely different possible world (as in
the case of visualising). Hence, neither episodic memories, nor imaginative
episodes can instantiate the subjective aspect of leftishness. If we want to
imagine an experience with that aspect, we therefore have to imagine a
perspectival perception. This conclusion can also be inferred more directly
from Martins additional claim mentioned above, namely that the presence
of leftishness suffices for the presence of perception. Indeed, the reasoning

322

The Dependency Account

put forward in support of that claim has been very similar to the one rehearsed in the second half of this paragraph. But, strictly speaking, the additional claim does not seem to be necessary for Martins argument.
Martin further illustrates this argument by comparing the subjective perspectivalness of perceptions to the subjective aspects involved in some
bodily sensations. His example are experiences of itchiness; but experiences of pain are equally good candidates (see Sections 11.5 and 14.3). 34
Experiences of itch instantiate the phenomenal aspect of itchiness: they are
sensations of itch. Moreover, having a sensation of itch is sufficient for
there actually being an itch and, hence, for experiencing an actual itch. If
we sense an itch at a certain location on our skin, then that part of our skin
does indeed itch independently of whether there is any skin irritation,
say. By contrast, merely imagining our skin as itching does not involve the
presentation of an actual itch. In particular, it does not induce us to scratch
the respective part of our skin. But this raises, again, the question of how
imagining an itch can still involve the phenomenal aspect of itchiness
which it clearly does, albeit possibly to a lesser degree of intensity and determinacy than real feelings of itch. For, otherwise, imagined itches would
not phenomenally resemble genuine itches and, moreover, be classified as
imagined itches in the first place. As above, the solution is to understand
imagining an itch as imagining a sensation of itch that is, as imagining an
experience which instantiates the phenomenal aspect of itchiness. This
concludes what are, in essence, Martins considerations in favour of the
truth of (ER) limited to (some) instances of visualising.

10.2. A Second Argument: Egocentric Orientation


However, the analogy with itch (or the similar analogy with pain) suggests
a second route to the conclusion that visualising the orientation of objects
in egocentric space requires imagining perceiving that orientation. Feeling
an itch is not only sufficient for the existence of itch, it is also necessary
34

Nothing here depends on whether the presented view on itches and sensations of itch
(or the comparable view on pains and pain experiences) is correct. The analogy is
merely meant to further illustrate Martins treatment of the involvement of subjective elements in imaginative experience.

Visual Imagining As Experiential Imagining

323

for the latter. Our skin does not really itch if we do not feel itch. Of course,
other things may distract us so that we do not always notice the itch. But if
we do not feel any itch at some location on our skin, despite being sufficiently attentive to the latter, it does not seem true to say that that part of
the skin in fact itches. In particular, by-standers cannot insist that we sense
an itch by pointing to some irritation of our skin. Such evidence cannot
trump our failure to feel an itch. Accordingly, the presence of an itch requires an experience of that itch and, presumably, as part of the same (actual) world. Hence, imagining an itch has to involve imagining sensing that
itch.
Now, egocentric orientational properties seem to be similarly subjective
opening up the possibility of formulating a similar argument in favour of
the idea that visualising means imagining seeing. Martin does not discuss
this second route to the conclusion; and it is not clear whether he would accept the subjectivity of egocentric orientations, or the argument exploiting
it. But even if not, it is still worthwhile to discuss both. When we see a
building as being to our left, it does not possess this orientation independently of being perceived by us as having it. Certainly, the objective location
of the building comes with the disposition of giving rise to a perception of
leftishness when viewed from a position to its North-West by a normal human being with a normal orientation in objective space (e.g., standing on
his feet, etc.) who faces South. But its perceived property of being to our
left cannot simply be reduced to this objective disposition. Instead, the instantiation of this egocentric orientation seems to depend on our actual perceptual awareness of it.
For one thing, which dispositional property is correlated to the property
of being to the left of us varies with changes in our location in objective
space. Once we being to move or turn around, the building may very well
cease to be to our left though it may also begin to be to the left of another
person who steps in and takes our previous spot. The disposition may
therefore constitute the property of being to the left of whoever occupies
the objectively specified location to its North-West with the respective objectively specified orientation. But it does not amount to the property of
being to the left of us (understood in first-personal terms). This is reflected
in the more general fact that egocentric space cannot be fully specified in

324

The Dependency Account

objective terms which is why the two are to be distinguished in the first
place. In particular, what we describe with the expression to our left is not
a cardinal direction in objective space (see Campbell (1994): especially ch.
4).
Without this lack of strict correlation between egocentric and objective
spatial features, it would also seem impossible to explain why we cannot
suffer an illusion with respect to perceiving something as being to our actual left. Of course, when facing South, we may perceive a building as being to our actual left while, in fact, it is located to the South-West of our
current location in objective space. But, as the previous considerations
have indicated, the objective orientational properties of the building are
neither sufficient, nor necessary for its instantiation of any subjective orientational property. What happens in cases like this is just that we fail to
track the former by perceiving the latter an error which is due to some
breakdown in our relation to our environment.35
But the subjectivity of egocentric orientations has consequences for our
attempts to visualise objects as having them. Objects can possess these
subjective features only when they are perceived as having them. Furthermore, this is true as much of imagined or other possible situations, as it is
true of actual ones assuming that they all contain the same ontological
kind of objects and properties. Finally, the dependency in question does not
range over different possible worlds, but is confined to a single one: the
perceived object and the perception are always part of the same world.
Hence, visualising an object as having such an experience-dependent property requires imagining a suitable perception of that object. Visualising a
35

The property of being to our left shares both discussed aspects with the property of
being here if ascribed to ourselves. The latter, too, cannot be specified in purely objective terms. And we cannot go wrong in being aware of ourselves as being here.
One difference between our awareness of us as being here and our awareness of other objects as being to our left is, however, that the latter may concern hallucinated
objects. This is not problematic for the present argument since it relies only on the
claim that the instantiation of egocentric orientations requires the occurrence of a
perception of them, but not on the reverse claim. Besides, as already implied in Note
9 above, there are perhaps other ways of reconciling the subjectivity of a property
with the possibility of hallucinating an instance of it for instance, if the instantiation of the features concerned is understood as depending on the mind of some imagined subject, rather than on the mind of the imagining subject.

Visual Imagining As Experiential Imagining

325

building as being to the left, for instance, has to involve imagining a perception of a building as being to the left.

10.3. Replies to Objections


It should now be easier to understand why some of the objections raised
against the considerations in support of the application of (ER) to visual or,
more generally, sensory imaginings have in fact been missing their target.
In many cases, this is due to the fact that the critics have overlooked or underappreciated the importance of subjectivity in the two presented lines of
thought. A good example for this is Tyler Burges discussion of Martins
argument. Burge seems to have no problem to accept that visualising an
object as being to the left requires the presence of a respective point of
view in the imagined situation. But he rejects the claim that there also has
to be a perception which occupies this location.
[Martin] begins by rightly noting that visualising an object involves taking
an imagined visual perspective on the object for example, visualising it
from a perspective according to which the object is to the left. [...] Martin
assumes that since the perspective is from some position in the imagined
scene, it must be the perspective of an imagined experience in the imagined
scene, or of an experience imagined to be in the imagined scene. This seems
tantamount to begging the question in an argument for the Dependency
Thesis. (Burge (2005): 63f.)

But it is not clear whether Burges charge really is one of begging the question. For he still seems to briefly discuss and dismiss Martins argument involving the subjective perspectivalness of our presentation of orientational features. Here is how the first part of the passage just quoted
continues:
It is quite true that one could have such a perspective on the object only if
one were to have an experience of the object. It does not follow that if one
imagines something from a perspective that one could have only if such and
such were the case (only if one were experiencing the object from that perspective), then in imagining something from that perspective one must imagine such and such to be the case. (Burge (2005): 63)

326

The Dependency Account

One problem with this passage is to understand the difference between


having a perspective on an object and imagining something from that
perspective. That Burge takes the two to be different becomes evident in
his claim that only the first requires having a perceptual experience of the
object. So what does having such a perspective on an object mean? If
what is meant is simply visually presenting an object as being to the left,
without further specifying whether this presentation is perceptual or imaginative, then the intended contrast collapses. For visualising an object to
the left involves such a presentation, too. Furthermore, since Burge wants
to deny that visualising requires imagining a perception, his claim that
having such a perspective requires having a perceptual experience turns
out to be false. So this cannot be the right interpretation of his words. If, on
the other hand, what is meant by having such a perspective is visually
perceiving an object as being to the left, then we get the desired contrast,
given that visualising something does not involve perceiving it. But then,
the noted dependency claim becomes trivial and has nothing to do with
subjectivity. Perceiving an object as being to the left obviously requires
having a perception of it. And it does so independently of the subjective
status of the perspectivalness or the perceived properties involved. So, if
this reading is correct, Burge does not really engage with Martins focus on
the subjective element in our perceptions of egocentric orientations.
Something similar seems to happen in Paul Noordhofs direct reply to
Martins paper.36 For example, he acknowledges that (ER) and the Dependency Thesis are plausible if not true in the case of subjective properties.
But he does not recognise that this is how the egocentric properties figuring in Martins examples should probably be understood. Moreover, he
does not properly address the fact that what really matters in Martins main
argument is not the ontological status of the perceived properties, but instead that of our perspective onto them.
This has, for instance, the result that he underestimates the resources of
proponents of the universal truth of (ER) limited to sensory imaginings
(among whom he seems to count Martin) for explaining the apparent fact
36

Currie and Ravenscroft (2002, 28) also do not address the issue of subjectivity when
they briefly sketch Martins motivation for endorsing (ER) with respect to visualising.

Visual Imagining As Experiential Imagining

327

that we find this thesis which is identical with the Dependency Thesis
more plausible in some cases than in others:
As I have already noted, the Dependency Thesis varies in plausibility depending upon the sensory modality we consider. It is more plausible when
we consider what is involved in imagining the feel of somebodys skin or
the taste of bacon. Our capacity to imagine these things seems to rest upon
our capacity to imagine our experiences of these things. The proponent of
the Dependency Thesis needs to explain why it is more plausible in these
cases given that the Dependency Thesis holds across the board. My guess is
that the proponent of the Dependency Thesis might try to argue that the
variation in plausibility depends upon contingent psychological facts about
what we find easier to consider independent of experience. The issue is
whether we should search for an explanation there rather than in the objects
and properties imagined. If the feel of someones skin or the taste of bacon
imply the existence of corresponding perceptual experiences in contrast
with other objects of imagination, then the Dependency Thesis cannot be
true for all sensory imaginings. (Noordhof (2002): 446)

Here, Noordhof insists that the perceived difference in plausibility should


be explained in terms of a difference in the objects and properties imagined. His suggestion is that we find it plausible with respect to certain
things, but not others, to conceive of imagining them as an instance of experiential imagination because we understand that they, but not the other
things, involve some experience-dependent element. So Noordhof, too,
proposes a limitation of the (ER) to cases pertaining to subjectivity. But he
fails to acknowledge that this is exactly Martins non-universalist take on
the issue. Besides, Martin can explain the difference between instances of
imagining which require experiential imagination and instances which do
not without having to refer to contingent facts about what we find easy to
conceive of as being experience-independent. For he can simply refer to
the involvement of subjective aspects of character such as the perspectivalness involved in the perception of egocentric orientation.
A very similar oversight becomes apparent in Noordhofs discussion of
the factors which determine what is imagined when we are visualising
something. Although he does not draw this connection, it will be helpful to
briefly consider the elements involved in fixing what is part of the situation

328

The Dependency Account

depicted by a painting. The marks on the surface (plus perhaps our general
recognitional abilities) determine whether the painting depicts a blonde
man or a brunette woman. But assuming that it is a blonde man, extrapictorial factors such as the stipulation of a title or the exploitation of
iconographic conventions decide whether it is a portrait of, say, Saint
John or Hercules. The pictorial element puts certain constraints on the extra-pictorial take on the nature of the depicted entities, which cannot be
overridden by the latter. An artist may turn his painting of a man into a portrait of Jean of Arc by labelling it as such. But it will then be a painting of
Jean of Arc in the disguise of a man.
Noordhofs observation is that very similar factors are responsible for
determining the nature of the objects of visualising. On the one hand, there
is the basic visual presentation and, on the other hand, the accompanying
intentions or suppositions which put a certain conceptual gloss on that
visual presentation. To take an example from Peacocke (1985), when we
are visualising a suitcase, we may think of it as a suitcase with a cat hidden
behind it, or as a suitcase which is merely hallucinated by some brain in a
vat. The visual presentation is limited to the presentation of the suitcase,
while the wider imaginative project concerns also other aspects of the imagined situation. And again, how we conceive of the visualised entities as
part of such a project is constrained by how these entities are visually given to us. We may use the visual presentation of the suitcase in order to imagine a car, but only by imagining a car with the visual appearance of a suitcase.37
According to Noordhof, this constraint on what is imagined as part of
the wider imaginative project is problematic for the application of (ER) to
visual imaginings. His starting point is the idea that it is possible to pursue
the project of imagining a certain object, and nothing else. The subject engaged in this project conceives of it in terms of the object, but not in terms
of any experience of that object. Hence, if the project is also taken to involve imagining an experience of the object, this cannot be due to any of
the accompanying intentions or suppositions of the subject. So a proponent
of (ER) with respect to visualising has to assume that the imagining of the
experience is part of the visual presentation of the object. But this seems to
37

See also Peacockes discussion (1985) of Wittgensteins example of Kings College


on fire (see Sections 3.6f. above).

Visual Imagining As Experiential Imagining

329

violate the constraint of the visual presentation on the extra-visual interpretation of it. For we cannot pursue the project of imagining nothing but
an object by means of an episode of visualising which involves imagining
more than that object. The only way out appears to be to consider the extravisual gloss on the visual presentation to be irrelevant or, even worse, misleading:
When we consider what characterises an imaginative project, it is clear that
there are cases where the project is to imagine merely an F. Proponents of
the Dependency Thesis dont have to resist this but, if they dont, they must
insist that certain facts about the mental image override a subjects own
characterisation of his or her imaginative project so that, in fact, what is
imagined is a perceptual experience of an F. This is quite a strong claim to
have to establish. (Noordhof (2002): 430)

The questionable assumption in Noordhofs line of reasoning, however, is


that imagining an object as having a certain feature is always distinct from,
and more basic than, imagining perceiving such an object. For this is not
true if that feature is the egocentric orientation of an object. Visualising
such an orientation requires the involvement, but not the instantiation, of a
perception-dependent phenomenal aspect (e.g., that of leftishness); and
what is thereby visualised is, presumably, a subjective state of affairs. Imagining the instantiation of an egocentric orientation therefore involves imagining a perception of that instantiation. As a consequence, it is a misconception to think that it is possible to imagine merely such a feature, if this
is meant to exclude imagining an experience of it. Notably, when we describe an episode of ours simply as one of imagining an object as being to
the left, our characterisation is at best incomplete. It may refer to the episode of imagining a perception of an object as being to the left. Or it may
denote the episode of imagining an object as being to the left of a specific
perceiver which has to include explicitly imagining that perceiving subject in addition to the object. But contrary to what Noordhof suggests, there
is no simpler episode of imagining to be picked out by that description.
Much the same response should be given to Burges insistence that it is
natural and coherent to say that we can visualise something without ima-

330

The Dependency Account

gining experiencing it:38


It seems to me that in a certain clear and natural sense, one can visualise an
object and not imagine visually experiencing the object. One imagines the
object from the perspective of a visual experience, but no experiencing of
the object (either by oneself or by anyone else) is imagined to be included
in the imagined scene. [...] Prima facie, there seems to be nothing contradictory in this claim, as there would be if (DT) [i.e., (ER) limited to visualising FD] were (constitutively) true. (Burge (2005): 63)

Again, the problems start once we focus on visualising orientational features that locate objects in egocentric space. Their imagination requires the
imagination of a perceptual perspective onto them. That it is still natural
for us to say that what we are imagining is just the object and its orientation may suggest that we either make use of an elliptical characterisation,
or have no full grasp of the nature of our episode of imagining and, in particular, of the subjectivity involved in the perspectival presentation of egocentric orientations. Indeed, it is to be expected that not all of our conceptions of subjective properties characterise them as subjective. For example,
we can discover that the phenomenal aspect of leftishness and the property
of being to our actual left are in fact experience-dependent. So we should
be able to conceive of those properties without conceiving of them as subjective. But this means that we can engage in imagining without fully
grasping what we are thereby engaging in. In particular, we may imagine
an instance of a specific egocentric orientation without realising that we
are thereby imagining a perception of it.
This helps to resolve another of Noordhofs worries. Granting his opponent, for the sake of argument, that imagining perceiving something constitutes sometimes the most basic episode of imagining available to us, he
still insists that it has to involve an explicit or implicit thought which
conceives of what one is imagining in terms of an experience. But that
visualising is therefore supposed to require the possession of the concept of
an experience casts doubt on whether, say, children under the age of four
can visualise something, given that they may lack the notion of an experi38

See also the similar point against the Dependency Thesis made by Currie &
Ravenscroft (2002): 27f..

Visual Imagining As Experiential Imagining

331

ence.
If the Dependency Thesis rests on the claim that imaginers, at least tacitly,
suppose that they are imagining a perceptual experience, then it links the
capacity to imagine with possession of the concept of perceptual experience. In which case, the attribution of imaginings to the autistic, young children, and animals, becomes as doubtful as their possession of the concept of
perceptual experience. (Noordhof (2002): 436)

While acknowledging that the evidence seems undecided on this last issue,
Noordhof is right in pointing out that this connection to the empirical question of when children acquire the concept of experience threatens to undermine the Dependency Thesis, and thus also (ER) applied to visual imaginings. For not only does it seem empirically far less controversial whether
young children can visualise something than whether they can conceive of
perceptions. But the issue of when children become able to conceive of experiences should not be expected to have such a strong bearing on the issue
addressed by (ER) namely the constitution (if not the concept) of imaginative experience.
These problems related to an over-intellectualisation of visualising arise,
however, only if (ER) indeed implies that imagining perceiving something
requires the possession or application of the concept of perception. Noordhof assumes that it does. But the proponents of (ER) need not and should
not follow him in this. As already noted, we need not fully grasp the
nature of our imagining when being engaged in it. And this includes the
fact that it may involve imagining an experience. But, more important, this
partial ignorance is possible because we imagine an experience, not by
thinking of it as part of the imagined situation, but by experientially imagining the instantiation of its character. That is, imagining an experience is
a form of object awareness, and not of thinking. What (ER) claims about
visual imagining is that the visual presentation involved in relevant cases
of visualising consists in the experiential presentation of the character of a
visual perception. And this kind of object awareness does not involve the
employment of the concept of a visual perception. At best, it relies on some
discriminatory or recognitional capacities with respect to experiences, as
well as perhaps some basic demonstrative ways of referring to the respect-

332

The Dependency Account

ive experiences as this or that. It is of course still possible for us to add


some additional thoughts to our episode of experiential imagination. For
instance, we can take the imagined experience to be a hallucination, or one
of the perceptions that Napoleon had when looking at the battle of Jena.
But the basic form of imagining a perception does not involve such conceptualisations, but instead constrains them (see Section 3.6).

10.4. The Nature of Experiential Imagining


The final criticism that I would like to address is the charge that understanding visualising in terms of imagining perceiving raises more questions
than that it answers. It is worthwhile to note that, if this charge is adequate,
it poses a general difficulty for all views about the nature of visualising, assuming that the argument concerning the visualisation of egocentric
orientations goes through. But it is true that more needs to be said about
the nature of experiential imagination and, in particular, how it can accommodate certain important features of visualising notably its involvement
of a visual and perspectival presentation of external things, combined with
a cognitive attitude. Part of this explanatory challenge can be restated by
the distinction between the exemplification and the representation of a subjective aspect of character. Episodes of seeing exemplify a perceptual character, while episodes of visualising if they amount to experiential
imagination involve the imaginative representation of such a character.
What then needs explaining is how the proposed kind of representation can
ensure that the character of representing (i.e., that of visualising) is very
similar to the represented character (i.e., that of seeing) in the mentioned
respects (i.e., visual presentation, perspectivalness and cognitive attitude),
without being identical to it.
In addition to the problems just mentioned, there is obscurity in the
explanation provided by the Dependency Thesis. It rests upon the idea that
the experience is represented in imagination. But it is unclear how to cash
this out. [...] It is hard not to think that all the explanatory work is being
done by the nature of imagination and the kind of representation which
serves it. (Noordhof (2002): 447)

Visual Imagining As Experiential Imagining

333

Noordhof is absolutely right about the last point. But the proposed kind of
representation is perhaps less mysterious than it might seem to him and
others. Consider the reproduction of a painting for instance, a postcard
hanging at your wall. This image does not itself amount to a painting and
differs substantially from one (e.g., it does not involve paint and has no
perceivable texture). But it none the less inherits important aspects of the
reproduced painting. Most of all, it depicts the same objects and features,
and from the same perspective, as the painting. Indeed, if the reproduction
is done well, its perspectivalness derives from that of the reproduced painting, and not from the perspectivalness of the photographic process involved in the reproduction. That is, the impact of the point of view occupied by the lens directed at the painting is typically negligible in comparison to the impact of the point of view inherent to the photographed depiction.39 Much more can surely be said about how the reproduction does end
up presenting the same situation from the same perspective as the painting.
But the absence of such further elucidations does not render the kind of
representation involved in photographic reproduction mysterious or completely unilluminating. We accept that this kind of representation exists.
And the description given above gives us some grasp of what it amounts
to. In fact, we know at least that the reproduction represents the painting
partly by representing the visual perspective of the painting; and that it represents the latters perspective by presenting the same objects and features
as they are presented to the point of view of the painting.
Imagining a perception involves the same kind of representation. It represents a perception partly by representing the latters perspective. And it
does this by presenting the same external objects and features as they are
presented to the point of view of such a perception. In imagining a perception, we thus imagine a possible perceptual perspective onto the world. 40
And, as in the case of the reproduction of a painting, the resulting episode
of visualising ends up with a character very similar in its visual, perThe process of photocopying, which does not involve any such perspective onto the
reproduced piece of paper, is perhaps an even better illustration of the kind of representation pertaining to experiential imagination (as proposed by Martin in a personal discussion about how best to understand Humes copy principle).
40
This idea is not new. In particular, Humes copy principle may be read as claiming
pretty much the same if applied to the case of imagining.
39

334

The Dependency Account

spectival and attitudinal character to that of an episode of seeing. Besides,


we also know that this imaginative representation of a perceptual perspective constitutes an experiential form of object awareness, which may be
spelled out in intentional terms an idea which, incidentally, disjunctivists
about perceptions may agree with. Although it leaves many issues open,
this characterisation of what experiential imagining amounts to, and how it
can inherit some of the features of the imagined experiences, should be illuminating enough to rebut the charge of obscurity. The proposed kind of
representation is involved in other phenomena as well. And we have some
understanding of how it can explain the presence of the important features
of reproductions and episodes of visualising noted.
To illustrate that explanatory power, it is worthwhile to have a brief look
at how this account of experiential imagination can answer a challenge
raised by Currie and Ravenscroft ((2002): 28).41 They ask for an account of
why it is possible that we may mistake an instance of seeing for an instance
of visualising, or that we may recall something as seen that we have merely
visualised in the past (e.g., as in the case of fabricated memories). Their
suspicion is that the defenders of (ER) with respect to visualising do not
have the resources to identify the underlying similarities, assuming that
they have to accept a difference between what we see and what we visualise as part of the respective experiences.
How could we explain, on this hypothesis, why people are prone to misrecall visualising as seeing, and in some circumstances will mistake perception for visualisation? According to the hypothesis, visualising an F has the
representational content, not F, but seeing an F. Thus the seeing and visualising have quite different contents. Why would states with such different
contents seem to us to be so similar? Content is just one dimension of similarity; perhaps states could differ in content and be similar in other ways.
41

Reference to the kind of representation at issue promises also to illuminate why


episodes of visualising often possess a lesser degree of repleteness, determinacy or
intensity than episodes of seeing. Just as the reproduction of a painting may lead to
the loss of some of these qualities, imagining perceiving something may have this
effect. Martins employment in Martin (2001) of the Dependency Thesis in his
account of the phenomenal differences between seeing, visually remembering and
visualising provides another example of the explanatory force of treating at least
some instances of visualising as an instance of experiential imagination.

Visual Imagining As Experiential Imagining

335

But the hypothesis offers us no account of what these other similarities


might be and how they could, in the face of content-dissimilarity, sustain
the overall phenomenological similarity that seeing and visualising enjoy.
(Currie & Ravenscroft (2002): 28)

One part of the answer to this challenge is to stress that, for a proponent of
(ER), there is contrary to what Currie and Ravenscroft suggest a substantial overlap in content between seeing and visualising. For imagining a
perception of an external object involves the visual presentation of that object as part of the imagined situation just as the reproduction depicts
whatever is depicted by the reproduced painting. In other words, experiential imagination has two objects: the imagined perceptual experience and
the external thing presented by that experience (and visualised as part of
the instance of experiential imagination concerned).42 The other part of the
reply to Curries and Ravenscrofts challenge is that, as already illustrated
by reference to the analogy with the reproduction of paintings, the
presentation of external things involved in imaginatively adopting the subjective perspective of a perception shares many important aspects with the
presentation of those things involved in perceiving them from such a perspective. Accordingly, seeing and imagining seeing resemble each other,
not only in what they make us aware of, but also in how they present it to
us. This is indeed best explained by reference to the truth of (ER) when applied to cases of visualising.

42

See Dorsch (2010c) for more on how both certain disjunctivist and certain intentionalist approaches to perception can accommodate the fact that imagining seeing involves two objects of awareness.

CHAPTER 11
Emotional Imagining
As Experiential Imagining

Discussions about imagining and, in particular, the truth of (ER) normally concentrate on the imaginative counterparts of perception and judgemental thought (or occurrent belief). Other forms of imagining such as
daydreaming, or the imaginative counterparts of bodily sensations and
episodes of emotion or desire are less often considered.43 In this chapter, I
aim to develop an account of emotional imagining as a specific instance of
experiential imagining. That is, I intend to defend the view that (ER) is true
of the imaginative counterparts of emotional feelings: emotional imagining
is essentially imagining emotional feeling. Accordingly, emotional imagining consists in non-propositionally (or non-intellectually) imagining the instantiation of the phenomenal character of an episode of emotion.
I motivate the proposed account in response to the theories of Kendall
Walton and Richard Moran. Waltons view stays too unspecific about the
nature of emotional imagining when it matters; and my own account may
be understood as supplementing Waltons by rendering it more specific.
Morans theory, on the other hand, is in conflict with both my own view
and that of Waltons; and I argue that it should be given up in favour of the
latter. Although, strictly speaking, not all affective imaginings are emotional (e.g., conative imaginings or imagined bodily sensations), it seems
fair to assume that many if not all of them possess the same nature as,
43

As already mentioned in the Notes 3f. in the Introduction, Scruton (1974), White
(1990), OShaughnessy (2002), Currie & Ravenscroft (2002) and McGinn (2004)
are recent examples of this kind of limited focus.

338

The Dependency Account

say, imagining feeling jealous or sad (see also Sections 11.5 and 14.3 on
imagining feeling pain). Hence, my discussion of emotional imaginings is
meant to apply to other types of affective imagining as well.
Both Walton and Moran discuss the connection between imagination
and emotion in the context of our responses to representational media.
Both pictures and texts, and possibly also pieces of music and other artefacts, portray fictional or in the case of didactic stories or thought experiments hypothetical worlds. One thing that is particularly interesting
about our engagement with such works is that it need not be concerned
with real persons, situations or events to help us to acquire knowledge
about reality. Reading about the adventures of a fictional character, being
confronted with a potential dilemma or envisaging a new possibility may
enable us to gain theoretical or practical insights into the actual nature of
ourselves and of aspects of the world. In the aesthetic and the moral cases,
these types of engagement with representations of fictional or hypothetical
worlds and the resulting instances of knowledge acquisition are often accompanied or facilitated by emotional responses. Some of these responses
constitute episodes of real emotion, while others amount to instances of the
affective imagination. The latter are therefore relevant for both aesthetics
and ethics.44 Although the subsequent considerations are focused exclusively on our aesthetic engagement with representations of fictional worlds,
they should equally apply to our moral assessment of hypothetical situations.45
The consideration of hypothetical scenarios is also central to science and theoretical
philosophy and, to some extent, also to theology and religion. We invent or use stories and models, say, when attempting to make sense of the structure of atoms, or the
nature of the universe. In the case of faith, this is likely to involve emotional episodes as well; in the case of scientific or metaphysical investigation, on the other
hand, probably less so.
45
Indeed, the affective imagination may be relevant for ethics in several respects. First,
as just described, we may come to determine what is morally required of us (or
someone else) to do in a given situation by imagining performing the different available actions and considering our emotional responses to those instances of imagining assuming here that emotional responses provide some indication of the presence of values (but see Dorsch (2007) for the restriction of this idea to subjective
values). Second, in order to assess (or even understand) the actions of another person, it may be necessary to empathise with them which, again, may require imagin44

Emotional Imagining As Experiential Imagining

339

The central disagreement between Walton and Moran is about whether


instances of the affective imagination involve emotional elements as part
of their content or as part of their manner (or mode) of representation. I
side with Walton on this issue and argue that what is characteristic of emotional imagining is that it consists in the imagination of an emotional feeling. The main challenge to this view is that this does not obviously suffice
for the respective imaginative episodes to count as affective (rather than,
say, cognitive or dispassionate).46 While Walton remains silent on this issue, I propose a way of how it may be successfully addressed whether as
an integral part of his view, or entirely independently of it.
The chapter is divided into five sections. In the first, I outline the puzzle
of fiction, in response to which Walton and Moran have developed their
views of the affective imagination. The second section is devoted to
Waltons theory of our emotional engagement with representational art,
and to the already noted challenge which this theory faces. In the third and
the fourth section, I critically discuss Morans alternative view and, especially, his distinction of emotional imagining from other forms of imagining in terms of an affective manner of representation. The fifth and last section presents my own account of emotional imagining. I aim to show that it
is capable of answering the challenge raised by proposing that occurrences
of emotional imagining count as affective precisely because they are representations of emotions namely non-propositional and experiential reping having some of their emotional feelings. Third, how we and others are inclined
to emotionally react whether in reality or in the imagination to a given situation
(including imagined ones) may reveal something about our character and, especially,
our moral character. And there are probably more scenarios in which imaginationbased emotions become morally relevant.
46
Affective episodes comprise both feelings of genuine emotion and merely emotionlike experiences. What both have in common is an affective phenomenal character. I
assume here that the difference between affective and non-affective mental episodes
(e.g., between episodes of felt jealousy or joy and episodes of perception and belief)
is subjectively salient; but not necessarily that we have any detailed grasp of the
nature of this difference. It does not seem implausible to describe the phenomenal
character of emotional episodes by reference to values and, in particular, the feature
of having a valence (i.e., being either a positive or a negative experience). But nothing in what follows depends on this or any other specific claim about what it means
for an episode to count as affective.

340

The Dependency Account

resentations of the affective character of emotions.

11.1. The Puzzle of Fiction


It is an uncontroversial and as such unproblematic fact that, when we
watch movies or read novels, we often become emotionally involved. Part
of these emotional responses are centred on the works themselves. The latter captivate, excite or bore us and thereby move us to continue or,
alternatively, stop our engagement with them. These reactions, which form
part of our aesthetic experience of the works concerned, are clearly genuine instance of emotion.47 Moreover, they also occur in the case of non-representational artworks or aesthetic objects in nature and are therefore not
directly concerned with the fictional story told by the movies or novels in
question (though of course the story still has an influence on whether, and
how, we enjoy the artworks concerned). In particular, the question whether
these responses are directed at aspects of the fictional world portrayed does
not arise: they are clearly focused on the artworks themselves. Hence, they
are not of interest for our current discussion.48
However, our emotional involvement with representational art may in
addition involve elements that are more directly related to the fact that the
works are representational and present us with a specific fictional world.
Indeed, we do not find it inadequate to describe affective experiences of
this second kind in terms of the fictional content of the artworks. For instance, we say that we rejoice or suffer with the characters or hope or regret the occurrence of certain events within the story. The debate about the
nature of our emotional responses to representational art focuses on these
reactions and asks whether they indeed amount to genuine emotions directed at fictional entities. What is at issue is thus whether our descriptions of
As I use the terms, if an emotion is genuine or real, it is actually true of that it is
an emotion. By contrast, merely emotion-like experiences do not actually belong to
the mental kind of emotions although this may still be fictionally the case.
48
In Dorsch (2007), I discuss in more detail the role of emotions in aesthetic evaluation, which is one central aspect of aesthetic experience. My conclusion there is
that they cannot justify objective judgemental ascriptions of aesthetic value, but may
none the less point us to the presence of subjectively important values, which may
very well include certain instances of aesthetic worth.
47

Emotional Imagining As Experiential Imagining

341

these responses are to be understood literally true, or instead in a different


manner say, as statements about what it is fictionally true about ourselves
and our engagement with the fictional worlds (see Walton (1990)). But
both sides accept that our appreciation of representational art involves
emotional elements that are concerned with and cannot be understood
without reference to particular aspects that characterise the represented fictional worlds (see Walton (1990, 1997) and Moran (1994)).
This close connection to the representationality of art ensures that our
fiction-directed affective responses differ in one or more crucial respects
from work-directed emotional responses. The most notable is perhaps that
the former but not the latter are imagination-based in the sense that
they are triggered by what we imagine about their objects. Our evaluative
emotional responses towards artworks are concerned with what we take to
be actual facts about those artworks, while our affective responses to fictional characters or situations are concerned with what we take to be fictional about them that is, what we imagine about them. The two types of
responses therefore have different targets in that they deal with different
worlds the actual or real world or a fictional one, respectively. And while
we access the actual world by means of perception, belief, and so on, we
access fictional worlds by means of the imagination.49 Our evaluations of
representational artworks may, of course, be imagination-based in a different, and weaker, sense: they are based on our experience and understanding of the work which may very well involve imagining the represented characters, locations, events, and so on. But this does not count as imagination-based in the sense just introduced above, since the instances of
imagining concerned are not about the object of evaluation, that is, the artwork itself. What we imagine to be imbalanced is Hamlet, the fictional
character and not Hamlet, the piece written by Shakespeare.
One recurring theme in this chapter is that this central difference
between the two types of affective responses gives rise to a second difference: namely that only our work-directed reactions are genuine instances
of emotion, while our fiction-directed responses are instances of a different
though still emotion-like kind of mental phenomenon. Walton, for instance, insists on this difference in mental kind, while Moran argues that
49

See Walton (1990) for a discussion of the link between imagination and fictionality.

342

The Dependency Account

both types of responses belong to the same kind (i.e., the kind of
emotions). Their disagreement is related to the issue of whether being a
genuine emotion is compatible with being imagination-based in the sense
described. Accordingly, while both sides accept that our engagement with
fiction is imagination-based and involves affective responses, they are at
odds with respect to whether these imagination-based responses constitute
real emotions, or merely emotion-like experiences. Here is a nice quote
from Walton describing an example of the central element of the kind of
reaction at issue (denying, in the process, that it constitutes a genuine emotion directed at fictional entities):
Charles is watching a horror movie about a terrible green slime. He cringes
in his seat as the slime oozes slowly but relentlessly over the earth, destroying everything in its path. Soon a greasy head emerges from the undulating
mass, and two beady eyes fix on the camera. The slime, picking up speed,
oozes on a new course straight toward the viewers. Charles emits a shriek
and clutches desperately at his chair. [...] Charless condition is similar in
certain obvious respects to that of a person frightened of a pending realworld disaster. His muscles are tensed, he clutches his chair, his pulse
quickens, his adrenaline flows. Let us call this physiological -psychological
state quasi-fear. [...] Afterwards, still shaken, he confesses that he was terrified of the slime. (Walton (1990): 196)

Quasi-emotions like these are real episodes in our mental lives. Moreover,
they are affective episodes which is reflected in the fact that they resemble
genuine emotions (e.g., fear) in various significant respects. First of all,
they are similar to genuine emotions in their involvement of actual emotional feelings and emotion-related physiological events. Charless experience is, from his subjective point of view, very similar to an experience of
genuine fear notably in that it involves a similar kind of unpleasantness
and makes him aware of similar bodily changes. 50 But quasi-emotions resemble genuine emotions also in being triggered by the same mechanisms.
Both types of emotional response are partly due to dispositions to react af50

Of course, his overall experience of the movie is pleasurable. Otherwise, he would


stop watching the movie in order to get rid of it. But his overall pleasure is compatible with and surely partly due to the fact that there is something unpleasant and
discomforting about seeing the slime on the screen seemingly moving towards him.

Emotional Imagining As Experiential Imagining

343

fectively and physiologically to certain mental representations whether


they are perceptions, thoughts, beliefs, or imaginings. Just as the recognition of the real danger presented by an approaching lion is crucial to the
occurrence of genuine fear, the recognition of the fictional danger presented by an approaching slime is crucial to the occurrence of a fear-like
quasi-emotion. This explains why it matters for our emotional responses to
artworks whether they are representational and, if so, what it prompts us to
imagine to be part of the represented fictional world.
However, what is controversial is whether quasi-emotions (perhaps together with the imaginings on which they are based) constitute genuine
emotions, or whether they are merely emotion-like experiences. There is a
long tradition of taking responses involving quasi-emotions to be puzzling.
At the heart of the respective discussions has been the so-called paradox
of fiction which can be formulated in the form of a triad of jointly inconsistent claims:51
(a) We have real emotions towards what we take to be fictional characters or situations.
(b) At least in some cases, real emotions are constitutively dependent
on belief in the reality of the relevant entities.

51

See the essays in Hjort and Laver (1997), and especially Levinson (1997). Walton
allows for the possibility of emotions which do not presuppose any kind of belief,
but still stresses we cannot have emotions towards something that is merely imagined (see Walton (1990): 245). The puzzle may also be formulated in two other
ways, namely in terms of belief in certain relevant propositions, or in terms of belief
in the existence of the entities in question. The first of these alternative option is
more specific than the other two (i.e., it implies, but is not implied, by the latter) and
is not required in all cases. Fearing something may presuppose believing it to be
dangerous, but hoping for something does not presuppose any belief (in particular,
we may be agnostic about whether the hoped-for state of affairs has already been
realised). The second alternative of spelling out the puzzle can allow for emotions
directed at the past or the future only if existence is understood as denoting reality,
that is, actuality. Furthermore, emotions may be said to be dependent, not on what
we believe to be real or existent, but what in fact is real or existent (see externalism
about thought contents).

344

The Dependency Account


(c) We do not believe in the reality of what we take to be fictional entities.

This paradox possesses considerable initial force. When watching a horror


movie or reading a tragedy, we undergo emotional experiences and
physiological changes which are very similar to those involved in real life
cases of fear or pity. Moreover, we are inclined to say that we fear the
monster or pity the heroine. So there seems to be good reason to take our
statements literally and ascribe to us emotions towards fictional characters,
situations, events, and so on. However, genuine fear or pity also appear to
require us to believe that the respective objects exist as part of reality.
When we think that there is really no lion in the room, it seems impossible
for us to genuinely fear one. But when we watch movies or read tragedies,
we usually do not take the represented characters and events to be real (unless we are concerned with some historical or biographical works). Hence,
we typically lack the belief that they exist as part of reality. And it therefore
becomes puzzling why we none the less seem to react with something like
an emotion of fear or pity to these fictional entities.
The described problem of emotions towards fiction has been widely debated, and many different solutions to it have been proposed. The claims
(a), (b) and (c) are jointly inconsistent. Similarly, (c) should be accepted as
a fact about our normal psychology if not even as a claim about our ordinary conception of what fictional entities are. Of course, there may be
cases of people who believe in the reality of fictional entities and develop
real emotional feelings towards them (e.g., when they fall in love with a
character of a telenovela). But it is doubtful that they then conceive of the
objects of their emotions as fictional and if they do, their response becomes even more problematic. In any case, it suffices for the presence of a
paradox if (c) is true of our normal engagement with fiction.
This leaves two broad strategies of how to deal with the puzzle. The first
is to deny (a). This means insisting that no genuine emotion involved in
our engagement with fiction can be directed at fictional characters or
events; and that no response to the latter including those involving quasiemotions can constitute an emotion, even if it may resemble one in certain important aspects. The main reason for such a view is to stress the cog-

Emotional Imagining As Experiential Imagining

345

nitive element present in emotion: they are concerned with how reality is
like and therefore require a specific take on the latter (see, e.g., Walton
(1990)). Perhaps proponents of this answer are also prepared or forced
to accept the additional claim that we are in some sense wrong or irrational
to treat our emotion-like responses to fictional entities as if they were
genuine emotions.
The second option is to reject (b). This may be done for several reasons.
For instance, it may be claimed that imagining the reality or existence of
objects rather than believing in it may already suffice for having emotions towards those objects (see Moran (1994)). Or it may be assumed that
there are two fundamentally different kinds of emotion, one exclusively
directed at real entities and the other exclusively at fictional ones; and that
(b) applies only to the first, but not the second kind. Or, finally, it may be
maintained that (b) does not express a constitutive dependency, but merely
a rational requirement, so that it is still possible albeit irrational to have
emotions without belief (see Radford (1975)). Again, the denial of (b) may
(have to) be accompanied by the postulation of a systematic form of irrationality inherent to our responses to fiction, in this case concerning the
fact that it is in some sense unreasonable to feel emotions towards entities
which we do not take to be real. At least, to respond with fear seems to be
more suitable when one perceives or believes a lion to be in the room than
when one merely visualises or imagines it to be there just as there is
something irrational about fearing real spiders which one takes to be completely harmless (see Goldie (2000) and Dorsch (2007)).
This is not the place to settle the debate between the two approaches.
Our concern is with the nature of affective imagining, and not with the
nature of our emotional responses towards representational artworks and
the fictional worlds that they portray. But since both Walton and Moran
the first of whom rejects (a), and the second (b) develop their views on
affective imagining in the context of this debate, it is worthwhile to a look
a bit closer on their views on how we react emotionally to representational
art.

346

The Dependency Account

11.2. Waltons Account of Emotional Imagining


Walton maintains that Charless real and affective experience of quasi-fear
alone does not suffice for genuine fear (see Walton (1990): 196). What is
missing, according to him, are the right kind of accompanying beliefs and
action tendencies.52 Charles does not believe in the existence or danger of
the slime, and he does not run away or shout for help. Instead, he merely
imagines the presence of the slime and desires to stay where he currently
is. Some of his more basic, instinctive inclinations (e.g., to freeze or to
grab hold of the person next to him) may still be the same as in a case of
genuine fear. But on the level of intentional agency, the two experiences involve with very different motivational profiles. Walton therefore concludes
that Charles is not really frightened by the fictional slime, given that his reaction lacks certain characteristic aspects of such a fear.
This does not necessarily prevent quasi-emotions from sometimes constitute real emotions, together with some respective beliefs and action tendencies. For instance, when watching a battle scene in a war movie, our
quasi-fear may combine with or even be partly brought about or intensified by our belief that a friend of us is actually in a very similar situation
right now. Our quasi-fear then becomes part of our real fear for our real
friend. This is possible because quasi-emotions are relatively unspecific.
The feelings and bodily events involved are compatible with many different emotions whether of the same kind or even of different kinds and in
need of determination by the accompanying representations and motivations which specify, among other things, the objects of the responses.
More important, however, if quasi-emotions are triggered by, and part
of, our imaginative experience of representational art, they may and typically do lead to affective imagining. According to Walton, our basic and
non-emotional engagement with representational art involves three distinct
elements (see Walton (1990, 1997)).
First, we experience the material qualities of the work. For instance, we
perceive the printed marks in a book, or the configurations of colours and
shapes on a canvas. These marks and configurations together with genrerelated conventions determine what the works represent and, hence, what
52

At least in this case; Walton allows that, in other cases, something else might be
missing (see Walton (1990): 245.

Emotional Imagining As Experiential Imagining

347

is true within the respective fictional world. One of the central ideas of
Waltons account of representational artworks is that the fictional truths related to such works are not only concerned with the represented objects
and their features, but also with us and our access to those objects and features. The worlds of paintings and novels include landscapes and battles.
But they also include our fictional perspectives on those landscapes and
battles for instance, our seeing or thinking about them. According to
Walton, this is part of our conventional rules of engagement with representational art.
Second, and on the basis of the first experience, we intellectually imagine whatever is part of the represented fictional world. Indeed, the demand to imagine the fictional truths created by a representational artwork
in accordance with the relevant conventions is an integral part of our engagement with such art. Walton is adamant that, without this element, our
experience would not really amount to an experience of something as representational art. But imagining the fictional truths in question requires that
we recognise the representational content of the work. 53 We thereby exploit
explicitly or implicitly known principles or conventions which link the perceived material configurations to the represented entities. This enables us
to recognise words and their meanings, or the three-dimensional arrangement of objects in a scenery, by perceiving the specific nature of the marks
on the respective surfaces.
And, third, we imagine, again on the basis of the preceding elements of
our engagement with the work, having a certain epistemic access or
standing in a certain epistemic relation to the imagined world of the
work. For instance, we may imagine seeing the landscape depicted by the
painting before us, or believing the propositions expressed by the sentences
in the novel. Again, this is an essential part of our imaginative engagement
with representational artworks and required by the fictional truths determined by the latter.
Since, for Walton, affective imagining works very similar to pictorial
53

It is interesting to ask whether the recognition precedes the intellectual imagining, or


whether the latter is identical with or, alternatively, part of the former. What
speaks in favour of the distinctness of the two phenomena is that imagination normally does not play a role in knowledge acquisition, while the recognition of what a
painting depicts or what sentences mean is a form of knowledge.

348

The Dependency Account

experience, it is worthwhile to dwell a bit longer on the latter (see Walton


(1990)). First of all, it is important to note that it is distinct from sensory
imagination, such as visualising or auditorily imagining it. Looking at a
picture and imagining seeing what it depicts does not involve visualising
the depicted scenery at least not in the same sense in which we can visualise something with closed eyes, say. None the less, imagining seeing that
scenery which is part of our pictorial experience is distinctively visual
in character and cannot be reduced to intellectual imagination. Waltons
proposal is that imagining seeing the depicted scenery amounts to imagining of our actual visual perception of the picture (i.e., the first kind of experience) that it is a fictional visual perception of what is depicted. If a
painting depicts trees, we imagine seeing those trees by imagining of our
perception of the material qualities of the painting (which is part of the actual world) that it is a perception of trees (which is part of the fictional
world depicted by the painting).
The kind of imagining at issue amounts to what Walton calls imagining
from the inside, meaning that we imagine things from our first-personal
perspective rather than imagining our point of view from a third-personal
perspective (e.g., when we imagine how we look like from the perspective
of our friend sitting opposite of us). In the case of pictorial experience, this
means that we imagine seeing the depicted scenery from the perspective of
our perception of the picture: we imagine an identity between our real and
our fictional point of view.54
The issue of whether pictorial experience really involves such a complex kind of imagining, and whether this form of imagining is indeed visual, has been subject of much debate (see, e.g., Hopkins (1998) and Dorsch
(2012c)), but need not be settled here. What is important to note is that the
first two elements alone do not suffice for becoming aware of a picture as a
picture. Simply perceiving its material properties and, in addition, intellectually imagining that there is a landscape does not give rise to experiencing
it as depicting such a landscape. Perception and imagination have to be
more intimately linked to each other to constitute pictorial experience. Perhaps Walton is wrong about the details of this close connection. But he is
54

It is in this sense that we or, more precisely, our subjective perspectives enter
the fictional world. This fits well with Waltons characterisation of imagining from
the inside as one (but not the only) form of imagining de se.

Emotional Imagining As Experiential Imagining

349

right that, if pictorial experience is indeed partly imaginative, the imaginative element has to be integrated with the perceptual one in a single visual
experience.55 What matters here is the twofoldness of pictorial experience.
Seeing something as a picture of something else involves two instances of
object awareness: our awareness of the picture and our awareness of what
is depicted. And although they are distinct, they are also inseparable from
each other. We can, at least to some extent, shift our attention from one object to the other. But we cannot stop being aware of one of them without
ceasing to have a pictorial experience. Moreover, we are aware of both
objects as part of a single and unified experience.56
According to Walton, our affective imaginative engagement with fiction
is similar to pictorial experience in that it involves the same kind of elements as the latter.
First, we experience some quasi-emotion concerned with some aspects
of what the respective work represents. This real emotional reaction is
thereby triggered by our more basic non-emotional and imaginative engagement with the work for instance, our recognition of the portrayal of
an approaching slime or lion, and our imagination of the danger posed by
the latter. For Walton, the occurrence of quasi-emotions brought about in
this way in conjunction with the conventions of our engagement with
representational art makes it fictional that we feel the respective genuine
emotion towards the fictional entities concerned. The quasi-fear triggered
by imagining a dangerous lion approaching does not amount to real fear of
the fictional lion. But it determines that it is fictionally the case that we are
frightened of that lion.
Second, in response to the general demand to imagine what is part of the
fictional world related to some representational artwork, we intellectually
imagine that we have a certain genuine emotion namely that corresponding to the quasi-emotion towards the fictional entities in question. In our
example, we imagine that we fear the approaching lion.
And, third, we imagine some corresponding form of access to the fictional world, this time an affective kind of access. More precisely, we imaSee Hopkins (1998) and Dorsch (2012c). This also explains why OShaughnessy
(2000) talks about imaginative perception when describing pictorial experience and
its relationship to the imagination.
56
See Hopkins (1998), following the writings by Richard Wollheim.
55

350

The Dependency Account

gine feeling the genuine emotion towards the fictional entities at issue. We
do so by imagining of our quasi-emotion that it is a real emotional response towards what is represented by the work. That is, we imagine being
frightened by the lion by imaginatively identifying our quasi-fear with an
instance of genuine fear of the lion. Again, the kind of imagining in question is imagining from the inside: the imaginative identification in question
involves an identification of two subjective emotional perspectives, one
real and the other fictional.
Our emotional responses towards fictional entities are not twofold: we
are not emotionally aware of two different objects. In particular, the quasiemotions concerned are not directed at the respective artworks. Indeed,
they do not have any (clear) object. They are triggered by the imagination
of some fictional entities or situations, but are not about them (at least according to Walton). Hence, the problem of guaranteeing that our imaginative response forms a unified experience is less pressing than in the case of
pictorial experience. Feeling the quasi-emotion and imagining it to be a
genuine emotion towards fictional entities need not form a single and unified experience. Instead, the main reason for assuming the third element
over and above the other two seems to be that the intellectual imagination
does not involve any affective elements. Imagining that one is feeling an
emotion does not suffice for having an affective reaction towards it. But
just as our awareness of what is depicted possesses a visual character, our
response to fictional entities possesses an emotional character. Assuming
that we also imagine feeling an emotion promises to introduce the required
affective element into the experience.
However, it is not clear how this supposed to work how imagining
feeling an emotion can really possess an affective character. As Moran
notes, the problem arises because the emotion is assumed to be merely part
of what is imagined.57 In the case of intellectual imagining, this is precisely
57

See Moran (1994). Note, however, that Moran seems to misunderstand Waltons
position by ascribing to him the view that what is central to affective imagining is
intellectually imagining that one has the emotion concerned. The reason for this misunderstanding appears to be the failure to see what is responsible, according for
Walton, for the fictional truth that we are feeling an emotion towards the fictional
entities in question. Moran seems to assume that this is due to intellectually imagining that proposition hence the view which he ascribe to Walton; while the latter in-

Emotional Imagining As Experiential Imagining

351

what prevents the episode of imagining from being affective. So why


should the situation be different in the case of the kind of imagining
Walton proposes? He maintains that imagining, from the inside, having an
emotion is more affective in character than intellectual imagining and,
hence, not an instance of the latter (see Walton (1990): 247). But he does
not say much to help us to better understand imagining feeling an emotion,
apart from the fact that it is imagining experiencing an emotion from the
first-personal perspective. What still needs to be explained is why or in
which sense this kind of imagining should count as affective.

11.3. Moran on the Various Forms of Imagining


Moran tries to provide an answer to this question of how we can have responses towards fictional entities that count as genuinely emotional. His
theory consists mainly of two claims (see Moran (1994)). The first is that
the affective character of our responses is due to their manner or attitude,
and not their content. Just representing having an emotion does not lead to
an affective experience, since we can represent having an emotion in a dispassionate way for instance, when we suppose, for the sake of an argument, that we are angry. Therefore, affective representations have to
amount to representing something in an affective manner just as visual
representations amount to representing something in a visual manner. The
second claim central to Morans view is that the quasi-emotions triggered
by our engagement with representational art are in fact constituents of our
emotional responses to fiction that is, for Moran, of our experiences of
imagining something in an emotional manner.
He combines these two claims with a third, namely that the resulting
emotional experiences towards fictional entities should count as genuine
emotions. Accordingly, he chooses the second strategy in dealing with the
paradox of fiction. That is, he rejects (b) by insisting that imagining something is already enough to give rise to full-blown emotions. Two elements
have motivated his choice: the observation that our engagement with fiction involves real affective elements; and the hypothesis that, partly for this
sists that the occurrence of a relevant quasi-emotion is the effective factor (see the
discussion in Walton (1997)).

352

The Dependency Account

reason, our emotional responses to fictional entities are very similar to, and
as unproblematic as, our more ordinary affective reactions to certain real
objects, situations or events such as those which are in the past or the future, or which constitute unrealised, but real possibilities (e.g., missed opportunities or alternative courses of action).
As discussed in the previous section, Walton rejects all three claims. He
thinks that emotional feelings are part of what we imagine, not of how we
imagine it. He also maintains that our imaginative emotional engagement
with fictional worlds is only prompted by, and about, the relevant quasiemotions, but does not include them as one of its constituents. And finally,
neither the quasi-emotions, nor our imagining feeling an emotion are, for
him, instances of emotion which is reflected by his acceptance of (b).
Again, the aim here is not to settle the debate about the third claim that
is, about how best to reply to the seeming paradox of fiction. But the first
two claims are relevant for the nature of affective imagining. We have
already considered Waltons position. It it now time to look into the details
of Morans view.
Moran understands our affective responses to fiction as instances of
what he calls emotional imagining. This label fits very well with the fact
that the form of imagining concerned is treated by Moran as being genuinely emotional. Episodes of emotional imagining are taken by him to be
instances of real emotion and, hence, on a par in this respect with normal
emotional feelings, in the same way in which episodes of visual imagining
are sometimes taken to be instances of visual experience and, hence, on a
par in this respect with visual perceptions.58 In the course of his paper,
Moran distinguishes emotional imagining from three other imaginative
phenomena: propositional imagining, dramatic imagining, and imaginativeness. While propositional, emotional and dramatic imagining have in
common that they occur in the form of mental episodes or activities, imaginativeness constitutes a mental ability or disposition. He does not explicitly talk about a fifth form imagining, namely sensory imagining. But
there is no reason to assume that he would not acknowledge its existence,
58

I follow Moran in reserving the expression emotional imagining for affective responses to fiction that are genuinely emotional albeit being imaginative, rather
than real, affective states. The debate between Moran and Walton is therefore about
the possibility (or at least actual occurrence) of such responses.

Emotional Imagining As Experiential Imagining

353

which is why I have added it to the list.59


Propositional or hypothetical imagining amounts to the simple imaginative entertaining of a proposition for instance, when we imagine or suppose that it rains, or that the Earth is flat (perhaps as part of some daydream, thought experiment or hypothetical reasoning; see Moran (1994):
104). Propositional imaginings or intellectual imaginings, as I have
called them are thus instances of conceptual or intellectual thought and as
such differ from sensory forms of representation, such as visual perceptions or memories, or bodily sensations. Moran leaves it open whether all
non-endorsing or non-judgemental entertainings of a proposition are imaginative, or whether instead there is a difference, say, between merely having the thought that it rains and imagining or supposing the same proposition. But he is clear about the fact that mere propositional imagining is dispassionate, that is, does not involve any real emotional feelings or affective
elements though of course it is possible to propositionally and dispassionately imagine that one has certain emotional feelings (see Moran
(1994): 89f.).
In contrast to propositional imagining, sensory imagining does not have
a propositional content and is therefore not an instance of thought. Instead,
what we sensorily imagine are objects or events and their perceivable features. While thoughts merely describe or name objects or events, sensory
episodes (including perceptions or episodic memories) show them (see
Dorsch (2010a)). Examples of instances of the sensory imagination are
visual, tactile or auditory imaginings. Like propositional imagining, sensory imagining is dispassionate and does not possess an affective character.
But again, it is possible to sensorily and dispassionately imagine someone
having or perhaps rather expressing specific emotions (e.g., when we
visualise someone crying).
Emotional imagining or imagining with respect to emotional atti59

Propositional imagining includes and is probably also meant to be identical with


what I have called intellectual imagining. By contrast, emotional imagining (in
Morans sense) is not an instance of affective imagining (in my sense), but instead
an instance of sensorily or intellectually imagining something with real emotion (see
below). Finally, dramatic imagining is a type of imaginative project which may involve experiential imagining and empathy. It seems to be very close to what Currie
and Ravenscroft have called perspective-shifting (see Section12.2).

354

The Dependency Account

tudes consists in imagining something with feeling or emotion, in contrast to imagining it dispassionately (see Moran (1994): 90 and 105). Moran concentrates on propositions as candidates for what we can imagine
with feeling. But just as with the existence of sensory imaginings, it is fair
to suppose that he would also allow for the emotional imagining of objects
or events. Morans examples for emotional imagining are imagining something with loathing, anticipation, apprehension or regret (see Moran
(1994): 86, 90 and 93). The affective aspect of the imaginative episode
consists thereby in a real and not merely in an imagined feeling. Accordingly, imagining something with regret involves really having a feeling
of regret. As a consequence, emotional imagining cannot or not exclusively be a matter of propositional imagining. In particular, imagining
something with, say, sadness cannot be reduced to imagining that one feels
sad. While the former involves a real feeling of sadness, the latter does not.
Now, given that the affective aspect of emotional imagining is real, and not
merely imagined, it should as Moran maintains be located in the manner (or mode), and not in the content, of the imagining (see Moran (1994):
90 and 93). The statement that something is imagined with feeling or emotion thus qualifies how it is imagined, and not what is imagined. It is therefore likened by Moran to the statement, say, that something is imagined
visually or auditorily (see Moran (1994): 93).
Both propositional and emotional imagining occur in the form of single
mental episodes. By contrast, dramatic or empathetic imagining is typically
more complex by involving several distinct episodes in a single imaginative project (see Moran (1994): 104). More specifically, dramatic imagining
consists in the imaginative adoption of, and identification with, a certain
point of view different from ones own. The adopted perspectives in question are typically characterised partly by a set of evaluative attitudes and
the related emotional or conative dispositions. Thus, imaginatively adopting such a point of view usually involves imagining having the respective
evaluative and affective responses to given situations, in addition to more
neutral propositional and sensory imaginings about those situations.
Morans description of dramatic imagination renders it very similar to if
not identical with the phenomenon of empathy, or the closely related phenomenon of imagining being in the place or shoes of someone else (see

Emotional Imagining As Experiential Imagining

355

Goldie (2000) for an extensive discussion of imaginative projects of this


kind).
Moran does not always clearly distinguish between emotional imagining
and dramatic imagining. In fact, he notes certain close links between the
two. Empathetic identification with a certain point of view different from
ones own often involves the dramatic rehearsal of emotions; while emotional imagining may require such things as dramatic rehearsal, it involves something ... like a point of view, a total perspective on the situation (Moran (1994): 105). However, the two are none the less quite different phenomena. Not only is dramatic imagining typically more complex
than emotional imagining (i.e., results in extensive mental projects rather
than in single mental episodes), but the two phenomena are also independent from each other. On the one hand, we can empathise with or enter the
mind of another person without actually having any real feelings, but instead only imagining them (see the proposal put forward in the last
section). And, on the other hand, we can respond with fear to imagining the
scenario of being pursued by a lion without thereby imaginatively adopting
a particular point of view different from ones own.60
Besides, Moran introduces the notion of imaginativeness which denotes
for him a complex ability covering, in particular: the ability to recognise
and link to each other the features of artworks which are responsible for
their emotional tone (i.e., their expressive features; the ability to emotionally and otherwise respond to these features and their links; and the ability
to empathise with or put oneself in the place of someone else (see Moran
(1994): 86f.). It thus includes or combines both the capacity to imagine
emotionally and the capacity to imagine dramatically. But it also involves
certain non-imaginative, cognitive abilities, such as recognitional capacities or sensitivities.

60

This is true even if, say, what is involved is imagining having certain sensory and affective experiences. For our act of imagining need not further specify the perspectivalness of the imagined experiences, or assume by default that it is our own
(see Martin (2002a) and Dorsch (2010a)).

356

The Dependency Account

11.4. Morans Account of Emotional Imagining


According to Morans picture, what is central to the affective imagination
and, in particular, our emotional responses to fictional entities, is what he
has labelled emotional imagining. By contrast, dramatic imagining is
linked to the affective imagination only in so far as it may include what he
calls emotional imagining; and imaginativeness is linked to the affective
imagination only in so far as it includes the capacity to engage in emotional imagining (as well as in dramatic imagining that includes emotional
imagining).
What he calls emotional imagining fits Morans two claims about the
nature of affective imagining. It involves both the imagining of certain aspects of the fictional world in question (e.g., that a character suffers unjust
treatment) and a really felt response towards these or related aspects (e.g.,
real feelings of sympathy towards the character and of anger towards the
unjust perpetrators). In accordance with Morans second claim, the latter is
taken to consist in a quasi-emotion triggered by the former. That is, imagining something with feeling or emotion consists in imagining something
with some quasi-emotion directed at it. The resulting episode is, for Moran,
an instance of genuine emotion. In accordance with this, what is responsible for the affective character of the resulting imaginative experience is
not the imagined content, but instead the really felt quasi-emotion as
Morans first claim maintains.
His main reason for taking emotional imagining to be central to affective imagining notably in the context of our experience of representational
art is his claim that imagining having an emotion is no exception to the
rule that it is always possible to imagine something in a dispassionate way
(just as it is always possible to imagine it with feeling). 61 We can imagine
61

Moran presents another motivation for assuming that our responses to fiction and the
involved affective imaginings are really and not merely imaginatively emotional:
namely that we are often held responsible for having or failing to have them. He
notes that we may be praised or blamed (morally or otherwise) in relation to whether
we react to fictional situations, say, with laughter or lust; and that how we react often
reveals something important about our personality (see Moran (1994): 93f. and 105).
Laughing at a racist joke, for instance, may reveal racist tendencies or beliefs. However, as Walton has correctly pointed out (see Walton (1997), what manifests our
convictions and is subject to assessment can equally well be our dispositions to ima-

Emotional Imagining As Experiential Imagining

357

that we feel sad or visualise ourselves as expressing our sadness through


crying, say, without thereby being in any affective state. That is, both propositional and sensory imagining can occur in a dispassionate manner,
even if they have a felt emotion as their object of imagining. As already indicated in the last section, Morans explanation of this fact is that having an
emotion as part of its content does not suffice for a representation to be
emotional. That is, real affectivity cannot simply derive from imagining of
an emotion. But if what we imagine has no impact on the affective dimension of imagining, the thought continues, it has to be due to how we imagine it. Hence, Moran concludes that affective imagining consists in, or
involves, emotional imagining.
This line of reasoning has three weaknesses, though. The first is that
Moran has not done enough to establish the claim that what is imagined is
always neutral on emotionality of the imaginative episode concerned. This
may be true of intellectual and sensory imagining. But there are perhaps
other ways of imagining feeling an emotion which are, by their very
nature, always affective in character. I return to this possibility in the next
and last section.
A second problematic aspect is that the passivity of the occurrence of
quasi-emotions casts serious doubts on the imaginativeness of any episodes
involving them as one of their constituents. The various forms of imagining might perhaps allow for passive instances such as spontaneously
arising images and thoughts in the case of sensory and intellectual imagining, or aimlessly floating daydreams in the case of more complex imaginative projects.62 But they would still have in common that they also allow for
voluntary instances and, moreover, permit us to take deliberate control of
their passive instances. We can actively sustain the fleeting spontaneous
images and thoughts and give their further development a particular direction (see Section 14.4). But what Moran takes to be emotional imagining
can never be subject to our direct voluntary control, given that the occur62

ginatively engage with fiction in certain ways rather than others.


See, for instance, the discussion of imaginings in OShaughnessy (2000). The idea
that instances of imaginings can sometimes be passive is compatible with their general subjection to the will (see, for example, Scruton (1974) and McGinn (2004)),
but not with their status as mental actions. I discuss the different positions and considerations and come to deny the possibility of passive imaginings in Chapter 13.

358

The Dependency Account

rence of quasi-emotions is not up to us. Of course, we may be able to bring


about quasi-emotions by exploiting our knowledge about our emotional
dispositions that representations of a certain kind give rise to those quasiemotions (e.g., we can induce quasi-fear in us by imagining something that
we know to scare us). But this does not render quasi-emotions subject to
our will at least not in the same direct way as imagining is (see Section
13.2 and Dorsch (2009b)). Hence, the emotional reactions that Moran focuses on and, in particular, the choice of representing whatever is imagined
in an emotional manner (rather than, say, in a visual manner) is never voluntary. So the challenge is to explain why we should count them as instances of imagining in the first place assuming that imagining is at least
in principle always subject to the will.63
The third weakness in Morans proposal is that it is not clear how to
make sense of his idea of emotional imagining. His talk of imagining with
feeling invites a certain ambiguity. If this form of imagining is meant to
consist just in the complex of an episode of propositional or sensory imagining and an additional episode of quasi-emotion triggered by the first,
then it is doubtful that it constitutes an instance of affective imagining at
all. Nothing ensures that the two episodes are more closely linked to each
other than by a causal connection, given that the very same quasi-emotion
can also occur in response to perceiving or believing something. Hence,
the overall complex is imaginative only in so far its dispassionate component is imaginative; while it is affective only in so far its non-imaginative
element is emotional. The emotional and imaginative elements in affective
imagining should be expected to be more unified. This suggests taking
Morans comparison of emotional imagining with visual imagining more
seriously. There is good reason to assume that the content and the manner
of representation are inseparable. Hence, if emotional imagining literally
63

See Chapter 13, as well as Scruton (1974) and McGinn (2004). A similar argument
may be formulated against the idea that our awareness of what a picture represents is
imaginative, given that it is usually not up to us what we experience a picture as depicting (with the exception, perhaps, of ambiguous pictures), or whether we experience it as depicting something in the first place (see Dorsch (2012c)). None the less,
Walton (1990) is not the only one who defends an account of pictorial experience in
terms of imagining. The same idea can be found in, for instance, Scruton (1974),
OShaughnessy (2000) and Stock (2008).

Emotional Imagining As Experiential Imagining

359

involves an affective way of representing something just as visualising


involves a visual way of representing something then the unity of the affective and the imaginative elements can be guaranteed. According to this
interpretation, emotional imagining is more than the mere conjunction of
some imaginative episode and some subsequent emotional response. Like
in the case of visualising, the content and the manner of emotional imagining are understood as aspects of a single and unified experience.
However, the postulation of an imaginative episode with an emotional
manner is problematic for its own reasons. First of all, how something is
represented puts a characteristic restriction on what can be represented. At
least, this is the case with all the widely accepted ways of representing
something. Visual representations are limited to visible entities: we can see
or visualise only objects and features which are visible. Something similar
is true of other sensory modes, such as representing something in an auditory or tactile manner. Intellectual representations come with conceptual restrictions: we can believe in or suppose the truth of only those propositions, which we possess the required concepts for (and, perhaps, also only
those propositions that are not logically inconsistent); and we can desire
the realisation of only those states of affairs that we can conceive of. Finally, representation in a conative or motivational manner is limited to possible courses of action: all our intentions, strivings and impulses are concerned with something to do. 64 By contrast, there are no distinctive restrictions on what we can imagine with emotion, or with specific emotional
feelings. If at all, such imagining inherits its limitations from the underlying dispassionate imagining, such as visualising or propositional imagining.
Moreover, the traditionally assumed manners of representation exclude
each other. Thus, we cannot represent something, in a single instance of
representation, visual-auditorily, or tactile-propositionally. Of course, our
episodes can involve two distinct representational elements which involve
different manners of representation for instance, when we see and hear a
64

Not all desires are conative. We may perhaps desire the occurrence of peace, or that
it will rain (in contrast to desiring to actively bring about peace or rain). But such
desires are not motivational states. Whether they are like emotions or preferences,
and whether they involve a distinctive manner of representation, are interesting
questions which, however, need not concern us here.

360

The Dependency Account

theatre production, or have thoughts about what we feel. But each of the
representational elements is still confined to a single manner of representation. However, as Moran acknowledges, it is possible to visualise or, indeed, propositionally imagine something with feeling. Again, this provides
a good reason to doubt that there is an emotional mode of imagining, in addition to and of the same kind as sensory and intellectual modes.
Moran therefore faces a dilemma in relation to his insistence on the existence of emotional imagining. If he conceives of the emotional element
as something in addition to sensory or intellectual imagining, he cannot ensure that the two components are unified in a single instance of affective
imagining. But if he understands the emotional element as a substitute for
the sensory or intellectual component in other instances of imagining, he
cannot accommodate the fact that the affective element behaves in a different way and, indeed, combines well with sensory or intellectual elements.
The conclusion should be that what Moran calls emotional imagining
that is, imagining something with emotion does not constitute a distinctive form of imagining. At best, it captures the fact that some of our imaginative representations give rise to quasi-emotions.

11.5. The Experiential Nature of Emotional Imagining


The discussion of Morans proposal has shown that locating the emotionality of affective imagining in the manner of representation is not a plausible
option. If there is such a thing as emotional imagining, its affectivity
should be due to what is imagined, and not how it is imagined. However,
the challenge for Walton has been precisely to say more about how it can
be possible that instances of imagining are affective episodes just in virtue
of their content that is, more specifically, just in virtue of being
representations of having an emotion. Moreover, the affective element of
emotional imagining cannot derive from any underlying quasi-emotions
as, again, the considerations about Morans view have illustrated. And
there are no obvious candidates for some other real emotional feelings that
might be involved in instances of the affective imagination, such as our
emotional responses towards representational art. Therefore, the challenge
for Walton can be formulated in a more refined way: how can emotional

Emotional Imagining As Experiential Imagining

361

imagining possess an affective character in virtue of representing an emotion, without actually including any real emotional feeling?
Moran has proposed two types of imagining concerned with emotion:
propositional imagining about emotions; and imagining something with
emotion. The first is characterised by the fact that the emotions form part
of the propositional content of the imagining for instance, when we imagine that we have or feel fear directed at an imagined lion in the room. By
contrast, Moran takes the latter to be an episode of imagining something in
an emotional manner. Propositional imagining is dispassionate and therefore no good candidate for affective imagining; while imagining something
in an emotional manner can be ruled out since there are good reasons to
doubt the existence of such a mode of representation. But independently of
what one thinks about the plausibility of imagining something emotionally,
there is at least a third alternative in which emotions may enter imagination: they may be the direct objects of non-propositional imagining.
Consider the case of imagining a pain (or a similarly subjective bodily
phenomenon, such as an itch; see Sections 10.1f.). This imaginative episode differs subjectively from real instances of pain. Most notably, we do
not come to find the former unbearable in the same way as the latter for
instance, we do not cry or faint as a consequence of experiencing it. In this
respect, imagined and remembered pains are much closer to each other
than to really felt ones. Moreover, this difference between imagined (or remembered) and really felt pains is not simply a matter of degree in determinacy or intensity. We sometimes have real pains which are not very intense or determinate, but which we still experience as real pains, and not
merely as imagined ones. And we also can imagine having rather strong
and specific pains, without thereby beginning to really feel pain. None the
less, imagined (and remembered) pains still involve the quality of pain.
This is reflected by the fact that we describe their subjective character in
terms of pain (e.g., that they feel similar to genuine pain) and group them,
from our first-personal perspective, together with real feelings of pain,
rather than, say, with thoughts about pain, or with sensations of itch. In
short, imagining a pain is an experience, but not a real pain experience.
That is, it involves the quality of painfulness, but does not instantiate it.
The best explanation of this situation is to assume that imagining (and

362

The Dependency Account

perhaps also remembering) a pain is an instance of non-propositional object awareness (or acquaintance) which takes the feeling of pain rather
than the felt pain as its direct object. 65 The idea is that, while a feeling of
pain involves painfulness by instantiating it, the imaginative (or mnemonic) awareness of such a feeling involves painfulness by representing it as
instantiated. As a result, feeling pain and imagining it are subjectively similar in that both their phenomenal characters involve the quality of
painfulness. But they differ from our first-personal perspective in that they
involve this qualitative aspect in different ways: the former is really an experience of pain, while the latter is an episode of representing pain. The involvement of painfulness in the case of imagining pain is thereby not a
matter of the non-propositional manner of representation. Rather, it is a
matter of what is imagined, namely a feeling of pain which instantiates the
quality of painfulness.
The proposed treatment of imagining pain therefore satisfies all the conditions on affective imagining: it is non-propositional; it does not involve
an emotional manner of imagining, but instead is a representation of emotion, and its affectivity does not amount to the real thing, that is, to a genuine instance of emotion. Emotional imagining may and should thus be
understood, along the lines of (ER), as an instance of experiential imagining: namely as imagining an episode with an affective phenomenal character (i.e., an episode of emotion). This guarantees that the affectivity of
emotional imagining arises from what is imagined, and not from how it is
imagined. But it also ensures that there is an affective element involved in
emotional imagining, but no real emotional feeling, only a represented one.
The proposed imaginative and non-propositional manner of representation does not face the same problems as Morans emotional manner. First
of all, it puts a restriction on what can be imagined, namely particulars and
their experienceable features in this case, mental episodes and aspects of
their phenomenal characters. And then, it excludes other manners of representation. This is true independently of how the precise nature and role of
the non-propositional manner involved in emotional imagining is specified.
There are basically three options. First, it may be held that all instances of
object awareness involve the same basic non-propositional way of repres65

See the comparable claims about itchiness and perspectivalness in Martin (2002a),
which I presented in Section 10.1.

Emotional Imagining As Experiential Imagining

363

enting something. Perception and imagination then differ in whether the


represented objects are external objects or mental episodes (i.e., representations of external objects). Second, it may be thought that, while perception
does not involve representation at all (but instead some relational form of
awareness, such as acquaintance), imaginative object awareness always
consists in the non-propositional representation of some episode with a
sensory, affective or similar character. This means taking visualising, say,
to be the imaginative representation of a visual perception. And third, it
may simply be claimed that the non-propositional manner of representation
is distinctive to the (imaginative, mnemonic or otherwise) representation of
mental episodes and their phenomenal characters we may call it an
experiential manner of representation (see Dorsch (2010a)). This option
leaves it open whether visualising is directed at external objects or
perceptions of them, as well as whether perception is representational or
relational. But in all cases, the various manners of representation including that taken to be involved in emotional imagining remain
mutually exclusive.
In addition to meeting the conditions on affective imagining which have
emerged during the discussion of Waltons and Morans views, the account
in terms of the non-propositional representation of emotions fits well with
Waltons approach to our engagement with representational art. The demand imposed on us by the occurrence of quasi-emotions may very well be
understood as involving the demand to imagine oneself as feeling the emotion concerned (see the third element). For this imagining is not only nonpropositional, it also happens from the first-personal perspective: we imagine the instantiation of the phenomenal character of an emotion, and this
character is identical with what the emotion is subjectively like (see
Dorsch (2010a)). This also clarifies the relation to the underlying quasiemotions: they are not constituents of affective imagining, but indirectly
give rise to the latter in so far as they establish the fictional truth that we
feel an emotion towards the fictional entities in question and, hence,
require us to engage in affective imagining as part our engagement with the
representational artwork in question.
This proposal may then be used to explain why the resulting affective
states do not motivate us in the same way as the real emotions, namely to

364

The Dependency Account

interact with the respective entities at which they are directed. The idea is
that the emotional aspect looses its motivational power, once it is experienced, not in an immediate way, but instead only mediated by a non-propositional representation of it. Thus, while the experience of fear felt towards a real lion has the power to move us to run away, the imaginative
representation of such an experience of fear directed at a fictional lion does
not possess this power any more. Similarly, the intensity and determinacy
of the feeling usually decreases when we move from a real experience to a
represented one. Thus our imaginations (and memories) of fear are typically though not necessarily always less vivid than the comparable experiences of fear in real life situations (see Section 3.2). The claim that our
emotional responses towards fictional entities amount to the non-propositional imagination of having emotions towards those entities may thus help
to explain some important characteristics of our engagement with representational art.
Finally, the provided account of emotional imagining is compatible with
the idea that it is distinctive of imagining that it allows for voluntary control. The occurrence of quasi-emotions and the subsequent establishment of
a fictional truth to be imagined by us may be beyond the direct influence of
our will. But whether we follow this demand and imagine feeling the respective emotion is at least in principle up to us. The truth of (ER) with respect to emotional imagining and other instances of affective imagining is
therefore compatible with the Agency Account of imagining.

Chapter 12
Semantic Dependency,
Simulation, and Pretense

The representational echo thesis (ER) should be given up since it does not
apply to intellectual imaginings and imaginative projects and, therefore,
cannot capture the shared nature of all central forms of imagining. The
most plausible version of the causal echo claim (EC), on the other hand,
has turned out to be irrelevant for the formulation of a unified account of
imagining, given that it takes imaginings to be dependent on cognitions
only in so far as the latter are needed to acquire the necessary means for
representing something in a sensory or intellectual manner in the first place
a claim which is too weak to be distinctive of imaginings, or to link imaginings to cognitive counterparts with the same content and mode of representation.
But what about other echo claims about imagining? In what follows, I
argue that the postulation of a different kind of dependency is equally unsuitable for capturing the nature of all or even most instances of imagining. In the second and the third section of this chapter, I consider the attempt to characterise imaginings in terms of simulation and pretense, respectively. Both views may be understood as endorsing the idea that imaginative phenomena fundamentally imitate cognitive ones:
(EI) Imaginings are essentially imitations (or simulations) of their
cognitive counterparts.
The next section, by contrast, takes a closer look at the idea that imagin-

366

The Dependency Account

ings are, as part of their fundamental nature, semantically dependent on


certain cognitions:
(ES) Imaginings are essentially semantically dependent on perceptions
(and other cognitions).
My main concern is with the question of whether this echo thesis can be of
any relevance for the provision of a unified account of imagining.

12.1. Semantic Dependency


Among many others, Currie and Ravenscroft (2002: 18f.) argue that we
cannot use natural kind or name-like concepts in intellectual imagining if
we do not already have beliefs involving those concepts, or at least had
judgements or beliefs involving them in the past. The idea is that intellectual imaginings cannot stand in the kind of relations to the environment or
the community which are commonly assumed to be required for the fixation of the reference and, hence, also for the acquisition of such concepts.66 For example, that we intellectually imagine that there is water in
London and not that there is twater in Tlondon needs to be explained in
terms of the fact that we have been in the past in cognitive contact with
water (i.e., H2O) and the city London and not with their twins twater (i.e.,
XYZ) and Tlondon in another possible world in such a way as to require
the two respective concepts. But this cognitive contact and the subsequent
acquisition of the concepts arguably require the perception-based formation of judgemental thoughts or beliefs about water and London. As a result, intellectual imaginings presuppose both perceptual and intellectual
cognitions.
Something similar seems to be true of sensory imaginings and their representational capacity. It might perhaps be possible to visualise a shade of
blue without having seen it before for instance, by interpolating what the
hue is like from closely resembling shades of blue that we did see in the
past (see Section 9.4). But it does not seem to be possible to visualise that
missing shade of blue without having had any perceptual exposure to oth66

See the extensive debates started by Putnam (1975) and Burge (1979).

Semantic Dependency, Simulation, and Pretense

367

er (blue) hues or, at least, to some other perceivable features of


whichever kind. A congenitally blind person may be able to sensorily imagine the bluishness of an object by evoking tactile or olfactory representations (e.g., of coolness, smoothness or freshness). But he is presumably unable to imagine it visually, and able to imagine it in a different sensory
modality only because he has had respective experiences in that modality
in the past.67 It is important to note that what is at issue here is not the
(non-accidental) veridicality of sensory imaginings, but their possibility. If
I ask you to sensorily imagine the taste of a Rambutan (which you have
never tasted, but know to be an exotic fruit), it is highly likely that you will
be able to conform to my request for instance, by combining or otherwise
relying on your episodic memories of the taste of other exotic fruits. But it
is indeed highly unlikely that there will be a match between how your gustatory imagining represents the taste of the Rambutan to be and how it
really is like. Besides, your ability to gustatorily imagine the taste of a
Rambutan presupposes in part some past gustatory (or perhaps also other)
experiences, on which you may rely in your imaginative activity.
What these considerations show is that both our sensory and our intellectual capacities to imagine something depend on our perceptual and conceptual capacities. Since we acquire the latter by cognitively interacting
with aspects of the world (including, say, abstraction from sensory input, or
conceptual analysis and construction), it follows that imaginings depend on
cognitions for their presentational power. This semantic dependency explains why we are, in many ways, limited in what we can imagine notably in that we can sensorily imagine only perceivable items and features,
and imaginatively refer to real individuals or natural kinds only if we stand
in the right cognitive relation to them (see Sections 2.3 and 7.1).
However, since the semantic dependency does not link imaginings to
specific cognitive counterparts (e.g., there are many ways of acquiring a
concept, or of combining past experiences), and also since many other representational mental episodes rely in this or a similar way on our previously gained perceptual and conceptual capacities, the claim that imagin67

See Thomas (1999) for apparent empirical evidence to the effect that congenitally
blind people can enjoy a kind of sensory imagining that comes very close to visual
imagery; and Scruton (1974: 104) for the more orthodox opinion that such people
cannot visualise anything.

368

The Dependency Account

ings are semantically dependent on cognitions and, hence, the Semantic


Account becomes relatively uninteresting for a theory of the distinctive
features of imaginings. (ES) may still help to elucidate the representational
side of the nature of imaginings, but it cannot contribute to an account of
what distinguishes imaginings from other kinds of mental episode.
That the semantic echo claim (ES) faces more or less the same difficulties as the causal echo claim (EC) is no accident. Indeed, (EC) is a direct consequence of (ES). The causal chain leading to the occurrence of a
particular instance of imagining includes the cognitive episodes which
have been crucial though merely contingently so in the provision of the
necessary means for sensory or intellectual representation. In other words,
some of our past cognitive episodes are among the causes of our present
imaginative episodes in so far as the cognitions have been, in part, causally
responsible for the acquisition and, hence, employment of the relevant representational capacities.

12.2. Imagination as Simulation


We have already seen that Currie and Ravenscroft endorse (ES). But, more
important, they make use of the semantic dependency in an argument for a
specific version of (EI), according to which imaginings imitate cognitions
in the sense of being simulations of them. 68 More specifically, their theory
of imagining contains three central elements.
First of all, they highlight some similarities between imaginings (or
what they also call states of recreative imagination: Currie & Ravenscroft
(2002): 11) and cognitions with regard to the way in which the two kinds
of representations may be integrated into a system of causally or rationally
related mental representations. The underlying assumption is that what
matters for the role of representations in inferential patterns (in the case of
intellectual representations) or similar sensory processing mechanisms (in
the case of visual and other sensory representations) is solely the content of
the representations, and not their attitude or other features. Hence, given
that imaginings and cognitions may share their contents, they can figure in
the same type of inferences and mechanisms whether these are real or
68

Similar ideas can be found in Heal (2003) and Goldman (2006).

Semantic Dependency, Simulation, and Pretense

369

online when cognitions are concerned, or merely hypothetical or offline


when imaginings are concerned (see Currie & Ravenscroft (2002): e.g., 49,
81, 93f. and 100; see also Currie (2000): 176).
Then, they take these causal and rational similarities to enable imaginings to simulate or to mimic, and relative to certain purposes, to substitute for perceptions, beliefs, decisions, and experiences of movements:
imaginings can take over the role of cognitions in certain mental processes,
such as reasoning or the processing of visual information only, again,
with the proviso that the processes involving imaginings run offline
rather than online as the processes involving cognitions (Currie &
Ravenscroft (2002): 11; 49).69 One particular consequence of this is that we
are able to engage in what Currie and Ravenscroft call perspective-shifting and what they are mainly concerned with in their book: namely our
ability to put ourselves in the place of another, or in the place of our own
future, past, or counterfactual self (Currie & Ravenscroft (2002): 8f.).
Finally, they assume that the relation of simulation entails not only a
partial and, presumably, symmetric similarity in causal or rational role, but
also some kind of asymmetric dependence (Currie & Ravenscroft (2002):
49). Although they are not entirely clear about it, Currie and Ravenscroft
seem to have in mind a constitutive or conceptual dependency of imaginings on cognitions. They maintain that imaginings are parasitic on cognitions and are adequately describable only by reference to some counterpart, that is, perception, belief, and so on (Currie & Ravenscroft (2002):
19; 32; see also Currie (2001): 254). 70 They also claim that the dependency
in question as well as the resulting relation of simulation holds only in
one direction: imaginings are dependent on and simulate cognitions, but
not the other way round (Currie & Ravenscroft (2002): 18f.; 49). Accordingly, Currie and Ravenscroft embrace an echo thesis about imaginings:
Currie and Ravenscroft note, however, some doubts about the simulative potential of
sensory imaginings (e.g., Currie & Ravenscroft (2002): 94 and 100).
70
Scruton seems to endorse a similar idea of dependency when he writes that [a man]
will feel that to describe his image in terms of sensory experience is appropriate, and
indeed inevitable (Scruton (1974): 100). But in apparent contrast to Currie and
Ravenscroft, he also accepts that the respective relation between sensory imaginings
and perceptions cannot really be further specified: [the] man will be unable to
indicate in what way his image is like a particular sensory experience (ibid.).
69

370

The Dependency Account

their account involves both an idea of similarity and an idea of constitutive


or conceptual dependency. Their characterisation of the dependency relation in terms of simulation and their simultaneous rejection of (ER) (see
Section 10.3) indicate that they, more precisely, endorse (EI).
Currie and Ravenscroft argue in detail for their similarity claim and
their thesis that imagining is central to perspective-shifting (Currie &
Ravenscroft (2002): sec. 1.3f., and chs. 3ff.). They also provide substantial
support for the idea that imaginings are semantically dependent on imaginings (see above). For the sake of argument, I assume here that Currie and
Ravenscroft are right about all.71 My focus is instead on their idea of an
asymmetric dependency, and on the fact that they spend hardly any time on
elucidating or supporting this idea. Indeed, the only claims that they appear
to argue for in relation to their postulation of such an asymmetric dependency are their claim that imaginings depend on cognitions for their representational power, and their claim that imaginings and cognitions may enter
the same types of inferential or informational processes.
The main problem with understanding (EI) in terms of (ES) is, however,
that the former but not the latter concern pairs of intellectual counterparts. The supposition that there is water in London is said to simulate the
belief (or judgemental thought) that there is water in London. If the two
episodes would not share their content (i.e., the proposition that there is
water in London) and their mode of representation (i.e., their intellectuality), they would not share their inferential role. By contrast, in order to be
able to intellectually imagine that there is water in London, it is not required to have (had) the belief that there is water in London. All that is
needed are some prior water-belief and some prior London-belief. Hence,
the semantic dependency cannot be central to the constitutive or conceptual
asymmetric dependence that Currie and Ravenscroft postulate in their
71

One obvious target for criticism is their ultimately incorrect rejection of the
Agency Account of imagining, as well as of the Representational Account with respect to sensory and affective imaginings (see Section 10.3 and Part Four). Their
postulation of desire-like imaginings is also problematic (see Note 16 in Chapter 1).
Carruthers (2003) and Soldati (2006) note further difficulties in their critical reviews
of Currie and Ravenscrofts book. Note, however, that the similarity claim is compatible with (ER): it suffices for resemblance that the content of the imaginings contains the content of the perceptions (see Sections 9.1 and 9.4).

Semantic Dependency, Simulation, and Pretense

371

specific simulationist version of (EI).


Their similarity claim with respect to the shared general causal or rational role, on the other hand, is just a consequence of their observation
that imaginings and cognitions share their contents and their assumption
that all what matters for informational or inferential processes is content.
This means that the similarity claim is not much more informative than the
claim about the shared contents, entailing that it cannot be at the core of a
unified account of imagining. Furthermore, it also means that the similarity
claim does not constitute an echo claim, let alone a simulationist version of
(EI).
This leaves only two plausible conclusions. One is that Currie and
Ravenscroft do not really intend to capture the distinctive nature of imagining when claiming that they are simulations of cognitions. Instead, what
they have in mind is just the similarity claim; while they take the asymmetric dependency to be completely independent of the similarity in informational and inferential role, and to consist in nothing more than the semantic
dependency of imaginings on cognitions.72 However, the way in which
they present their simulationalist view (see, for instance, Currie &
Ravenscroft (2002): 49ff.) strongly suggest that they are more ambitious
with their simulationist approach, and that the asymmetric dependency is
goes beyond semantic dependency not the least because many non-simulating mental states are also semantically dependent on perceptions or beliefs (e.g., memories, emotions, desires, hopes, and the like). This is in fact
the second plausible conclusion: that they actually aim to defend a simulationist version of (EI) which assumes a genuine constitutive or conceptual
dependency of imaginings on cognitions, but that this central claim of their
view remains unelucidated and unmotivated.
Their comments about perspective-shifting, too, do not introduce other
considerations that might further illuminate or even speak in favour of
their endorsement of the idea of a pairwise asymmetric dependency of imaginings on their cognitive counterparts. It might be suggested that Currie
and Ravenscrofts true intention is to identify imaginings with precisely
those mental episodes that are essential to simulation in the sense of perspective-shifting. But there are three problems with this proposal. First, it
72

This would help to explain a certain passage in Currie (2000: 176), where he describes his simulationist view without at all mentioning the asymmetric dependency.

372

The Dependency Account

would fail to establish the desired asymmetric dependency of imaginings


on cognitions. Second, it would permit for an account of imaginings in
more fundamental terms say, by reference to the idea that they, and only
they, are voluntary with respect to what they represent, and to the idea that
perspective-shifting requires mental representations which are voluntary in
this respect. Third, the proposed characterisation of imaginings cannot
merely claim that perspective-shifting has to involve imaginative episodes.
For the same is true of certain non-imaginative representations as well. In
particular, imagining being in the position of someone else and simulating
their state of mind requires some beliefs about the other perspective (e.g.,
about the respective persons situation, mental states or character traits),
and the desire or intention to simulate occupying that perspective.
The proposal might be strengthened by adding the idea that imaginings
are to be defined by their unique role in perspective-shifting. For example,
they might be identified with precisely those episodes that constitute the
adoption of a perspective distinct from our actual one rather than merely
enabling such an adoption, as the underlying beliefs and desires do. However, this new suggestion does not really tell us anything about the distinctive nature of imaginings, other than perhaps that they are all concerned
with our possible rather than our actual occupation of a perspective. 73
Consequently, the proposal to characterise imaginings in terms of their
unique role in perspective-shifting amounts to the view that what is distinctive about imaginings is that they are concerned with something merely
possible. It therefore falls victim to the same objections as the Modal Account discussed in Section 5.4. Accordingly, the connection between imagining and perspective-shifting seems looser than perhaps originally
thought.

73

It does not matter for this characterisation of imaginings whether the perspective,
that we merely possibly adopt, is actually occupied by someone else, or was or will
be by our own past or future self. It is true that perspective-shifting may help us to
get clearer about the actual situation of others in this sense. But it always does so, at
the basic level, by considering a perspective which we do not actually, but merely
possibly adopt.

Semantic Dependency, Simulation, and Pretense

373

12.3. Imagination as Pretense


But if we cannot satisfactorily spell out the idea of an asymmetric relation
of imitation involving a constitutive or even conceptual dependency of
imaginings on their cognitive counterparts in terms of simulation, it is difficult to see how else we can make sense of that idea. One possible alternative is to maintain that imaginings are mental acts of pretending to see,
believe, and so on which is perhaps a plausible interpretation of Ryles
view on imaginings.74 The proposal implies a constitutive link between
imaginings and cognitions which is, moreover, modelled upon a fairly
common and well-understood phenomenon, namely pretense. There are,
however, three different ways of understanding this view, none of which
leads to a tenable account of imagining.
First, the idea that imaginings pretend to be cognitions (or, for that matter, simulate cognitions) might be understood in terms of make-believe.
The idea is that to sensorily or intellectually imagine something is to makebelieve that one sees or believes. But this would just shift the problem to
an analysis of what it means to make-believe that something is the case.
The natural proposal of taking make-believe to be identical with (some
form of) intellectual imagining is bound to fail. For this identification
would amount to the indefensible denial of the existence of sensory or affective imaginings, given that it would lead to the reduction of all forms of
imagining to intellectual imagining. Moreover, the nature of the form of intellectual imagining concerned would still be in need of elucidation (and of
differentiation from other forms of intellectual imagining), given that it
could not itself be elucidated in terms of make-believe without the occurrence of an infinite regress.
Second, imagining might be said to literally constitute a form of pretense. That is, imagining just means pretending and is of the same kind of
phenomenon as the following cases: pretending to be someone else (e.g., in
order to commit some fraud); pretending that one has forgotten a friends
birthday (e.g., just to surprise him or her later on); or pretending to participate in the game of make-believe that tree stumps are bears (e.g., to make
74

See Ryle (1949/1963): sec. 8.6. See Ishiguro (1966), Shorter (1952) and White
(1990) for critical assessments of Ryles position.

374

The Dependency Account

ones children happy).75 However, this proposal should be rejected, given


that imagining can occur without pretending, and vice versa (see White
(1990): 150ff.; and Currie & Ravenscroft (2002): 33; 153f.).
For instance, I may pretend to be dead, or pretend to participate in the
game of make-believe mentioned above and pretend that tree stumps are
bears (e.g., call them bears, run away from them, warn other people about
them, and so on), without thereby having to imagine that I am dead, or that
the stumps are bears. It seems that I just have to believe that, as part of my
pretense, I have to lie completely still, or to treat tree stumps in a certain
way. Similarly, an imposter may pretend to be an aristocrat without imagining that he is one, or without visualising himself as one. Instead, he may
simply change his appearance and behaviour in accordance with (what he
knows about) how aristocrats typically look and act. A different person, on
the other hand, may imagine being an aristocrat and picture himself entertaining noble guests in an old and elegant country house. But he need not
thereby perform any bodily actions, and in particular none concerned with
establishing a likeness in appearance or behaviour with aristocrats, which
may then be picked up by a receptive audience.76
I will not try to provide an intensional characterisation of the notion of pretense that
I have in mind. But the same or at least a very similar notion seems to be prevalent in Ryle (1949/1963): sec. 8.5., Shorter (1952), White (1990): ch. 17, and Nichols
& Stich (2000). Their examples are very similar to mine, e.g., pretending to be a
bear or a corpse (Ryle (1949/1963): 245); or pretending that a banana is a telephone
(Nichols & Stich (2000): 117). Furthermore, I assume that genuine participation in
games of make-believe but not pretended participation requires imagining (or at
least trying to imagine) according to the rules of the game (see Walton (1990): ch. 6,
especially 209ff.). When Walton speaks of pretense as the verbal participation in a
game of make-believe (Walton (1990): 220), he does not seem to claim that pretense
should or has to involve imagining in general, but only if it becomes part of (and an
expression of the respective imaginings involved in) ones participation in a game of
make-believe.
76
That these particular instances of pretense should not be taken to be imaginative
does not mean that other instances (perhaps even of the same kind of pretense) cannot involve one of the types of imagining outlined above. For instance, I may really
enter the game of make-believe, imagine stumps to be bears and, as a consequence,
get emotionally involved in the game (e.g., by feeling anticipatory fear). And imagining being another person may help one better pretend to be that other person
(something which is perhaps exploited by method actors).
75

Semantic Dependency, Simulation, and Pretense

375

The underlying reason for why imagining and pretending are distinct
and fairly independent phenomena is, of course, that, while instances of
imagining are mental phenomenon, instances of pretense are just like instances of assertion or stage-acting inseparably linked to publicly observable behaviour and, relatedly, require some actual or at least potential audience (see Ryle (1949/1963): sec. 8.5, and Currie & Ravenscroft (2002):
32). Pretending is aimed at behaving as if one were a particular person, or
as if one were having certain mental states. Moreover and, again, just like
assertion or stage-acting pretense may be used for deception. Consequently, it is no surprise that imagining cannot literally amount to (a kind
of) pretending.
Third, it might be attempted to elucidate imagining by means of drawing an analogy with pretense for instance, by claiming that imagining is a
form of inner pretense, while not really amounting to pretense proper
(i.e., publicly observable pretense). But this last proposal is not much more
plausible than the previous two. This time, the problem is that imagining
and pretending do not have any significant aspects in common, apart from
their being instances of agency.77 In particular, while pretending to do or to
be something is modelled on a specific kind of action, or on some condition or state closely related to specific behavioural dispositions, imagining
is not (see White (1990): 157f.).
This means, first of all, that pretense involves the same, or very similar,
actions as the object of pretense (i.e., the mimicked action, condition or
state), but without becoming the real thing. Pretending to have a fight with
someone involves performing some of the bodily movements pertaining to
real fights, but does not amount to fighting; while pretending to be an aristocrat consists in acting in ways in which real aristocrat behave, without
thereby really becoming an aristocrat. Moreover, acts of pretense can be
appropriately described only by reference to what their object is: sham77

White lists certain other similarities for instance, that both imagining and pretending are rather unspecific with respect to their objects, or that both establish a contrast
with what is real (White (1990): 153ff.). But this is just to say that both lack a cog nitive attitude and may be relatively indeterminate which does not, however, distinguish them from many cognitions (in the case of indeterminacy; see Section 3.3)
or other non-cognitive phenomena (in the case of a lack of cognitive attitude; see
Sections 5.5 and 8.1).

376

The Dependency Account

fighting has to be elucidated in terms of real fighting. Finally, the pretending person has to rely on her beliefs about the object of pretense and the related ways of acting when engaging in her pretense: she can pretend to
fight only if she has some idea of what it means to really fight, and only if
she lets her respective beliefs inform and influence her behaviour in a manner appropriate for pretense (see Ryle (1949/1963): sec. 8.5; see also White
(1990): 155ff.).
In contrast, acts of imagining are not modelled on certain actions, conditions or states in such a way that they mimic the latter by involving the
same type of activities as the latter; that they have to be accounted for by
reference to the latter; or that the imagining person has to rely on her beliefs about the latter in her imaginative agency. If at all, imaginative episodes should be modelled on cognitive episodes: that is, they should be
taken to be pretend perceptions or pretend beliefs. But the only relevant
activities which may pertain to both imaginings and cognitions are attention and inference; and neither suffices to establish that imaginings mimic
cognitions. Paying attention to what we sensorily imagine (e.g., to certain
aspects of a visual image) already presupposes that the respective sensory
imagining has been formed. Inferring a certain proposition from a set of
others, on the other hand, is not necessary for the active formation of intellectual imaginings.78 Hence, neither form of agency can constitute the kind
of activity at the heart of active imagining.
Relatedly, the elucidation of what it means to imagine something does
not have to make reference to attention or inference. And the imagining
person also does not have to make instrumental use of her beliefs about attention or inference while being engaged in active imagining. Consequently, imaginings are not pretend cognitions. At best, one may perhaps speak of pretend inferences which mimic real inferences by using
and resulting in intellectual imaginings instead of judgements or beliefs;
78

It is maybe also not sufficient. For while inferences on the basis of beliefs seem to
compel us to endorse the resulting propositions, it seems ultimately up to us whether
we (continue to) imaginatively entertain a proposition which we have recognised to
rationally follow from other imaginatively entertained propositions (plus perhaps
some believed ones). This difference may also be related to the fact that there are no
dispositional imaginings which could compel us to imaginatively endorse a proposition simply by becoming manifest independently of our will (see Section 1.3).

Semantic Dependency, Simulation, and Pretense

377

and of pretend attention which mimics real attention by focusing on sensorily imagined objects instead of really perceived ones. But the fact that
we can pay attention to what we sensorily imagine and employ intellectual
imaginings in inferences appears to be primarily a consequence of the fact
that imaginings are of the same types of representation (e.g., visual or intellectual) as their cognitive counterparts. It does not pertain to a pretend
relation between imaginings and cognitions that is structurally similar to
that holding between acts of pretense and their objects.

12.4. Conclusion and Outlook


It is not very promising to attempt to formulate a substantial and satisfactory echo thesis by reference to the ideas of simulation or pretense. Indeed,
it seems rather difficult to make sense of an echo thesis otherwise than in
causal, representational or semantic terms. But (EI) has also difficulties to
capture imaginative projects, and for roughly the same reasons as (ER).
Imaginative projects cannot be construed as echoes of cognitive ones (e.g.,
daydreaming about walking on the moon does not simulate, or causally depend on, any cognitive project). And, they also cannot be characterised by
reference to the involvement of episodes which echo cognitive counterparts.
Moreover, with respect to intellectual imaginings, there is always the
simpler alternative of characterising all forms of intellectual representation
(whether they are cognitive, imaginative, or neither) in terms of one and
the same representational element (i.e., the entertainment of a proposition)
and an independent and perhaps not always necessary involvement of
an attitude towards the proposition in question. One advantage of this simpler view over (EI) is that it can characterise intellectual imaginings
without having to make substantial reference to intellectual cognitions. For
instance, it may be claimed that imaginatively entertaining a proposition
involves a specifically imaginative attitude (see Sections 2.1 and 5.4f.).
Now, the overall conclusion of this part of the book is that the Dependency Account of imagining cannot provide a unified theory of imagining.
The various echo theses, which take imaginings to be causally, representationally or otherwise dependent on cognitive phenomena, are at best unillu-

378

The Dependency Account

minating with respect to the distinctive nature common to all imaginings,


and at worst false with respect to certain central forms of imagining. Since
the Epistemological Account and the various minor accounts discussed in
Chapter 5 also fail as unified theories of imagining, it is necessary to look
somewhere for a satisfactory account of what the fundamental imaginativeness of the different forms of imagining consists in.
One option is to build upon OShaughnessys Argument from Origin for
(NC*) in the following way. The main idea is to say that the scope of the
argument covers too much ground because it concerns both imaginative
and non-imaginative phenomena. For most, if not all, of OShaughnessys
examples of passively formed imaginings seem to be controversial and, at
best, non-central cases of imagining: visual hallucinations which are either
phenomenologically indistinguishable from perceptions, or had by the insane and the overheated and the drug-laden, or both (OShaughnessy
(2000): 479; see also 349); hypnagogic images (ibid.: 349); pictorial
experiences (ibid.: 346ff.); dream experiences and dream beliefs (ibid.:
344f.; 349); and delusional representations (ibid.: 344f.). A promising
strategy may therefore be to argue that the scope of OShaughnessys account is not only too narrow (since it cannot really capture imaginative
projects, or cognising visual imaginings), but also too wide, given that it
captures these contentious passive cases.
In response, it may be maintained that the examples just listed should
not count as instances of imagining; and that there are, as a result, only active instances of imagining. The hope is then to be able to identify a specific
type of agency that is common to all the central cases of imagining, and
which is responsible for their imaginativeness. Although it remains true
that sensory and intellectual imaginings are non-cognitive in the sense specified by (NC*), this can be explained by a more fundamental feature: their
status as mental actions. At the same time, this promises to disprove
OShaughnessys belief that imaginings do not share a common essential
feature, over and above their property of satisfying his definition of imagining. For the idea is that the kind of mental agency, which has been identified as being at the heart of imagining, is essential to all instances of imagining. This is a very rough sketch of the strategy I aim to adopt, and the
position I aim to try to defend, in the remainder of the book.

Part Four
The Agency Account

Introduction to Part Four

In the preceding two parts of the book, I discussed the Epistemological and
the Dependency Account. While the first focused on the non-cognitivity of
imaginings, the second highlighted their descendance from, or constitution
by, cognitions. However, I found both theories wanting in respect of their
prospects of providing a unified account of imagining. It is therefore necessary to look for a different feature that is common to, and distinctive of, all
imaginings, whether they are sensory or intellectual, episodes or projects.
The feature that I would like to put forward in this part is their intimate relationship to the will.
It should be uncontroversial that many instances of imaginings are voluntary: they are actively formed and, as such, deliberate expressions of
what we want. In particular, it makes sense to ask or order someone to imagine, and also to answer to such a demand. 1 This aspect of imagining has
already been emphasised in the context of the discussion of Wittgensteins
and OShaughnessys writings on imagining (see Section 4.7 and Chapter
7, respectively), who both take the active character of the respective imaginings to be responsible for their non-cognitivity (e.g., their uninformativity, or their lack of a power to cognise reality). I agree in general with their
explanatory strategy of saying that voluntary imaginings are not in the
position to provide knowledge because of their origin in the will. But I disagree with them in two important respects. First, I assume that all and not
only some instances of imagining are actions. Second, I assume that it is
1

See Wittgenstein (1984c): vol. II, sec. 125; Scruton (1974): 94f.). See also
OShaughnessy (1980): vol. I, 1ff. for a detailed discussion of the link between voluntariness and the applicability of orders.

382

The Agency Account

precisely their active character which allows some imaginative episodes to


constitute knowledge under the right kind of circumstances (see the examples discussed in Chapter 4).
In this final part of the book, I shall develop and defend these ideas and
present them as the core of a unified account of imagining of what I shall
call the Agency Account of imagining. This theory of imagining is characterised by the claim that all central cases of imagining consist, essentially
and fundamentally, in voluntary actions of a particular kind; and that they
strictly differ in this from all paradigm cases of non-imaginative phenomena. As mentioned before in the introductory chapter, the Agency Account
deserves to be treated at monograph length, and my discussion of it here, in
the context of the issue of how best to account for the unity of imagining,
is bound to remain incomplete.
I already noted two important motivations for switching the focus of the
discussion to the idea that imaginings are a special kind of mental action.
One is that many of them are voluntarily formed and controlled; and the
other is the inadequacy of the various alternative accounts as unified theories of imagining. In particular, the idea of imaginings as something that we
are actively doing may be a more promising candidate for the foundational
element of a unified account because it does not readily exclude affective
imaginings or imaginative projects. To the contrary, mental projects are
already by their very nature instances of mental agency. And it is natural to
assume that, if purely representational imaginings turn out to be mental actions of a certain kind, imaginings that additionally involve an affective
element will do so as well.
A third motivation for switching focus is that several philosophers have
voiced support for the idea that (many) imaginings are voluntary often,
however, without elucidating or defending their claim in any detail. 2 This
2

The Agency Account has been endorsed by Wollheim (1973): 69, Scruton (1974): 95
(see in general 94-100), Levinson (1998): p. 232, n. 3, Kind (2001): 90ff., and McGinn (2004): 12ff.; 131f. Hopkins (2011a) also assumes the truth of the Agency Account, though perhaps only for the sake of argument. Most of these endorsements
have been inspired by Wittgensteins comments on sensory imaginings (see his
(1984b): vol. II, sec. 63 and 627; see also Budd (1989): 104ff.).
At first sight, it might seem that Sartre and Collingwood also present versions of
the Agency Account. Sartre characterises sensory imaginings as spontaneous and

Introduction to Part Four

383

alone renders it worthwhile to consider the Agency Account of imagining,


and to try to develop a more detailed version of it in order to assess its general merits as a unified account of imagining. While the following chapter
is concerned with the concrete formulation of the Agency Account to be
defended and an independent argument in favour of its endorsement, the final chapter illustrates how it meets the two desiderata for unified theories
of imagining, and why its denial of the possibility of non-representational
and passive imaginings is unproblematic.

creative (Sartre (1940): 14) and always involving, or being accompanied by, an
intention (Sartre (1940): 11; 19; 32), which distinguishes them from passive
perceptions (Sartre (1940): 14; 33). And Collingwood claims that imaginings are
active (Collingwood (1938): 195ff., and chs. 9f. in general). But, as noted in
Section 8.2, they both allow for involuntary imaginings (Sartre (1940): 19;
Collingwood (1938): 179; 195); and, as already discussed in the Sections 4.1 and
5.6, Sartre takes many other intentional states to be spontaneous, among them
sensory memories and pictorial experiences (Sartre (1940): 10). Therefore, although
many ideas to be found in Sartres and Collingwoods writings notably the idea
that imaginings are active are congenial to the Agency Account, they do not
really constitute endorsements of it. In particular, Collingwood seems to favour the
weaker Subjection to the Will Account (see Section 13.5).
Finally, OShaughnessy (2004: ch. 11f.) would be a proponent of the Agency
Account, if he would not wrongly assume that hallucinatory, illusory, pictorial and
similar experiences are imaginative, rather than perceptual (see Section 14.4).

CHAPTER 13
Imaginative Agency

The central idea of the Agency Account is to identify imaginings with a


special kind of mental actions. Not all actions are mental; and not all mental actions are imaginative. For instance, we may deliberately make up our
mind and form a judgement or intention with respect to a certain theoretical or practical question; or we may voluntarily try to recall the appearance or name of a particular person. Both cases are instances of non-imaginative, cognitive mental agency. Moreover, I have argued elsewhere that
improving ones mood by recalling a pleasant event or by daydreaming
about something enjoyable, say, is an action albeit a less direct one than
the recollection or daydream involved.3 Again, the resulting action as a
whole is not imaginative, even though some parts of it (e.g., the daydreaming) may be instances of imagining. One main task is therefore to provide a
satisfactory delineation of the class of imaginative mental actions; the other
to defend the claim that there are no instances of imagining outside of this
3

See Dorsch (2009b) for a discussion of all these examples of non-imaginative mental actions. The main point raised there is that the connection between the basic action of thinking of something pleasant and its causal consequence of an improved
mood does not significantly differ from the connection, say, between the basic action
of pulling the trigger of a gun and its causal consequence of the death of a person. In
particular, in both cases, we engage in the basic action because we intend to bring
about a certain result (i.e., an improvement in our mood, or the death of the person),
and because we know that we can cause that result to occur by engaging in the basic
action in question. In other words, killing a person is as directly or indirectly linked
to agency as improving ones mood. Hence, there is no reason to deny the latter the
status of an action, if it is assigned to the former. Killing, however, is often treated as
a paradigm instance of agency.

386

The Agency Account

class. This chapter takes on both tasks in turn.

13.1. Mental vs. Bodily Agency


Kinds of agency differ primarily in two main dimensions: in their purpose
(or in the description(s) under which they are intentional); and in whether
they are mental or bodily. The action of calculating a sum in ones head
possesses the same (general) purpose with the action of calculating a sum
with the help of a piece of paper and a pencil. But only the first action
counts as mental, given that the second involves some actively controlled
bodily movements. The action of forcing a conclusion when deliberating
about what to believe or do, on the other hand, differs from both instances
of calculating a sum in its goal. By treating imaginings as mental actions of
a certain kind, the Agency Account has therefore to outline the difference
between mental and bodily actions and identify the specific purpose distinctive of imaginative mental activity.
Actions count as mental, first of all, because they aim at altering some
facts about ones own mind. The cognitive mental actions listed above, for
instance, have the end to elicit the occurrence of a judgement, an intention,
a guess or a conscious manifestation of an episodic memory. Similarly,
imaginative activity is concerned with the production of mental representations, such as the image of a unicorn or the thought of the Earth as flat. In
contrast, most non-mental actions aspire to bring about facts that are either
non-mental or concern the mind of other persons. Walking is aimed at the
relocation of ones body or simply at the movement of ones legs; while insulting someone is focused on hurting him or making him angry. But, as
the example of calculating a sum shows, the mentality of the facts or
changes to be brought about does not suffice to render an action mental. It
seems to be a further requirement on mental action that they occur from
their conception until their completion exclusively in the mind. Calculating the sum entirely in ones head counts as a mental action because it does
not, so to speak, leave the mind; while the additional use of paper and
pencil would render the action non-mental since it would introduce bodily
agency.
The outlined border between mental and non-mental actions is perhaps

Imaginative Agency

387

not absolutely precise. While mental actions at least typically consist solely
of mental episodes (and perhaps also dispositions) and the mental processes linking these together, it might be argued that some mental actions
involve also certain forms of relatively non-interfering bodily actions. For
instance, it does not seem to matter much for the action of finding the best
move in a game of chess whether one scans the position on a chess board
with ones eyes or merely visualises it in ones mind. Arguably, it might
also not matter much whether one actually tries out the moves on the board
(say, during an analysis) or visualises them in ones mind. Only once one
actually makes a committal move in the game is ones action interfering
with the world in a relevant or significant manner with respect to the situation in question that is, in this case, changing the actual course of the
ongoing game of chess in question.
However, I assume in what follows that, in most cases, the distinction
between mental and non-mental actions is precise. I also take it for granted
that imaginative actions as instances of imagining are realised entirely
in the mind. Their pursuit may require the performance of certain bodily
actions as enabling conditions (e.g., we may have to find a comfortable
position in a quiet corner and to close our eyes in order to be able to visualise something). And imaginative actions may figure in wider projects that
are neither purely mental nor purely imaginative (e.g., instances of aesthetic appreciation, or of prop-involving games of make-believe; see
Walton (1990): especially chs. 1 and 6). But imaginative actions do not by
themselves involve any events which occur outside the mind of the imagining person. In what follows, I therefore focus on mental actions consisting
solely of conscious episodes and the mental processes linking them together, and ignore the possibility of other types of mental agency.

13.2. The Nature of Imaginative Agency


By being identified with mental actions, instances of imagining are thus
said to aim at the production of mental phenomena. But, as already noted,
this is true of most if not all other kinds of mental agency. What distinguishes imaginative from other mental actions are three particular aspects
of their purposes. First, they are meant to lead to the occurrence of epis-

388

The Agency Account

odic representations. Second, the content of these representations is to be


determined in a direct or, as I say in Dorsch (2009b), straightforward
manner. That is, the content is to be determined by our agency without any
reliance on epistemic or merely causal processes as means. Third, this determination has to be specific with respect to what is to be represented.
This means that it has to concern not only the general subject matter (e.g.,
the colour of a friends car, or the weather), but also some specific states of
affairs, that is, the instantiation of specific properties by the represented
objects (e.g., the redness of the car, or that it is raining).
The first aspect differentiates imaginative activity from mental agency
that is concerned with bringing about some non-representational mental
phenomenon (such as, perhaps, a change in mood), as well as from mental
agency that is concerned with bringing about non-episodic mental representations (such as, perhaps, the formation of a future-oriented intention).
The other two aspects are needed to discriminate between imaginative
mental agency and cognitive or similarly representational and non-imaginative mental agency. To see this more clearly, it is necessary to say a
bit more about the nature of, and the need for, the directness and the specificity conditions; and I start with the latter.
As mentioned above, we can voluntarily force a conclusion to a certain
theoretical or practical deliberation, or actively attempt to remember an
event or fact. Whether we start to consider the question of what to believe
about the regulation of financial markets, or the question of what to do tonight, is entirely up to us (even though, of course, external factors or incentives may prompt us to do so). Similarly, we may decide, after a while,
to stop our deliberation and come to a conclusion for instance, because
we need to apply our limited resources to other issues or problems. As a
result, we may deliberately trigger the occurrence of the judgement (or at
least informed guess) that the current banking regulations are too weak, or
of the decision to see a particular movie. However, which proposition or
course of action we end up endorsing is determined by the theoretical and
practical reasons gathered during the process of deliberation or already previously available to us.4 We enjoy the same limited control over our
4

See, for instance, Pink (1996) and Owens (2000) for detailed discussions of, respectively, the involvement of theoretical considerations in the formation of intentions,
and the involvement of practical considerations in the formation of beliefs.

Imaginative Agency

389

memories. We may actively trigger the conscious occurrence of an episodic


or semantic memory by, say, deliberately considering the respective issue
(e.g., how our friend looked like yesterday, or what the capital of Ecuador
is). But it is beyond our influence how specifically we remember the entities concerned as being (e.g., whether we recall our friend as being happy or
sad, or whether we remember the capital of Ecuador to be Quito or Lima).
What these examples illustrate is that we may have voluntary control over
the occurrence and subject matter of cognitive episodes, but not over their
specific content. By contrast, imaginative activity involves the voluntary
determination of which entities are represented as instantiating which
properties. This difference is captured by the third aspect of the purpose of
imagining.
The directness condition, on the other hand, is needed to distinguish
imaginings from cases in which we deliberately form cognitive episodes
with specific contents by exploiting the workings of certain epistemic or
merely causal mechanisms. We may, for instance, come to know that taking a certain pill or going to the hypnotist has the consequence of acquiring
the (occurrent) belief that it is raining, or an (occurrent) episodic memory
of the blueness of our friends car. Furthermore, we may exploit this knowledge when acting on the desire or intention to judge that it is raining, or to
recall the car as being blue: namely by relying on the causal effects of the
pill or the hypnosis as means. Similarly, we may learn that listening exclusively to experts or commentators with a strong leaning towards one side of
the debate and avoiding any confrontation with dissenting opinions is a
good way of forming a certain desired conclusion (e.g., that the financial
markets need not be regulated at all). In such cases of evidence manipulation or avoidance, we rely on the fact that what we judge to be the case is
rationally determined by the evidence available to us. In a more benign
case, we may simply know that we are disposed to visually recall the colour of our friends car as being red and use this particular piece of knowledge when desiring to produce a visual representation of some red car
(e.g., when aiming to paint a red car). None of these activities is imaginative. But they all involve the active and specific determination of how certain objects are represented as being. Hence, the specificity condition is not
sufficient to individuate imaginative activity. In addition, what is required

390

The Agency Account

is the directness condition: the voluntary determination of what is represented has to happen without any reliance on epistemic or merely causal processes as means. The main thesis of the Agency Account of imagining
should accordingly be expressed by means of the following thesis about
the fundamental nature of imaginings:
(ACT)

Imaginings are essentially mental actions with the ultimate


intrinsic purpose of forming one or more representations
with specific and directly determined contents.

Imagining may also serve purposes extrinsic to it. For instance, we may
imagine something in order to improve our mood, or in order to solve a
problem. But what identifies instances of imaginings like this as imaginings are not those contingent and external ends, but instead the necessary
and internal goal of directly producing one or more specific representations. Similarly, this intrinsic imaginative purpose end may give rise to further instrumental (and intrinsic) ends. Imagining something may require,
say, shifting ones attention away from what one is perceiving (e.g., by
closing ones eyes) and concentrating on the imaginative task in question;
or recalling the appearance of something; or imagining separately each of
the parts of something. But, again, what marks the resulting formation of
representations as imaginative are not these contingent or necessary means,
but instead its ultimate intrinsic goal of forming them in a direct way and
with the specific determination of which states of affairs are represented.
This is why (ACT) is concerned with the specification of the ultimate intrinsic purpose of imagining.
The thesis (ACT) is meant to apply primarily to the central cases of imagining described in Section 1.2. It may also be understood as capturing a
core notion of what it means to be an imaginative mental phenomenon. Accordingly, to imagine something may simply mean to directly form one or
more specific representations of it. But the main intention of the Agency
Account is to take (ACT) to be a claim about the nature of the paradigm instances of imagining, and not about our concept of imagining. Although
(ACT) claims that all imaginings share the same essence and, in this sense,
form a particular kind of mental phenomenon, it does not assume that they

Imaginative Agency

391

form a fundamental natural kind. Instead, they are said to be instances of a


more basic mental kind, namely that of (mental) actions. In addition, the
presented Agency Account of imaginings has to be further qualified, partly
to forestall certain potential counterexamples to (ACT).

13.3. Qualifications
(i) Concreteness
One set of qualifications concern the nature of the concrete goals of imaginative agency. First, the purposes of imaginative mental actions or the
relevant underlying motivational states need not determine the type or
mode of the representations to be formed. We may decide to imagine a
square box, say, without any particular intention to do so by means of representations of a particular type (e.g., tactile rather than visual representations, or intellectual rather than sensory representations). As a result, we
are free in our pursuit of the respective project to use appropriate
representations of any modal type.
Second, the purpose pertaining to an instance of imagining may be
more, or less, general in its description of the representations to be formed.
For instance, it is possible to imagine a garden party on the basis of intending to represent one, sensorily imagine one, or visually and auditorily imagine one.
Third, the determination of the states of affairs to be represented may be
more, or less, specific. We may aim to simply visualise a tree, or to visualise a Japanese cherry tree in full blossom, or to visualise such a tree in a
large garden with bridges, lakes and tea pavilions. In the same way, we
may just want to imagine being Caesar; or we may want to imagine being
Caesar during his successful political career in Rome; or we may want to
imagine being Caesar at the moment when he realises that his son is among
his assassins. The degree of specification can thus vary greatly as long as
there is at least some active determination of which properties the represented objects are represented as having.

392

The Agency Account

(ii) Voluntariness
Another set of qualifications pertains to the fact that the active and direct
determination of what is represented need not be complete or entirely voluntary. Four considerations are of relevance here.
First, it may very well be that there are instances of mental and, in particular, imaginative agency that we voluntarily pursue without noticing it.
That is, there might be forms of imagining that are sub-intentional in the
same way in which, say, absent-mindedly scratching ones back is (see
OShaughnessy (1980): vol. II, ch. 10; Kind (2001): 91). This possibility is
perhaps more plausible with respect to intellectual representations than
with respect to sensory ones. For instance, we may continue to think about
a certain problem while being occupied by a completely different task; and
we may actually come up with a solution to the problem in this way. Or we
may discover that we have been worrying all along about what a person
has meant by something she said to us, only realising afterwards the effects
that these worries have had on our feelings and behaviour. Or we may find
ourselves imagining about meeting a person again who we encountered the
other day at a party. Perhaps not all types of mental or imaginative activity
allow for such an engagement without attention. But some might do.
Second, there are certain general limits on what we can imagine. I have
already noted that intellectual imaginings especially if they do not occur
in isolated form, but are embedded in a wider imaginative context may
be subject to moral or logical constraints (see Section 1.2). Thus, we may
have difficulties of adopting the moral point of view adopted by the (implied) narrator of a story (Gendler (2000)). In addition, when we imagine
something about a certain character in a particular situation, while the
nature of both is already determined in rich detail by our previous imaginings and other representations, our imaginative activity may gain its own
momentum: that is, we may more or less automatically imagine the person
to behave in a way which follows logically from what we take her situation
and her personality to be (Wollheim (1973): 69f.; see also Eco (1994) and
Carroll (2000): 124f.). But the restrictions on what we may be able to imaginatively represent are not due solely to our moral attitudes or our rationality. They may also arise from limitations in the scope of our past experiences, our conceptual capacities, our imaginative skills, our ability to con-

Imaginative Agency

393

centrate, and many other factors (see also Sections 9.1 and 9.3 on causal,
and Section 12.1 on semantic dependency).
Third, not all aspects of the contents of the representations formed in accordance with the specific imaginative purpose have to be actively and directly determined by the underlying motivational states. For instance, what
we are sensorily imagining is often partly determined by how our mind
passively and spontaneously fills in the details of the images that we bring
about (see Section 14.4 for more on spontaneous representations). When
we act just on the general aim to visualise a building with no specific shape
or number of windows, we may none the less end up visualising it as having a determinate shape and number of windows possibly due to the influence of association or generic memory (see Sections 4.6 and 7.2). But
this does not undermine the imaginative status of the respective episodes
since at least some of the represented states of affairs (e.g., that the imagined entity is a building) remain determined actively and directly.
Fourth, there may be cases of imaginative episodes in which we actively
imagine something, but ultimately against our will in the sense of akratic
or obsessive agency: that is, against what we really want to do or what we
take to be the best thing to do. In particular, we may produce images and
thoughts, which we would in fact prefer to banish from our minds, but cannot help to (continue to) produce. For instance, when we suspect that our
partner has an affair with someone else, we may not be able in our jealousy
to stop imagining the two together despite the fact, for instance, that we
would prefer not to picture or think about them because it would make us
feel much better; or despite the fact that we know our suspicion to be
without any serious evidential foundation. Similarly, our perception of the
physical attractiveness of a person may lead to our active engagement in
some sexual fantasy concerning him or her, although we would rather like
not to follow this course of action (e.g., because we take it to be morally
inappropriate, because we very much dislike the personality of that person,
or simply because it is not the right moment or leads to an awkward situation). In both examples, we would like to get rid of the respective images
and thoughts. Although these imaginative episodes are still actively produced by us, we may not succeed or only with great effort in our attempt to make them disappear again.

394

The Agency Account

It is important to note that such weak-willed or compulsive instances of


imagining still count as active and, in particular, as voluntary. It is true that,
in some sense, it is adequate to describe our imaginative activity involved
as involuntary: we are compelled to engage in it, and the resulting representations are forced upon us. But, in another and more fundamental
sense, the activity is also clearly voluntary and motivated by some of our
motivational states.5 Indeed, what seems to occur in such cases is some
kind of conflict among our various (and not always conscious) motivational states, or between our motivational states and our evaluative judgements (if the latter do not possess motivational power themselves). Independently of whether these cases should be described as instances of
akrasia or obsession, they have to be strictly distinguished from spontaneous representations which occur passively and, especially, without any
imaginative agency (see Section 14.4 for more on spontaneous images and
thoughts and their relation to imagining).

(iii) Directness
The final set of qualifications is concerned with the status of imaginings as
direct actions. This time, two considerations are of particular importance.
First, the fact that imaginative activity is direct with respect to the specific determination of which states of affairs are represented is completely
independent of the issue of whether imaginative activity is direct or indirect in other respects, or whether it differs in this from other kinds of mental
5

It seems that Wittgenstein has cases like these in mind when he writes that imaginings often force themselves upon us and stay against our will, do not let themselves
be scared away. But the will is able to fight against them (Wittgenstein ((1984b):
vol. II, sec. 86). (The translation is mine. The original reads: da Vorstellungen oft
gegen unsern Willen sich uns aufdrngen und bleiben, sich nicht verscheuchen
lassen. Doch aber kann der Wille gegen sie ankmpfen.). Budd endorses the same
interpretation in his book on Wittgenstein (see Budd (1989): 105 and 109; see also
Section 4.7). Armstrong (1969): 298, and Scruton (1974): 94f., make very similar
statements about the relation between imagining and the will. Compare finally the
observation that obsessive or intrusive images and thoughts are actively produced by
us against our will though, in the latter case, not acknowledged by us as such (see
Sartre (1940): 148ff., and Roessler (2001)).

Imaginative Agency

395

agency. In particular, the Agency Account stays neutral on the issue of


whether the determination of the occurrence or subject matter of imaginative representations may be indirect, that is, may involve the reliance on
epistemic or merely causal processes as means. Similarly, the view can allow that the occurrence or subject matter of non-imaginative representations for instance, episodes of recollection may be directly determined
by our will (e.g., when we deliberately try to recall the appearance of a certain person). The Agency Account just claims that what is distinctive of
imaginative agency is its directness with respect to the specific determination of what is represented.
Second, the direct active determination of contents distinctive of imaginative agency still permits that acts of imagining may be non-basic in the
sense that they are pursued by pursuing another and hence more basic
action.6 In short, while imagining need to be direct in its determination of
what is represented, it need not be basic. The only condition is that the
more basic actions, by means of which we imagine something, are themselves direct, that is, do not make use of epistemic or merely causal mechanisms in such a way that the overall project becomes indirect in its determination of the contents of the desired representations. One example of
a non-basic instance of imagining is [visualising] Wellingtons face by
visualising Goyas picture of him (see Currie & Ravenscroft (2002): 32,
referring to a suggestion made by Jerrold Levinson). Another case is perhaps that of acting on ones intention to visualise the appearance of a complex object by visualising its parts one after the other. Visualising all sides
of a house may be such a case; or visualising a face close up and in great
detail.
Third, subjects engaging in imaginative activity need not possess the
concepts of specificity and of directness used to formulate (ACT). This
claim identifies the distinctive purpose of imagining with the goal to produce one or more representations with a specific content in such a way that
the determination of which states of affairs are represented is specific and
does not rely on any knowledge of epistemic or merely causal mechanisms.
6

I follow here the traditional understanding of actions as non-basic just in case they
are done by performing another action (see: Danto (1963); OShaughnessy (1980):
vol.1, xiii; Searle (1983): 98ff.; McCann (1998): 4; Audi (1993): 171; Lowe (2000):
234).

396

The Agency Account

As a consequence, the engagement in imaginative activity presupposes


some grasp of direct and specific ways of forming and determining representations. But it does not require the ability to conceive of them as direct
and specific ways. It suffices that the subject in question has some conception of those ways of acting, can single them out and knows how to realise
them in his or her own agency. It is with all these qualifications in mind
that (ACT) should be taken to provide the most fundamental characterisation of the shared and distinctive nature of imaginings.

13.4. The Essentiality of Imaginative Agency


This concludes the exposition of the Agency Account of imagining that I
want to put forward. The remainder of this final part of the book is concerned with its defense. Before I address, in the chapter to come, the explanatory power and extensional adequacy of (ACT), I present a general
argument for the truth of this thesis.
Indeed, several reasons can be put forward in support of the Agency Account of imagining. The first is simply the fact that as the preceding parts
of the book have shown all alternative attempts at providing a unified account have failed. A second reason which is to be substantiated in the
next chapter is that the Agency Account meets the two desiderata for unified theories of imagining and, especially, is able to explain the main features common to imaginings (such as their lack of a cognitive attitude and
function, or their immunity to error through misidentification with respect
to what is imagined). A closely related third reason is that the Agency Account promises an elucidation of the nature and imaginativeness of the two
examples central to my case against the Epistemological Account. The relevant episodes of visualising are formed as part of more complex imaginative projects; and the thought is that they have the power to cognise reality
and, therefore, constitute counterexamples to OShaughnessys and similar
positions precisely because of their origin in imaginative activity. A final
reason already discussed in Section 3.6 is that the postulation of an intention-in-action provides the best explanation of the quasi-observationality of sensory imaginings.
My focus in this section is, however, on a fourth reason for endorsing

Imaginative Agency

397

(ACT) a reason which may be expressed in a simple argumentative form.


The argument in question involves two main premisses. The first is simply
the widely accepted observation that at least some imaginings are actively
and, indeed, voluntarily formed with the direct determination of their
occurrence and content. The second premiss claims that actively and directly formed episodes are essentially active, that is, differ in nature and
kind from passively formed episodes. It follows from the two premisses
that, if there were also passively formed imaginings, they would differ in
nature and kind from actively formed imaginings.7
This means that there are no passive imaginings, despite the fact that we
might be tempted to use the same notion (i.e., imagining) to refer to both
actively formed imaginings and some presumably very similar passively formed representations.8 In other words, philosophers, who speak of
the existence of passive imaginings, in fact misleadingly use one and the
same word to pick out two completely different kinds of mental phenomena: active imaginings and passive imaginings. 9 The conclusion of the
presented argument should therefore be that (ACT) is true. Assuming that
the truth of the first premiss should be rather obvious and uncontroversial,
my main concern in what follows is with elucidating what it means for
imaginings to be essentially active, and with providing argumentative support for the second premiss.
The active and direct formation of an imaginative episode consists of
two elements: the act of forming the episode and the actively formed epis7

The same conclusion follows also with respect to indirectly determined


imaginings, in contrast to directly determined imaginings. Deliberately inducing a
visual (and non-perceptual) representation of tree by means of hypnosis or drug-intake is therefore no instance of imagining, either.
I return to this similarity during the discussion of spontaneous images and thoughts
and their comparison with imaginings in Section 14.4.
There are no good reasons to reserve the word imaginings for the involuntary
rather than the voluntary mental representations. For the relevant disagreement is
exclusively concentrated on the issue of whether there are, or can be, any passive instances of imagining, while it is completely uncontroversial that there are, or can be,
imaginings formed at will. In this respect, imaginings differs from jade: although
the latter has also been used to refer to two different kind of phenomena, there is no
obvious priority of one reference over the other.

398

The Agency Account

ode itself.10 The relation between these elements may be understood in two
different ways. The resulting views the Causal and the Constitutive View
differ in whether they treat the two relata as distinct mental phenomena,
in the sense that they could occur independently of each other. They are
therefore concerned with the internal structure of the active and direct
formation of imaginative episodes. More specifically, the two views disagree about whether the active origin of the relevant imaginative episodes
is essential to them, given that that actively formed representations are essentially active in so far as they are inseparable from and form an episodic unit with the event of their active formation.
The Causal View assumes that the occurrence of the imaginative episode is merely a causal result or product of the imaginative activity. Accordingly, the two elements are treated as distinct entities: the activity could fail
to bring about any imaginative representation; and the imaginative episode
could occur in a different, purely passive manner. The two elements merely
happen to be part of a more complex entity, namely the active formation of
an imaginative episode. Given that the very same imaginative episode
could have occurred without the involvement of agency, passive imaginings are possible. It is in this sense that the active origin of imaginative
episodes is assumed to be inessential to them.
The Constitutive View, by contrast, maintains that the two elements are
constitutively dependent on each other. On the one hand, actively formed
imaginative episodes could not have occurred without imaginative activity:
the active origin of imaginative episodes is essential to them. On the other
hand, imaginative activity could not occur without actively formed imaginative episodes: the occurrence of an imaginative representation is essential
to imagining. It is therefore no accident that the two elements form a
whole. Moreover, active imaginings differ essentially from passively occurring images and thoughts. That is, the two belong to different mental
kinds. Assuming that the actively formed episodic representations are central instances of imagining, it follows that passive imaginings are impossible.
10

In what follows, I assume but do not always explicitly mention that the active
formation of the imaginative representations under consideration is direct: that is, involves an active determination of content, without any reliance on epistemic or
merely causal mechanisms.

Imaginative Agency

399

Of course, which of the two mental phenomena should be labelled imaginings is primarily a terminological issue. Perhaps there are good reasons
to use that term ambiguously and refer to both active and passive cases as
imaginings. But this does not undermine the claim that there is an essential difference between actively and passively produced mental episodes;
and that, hence, a unified account of imagining can hope to accommodate
at best one of them, given that it aims to identify the essence common to all
instances of imagining. No one denies that there are voluntary instances of
imagining; while the possibility of involuntary imaginings is controversial.
Hence, the second view should be read as entailing that there are no passive imaginings.
So far, the characterisation of the two views has been neutral on the ontological categories of the two elements concerned. But it is natural to assume that both the episodic representations and their active formation are
event-like aspects of the stream of consciousness. For instance, actively
visualising a tree seems to consist of an event of trying or willing to visualise a tree (what I have called in Section 3.5 an intention-in-action), and of
an event of visually experiencing a tree. The two views therefore differ in
whether they treat the two elements as two distinct events, or rather as two
dependent parts of a single event. One particular issue is whether the resulting imaginative episodes constitute an inseparable part of the mental actions concerned, or whether they are merely causal consequences of those
actions. The other main issue is whether imaginative activity can fail to
result in imaginative representations.11
The view that I would like to defend is, of course, the Constitutive View,
given that it includes the second premiss of the argument outlined above,
according to which imaginative activity is essential to actively formed imaginative episodes. But the Constitutive View assumes more than this con11

The disagreement between the two views therefore parallels the debate about whether bodily movements are part of, or merely the causal product of, bodily actions
and whether, therefore, bodily actions can be reduced to tryings or similar mental
events, or not (see, for instance, the respective discussions in OShaughnessy
(1980), Hornsby (1980) and Ginet (1990). It also parallels the debate about whether
disjunctivism about actions is true: that is, whether there is an essential difference
between acting, on the one hand, and merely trying but failing to act, on the other
(see, for instance, Dancy (2008) and Hornsby (2008), as well as Dorsch (2011a)).

400

The Agency Account

stitutive dependency of actively produced imaginings on the underlying


agency. It also accepts a constitutive dependency in the other direction: actively formed imaginative episodes are taken to be essential to imaginative
activity.
Strictly speaking, the defense of the second premiss does not require the
defense of this additional claim. But there is a good reason to take the constitutive dependency between the two elements to be symmetric. The central thought is that the simplest instances of active imagining constitute selfcontained episodes that can occur on their own, that is, independently of
other mental phenomena. But this would not be the case if actively formed
imaginative representations would depend on independent episodes of trying to imagine something. The whole formed by the two elements would
count as complex, since it would include a smaller episode (i.e., the trying)
which could occur on its own; while the imaginative representation, taken
on its own, of course fails to be self-contained due to its dependency on the
activity of trying to imagine something. Hence, active imaginings would
turn out to be complex, rather than episodic. To avoid this conclusion, it is
necessary to maintain that both elements are dependent parts of a simple
and indivisible mental episode. This is why the Constitutive View assumes
the constitutive dependency to be symmetric.12 Now, there are three considerations that speak against the Causal and for the Constitutive View.
First, there is the analogy with simple bodily movements. For instance,
there are no passive instances of walking or jumping: the respective bodily
movements are essentially active. Of course, some external factors such
12

Another motivation for this assumption is that it is unclear what kind of event-like
entity could depend on an event of trying (or willing) without itself being part of that
event. The picture just developed is, however, in tension with the idea that actions
should be identified simply with tryings or willings, and that none of the immediate or distant results of such tryings should count as part of the actions, let alone
necessary parts (see Hornsby (1980); Pietroski (2000)). But not only is this very restrictive view of what actions are problematic in its own right (see, for instance,
Audi ((1993): ch. 3) and Noordhof (2001)). Its proponents also permit that action
descriptions can apply to complex events consisting in tryings and their causal results if there is a suitable or non-deviant causal link between the two (Hornsby
(1980): 122f.). Hence, I continue to characterise the respective events of thinking,
imagining, moving ones body, and so on, as actions with two dependent parts,
namely a trying and a representation or bodily movement.

Imaginative Agency

401

as another person or a machine may change the position of our limbs in


the same way as acts of walking or jumping would do. But a change in the
bodys position does not suffice for the occurrence of a bodily movement.
That the body of a base jumper moves in the sense of falling towards the
ground does not mean that an event of bodily movement occurs. What is
missing is agency on behalf of the subject. That there are some difficult
borderline cases such as the beating of the heart, or the reflex of breathing does not necessarily undermine this particular view on bodily movements. Perhaps, such cases do not involve bodily movements; or, alternatively, they involve some sub-intentional form of agency.
Second, the Constitutive View but not the Causal View captures the
fact that there can be no thinking without a thought, and no visualising
without a visual image. It is, of course, possible to try and fail to think a
thought or form an image. But, in such cases, we do not count as having
thought a thought or formed an image: that is, we cannot be said to have
been engaged in thinking or visualising. 13 At best, we count as having performed another kind of action, namely that of merely willing or trying to
form a mental representation of a certain kind. Thinking and visualising
are, in this sense, success activities: we do not end up performing them if
we completely fail to produce any intellectual or visual representation.
Hence, what we may describe as the immediate results of direct mental actions appear to be in fact constitutive parts of these actions. 14 When we
form a thought or visualise something, the resulting and directly determined thought or image is partly constitutive of our mental action. Without
This is compatible with the fact that intending and trying to do think something involves thinking in a wider sense, namely thoughts about what one is about to do.
The kind of thinking that we are concerned with here, and which is missing in failed
attempts to think something, consists of thoughts that are the result of such attempts.
14
See McCann (1998): 76ff., and also Audi (1993) who specifies the result of an active
bodily movement as the movement intrinsic to the action in the way a hands rising
is intrinsic to ones raising it (ibid.: 77). These considerations gain further support
from the fact that the same considerations apply to simple bodily actions (see the
discussion above, including Note 9). When we move our arm, the resulting movement of our arm is part of the action, and not merely caused by it. Without the occurrence of the arm movement, this action would not have been performed, but at best
an action of willing or trying to move our arm just as nothing is seen in the case of
perceptual hallucination (see Dorsch (2010a)).
13

402

The Agency Account

the occurrence of such a thought or image, the action would not have been
performed. This still allows for partially successful direct mental actions
for example, when we manage to only partly think or visualise what we
have intended to, or maybe also when we manage to think or visualise only
something completely unintended. Solely in cases of complete failure does
this view assume that no action of the desired kind has been performed.
Third, only the Constitutive View can account for the specific phenomenal character of actively formed representations.15 From the inside, we
usually experience actively formed images and thoughts differently from
passively formed ones. A natural explanation of this fact is that our experience of actively formed representations involves an awareness of agency.
That is, it is part of the phenomenal character of these representations that
they are experienced as actively formed. This normally enables us to tell
whether we have brought about a certain visual image by means of visualising something, or whether it has occurred spontaneously and unwilled.
Similarly, we can usually distinguish thoughts which simply cross our
minds from those which we actively produce.
Now, we can become aware of the active formation of a mental representation in an immediate and non-inferential way. It is not that we infer the
presence of a causal link between the action and its result in response to
two independent instances of introspective awareness, one concerned with
the action of forming a representation, and the other with an occurrence of
a representation of that type. Instead, we are typically able to tell whether a
mental episode has been actively formed or not simply by introspecting its
phenomenal character. In other words, the phenomenal character of mental
episodes reflects their active or passive origin.
The Causal View cannot account for this fact about our self-knowledge
of direct agency. If representations were the result of agency solely due to
their contingent and subjectively inaccessible causal origin in some preceding activity, their phenomenal character would not involve a marker of
their active character. This is due to the fact that our awareness of agency
does not extend to mere causal results of actions. For instance, if we deliberately attempt to improve our mood by humming a tune in our mind, we
15

See Dorsch (2009b) for a more detailed discussion of the following phenomenological description of mental agency and the presented argument in favour of this description.

Imaginative Agency

403

experience only our imaginative humming, but not our resulting change in
mood, as an instance of agency. It thereby does not matter whether the
causal results concerned are of such a kind that, under different circumstances, we could experience them as active. When we use our right hand
to move one of the fingers of our left hand, say, we are not aware of the
movement of the latter as actively formed; while we are so aware if we
move the finger in the normal direct way.
In addition, the phenomenal character of mental episodes reflects only
aspects of their essences, but not also some of their contingent features
such as their causal origin, according to the first view. This is due to the
fact that the phenomenal character is itself part of the nature of the episodes concerned and, hence, cannot be determined by their non-essential aspects (see Dorsch (2012a) and Soldati & Dorsch (2011)). That we experience imaginative episodes as actively formed indicates that their phenomenologically salient activity is therefore essential to them. Hence, the
fact that we experience the immediate results of mental actions as actively
produced can be satisfactorily accounted for only by reference to the fact
that these results (i.e., the formed representations) are part of the respective
actions, and not merely their immediate causal results. The Constitutive
View on the relation between the active and the representational element in
the active and direct formation of imaginative representations should therefore be preferred over the Causal View: the two elements are dependent
constituents of the mental actions in question.
The preceding considerations have been concerned with establishing the
claim that the two elements of the active formation of imaginative episodes
the imaginative representations themselves and the acts of forming them
are constitutively dependent on each other. As a consequence, actively
formed imaginative episodes are essentially instances of agency: they
could not have had a passive origin. On the assumption that at least some
imaginings are actively and directly formed, it follows that there are no
passive instances of imagining. Furthermore, given that their active character cannot be traced back to some more fundamental aspect of their nature
(see Section 14.1), imaginings are fundamentally instances of agency.
Hence, the Agency Account of imaginings should be endorsed.

404

The Agency Account

13.5. The Subjection to the Will Account


The presented line of argumentation also has the consequence that the idea
that imaginings are mental actions of a certain kind is to be distinguished
from from the weaker idea that they are subject to our will (or, if one
prefers, to our agency or active control). For subjection to the will requires
merely the possibility of voluntary control, and not also its actuality. What
we attend to, or which position our limbs occupy, allows for the influence
of direct agency. That is, we can directly control at least within certain
limits which perceived features we pay attention to, or whether our legs
are stretched or bent. But we do not always actually exert this control.
Sometimes, we just give in or have to give in to external factors, which
then determine what we notice, or how our limbs are arranged. Some
events may grab our attention, or someone else may move our legs. Direct
mental actions, in contrast, cannot occur as the result of such unwilled external influence: they are always actively performed.
Moreover, being subject to the will does not require being open to direct
agency: it suffices if the respective kind of agency is indirect. For instance,
what we actually perceive in the first place, or which position the chairs in
a room occupy, is also subject to our will, even though it allows merely for
the influence of indirect agency. We can change what we see by turning our
head or closing our eyes, or move the chairs by pushing them around by
means of movements of our own bodies. Both courses of action are indirect in so far as they rely on knowledge of how the eye functions, or how
middle-sized objects behave within the gravitational field. If we did not
know that closing or turning away ones eyes leads to a change in what one
sees, for instance, we would not close or turn away our eyes when acting
on the intention to see something different. Again, imaginative activity is
essentially different in that it always involves direct mental agency.
That being a direct mental action differs from being subject to the will
even if the latter feature is restricted to the possibility of direct control
has the consequence that the Agency Account of imagining differs from the
Subjection to the Will Account. Central to the Subjection to the Will Account is the claim that it is distinctive of imaginings that they are subject to
the will in the sense of allowing for the direct voluntary determination of
their content. This leaves room for imaginative episodes that are brought

Imaginative Agency

405

about in a passive manner (e.g., due to association or some other causal


factors), for imaginative episodes that remain in passive existence and are
never influenced by mental agency, and for imaginative episodes that actually resist any voluntary control (e.g., compulsive or otherwise inescapable
images or thoughts) as long as there is in principle the possibility that
episodes of the kind concerned can be subjected to imaginative activity.
When I am asked not to think of or visualise my partner, say, I may be unable to satisfy this request for emotional or other reasons. But, according to
the Subjection to the Will Account, the resulting thoughts or images still
count as imaginative if they could also have occurred in voluntary and direct response to the desire or intention to think of or visualise my partner.
Now, the argument from the last section shows not only that the Subjection to the Will Account endorses a weaker claim about imaginings than
the Agency Account, but also that the Subjection to the Will Account is
false. Given that (almost) all mental episodes and states are subject to indirect control, the view should and, indeed, does claim that what is distinctive of imaginings is that they are, essentially and fundamentally, subject to direct control and, moreover, subject to direct control with respect to
which specific states of affairs are imagined.16 Accordingly, passive imaginings are said to differ from certain other unbidden representations in
that it is still possible to actively determine their content in a direct and
specific way. But, as argued above, the active and direct determination of
their content is essential to the mental episodes concerned. Hence, an episode, that occurs passively under one set of circumstances, could not be actively and directly controlled under different circumstances. The corresponding switch from a passive to an active and direct episode (or the other
way round) would have to be accompanied by a change in constitution and,
hence, lead to a completely new episode. It therefore does not make sense
to maintain that one and the same kind of episode, which is actually passive, might have been active and direct in a different context.
In addition, the Subjection to the Will Account is also to be rejected because it cannot provide a satisfactory specification of the kind of possibility concerned (i.e., of the possible worlds in which active and direct determination has to be present for there to be subjection to the will in the ac16

This view has been put forward by Casey (1976): 34f. and 63ff.; and perhaps also
Walton (1990): 13ff. and Collingwood (1938): chs. 9 and 10.

406

The Agency Account

tual world). People differ in their abilities, and what is possible for some
may be impossible for others. Abilities are also open to evolutionary developments and to individual training, so that what is impossible for all or individual human beings today might be possible for them in the future. It
therefore needs to be asked which kind of possibility of voluntary and direct control should be taken to be distinctive of imagining. For instance,
does what is possible for other people matter for whether one of my mental
episodes is an instance of imagining? And should we consider what is possible for the relevant people relative to how they actually are, or instead
relative to how they would be after five years of intensive training or after
the next step in the evolution of the human species? Unless the Subjection
to the Will Account does not answer these and similar questions and specifies the kind of possibility concerned, its characterisation of what it
means to imagine something remains unacceptably incomplete and inapplicable as a unified account.17 But it is doubtful that these issues can be
satisfactorily resolved. For the Subjection to the Will Account appears to
mix up the question of what the nature of imagining is with the question of
when it is possible whether for human beings or for other creatures to
engage in imagining. The act of forming an image of a chiliagon (i.e., a
regular polygon with a thousand sides) in a voluntary and direct way, for
example, seems to be a clear instance of imagining, independently of
whether any one will or would ever be able to deliberately perform or influence it.
Finally, even if the kind of possibility of voluntary control invoked by
the Subjection to the Will Account could be further specified in such a way
that it might become relevant for the question of whether mental episodes
count as imaginative or not, it would presumably require the potential answers to that question to be empirical. Whether it is possible for us to visualise a chiliagon for instance, after suitable and sufficient training, or
after changes to our environment, improvements in our diet, increases in
our brain size, and so on is by and large an empirical (i.e., experimental)
question. By contrast, we can normally settle the issue of whether one of
our own mental episodes is imaginative from our first-personal perspect17

Note that the problem here is not simply one of vagueness. There might be no clear
answer to the issue of how much training or evolution may make a difference. But
whether training or evolution matter in the first place is not a question of degree.

Imaginative Agency

407

ive: that is, by means of introspection of and reflection on the episode in


question and its conscious mental context. Accordingly, even if the Subjection to the Will Account would manage to provide a complete and at least
initially plausible unified account of imagining, it would turn out to be inadequate because of its incapacity to account for our (probably canonical)
non-empirical access to the presence or absence of an instance of imagining.18 Indeed, the view would be unable to accommodate the phenomenologically salient attitudinal difference between cognitions and imaginings.
The Subjection to the Will Account could perhaps avoid this unwelcome
consequence by claiming that what matters for which (of my) mental episodes count as imaginative is solely what I am capable of directly controlling right now, without any further training or other significant change
to my current mental setup a feature which I may be able to access from
the first-personal perspective simply by introspecting my attempts to directly control episodes of the kind concerned. But this version of Subjection
of the Will Account would (absurdly) imply that different mental episodes
would count as instances of imagining for different people; and it would
therefore have to drop any ambition of providing a unified account of imagining.

18

Of course, there is a sense in which introspection may be empirical as well, since it


may go beyond mere intellectual intuition or reasoning and involve some form of inner perception or phenomenal awareness (see Dorsch (2012a)). What I mean here by
empirical is experimental, as specified in the Introduction to this book (notably in
Note 15).

CHAPTER 14
Meeting the Desiderata

At the beginning of this book, I specified two desiderata for any unified account of imagining (see Section 1.1). Such a theory should be extensionally adequate with respect to all central cases of imaginings, as well as all
paradigm instances of non-imaginative phenomena. And it should provide
an explanatorily powerful account of the imaginativeness common to all
imaginings. In this final chapter, I illustrate how the proposed Agency Account of imagining can satisfy these two requirements on a unified theory
of imagining, thereby discussing them in reverse order. While the first section of the chapter highlights the explanatory power of (ACT), the remaining three sections discuss the scope of its applicability. Two of these three
sections are concerned with the discussion of apparent instances of imagining that appear to be problematic for the Agency Account, and which may
therefore constitute counterexamples to it. The particular focus is on the
possibility of non-representational imaginings and on the issue of whether
certain kinds of passive representation namely spontaneously occurring
images and thoughts should count as imaginative.

14.1. Explanatory Power


Three aspects are relevant for the explanatory power of a unified account
of imagining: (i) whether it provides a substantial and informative description of the nature of imaginings and, notably, of their imaginativeness; (ii)
whether its description is the same for all forms of imagining; and (iii)
whether it describes their nature by reference to the most fundamental

410

The Agency Account

property or set of properties distinctive to imaginings. Since mental actions can consist both in simple mental episodes and in complexes thereof,
(ACT) applies equally well to imaginative episodes, to experiential imaginings and to imaginative projects. I therefore focus on the other two aspects
of the explanatory power, namely informativity and fundamentality.

(i) Informativity
There should be no doubt that the main thesis of the Agency Account that
all imaginings are essentially mental actions of a particular kind, as specified by (ACT) is a substantial and illuminating claim. It says something
very specific about the nature of imaginings and their shared imaginativeness: it describes what it means for something to be an instance of imagining. Moreover, in contrast to, say, negation claims, it is not in obvious need
of supplementary theses about the constitution of imaginings. Indeed, reference to the particular kind of activity distinctive of imaginings and, especially, the presence of a conscious and concurrent motivational state (e.g.,
an intention-in-action, volition or trying) suffices not only to distinguish
imaginative phenomena from non-imaginative ones (see the next section),
but also to explain why imaginings share certain features with cognitions,
but lack the cognitive features that are unnecessary for the provision of
knowledge (e.g., a cognitive attitude, the status as providers of prima facie
reasons, and the function or aim of cognising reality).19
That imaginings can refer to reality, for instance, is partly due to their
19

In what follows, I focus on features shared by all imaginings. Of course, important


features that are limited to particular forms of imagining cannot be explained in
terms of the kind of mental agency common to all instances of imagining and,
hence, require a different explanation. For instance, the typically diminished vivacity
and determinacy of sensory imaginings is due to the fact that the latter are merely
representations (or copies) of vivid and determinate perceptions (see the Chapters
9f., especially Section 10.4). The quasi-observationality of sensory imaginings, on
the other hand, derives from the fact that they represent perceptions with single perspectives onto the objects concerned, as well as from the fact that the nature of imagined entities is determined by our imaginings and intentions (see Section 3.6)
which again may be determined by underlying fictions or games of make-believe
(see Walton (1990).

Meeting the Desiderata

411

semantic dependency on cognitions (see Section 12.1) and partly due to


their being intended to refer so (see Section 3.6). Similarly, imaginings
count as successful when they conform to the underlying desires or intentions to imagine something. These guiding motivational states are also ultimately responsible for the immunity to error through misidentification with
respect to what is sensorily imagined, even if only by determining the imaginative thoughts accompanying the imaginative visual representations.
What we visualise is Kings College, and not Hertford College, precisely
because we mean to visualise the former, and not the latter, and think of the
visualised object accordingly. Hence, since we cannot err about what we
intellectually and imaginatively take the identity of the imagined object to
be (i.e., about what the content of our imaginative thoughts are), we also
cannot err about which object we actually do visualise despite the possible fact that the visualised building may resemble the real Hertford College more than the real Kings College (see Section 3.5).
Furthermore, since what our imaginings represent is largely determined
by our underlying motivational states, our imaginings do not provide us
with prima facie reasons for first-order beliefs and do not possess the related function to cognise reality. For our activity of imagining does not
normally bring it about that the contents of the formed imaginings are reliably (or otherwise epistemically appropriately) linked to whichever aspects
of reality they may be about. Hence, they do not possess any justificatory
power with respect to the endorsement of their content in belief. Instead, as
illustrated in Chapter 4, we have to ensure ourselves that the circumstances
are suitable for the formation of cognising imaginings. But this also means
that they lack the (evolutionary or otherwise grounded) epistemic function
distinctive of cognitive episodes, given that this function is partly constituted by the power to provide prima facie reasons for first-order beliefs.
Imaginings also lack the other aspect of the epistemic function of cognitions: namely their motivational power with respect to first-order beliefs.
This fact can, too, be explained by reference to the essential activity of
imaginings. Because imaginative episodes are direct mental actions, they
or at least their relevant aspects, including their content are experienced
as actively determined (see Section 13.4). In particular, we take their contents to be directly motivated by our underlying desires or intentions. But

412

The Agency Account

that we recognise imaginings and their contents as the result of our own
agency helps to explain why we normally do not take them to reliably represent the world. It is not surprising then that, in most cases, we do not rely
on them in belief (nor, presumably, in action); and, hence, that they normally do not play the same epistemic role as cognitions.20
Finally, the absence of a cognitive attitude can derived from the presence of agency in a closely related way. In the Sections 2.1f., I suggested
that the cognitive attitude of perceptual experiences, judgemental thoughts,
and so on, constitutes one of the aspects of their phenomenal character that
reflect their function to cognise reality. That we experience cognitive episodes as making a claim about how things are is part of what renders it phenomenologically salient that their aim is to constitute knowledge. Hence, it
is not surprising that imaginative episodes, which do not possess this aim,
also lack a cognitive attitude: the functional difference between cognitive
and imaginative episodes is mirrored by their difference in attitude. This is
also part of the explanation of why the experience of agency involved in
imagining and the experience of endorsement involved in judging are incompatible with each other: while the second reflects the presence of the
function to cognise reality, the first reflects its absence (see Dorsch (2009b)
for more on this point).
The Agency Account is therefore in the position to provide a satisfactory
and illuminating explanation of why imaginative episodes share some of
the features characteristic of cognitions (e.g., the ability to refer to reality
and the possession of success conditions) and are without others (e.g., the
epistemic function, the cognitive attitude, the actual role in the acquisition
of knowledge and the lack of an immunity to error through misidentification of what is represented). Imaginings possess or lack these features exactly because of the active and direct determination of their occurrence and
their content.

20

This normative role of phenomenal character (i.e., how we experience our mental
episodes) for rational motivation is further discussed under the heading of Experiential Rationalism in Dorsch (2009b), (2012a) and (2013).

Meeting the Desiderata

413

(ii) Fundamentality
It is also reasonable to maintain that the thesis (ACT) assuming that it is
true identifies the most fundamental feature which imaginings have in
common and which is responsible for their imaginativeness. Indeed, the
basicness of the property of being a mental action with the purpose of directly producing specific representations is very difficult to deny, given that
there does not seem to be any other, more fundamental, property that all
the bearers of the complex feature in question have in common, and which
is responsible for their exemplification of direct agency with respect to the
specific determination of what is represented.
In particular, as I have argued in Chapter 8, the property of lacking the
capacity to provide knowledge or to show any of the cognitive features
essential to the provision of knowledge (e.g., reliability or veridicality)
cannot account for the common imaginativeness of imaginings, given that
some imaginings (such as the two examples introduced in Chapter 4) can
cognise reality, while others (affective imaginings and imaginative projects) do not differ from their non-imaginative counterparts in their noncognitivity. On the other hand, the cognitive features that imaginings lack,
but which are not necessarily linked to the provision of knowledge, are not
responsible for the distinctive active character of imaginings, either. There
is no good reason to maintain that imaginings involve the active and direct
determination of what they specifically represent because they do not possess a cognitive attitude, do not provide us with prima facie reasons for belief, or do not have the function or aim of providing knowledge.
Similarly, the fact that sensory and affective imaginings are perhaps best
understood as representations of cognitive episodes does not help to explain why (ACT) is true of all imaginings. First of all, it cannot account for
the active character of intellectual imaginings or imaginative projects, say,
because they do not represent their cognitive counterparts. But even if this
problem of scope is ignored, the representational link essential to some
forms of imaginings cannot account for their equally essential voluntariness. For being a representation of a cognitive episode is fully compatible
with being a passive episode. Sensory memories are arguably representations of perceptual experiences as well (see Section 9.4). But it is not up to
us how we recall people, situations or other entities as being. Whether I re-

414

The Agency Account

member my friend to have blond or brown hair depends on which colour I


perceived him as having in the past, and not on what I want his colour to
be like.
Hence, the complex feature that (ACT) identifies as being distinctive of
imaginings namely the feature of being a mental action with the purpose
of directly forming representations with particular contents resists any
elucidation in terms of a more fundamental property.

14.2. Extensional Adequacy


The other desideratum for a unified account of imagining is extensional adequacy. It requires of such a theory that it is true of all central cases of imagining, but not of any paradigm instances of non-imaginative phenomena,
in particular not of cognitive ones. Let me begin with the second and negative part of this requirement.

(i) Non-Imaginative Phenomena


In the last chapter, I have already observed that we have no direct control
over what we perceive, remember or judge to be the case (see Section
13.2). Accordingly, (ACT) suffices to distinguish imaginative episodes
from their cognitive counterparts. The same is true of imaginative and cognitive projects: while the former aim at the formation of representations
with directly determined contents, the latter aim at the acquisition of knowledge. The two cases introduced in Chapter 4 are good examples. Both the
project of mentally rotating the sofa before the background of the door and
the project of mentally changing the appearance of the friends face are
imaginative projects: their ultimate intrinsic end is to produce and directly
determine relevant visual representations. But both are also embedded in
and pursued because of wider cognitive projects: namely the project to
acquire (factual) knowledge about the relative sizes of the sofa and the
door, and the project to acquire (recognitional) knowledge of the current
appearance of the aged friend, respectively. This difference in purpose
between imaginative and cognitive projects is captured by (ACT): while
the latter aim at the production of cognising representations, the former

Meeting the Desiderata

415

aim at the production of representations with directly determined specific


contents.
Equally, (ACT) manages to differentiate imaginings from non-cognitive
phenomena. First of all, non-cognitive (and non-imaginative) kinds of representation such as emotions, desires, intentions, and so on also resist
any attempt at actively and directly determining what they represent. We
cannot bring ourselves to feel love for a certain person or to desire her
well-being simply by merely willing to do so, just as we cannot bring
ourselves to believe that she is nice or to remember her as beautiful by
mere decision to do so (see, for instance, Pink (1996) for a defense of the
involuntariness of decisions). In all these cases, we can only indirectly influence the occurrence of the respective mental states for instance, by going to a hypnotist or manipulating the evidence available to us. The proposed Agency Account explains the non-imaginativeness of these states
precisely by reference to this aspect of their nature. But it also captures the
central cases of non-imaginative and non-cognitive mental projects, whether they aim at the appearance of certain feelings or moods, the making of a
practical decision, or some other end. Again, these mental projects are nonimaginative because they do not possess the imaginative purpose of directly producing representations as their ultimate intrinsic end even
though they may embed, or be embedded in, an imaginative project with
such a purpose.
There are two other mental phenomena that resemble intellectual imagining, but escape easy classification. The phenomena in question are entertaining a proposition and wondering whether something is the case.
While the second should presumably not count as imaginative and, hence,
need to be distinguished from imagining in one way or another, which
treatment is most adequate with respect to the first is less clear. Although
both phenomenal would ideally require a more thorough investigation, I
confine my discussion here to some suggestions about how the Agency Account may accommodate them.
Entertaining a proposition (or a thought content) may be understood in
two ways. On the one hand, it may simply mean having a certain propositional content in ones mind as part of an intellectual episode (see Walton
(1990): 20). On the other, it may mean to actively bring about an episode

416

The Agency Account

with such a content. Entertaining a proposition in the first (and perhaps


more natural) sense seems to be involved in any kind of intellectual representation, whether it is a judgement, occurrent desire, supposition, and so
on. But it does not seem to constitute a self-contained mental phenomenon
(e.g., a single mental episode), let alone an imaginative one, but rather just
a dependent representational element of such a phenomenon. In contrast,
entertaining a proposition in the sense of actively thinking something involves more than having a proposition in mind: it also includes at least the
activity of forming the respective representation. The resulting instance of
active thinking may very well constitute an imaginative episode or simple
imaginative project. This is compatible with the idea that merely actively
imagining (or thinking) something is less committal or constrained than
supposing it as part of, say, the imaginative engagement with a fictional
story, a hypothetical inference, a thought experiment or a game of makebelieve (see Section 1.2).
Merely to wonder whether something is the case, on the other hand,
means to consider some or all of the alternative propositions, without actually endorsing any of them. Someone may thus ask himself whether we
have knowledge about the external world, or whether God exists, but
prefer not to form any fixed opinion about these matters. Such cases of
wondering seem to involve the active determination of what is represented.
But it is not absolutely clear whether wondering should be analysed as having intellectual imagining as one of its constituents (as Frege (1884) seems
to believe); or whether wondering should rather be seen as an independent
propositional attitude. My own proposal is to treat wondering as the engagement with a mental project, the purpose of which is to do whatever is
involved in the pursuit of a cognitive project (e.g., considering the different
propositions and the evidence available for them), without actually forming
a conclusion. Accordingly, cognitive projects differ from projects of wondering about something in that only the former, but not the latter, are explicitly aimed at making up ones mind about a certain theoretical issue. Indeed, it may be part of the purpose of some projects of wondering to deliberately avoid, or abstain from, any such judgement.

Meeting the Desiderata

417

(ii) Imaginative Phenomena


Now, (ACT) is extensionally adequate not only with respect to the
paradigm instances of non-imaginative phenomena, but also with respect to
the central cases of imagining. Imaginative projects obviously satisfy the
account of imagining expressed by (ACT). Mental projects are, in general,
instances of agency; and, just as mental actions, they differ primarily in
their purposes. While cognitive projects have the aim to gain knowledge,
imaginative projects have the aim to directly form representations with
specific contents.21 Daydreaming about being rich, say, means producing
sensory or intellectual representations of oneself being rich (e.g., of owning a mansion with a large swimming pool, of dining in expensive restaurants and shopping in exclusive shops, etc.) without the help of memories of
being rich, thought-inducing drugs, hypnosis, or similar means. Similarly,
imagining being Napoleon involves the production, in a direct fashion, of
representations which identify oneself with Napoleon, or which represent
the world as it might have been perceived and thought of by him.
Accordingly, the Agency Account characterises imaginings whether
they are episodes or projects by reference to the same feature: namely,
their activeness and purposiveness with respect to the direct formation of
specific representations. This means that there is no fundamental difference
between, say, visualising a static tree and visualising a galloping unicorn,
or between supposing that one is rich and daydreaming about the consequences which that state of affairs would have for ones life. Of course,
these mental actions differ in various significant respects, not the least in
whether they are episodic or complex. But they are all concerned with the
same basic purpose, namely the active formation of representations with
directly determined contents, and involve corresponding motivational
21

Given that the only way of directly forming representations with specific contents is
to generate imaginative episodes, imaginative projects may also be individuated as
projects that aim at the production of imaginative episodes. But their description in
terms of the direct production of representations with specific contents is to be preferred, not only because it reflects the fact that imaginative episodes and imaginative
projects count as imaginative for the very same reason, but also because it is more
fundamental. Imaginative projects aim at the production of imaginative episodes because they aim at the production of representations with directly determined specific
contents; and not the other way round.

418

The Agency Account

states. Their main difference (apart from particular differences in content,


type, time of occurrence, and so on) consists just in the complexity or number of the episodic representations to be formed it is, in some sense,
merely a difference in degree.
Note, however, that not all episodes and projects produced or included
during the pursuit of an imaginative project have to possess an actively and
directly determined content in order for imaginative projects to count as
satisfying (ACT).
On the one hand, our imaginative projects may include episodes, such as
perceptions or (perhaps non-representational) feelings, which have occurred passively and independently of the pursuit of the imaginative project in question. What we perceive or feel may often prompt us to imagine
certain things. Furthermore, at least sometimes, it is reasonable to say that
the respective episodes thereby become part of our imaginative project because of their contribution to its pursuit. For example, the perception of the
sea and the related feeling of joy may give rise to and further guide a daydream about a happy day spent with another person at the beach. In virtue
of their contribution to the representation of such a day, it may also be
plausible to take them to be part of the respective imaginative project, despite their passivity.
On the other hand, imaginative projects may involve the active occasioning of the indirect and passive generation of representations, for instance by means of eliciting the manifestation of mental dispositions, or by
using rational mechanisms, such as inference or conceptualisation, which
are passive with respect to the determination of what is represented. As a
result, we can remember something, come to feel emotions, judge something on the basis of evidence, and so on, as part of the pursuit of an imaginative project. The daydream about the day at the beach may prompt old
memories and new feelings. For instance, it may give rise to the conclusion
that one is in love.
Furthermore, it may not always be easy to decide both for the subject
in question and for us whether a given episode is part of a certain imaginative project, or only closely related to it. That is, it may be undecided to
what extent an episode contributes to the desired representation of certain
people, situations, events, and so on. For instance, daydreaming about the

Meeting the Desiderata

419

day at the beach may actually trigger some unexpected, ambivalent feelings or thoughts about the prospect of spending so much time with the other person. Given that these feelings or thoughts may run contrary to the intention to imagine a happy day, it is not clear to what extent they still belong to the daydream, or to what extent they actually terminate it or turn it
into a different daydream. Imaginative projects are by no means unique in
this respect. Which feelings, memories or associated thoughts actually contribute to and not merely accompany ones deliberation about what to
believe or do in a particular situation may sometimes be difficult to decide
too. Equally, the question of which particular acts of thinking are part of
the project of solving a certain problem in ones head may resist conclusive
settlement as well.
Many experiential imaginings are imaginative projects (see Hopkins
(1998): ch. 7, and Noordhof (2002)) and therefore fit the proposed Agency
Account as well. The experiential imaginings in question are those consisting of at least two distinct episodes: an imagining of a mental episode together with some higher-order representation about the imagined nature of
that episode. For instance, I may imagine a visual perception as of some
land appearing on the horizon and simultaneously imagine that my imagined representation is the perception of Columbus when he thought he
had reached India. We actively engage in such experiential imaginings, and
we do so for reasons related to the formation of specific representations.
My experiential imagining may be motivated by, say, my intention to imagine being Columbus when he first saw land.
But not all experiential imaginings of mental episodes are complex in
the way described: there are also episodic instances of experiential imagining. For instance, visualising something is best analysed as imagining seeing it, while the latter is understood as an instance of object awareness concerned with imagining an instance of a certain phenomenal type of perceptual experience (see Chapter 10). Something similar is true of emotional
imagining (see Chapter 11). Contrary to what Peacocke seems to think (see
his 1985, and also Section 3.6), these cases of experiential imagining do
not have to involve any thought-like element specifying the nature of the
imagined experience, given that this nature is determined by default. Imagining a perceptual experience as of a tree, for instance, amounts, by de-

420

The Agency Account

fault, to imagining a veridical perception of a tree.


This leads to directly to the consideration of imaginative episodes and
their prospect of satisfying (ACT). It should be uncontroversial that at least
many imaginative episodes are in conformity with the Agency Account in
virtue of their active and purposive nature. When we suppose that it is raining, visualise a palm tree, imagine hearing a melody, imagine seeing a face,
and so on, we normally act on the desire or intention to produce mental
representations of this kind. The same is true of imaginative episodes that
are not purely representational and also involve some affective elements.
They are the result of a desire or intention to directly produce representations of certain affective episodes, or at least representations with a specific
content and certain affective aspects. The active character of affective imaginings distinguishes them strictly from non-imaginative episodes that occur passively and involve real and not imagined affection. In particular,
there do not seem to be instances of spontaneously or otherwise passively
caused imagined feelings: we are not suddenly bothered by an imagined
itch, or overcome by imagined jealousy.
What these examples of imaginative episodes and projects illustrate is
that the general purpose of imagining that is, the aim of directly forming
representations with specific contents can be realised in many different
forms: by referring to more or less specific contents to be imagined; by including a determination of the type of representation; by referring to
particular types of representation; by aiming at episodic or more complex
representations; and so on. The examples also show that most of the central
cases of imagining all instances of imaginative projects and at least many
instances of sensory, intellectual, affective and experiential imagining are
captured by the proposed Agency Account of imagining.
What remains to be shown is whether it indeed applies to all imaginative episodes, and not only to many. Two questions are of particular relevance here: whether there are non-representational instances of imagining;
and whether there are passive instances of imagining. In the remaining sections, I intend to show that the Agency Account is in the position to answer
both questions negatively, thus completing the case for the exception-less
extensional adequacy of (ACT).

Meeting the Desiderata

421

14.3. Non-Representationality: the Case of Imagined Pains


To begin with, I consider the issue of whether there could be non-representational imaginative episodes. If there were such states, some of them
would presumably constitute central cases of imagining and, hence, be
problematic for the Agency Account, which takes all imaginings to be the
result of the purposive activity of directly forming representations of a certain kind. If there were non-representational imaginings, their formation
could not be linked to such a purpose.
The only plausible examples of non-representational imaginings are affective imaginings corresponding to non-imaginative and seemingly nonrepresentational affective episodes, such as feelings of pain or orgasm, or
feelings or moods of anxiety or loneliness. It is certainly controversial
whether there are any non-representational (or non-intentional) mental phenomena (see Crane (2001): ch. 24 for a defense of the intentionality of the
mental). But, for the sake of argument, let me assume that pains and certain
other feelings or moods are non-representational. What I intend to argue is
that, even if there are non-representational affective episodes, the corresponding affective imaginings are still representational. More precisely,
they are experiential imaginings of real episodes of pain or similar affective episodes and, in this sense, show immediacy. I make use of a similar argumentative strategy as in my discussion of visualising and emotional imagining (see Chapters 10f.) and confine my considerations to pains. But I
take it that they apply to other candidates for non-representational episodes
and their respective experiential features as well.
Remind yourself again of the case of an imagined pain (see Section
11.5; see also the Sections 10.1f. on imagined itches). Such an imaginative
episode differs phenomenologically from a real instance of pain. For instance, we do not come to find the former unbearable in the same way as,
sometimes, the latter. We do not cry or faint as a consequence of experiencing imagined pain, for instance. In this respect, imagined and remembered
pains are much closer to each other than to real ones. None the less, imagined (and remembered) pains still involve some form of experiential
awareness of pain: they have some experienced painfulness about them. In
particular, they differ in this respect in their involvement of an affective
element from merely thinking or intellectually imagining that oneself (or

422

The Agency Account

another person) has a pain.


Now, the difference between imagined and really felt pains is not a matter of degree (e.g., in determinacy, intensity or vivacity: see Sections 3.1
an 3.3). On the one hand, we often do have real pains that are not very intense or determinate, but which we none the less experience as real pains,
and not merely as imagined ones. Similarly, we can imagine rather strong
and specific pains, without thereby beginning to really feel pain. On the
other hand, real and imagined pains possess different functions or rational
roles: only real feelings of pain have the function to ground first-order beliefs about the body parts concerned, and to rationally move us to do something about the pain or the underlying body damage (e.g., to take an aspirin, apply a plaster, or visit the doctor). But such functional differences
are qualitative, not quantitative.
In addition, the difference between the two kinds of affective episode
cannot be one in attitude. If pains are indeed non-representational, it does
not make any sense to speak of a pain-related attitude, given that there is
no content (or intentional object) at which the attitude might be directed.
Non-representational feelings may be experienced as being actually
present. But they cannot be experienced as indicating the actual presence
of something other than themselves.
As already noted in Section 11.5, a much better alternative is to assume
that the difference between real and imagined pains is due to the fact that
imagining (and perhaps also remembering) pain involves a different
experiential element from really feeling pain. The idea is that, while having
a real pain involves (or consists in) the experience of real pain, imagining a
pain involves (or consists in) the experience of an echo of real pain. Accordingly, the fact that imagined pains are not real pains (and not experienced as such), while still involving some kind of awareness of painfulness, is explained by reference to the fact that imagined pains represent
painfulness, without really being painful. In other words, instances of imagined pain are instances of experientially imagining pain: that is, instances
of imagining the experience of real pain. Given that real pains involve the
experiential feature of painfulness, imaginative representations of them involve this experiential aspect as well however, only by representing it,
and without themselves instantiating it (i.e., without themselves becoming

Meeting the Desiderata

423

real pains).22
Given that there is no reason to doubt that the situation is the same with
respect to other examples of affective imaginings, that correspond to nonrepresentational feelings or moods, it should be accepted that there are no
non-representational imaginings. Accordingly, instances of imagining pain
or similar affective imaginings can be accommodated by (ACT): they are
imaginings formed with the purpose of directly producing a representation
of how it feels, say, to be in pain. This may also explain why discussions of
imagining typically do not address the issue of the actuality or possibility
of non-representational imaginings.

14.4. Passivity: the Case of Spontaneous Images & Thoughts


By contrast, reference to supposedly passive imaginings is perhaps the
most prominent and promising way of criticising the Agency Account
of imagining. Two potential counterexamples to (ACT) stand out in the literature: pictorial experiences, and spontaneous or unbidden images and
thoughts.23 Following Hopkins ((1998): ch. 7), I have argued elsewhere
(see Dorsch (2012c)) that pictorial experience does not involve imagining.
More important, the argument presented at the end of last chapter (see Section 13.5) should already be enough to convince skeptics about (ACT) of
the truth of the claim that imaginative episodes are essentially active. Accordingly, pictorial experiences and spontaneous representations do not
really pose a threat to the Agency Account, since they are clearly passive in
their representationality and do not involve imaginative activity. But it is
perhaps none the less helpful to spell out in a bit more detail how the
Agency Account can accommodate cases that have been suggested as initially plausible instances of passive imagining.
See Section 10.1 for a discussion of Martins similar considerations about imagined
itches (see Martin (2002a): sec. 3)).
23
Scruton (1974: ch. 8, especially 109ff.), Walton (1990: 293ff.), OShaughnessy
(2000: 346ff.) and Stock (2008) argue that our passive awareness of what pictures
depict is imaginative, while Collingwood (1938: 195) and OShaughnessy (2000:
350f.) maintain the same with respect to spontaneous images or thoughts. Especially
the latter two make clear that the existence of such passive imaginings poses a serious threat to the Agency Account.
22

424

The Agency Account

In the remainder of this chapter, I aim to do this with respect to spontaneous images and thoughts. In line with the specification provided in Section 5.6, spontaneous representations can be characterised as non-cognitive
sensory or intellectual representations that occur entirely unbidden or unwilled in our minds. This means, minimally, that they lack a cognitive attitude and the epistemic function characteristic of cognitions; and that they
arise in an unsolicited manner, that is, do not involve any agency or effort
on behalf of the subject (see Section 5.6; see also Casey (1976): 34f.; 68f.).
They are thus passive through and through, and not determined by any of
our motivational states.
Examples are thoughts that simply cross our minds, or images that appear unbidden when we close our eyes (such as hypnagogic representations). In this respect, they resemble perceptual experiences and strictly
differ from compulsive or seemingly involuntary imaginings, which we
actively bring about, though possibly sub-intentionally and contrary to
what we really want or take to be the best course of action (see Section
13.3). As Casey has observed, the fact that spontaneous episodes arise unwilled and without the involvement of any effort has also the consequence
that they have the capacity to surprise us. For we do not know in advance
what they will represent and of what type of representation they will be; or
even that they will appear in the first place (Casey (1976): 34; 63).
Spontaneous representations may still originate in some of our motivational states, as long as these merely cause their occurrence, that is, determine them only by some non-rational and possibly unconscious process. More often, however, the occurrence of spontaneous representations
is due to associative links, mental dispositions, and similar phenomena.
Non-mental causes, such as certain drugs or neural processes, may bring
them about as well.
Another important feature of spontaneous images and thoughts is that
they are by their very nature fleeting and elusive in character or, as Casey
puts it, they occur instantaneously [...] and without any sense of drawn-out
duration (Casey (1976): 34). In particular, they are typically transitory and
do not involve any of the development for instance, with respect to which
entities they represent, and how that active and deliberate imagining may
include (see Casey (1976): 34f.; 70f.). As a consequence, they usually dis-

Meeting the Desiderata

425

appear as quickly as they come, and as soon as some new representations


enter our stream of consciousness spontaneously or are actively formed by
us. When we think about a certain problem, say, a specific thought may
suddenly occur to us. But if we do not recognise it as relevant for our considerations, we do not dwell on it, and it soon vanishes again. Similarly,
when we close our eyes, we may just let the spontaneously arising images
follow each other for a while. They promptly fade away and are replaced
by other images, until we stop their movie-like sequence by actively holding on to a particular image, or by opening our eyes again and paying attention to what we can see around us. Their fleeting character may also be
the reason why spontaneous images and thoughts do not possess any distinctive function in our conscious mental lives (see Dorsch (2010a)).
Under certain circumstances, however, spontaneously appearing contents remain in existence for more than a very short period. Sometimes,
this happens in the context of psychologically or physiologically unusual
or pathological cases, in which the underlying causes continue to be effective and may even prevent us from banishing the respective representations.
Taking a hallucinogenic substance typically has such a result. But the
causes may also lie in ones unconsciousness or in some psychological disorder, as in the case of Macbeths visions of the dagger.24
In normal circumstances, however, spontaneous images and thoughts
are (if at all) sustained and further developed by our own agency. For instance, we may find a thought coming to us worthy of further consideration
and may thus decide to investigate its truth or justification. As a consequence, we may begin to actively entertain the proposition perhaps in
24

See OShaughnessy (2000: 346ff., 479), who also mentions Macbeths vision of
Banquo which, however may instead be taken to be a perfect perceptual hallucination with a cognitive attitude (or a full sense of reality in OShaughnessys
terminology; see Dorsch (2010a) and (2012a)). It may not always be easy to decide
whether cases like Macbeths visions indeed lack any agency on behalf of the subject. Some seemingly spontaneous representations may in fact involve such activity
which is, however, not acknowledged by the subject as agency, or not as his own
agency (see Roessler (2001) on inserted thoughts). Cases in which the agency is acknowledged as ones own, but is in conflict with some of ones further desires or intentions, are different. Someone may be obsessed by a certain thought and not be
able to banish it from his mind, but still recognise himself as actively thinking it, or
even wanting to think it (see the involuntary imaginings discussed in Section 13.3).

426

The Agency Account

the background of our minds while looking at some relevant evidence.


Or we may try to arrest some of the fleeting visual images occurring in us,
or to further modify, develop or otherwise work with them, by engaging in
activities of visualising (see Casey (1976): 63ff.). Our purpose in doing this
may be, say, to hold on to the original images or very similar visual representations, or to change them in interesting ways. What happens in both the
intellectual and the sensory case is that we sustain the spontaneously arisen
contents beyond their usual short duration by including them in, and adjusting them to, certain mental actions or projects. This presupposes, of
course, that the respective contents are subject to the will (i.e., that we can
actively influence or terminate them). But apart from unusual or pathological situations, this should normally be the case.
However, actively sustaining spontaneously occurring representational
elements has the further consequence that they thereby lose their spontaneity because they cease to be passively determined. Again, what is crucial here is that passively occurring episodic representations differ essentially from actively and directly determined ones (see Section 13.5). Accordingly, as soon as spontaneous images or thoughts are subjected to an
activity of sustaining or changing, the spontaneous representations are
turned into or, perhaps better, succeeded by mental states of a very different kind, namely active thoughts and images (see Casey (1976): 64ff.).
The spontaneously arisen contents may continue to exist; but not their
spontaneity. The representational elements cease to be spontaneous because they become actively sustained as part of their inclusion in the mental action or project in question (see Casey (1976): 64ff.). As a result, the
change from passivity to activity is accompanied by a change in essence
and kind of mental episode: a spontaneous representation is replaced by an
active representation which involves the same representational element as
the former, to the extent to which what is represented is not actively
changed by us. Once we begin to actively sustain spontaneously appearing
representational elements, they undergo a significant change in their
nature.
One consequence of this is that spontaneous images and thoughts despite being non-imaginative are more similar to imaginings than to cognitions (or other kinds of episode) in respect of their susceptibility to imagin-

Meeting the Desiderata

427

ative activity. If we want to include spontaneous representations into our


imaginative actions or, alternatively, use (aspects of) their representational
elements in such actions or projects, we have to actively sustain the spontaneously occurring representational elements by means of imaginative
activity. The required activity has to concern both the occurrence of the
representations and the determination of their content. For, except in unusual or abnormal cases, whatever originally causally determined their occurrence and specific nature becomes ineffective very quickly. But the impact of the imaginative agency has the result that the nature of the episodes, of which the representational elements in question are part, change
essentially from being passive to being active.
By contrast, cognitions do not become active when they are included
into imaginative actions or projects. Perceptions, judgements or memories
are not altered in their passive character when they become part of imaginative activity. Similarly, when we borrow the representational elements of
perceptions, memories or judgements for our formation of imaginative
episodes, our use of (aspects of) the respective representational elements
leaves the existence or nature of the original cognitions unchanged as well.
Hence, spontaneous representations are closer to imaginings than to cognitions in that they provide easily sustainable or influenceable representational material for the activity of imagining and can be turned into imaginative episodes, or included in imaginative actions or projects, simply by
becoming actively sustained. Indeed, it seems that no kind of nonimaginative representation, apart from spontaneous ones, allows for this
active and transformatory influence. This does not mean, however, that
spontaneous representations are imaginative. In fact, they are not, given
that they are passive with respect to the determination of their contents.
Only those different representations, that are the result of the active sustaining of spontaneous images or thoughts, should count as imaginative.
Despite these considerations, it has not been uncommon to take spontaneous representations to be imaginative (see e.g., Sartre (1940): 19; Casey
(1976): 34 and 63ff.; Peacocke (1985): 26f.); OShaughnessy (2000): 344f.
and 349ff.). But there is no good reason to embrace this conclusion. The
main motivation for the categorisation of spontaneous representations as
imaginings seems to be their non-cognitivity: for instance, that they cannot

428

The Agency Account

cognise reality, at least not in the same ways as perceptions and beliefs (see
OShaughnessy (2000): 344f. and 349ff.); or that they do not involve a
cognitive attitude (see Sartre (1940): 12f.; 181f. who distinguishes there
sensory imaginings from both perceptions and sensory memories).25 The
respective negation claims about spontaneous representations may very
well be true. As a consequence, spontaneous images and thoughts may
resemble imaginings more than cognitions, not only in their easy
subjection to the will, but also in their phenomenal character and their role
in cognition.
But, as I have argued at length in Part Two, the non-cognitivity of imaginings is neither unique to them, nor responsible for their imaginative
status. Hence, the similarity between spontaneous and imaginative representations in this respect does not suffice to justify the classification of the
former as a subclass of the latter. Moreover, spontaneous representations
do not possess any other distinctive features that might plausibly be said to
be sufficient for imaginativeness. In particular, they share their passive occurrence and their origin in mental or neural causes with many cognitive
representations, such as perceptual hallucinations and associatively
triggered episodic memories. The claim that spontaneous representations
are imaginative remains, therefore, completely unfounded. Instead, they
should just be taken to be very similar to imaginings in important respects
a fact which the Agency Account of imagining can satisfactorily explain.
It is important to note that this line of reasoning applies not only to normal and short-lived instances of spontaneous representation, but also to
spontaneous images and thoughts that are the result of mental disorders,
subconscious forces or hallucinogenic substances and which may stay in
existence for longer periods of time, and usually against our will. The latter
are equally non-cognitive in virtue of their lack of a cognitive attitude and
their incapacity to cognise reality. But again, and for the same reasons, this
25

Casey does not provide any specific support for his position. Sartre notes also similarities between spontaneous images and cases of active visualising (e.g., their quasiobservationality, or their negating attitude towards their objects; see Sartre (1940):
ch. 1). But these similarities hold between visual imaginings and visual memories,
too (as Sartre accepts at Sartre (1940): 10; 181), and hence are not relevant for the
imaginativeness of episodes.

Meeting the Desiderata

429

is not enough to render them imaginative. 26 The Agency Account is therefore able to accommodate, not only the imaginative counterparts of nonrepresentational and non-imaginative phenomena, but also cases of what
might wrongly seem to be passive imaginings notably spontaneous images and thoughts.

Conclusion

26

Wittgenstein also seems to embrace the view that these kinds of states are non-imaginative when he writes: Vorstellungen sind nicht Halluzinationen, auch nicht Einbildungen [Imaginings are not hallucinations, nor figments the author] (Wittgenstein (1984d): sec. 621, and (1984b): vol. II, sec. 63). See Dorsch (2010a) for more
on the nature of, and distinction between, short-lived and persistent spontaneous images and thoughts, as well as their relationship to imaginings and cognitions.

Conclusion

(i) The Agency Account


The principal aim of this book has been to investigate the prospects of formulating a satisfactory unified account of imagining. Its main conclusion
has been partly negative and partly positive: while most theories notable
among them the Epistemological and the Dependency Account fail to explain the unity of imagining, the Agency Account in its presented
manifestation promises to be true. Although the presented defense of this
view has by no means been complete and needs to be continued in more
detail elsewhere, it put forward four arguments for endorsing the Agency
Account: first, that all alternative theories cannot account for the unity of
imagining (see especially Chapters 5, 8f., 12, and Section 13.5); second,
that the Agency Account satisfies the chief desiderata for a unified theory
of imagining (see Chapter 14); third, that it promises the best explanation
of the quasi-observationality of sensory imaginings (see Section 3.6); and,
fourth and most important, that the imaginative agency involved in voluntary instances of imagining is essential to them and, hence, rules out the
possibility of involuntary instances (see Section 13.4).
The championed Agency Account takes imaginings to be mental actions
of a particular kind. More precisely, it claims with certain qualifications
that imaginings are mental actions aimed at the formation of one or more
mental representations of specific states of affairs, that are directly determined by the mental agency involved (see Chapter 13). That the determination of which states of affairs are represented is direct means thereby that

432

Conclusion

our voluntary control over what is represented is not mediated by the reliance on epistemic or merely causal processes as means. The resulting account can distinguish imaginative episodes from their cognitive counterparts and other non-imaginative episodes, given that the latter allow at best
for an indirect voluntary determination of their content. It also can distinguish imaginative projects from other kinds of mental projects, notably
from cognitive ones, which aim instead at the formation of cognising representations.
More generally, while the explanatory power and fundamentality of the
Agency Account with respect to the unity of imagining should be fairly uncontroversial, its extensional adequacy needs to be further defended against
several potential counterexamples. On the assumption that cognitive and
other non-imaginative representations do not allow for the voluntary and
direct determination of their content, my discussion has been primarily
concerned with cases that appear to be either non-representational, or passively or indirectly determined instances of imagining.
The best candidates for non-representational imaginings are imaginings
with non-representational counterparts. Plausible examples are imagined
pains or imaginatively felt anxiety assuming, for the sake of argument,
that genuine feelings of pain or anxiety do not show any form of
representationality. However, such cases are most fittingly understood as
instances of experiential imagining: namely as representations of their nonimaginative counterparts. Imagining a pain thus amounts to imaginatively
representing the occurrence of a real pain.
The array of potential examples of passively or indirectly determined
imaginings, on the other hand, is more varied, ranging from spontaneous
images and thoughts via obsessive mental representations to pictorial experiences or non-purposive associations (although the last two are not
really discussed in this book). The already mentioned argument to the effect that the direct mental agency involved in voluntary imagining is essential to its instances entails, however, that there is a basic difference in
nature between active imaginings and all passive representations, indicating that the latter belong to a different and, presumably, non-imaginative
mental kind. Hence, none of the listed phenomenal poses a threat to the
Agency Account of imagining. Spontaneous representations, for instance,

Conclusion

433

are simply very closely related to imaginings in so far as and in contrast,


say, to perceptual images they allow for their voluntary sustainment and
transformation, albeit with the major consequence that they change in
nature and turn from spontaneous into imaginative episodes, once they are
subjected to imaginative activity. I return to the issue of how the Agency
Account might deal with obsessive images and thoughts further below.
The explanatory power and extensional adequacy of the Agency Account also ensures that it has the resources to satisfy the basic need that has
motivated the search for a unified account of imagining: namely the desire
to explain the fact that we group together a variety of apparently very different phenomena under the heading imaginings, as well as the fact that
our corresponding categorisations of phenomena as either imaginative or
non-imaginative are typically immediate, assured and stable. The idea is, of
course, that our classifications track a certain feature that is responsible for
the unity of the class of imaginings, and the presence of which we can normally become aware of by means of introspection. According to the
Agency Account, the feature in question is the distinctive and phenomenologically salient active character of imaginings, as specified by the thesis
(ACT).1
There is a closely related and particularly noteworthy respect in
which the approach of the Agency Account differs from the other theories
of imagining, notably the Epistemological and the Dependency Account.
The strategy of the alternative views has usually been to investigate the
nature of imaginative episodes (e.g., visualisings or suppositions) and to
compare them with, and set them apart from, cognitive episodes. As a result, they fail to pay attention or do justice to imaginative projects or other
mental complexes involving imaginative episodes. In contrast, the Agency
Account deliberately centres on mental activity, of which imaginative projects are a prime example, and contrasts it with mental passivity (independently of whether it pertains to cognition or not).
Before I continue with this last point in a short moment, it is worthwhile
to briefly pause and consider one especially significant consequence of the
active nature of imagining: namely that at least some visual imaginings
manage to constitute quasi-perceptual knowledge about the external world.
1

See Dorsch (2009b) for further and independent arguments and observations in
support of the phenomenological salience of direct (but not indirect) mental agency.

434

Conclusion

For they do so if and, presumably, only if they occur in the context of


mental projects, the successful pursuit of which guarantees that the visual
imaginings concerned turn out to be reliable and possibly non-arbitrarily
veridical representations of certain aspects of reality (see Chapter 4). The
projects in question involve the active imaginative manipulation of perceptually acquired or mnemonically stored information; and they thereby take
into account the relevant ways, in which entities like the originally perceived or remembered objects behave in the actual world. Thus, despite the
fact that imagining cannot provide us with new evidence about reality, it
can none the less provide us with (a priori) access to new information
about the external world, that could not be read off of the underlying cognitive states by non-imaginative means (e.g., simply by means of introspection and reflection). The existence of such cognising imaginings is one
reason why the Epistemological Account does not come to much. But there
are others, to which I turn now.

(ii) The Epistemological and the Dependency Account


While the Agency Account is able to satisfy the two desiderata for a unified account of imagining, its two main rivals remain wanting in this respect. This is primarily due to the fact that both the Epistemological and
the Dependency Account characterise imaginings in terms of their noncognitivity. While the Epistemological Account takes imaginings to be
non-cognitive in so far as they lack one or more features distinctive of cognitive or cognising phenomena (e.g., an epistemic function, or the power to
constitute knowledge), the Dependency Account assumes imaginings to be
non-cognitive in so far as they are constitutionally dependent on and,
hence, distinct from cognitive phenomena (namely in some substantial way
that reaches beyond the general restriction on what we can imagine by our
conceptual capacities and our past experiences). Both the Epistemological
and the Dependency Account are therefore first and foremost concerned
with differentiating imaginings from cognitions. By contrast, and as
already mentioned, the Agency Account does not particularly concentrate
on the relationship between these two kinds of mental state, but instead
takes the distinction between mental actions and mental passions to be fun-

Conclusion

435

damental.
The focus on the non-cognitivity of imaginings is problematic since
characterising imaginings solely in terms of, and in contrast to, cognitive
phenomena significantly limits the scope of the Epistemological and the
Dependency Account. For certain forms of imagining notably affective
imaginings and imaginative projects do not have cognitive counterparts
and, hence, resist specification in terms of them. As a result, the relevant
theories turn out to be explanatorily unilluminating if not extensionally
inadequate with respect to certain central cases of imagining. For instance, it is not a substantial truth about affective imaginings that they cannot constitute knowledge, because this applies to all kinds of affective
states, including non-imaginative ones. And that emotional imaginings are,
indeed, representations of non-imaginative experiences does not distinguish them from, say, episodic memories and, moreover, does not amount
to a characterisation of them in terms of cognitive phenomena.
However, accounting for imaginings instead by reference to both cognitive and non-cognitive phenomena threatens to lead to a disjunctive theory:
while the nature of some imaginings (e.g., sensory or intellectual ones)
would be elucidated in terms of how they differ from or depend on cognitions, the nature of others (e.g., affective imaginings) would be described
by mention of how they differ from or depend on cognitions. The pair of
intellectual and affective imaginings shows that this would result in some
form of disjunctivism, given that intellectual imaginings depend merely
generally and semantically on judgements or beliefs, whereas affective
imaginings are representationally dependent on specific emotional feelings.
Such a disjunctive treatment is, by itself, unproblematic but becomes untenable once it is presumed to provide a unified account of imagining.
In addition, the Epistemological Account and, in particular, OShaughnessys rich and sophisticated version of it faces objections that are more
specifically directed at its central claim that imaginings lack a cognitive
concern with reality, or a closely related cognitive feature. To begin with,
the Epistemological Account possesses only limited explanatory power.
For postulating such an absence of a cognitive property does not say anything positive about what it means to imagine something which is in part
why OShaughnessy, as the principal proponent of the Epistemological Ac-

436

Conclusion

count, supplements his theory with a representational version of the Dependency Account.
Another problem is that, depending on which cognitive feature imaginings are said to lack, the respective versions of the Epistemological Account are either false, or uninteresting, with respect to the ambition to
provide a unified account. Some are interestingly false because they cannot
allow for the already noted possibility that sensory imaginings may, under
suitable circumstances, constitute knowledge and possess the required cognitive features (see above and Chapter 4). Accordingly, imaginings are not
always deprived of the properties of being reliable, informative, grounded
in perception, rationally integrated with our picture of the world and, possibly, constitutively linked to the relevant aspects of reality. The other versions of the Epistemological Account, by contrast, are uninterestingly true
because their denial that imaginings do not show certain other cognitive
features does not distinguish imaginings from other non-cognitive phenomena. The observation that imaginings lack a cognitive attitude, have no epistemic function and possess no prima facie justificatory power with respect
to first-order beliefs does not reveal much about their nature, given that the
same may be discerned, say, in emotional or conative states and in nonimaginative mental projects.
On the other hand, much of the Dependency Account notably in its
representational version can be preserved, not the least its identification
of a core divide, on the level of sensory and affective episodes, between
original or genuine occurrences and mere representations or imitations of
such occurrences. This crucial distinction has already constituted a central
element of Humes approach to (non-intellectual) imaginings and is as illustrated in Part Three still present in many contemporary theories, the
history of which can often be traced back to Humes theory of the mind.
However, the Dependency Account does not apply to intellectual imaginings and imaginative projects and should therefore be rejected as a theory
of the unity of imagining.
Moreover, the main truths noted by the Dependency Account as well
as by the Epistemological Account can be accommodated and explained
by the Agency Account. For example, visualising a tree amounts to imagining seeing a tree in part because the only direct way of actively bringing

Conclusion

437

about a visual image of a tree is precisely to represent the instantiation of


the phenomenal character of a perceptual image of a tree, given that we do
not have any voluntary control over perceptual or mnemonic images, and
only some limited active influence on spontaneous images (see Section
14.4). Similarly, imaginings do not normally play a role in cognition precisely because they originate in epistemically unreliable imaginative activity (see Section 14.1). Because of this explanatory power, the Agency Account is also more fundamental as a unified account than its two rivals.

(iii) Other Theories of Imagining


In Chapter 5 and Section 13.5, I discussed further alternatives to the
Agency Account of imagining and argued that they are equally untenable
as theories of the nature shared by, and unique to, imagining.
The Quantitative Account, which takes the basic difference between
imaginative and cognitive episodes to be merely one in degree, cannot capture the various phenomenologically salient differences in kind between
the two types of phenomena, notably the episodic differences in attitude
and epistemic function.
By contrast, the Phenomenal Account is wrong in assuming the fundamental difference between imaginative and non-imaginative episodes to be
phenomenal in nature, primarily because phenomenal differences among
mental episodes (such as the difference in attitude and seeming origin
between imaginings and cognitions) should and can be explained in
more fundamental terms.
One problem for the Ontological Account which claims that the difference between imaginative and cognitive phenomena consists fundamentally in the fact that they are concerned with entities of two very different
ontological categories is that its postulation of the two distinct categories
of objects lacks independent motivation. In particular, neither semantic
considerations about the status of fictional entities, nor phenomenological
considerations about the difference in attitude can be of help here. More
important, however, the Ontological Account fails to accommodate the fact
that imaginings can and often do refer to precisely those entities which

438

Conclusion

we can cognise, namely aspects of the actual world.2


The best version of the Modal Account of imaginings assumes that what
is distinctive of them is that they are committed to things being a certain
way in some possible world, rather than the actual world. The resulting
view is, in fact, an instance of the Attitude Account and faces the same
general objections as the latter. On a more specific level, the Modal Account cannot satisfactorily deal with cases in which we imaginatively take
some state of affairs to be impossible.
The idea of one or more distinctive imaginative attitudes, as proposed
by the Attitude Account, is problematic for two main reasons. First, it has
difficulties to avoid the conclusion that non-imaginative phenomena like
episodic memories, spontaneous images or pictorial experiences also involve the proposed kind(s) of attitude. Second, it does not identify the most
fundamental difference between cognitions and imaginings, given that the
phenomenologically salient attitude of mental episodes is further constituted by some underlying aspects of their nature. Besides, some versions of
the Attitude Account apply only to some forms of imagining, but not others.
The Spontaneity Account, too, is forced to treat spontaneously occurring
images and thoughts as imaginative. For it specifies the nature of imaginings in terms of their special origin in the mind, namely in subjective responses or doings rather than in externally imposed mere happenings
which may be either voluntary (in the shape of agency) or involuntary (in
the shape of spontaneity). But the views attempts at characterising spontaneity do not enable us to distinguish imaginings from hallucinations or
episodic memories. Moreover, the account fails to unify the postulated voluntary and involuntary forms of imagining and, relatedly, wrongly entails
that spontaneous representations are phenomenologically closer to rational
judgemental thoughts and mental actions than to passive perceptions and
sensations.
2

Besides, any ontological difference between imagined and cognised objects would
very likely to be linked to and perhaps even due to some co-extensional difference in how imaginings and cognitions relate to the world or the will (e.g., if imagined objects would turn out to be mind-dependent). The investigation of the latter
difference might thus render the investigation of the former at best supplementary,
and at worst superfluous.

Conclusion

439

Finally, the Subjection to the Will Account (which was discussed in Section 13.5 rather than Chapter 5) is a watered down cousin of the Agency
Account, given that it demands the possibility rather than the actuality
of voluntary and direct control over what imaginings represent. But, partly
because of this weakening of the requirement on the involvement of imaginative agency, it is to be rejected for three reasons. First, it cannot pay
justice to the fact that directly determined mental actions are essentially
active in this way. Second, it wrongly locates the difference between imaginative and non-imaginative phenomena in the ability to engage in imaginative activity, rather than in the nature of that activity. Third, it wrongly
turns the question of whether an episode is an instance of imagining into an
essentially empirical question, thus ignoring the phenomenological salience of the difference between imaginative and non-imaginative phenomena.

(iv) Some Loose Ends


The acceptance of the Agency Account as the most promising unified theory of imagining does not mark the end of inquiry. As already pointed out
in the Introduction to this book, it is indeed necessary to provide further
support for the Agency Account and to elucidate in more detail how it can
accommodate a whole range of mental phenomena, in particular those
which could not be discussed in this book a task which I have to leave for
a future occasion.
As it stands, the Agency Account is true only if pictorial experiences,
non-purposive associations and similar passive phenomena are indeed not
imaginative; and only if we really cannot actively and directly determine
the content of our judgements, beliefs, memories, intentions, and other
non-imaginative representations. Although these two assumptions are presumably less controversial than many other claims made in the book, more
work needs to be done to furnish them with a satisfactory argumentative
grounding (see Dorsch (2009b) and (2012c)). Similarly, some of the accounts discussed in Chapter 5 probably deserve more attention, not the
least with respect to how they are related to the idea that imaginings are
mental actions of a certain kind. Especially the Modal and, more generally,

440

Conclusion

the Attitude Account may have much to say about the nature of imaginative
episodes (if not also of complexes thereof) and, in particular, about the distinctive commitment of imagining.
It would also be interesting to investigate the extent to which mental
phenomena, that are neither clearly imaginative nor clearly non-imaginative, involve or are at least closely related to imagining. I already briefly
listed examples of such phenomena in Section 1.3 and discussed there also
why their status is rather undecided. But the truth of the Agency Account
may have interesting consequences for them, which are worthwhile of further exploration. Pathological or psychologically unusual phenomena (e.g.,
intrusive or inserted thoughts, compulsive representations, delusions, hallucinations, etc.) may turn out to involve some form of suppressed, non-deliberate or unacknowledged imaginative activity. Hopes, wishes, expectations, anticipations or speculations may likewise involve imagining, perhaps to the extent to which they mark or recognise the concerned non-actual states of affairs as unrealistic or unlikely to occur (i.e., to the extent to
which they fail to take the represented states of affairs to be present in the
actual future). Finally, although experiences of grouping phenomena, or of
noticing aspects, allow for voluntary switches between experiencing one
thing and experiencing another (or, perhaps more precisely, between
experiencing something as one thing or experiencing it as another), it is unlikely that they involve some imaginative activity as part of, or in addition
to, active attention. For which groupings or aspects we can choose between
does not seem to be up to us. In particular, we cannot decide to see a horse
in the famous duck-rabbit picture, just as we cannot decide to see a horse
in an unambiguous picture of a duck (see Dorsch (2012c)). And this list of
applications of the Agency Account to borderline cases can surely be extended much further.
Besides, it is very appealing to bring the conclusions about the distinctive nature of imagining to bear on relevant issues in other areas of philosophy. As mentioned in the Introduction to this book (see especially Note
2), the imagination is a central ingredient in accounts of such diverse phenomena as thought experiments, modal knowledge and knowledge of
counterfactuals, our simulation of and empathy with other people, pretense
and acting, games of make-believe, the aesthetic appreciation of artworks

Conclusion

441

and other linguistic or pictorial representations, or moral evaluation. The


gained insight that imaginings are mental actions of a certain kind may
very well help to clarify the nature of some of these phenomena even if
only by disclosing that they involve imagining to a far lesser extent than
originally thought.3
However, the general limits on the length of a book like this meant that
the discussion of the nature and unity of imagining had to be restricted to
the critical presentation of the various candidates for a unified account of
imagining to be found in the literature, as well as to an elementary defense
of the Agency Account as the best available unified theory. As the main
thesis (ACT) of this theory states, a mental phenomenon is imaginative just
in case and, fundamentally, because it constitutes a mental action with
the ultimate intrinsic purpose of actively forming one or more representations with specific and directly determined contents.

Bibliography

In this book, I discussed three specific applications of the imagination to other philosophical issues: to the acquisition of factual knowledge (see Chapter 4); to the acquisition of conditional and modal knowledge (see Sections 2.1 and 2.3, as well as
Williamson (2008)); and to our (emotional) engagement with art and fiction (see
Chapter 11, as well as Walton (1990) and Moran (1994)). I intend to elaborate on
these and other utilisations of the imagination in epistemology and aesthetics in a
future monograph.

Bibliography

Armstrong, David (1968). A Materialist Theory of the Mind. London:


Routledge.
Audi, Robert (1993). Action, Intention, and Reason. Ithaca: Cornell University Press.
Bayne, Tim & Pacherie, Elizabeth (2005). In Defence of the Doxastic
Conception of Delusion, in: Mind and Language, vol. 20, pp. 163-188.
Block, Ned (1994). Qualia, in: S. Guttenplan (ed.), A Companion to the
Philosophy of Mind. Oxford: Blackwell.
(1996). Mental Paint and Mental Latex, in: E. Villanueva (ed.), Philosophical Issues Vol. 7. Northridge: Ridgeview Publishing Company.
Bodrozic, Davor (2009). Fiktive Gegenstnde, fiktionale Rede und
berzeugungszuschreibungen. PhD Thesis submitted for publication,
University of Fribourg.
Boghossian, Paul (1989). Content and Self-Knowledge, in: Philosophical
Topics, vol. 17, pp. 5-26.
Braddon-Mitchell, David & Jackson, Frank (1996). Philosophy of Mind
and Cognition. Oxford: Blackwell.
(1997). The Teleological Theory of Content, in: Australasian Journal
of Philosophy, vol. 75, pp. 474-489.
Budd, Malcolm (1989). Wittgensteins Philosophy of Psychology. London:
Routledge.
(1992). Review of Kendall Waltons Mimesis as Make-Believe, in:
Mind, vol. 101, pp. 195-198.
(1995). Values of Art. London: Penguin.
Burge, Tyler (1979). Individualism and the Mental, in: French, Uehling &
Wettstein (eds.), Midwest Studies in Philosophy IV. Minneapolis: University of Minnesota Press.

444

Bibliography

(2003). Perceptual Entitlement, in: Philosophy and Phenomenological


Research, vol. 67, pp. 503-48.
(2005). Disjunctivism and Perceptual Psychology, in: Philosophical
Topics 33, pp. 1-78.
Campbell, John (1994). Past, Space and Self. Cambridge (Mass.): MIT
Press.
Carroll, Noel (2000). On the Narrative Connection, in: N. Carroll (2001)
Beyond Aesthetics. Cambridge: Cambridge University Press.
Carruthers, Peter (2000). Phenomenal Consciousness. Cambridge: Cambridge University Press.
(2003). Review of Currie & Ravenscrofts Recreative Minds, in: Notre
Dame Philosophical Reviews (at http://ndpr.icaap.org), no. 12.
Casey, Edward (1976/2000). Imagining. Bloomington: Indiana University
Press.
(1987). Remembering. Bloomington: Indiana University Press.
Chisholm, Roderick (1976). The Agent as Cause, in: M. Brand & D.
Walton (eds.), Action Theory. Dordrecht: Reidel.
Cocking, J. M (1991). Imagination: A Study in the History of Ideas. London: Routledge.
Collingwood, R. G. (1938/1958). The Principles of Art. New York: Galaxy.
Crane, Tim (1988). The Waterfall Illusion, Analysis, vol. 48, pp. 142-47.
(1992). The Non-Conceptual Content of Experience, in: T. Crane
(ed.), The Contents of Experience. Cambridge: Cambridge University
Press.
(2001). Elements of Mind. Oxford: Oxford University Press.
Crowther, Thomas (2010). The Agential Profile of Perceptual
Experience, in: Proceedings of the Aristotelian Society, vol. 110, pp.
219-242.
Currie, Gregory (2000). Imagination, Hallucination and Delusion, in:
Mind and Language, vol. 15, pp. 124-138.
(2001). Imagination and Make-Believe, in: B. Gaut & D. McIver
Lopes, The Routledge Companion to Aesthetics. London: Routledge.
(2002). Imagination as Motivation, in: Proceedings of the Aristotelian
Society, vol. 102 (3), pp. 201-216.
Currie, Gregory and Ian Ravenscroft (2002). The Recreative Mind: Imaginations in Philosophy and Psychology. Oxford: Oxford University
Press.
Dancy, Jonathan (2008). On How to Act Disjunctively, in: A. Haddock
& F. Macpherson (eds.), Disjunctivism: Perception, Action, Knowledge.

Bibliography

445

Oxford: Oxford University Press.


Danto, Arthur (1963). What We Can Do, in: The Journal of Philosophy,
vol. 60, pp. 435-445.
Davidson, Donald (1989). A Coherence Theory of Truth and Knowledge,
in: E. LePore (ed.), Truth and Interpretation. Oxford: Blackwell.
Davis, Lawrence (1979). Theory of Action. Englewood Cliffs (NJ): Prentice-Hall.
Dennett, Daniel (1988). Quining Qualia, in: A. Marcel & E. Bisiach
(eds.), Consciousness in Contemporary Science. Oxford: Clarendon
Press.
Divers, John (2002). Possible Worlds. London: Routledge.
Dorsch, Fabian (2000). The Nature of Aesthetic Experiences. MPhil
Thesis, University College London (published at: http://discovery.ucl.
ac.uk/1348583/).
(2007). Imagination and the Will. PhD Thesis, University College
London (published at http://discovery.ucl.ac.uk/1300296/ or http://sasspace.sas.ac.uk/2817/).
(2007) Sentimentalism and the Intersubjectivity of Aesthetic Evaluations, in: dialectica, vol. 61, pp. 417-46.
(2009a). Die Natur der Farben. Frankfurt: Ontos.
(2009b). Judging and the Scope of Mental Agency, in: L. OBrien &
M. Soteriou (eds.), Mental Actions. Oxford: Oxford University Press.
(2010a). The Unity of Hallucinations, in: Phenomenology and the
Cognitive Sciences, vol. 9, pp. 171-191.
(2010b). Colour Resemblance and Colour Realism, in: Rivista di Estetica, vol. 43, pp. 85-108.
(2010c). Transparency and Imagining Seeing, in: Philosophical Explorations, vol. 13, pp. 173-200.
(2011a). The Diversity of Disjunctivism, in: European Journal of
Philosophy, vol. 19, pp. 304-314.
(2011i). Emotional Imagining and Our Responses to Fiction, in: Enrahonar, vol. 46, pp. 153-176.
(2012a). Experience and Introspection, in: F. Macpherson & D. Platchias (eds.), Hallucination. Cambridge (Mass.): MIT Press.
(2012b). Non-Inferentialism about the Justification of Aesthetic Judgements. Manuscript.
(2012c). Pictorial Experience, Imagining De Re and Imaginative Penetration. Manuscript.
(2013). The Phenomenal Presence of Reasons, in: F. Macpherson, M.

446

Bibliography

Nida-Rmelin & F. Dorsch (eds.), Phenomenal Presence. Oxford: Oxford University Press.
Dretske, Fred (1979). Simple Seeing, in: F. Dretske (2000), Perception,
Knowledge, and Belief: Selected Essays. Cambridge: Cambridge University Press.
(1981). Knowledge and the Flow of Information. Cambridge (Mass.):
The MIT Press.
(1986). Misrepresentation, in: S. Stich & T. Warfield (eds.) (1994),
Mental Representation. Oxford: Blackwell.
(1995). Naturalizing the Mind. Cambridge (Mass.): The MIT Press.
Eco, Umberto (1994). Reflections on The Name of the Rose. London:
Minerva.
Evans, Gareth (1982). The Varieties of Reference. Oxford: Clarendon
Press.
Fish, William (2009). Perception, Hallucination, and Illusion. Oxford: Oxford University Press.
Fodor, Jerry (1990). A Theory of Content, in his: A Theory of Content
and Other Essays. Cambridge (Mass.): The MIT Press.
Frege, Gottlieb (1884/1980). The Foundations of Arithmetic. Evanston:
Northwestern University Press.
(1892). ber Sinn und Bedeutung, in: Zeitschrift fr Philosophie und
Philsophische Kritik, NF 100, pp. 25-50.
(1918), Der Gedanke. Eine logische Untersuchung, in: Beitrge zum
deutschen Idealismus 2, pp. 58-77.
Furlong, E. J. (1961). Imagination. London: George Allen & Unwin.
Gendler, Tamar (2000a). Thought Experiment: On the Powers and Limits
of Imaginary Cases, New York: Garland Press.
(2000b). The Puzzle of Imaginative Resistance, in: Journal of Philosophy, vol. 97, pp. 55-81.
(2006). Imaginative Resistance Revisited, in: S. Nichols (ed.), The Architecture of the Imagination: New Essays on Pretence, Possibility, and
Fiction. Oxford: Oxford University Press.
(2011). Imagination, in: E. Zalta (ed.), The Stanford Encyclopedia of
Philosophy (Fall 2011 Edition), http://plato.stanford.edu/archives/
fall2011/entries/imagination/.
Gendler, Tamar & Hawthorne, John (2002). Conceivability and Possibility.
Oxford: Oxford University Press.
Giaquinto, Marcus (1992). Visualizing as a Means of Geometrical Discovery, in: Mind and Language, vol. 7, pp. 382-401.

Bibliography

447

Ginet, Carl (1990). On Action. Cambridge: Cambridge University Press.


Goldie, Peter (2000). The Emotions. Oxford: Oxford University Press.
(2002). Imagination and Emotion in Fiction, in: M. Kieran and D.
Lopes (eds.), Imagination, Philosophy and the Arts, London: Routledge.
Goldman, Alvin (2006). Simulating Minds. Oxford: Oxford University
Press.
Gordon, Robert (1987). The Structure of Emotions. Cambridge: Cambridge
University Press.
Harris, Paul (2000). The Work of the Imagination. Oxford: Blackwell.
Hawley, Katherine & Macpherson, Fiona (2011). The Admissible Contents
of Experience. Oxford: Wiley-Blackwell.
Heal, Jane (2003). Mind, Reason and Imagination. Cambridge: Cambridge
University Press.
Hobbes, Thomas (1651/2010). Leviathan: Revised Edition. Peterborough:
Broadview Press.
Hjort, Mette & Laver, Susan (eds.) (1997). Emotion and the Arts. Oxford:
Oxford University Press.
Hopkins, Robert (1998). Picture, Image and Experience. Cambridge: Cambridge University Press.
(2011a). Imagining the Past, in: F. Dorsch & F. Macpherson (eds.),
Perceptual Memory and Perceptual Imagination. Manuscript submitted
for publication.
(2011b). Imagination and Observation. Manuscript submitted for publication.
(2011c). What Perky Did Not Show. Manuscript submitted for
publication.
Hornsby, Jennifer (1980). Actions. London: Routledge.
(2008). A Disjunctive Conception of Acting for Reasons, in: A. Haddock & F. Macpherson (eds.), Disjunctivism: Perception, Action, Knowledge. Oxford: Oxford University Press.
Hume, David (1739/2007). A Treatise of Human Nature: a Critical Edition.
Oxford: Clarendon Press.
Husserl, Edmund (1901/2009). Logische Untersuchungen. Hamburg:
Meiner.
(1905/2006). Hauptstcke aus der Phnomenologie und Theorie der
Erkenntnis: 3. Phantasie und Bildbewusstsein, in: Husserl (2006).
(1912/2006). Reproduktion und Bildbewusstsein, in: Husserl (2006).
(1918/2006). Zur Lehre von den Anschauungen und ihren Modi, in:
Husserl (2006).

448

Bibliography

(2006). Phantasie und Bildbewusstsein. Hamburg: Meiner.


Ishiguro, Hide (1966). Imagination, in: A. Montefiore & B. Williams
(eds.), British Analytic Philosophy. London: Routledge and Kegan Paul.
James, Williams (1890/1981). The Principles of Psychology. Cambridge
(Mass.): Harvard University Press.
Johnson, Mark (1993). Moral Imagination. Chicago: University of Chicago Press.
Kant, Immanuel (1781/2010). Kritik der reinen Vernunft. Hamburg:
Meiner.
Kearney, Richard (1988). The Wake of Imagination. London: Hutchinson.
(1991). Poetics of Imagining. London: Harper-Collins.
Kind, Amy (2001). Putting the Image Back in Imagination, in: Philosophy and Phenomenological Research, vol. 62, pp. 85-109.
(2011). The Puzzle of Imaginative Desire, in: Australasian Journal of
Philosophy, vol. 89, pp. 421-439.
Kosslyn, Stephen (1980). Image and Mind. Cambridge (Mass.): Harvard
University Press.
Levinson, Jerrold (1997). Emotion in Response to Art: A Survey of the
Terrain, in: M. Hjort & S. Laver (1997).
(1998). Wollheim on Pictorial Representation, in: The Journal of Aesthetics and Art Criticism, vol. 56 (3), pp. 227-233.
Lewis, David (1980). Veridical Hallucination and Prosthetic Vision, in
his (1986): Philosophical Papers: Vol. 2. Oxford: Oxford University
Press.
Lowe, E. J. (2000). An Introduction to the Philosophy of Mind. Cambridge:
Cambridge University Press.
Macpherson, Fiona (2010). Impossible Figures, in: E. B. Goldstein (ed.),
Encyclopedia of Perception. Sage Publications.
Macpherson, Fiona & Nida-Rmelin, Martine & Dorsch, Fabian (eds.)
(2012). Phenomenal Presence. Oxford: Oxford University Press.
Marbach, Eduard (2006). Einleitung, in: Husserl (2006).
Martin, M. G. F. (1992). The Rational Role of Experience, in: Proceedings of the Aristotelian Society, vol. 93, pp. 71-88.
(1994). Perceptual Content, in: S. Guttenplan (ed.), A Companion to
the Philosophy of Mind. Oxford: Blackwell.
(1998). An Eye Directed Outward, in: C. Wright, B. Smith & C. Macdonald (eds.), Knowing Our Own Minds. Oxford: Oxford University
Press.
(2001). Out of the Past: Episodic Recall as Retained Acquaintance, in:

Bibliography

449

C. Hoerl & T. McCormack (eds.), Time and Memory. Oxford: Clarendon


Press.
(2002a). The Transparency of Experience, in: Mind and Language,
vol. 17, pp. 376-425.
(2002b). Particular Thoughts and Singular Thought, in: A. OHear
(ed.), Logic, Thought and Language. Cambridge University Press: Cambridge.
(2004). The Limits of Self-Awareness, in: Philosophical Studies, vol.
120, pp. 37-89.
(2006). On Being Alienated, in: T. S. Gendler & J. Hawthorne (eds.),
Perceptual Experience. Oxford: Oxford University Press.
(2010). Whats in a Look?, in: B. Nanay (ed.), Perceiving the World.
New York: Oxford University Press.
McCann, Hugh (1998). The Works of Agency. Ithaca: Cornell University
Press.
McDowell, John (1977). On the Sense and Reference of a Proper Name,
in: Mind, vol. 86, pp. 159-185.
(1994). Mind and World. Cambridge (Mass.): Harvard University Press.
McGinn, Colin (2004). Mindsight. Cambridge (Mass.): Harvard University
Press.
Meinong, Alexius (1904). ber Gegenstandstheorie, in: A. Meinong
(ed.), Untersuchungen zur Gegenstandstheorie und Psychologie.
Leipzig: Barth.
Millar, Alan (2010). Knowledge and Recognition, in: D. Pritchard, A.
Millar & A. Haddock, The Nature and Value of Knowledge: Three Investigations. Oxford: Oxford University Press.
Millikan, Ruth (1989). Biosemantics, in: Journal of Philosophy, vol. 86,
pp. 281-297.
Moran, Richard (1994). The Expression of Feeling in Imagination, in:
Philosophical Review, vol. 103, pp. 75-106.
Mulligan, Kevin (1999). La Variet et lUnit dellImmaginazione, in:
Rivista di Estetica, vol. 11, pp. 53-67.
Nichols, Shaun (2004a). Imagining and Believing: The Promise of a
Single Code, in: Journal of Aesthetics and Art Criticism, vol. 62, pp.
129-139.
(2004b). Review of Currie & Ravenscrofts Recreative Minds, in:
Mind, vol. 113, pp. 329-334.
(2006). Introduction, in: S. Nichols (ed.), The Architecture of the Imagination. New Essays on Pretence, Possibility, and Fiction. Oxford: Ox-

450

Bibliography

ford University Press.


Nichols, Shaun & Stich, Stephen (2003). Mindreading: An Integrated Account of Pretense, Self-Awareness and Understanding Other Minds. Oxford: Oxford University Press.
Noordhof, Paul (2001). Getting Personal: Pietroskis Dualism, in: E-Symposium on Causing Actions by Paul M. Pietroski, http://www.uniroma3.it/kant/pietroskisymp_ noordhof.htm.
(2002). Imagining Objects and Imagining Experiences, in: Mind and
Language. vol. 17, pp. 426-455.
OBrien, Lucy (2007). Self-Knowing Agents. Oxford: Oxford University
Press.
OBrien, Lucy & Soteriou, Matthew (2009). Introduction, in: L. OBrien
& M. Soteriou (eds.), Mental Actions. Oxford: Oxford University Press.
OShaughnessy, Brian (1980). The Will. Cambridge: Cambridge University
Press.
(2000). Consciousness and the World. Oxford: Oxford University Press.
(2002). Dreaming, in: Inquiry, vol. 45, pp. 399-432.
Owens, David (2000). Reason Without Freedom. London: Routledge.
Papineau, David (1993). Philosophical Naturalism. Oxford: Blackwell.
(1999). Normativity and Judgement, in: Proceedings of the Aristotelian Society, suppl. vol. 73, pp. 16-43.
Peacocke, Christopher (1983). Sense and Content: Experience, Thought,
and Their Relations. Oxford: Oxford University Press.
(1985). Imagination, Experience and Possibility, in: J. Foster & H.
Robinson (eds.), Essays on Berkeley. Oxford: Clarendon Press.
(1987). Depiction, in Philosophical Review, vol. 96, pp. 383-410.
(1992). A Study of Concepts. Cambridge (Mass.): The MIT Press.
(1998). Conscious Attitudes, Attention, and Self-Knowledge, in: C.
Wright, B. Smith & C. Macdonald (eds.), Knowing Our Own Minds.
Oxford: Oxford University Press.
(2003). Action: Awareness, Ownership, and Knowledge, in: J.
Roessler & N. Eilan (eds.), Agency and Self-Awareness. Oxford: Clarendon Press.
(2008). Truly Understood. Oxford: Oxford University Press.
Pietroski, Paul (2000). Causing Actions. Oxford: Oxford University Press.
Pink, Thomas (1996). The Psychology of Freedom. Cambridge: Cambridge
University Press.
Pollock, John & Cruz, Joseph (1999). Contemporary Theories of Knowledge. Lanham: Rowman & Littlefield.

Bibliography

451

Priest, Graham (2008). An Introduction to Non-Classical Logic: From If to


Is. Second Edition. Cambridge: Cambridge University Press.
Pryor, James (2001). Highlights of Recent Epistemology, in: British
Journal for the Philosophy of Science, vol. 52, pp. 95-124.
Putnam, Hilary (1975). The Meaning of Meaning, in his: Philosophical
Papers, Vol. II: Mind, Language, and Reality. Cambridge: Cambridge
University Press.
Pylyshyn, Zenon (2002). Mental Imagery: In Search of a Theory, in: Behavioral and Brain Sciences, vol. 25, pp. 157-238.
Radford, Colin (1975). How Can We Be Moved by the Fate of Anna
Karenina, in: Proceedings of the Aristotelian Society, suppl. vol. 49, pp.
67-80.
Reicher, Maria (2010). Nonexistent Objects, in: E. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Fall 2010 Edition), http://plato.stanford.edu/archives/fall2010/entries/nonexistent-objects/.
Roessler, Johannes (2001). Understanding Delusions of Alien Control, in:
Philosophy, Psychiatry and Psychology, vol. 8, pp. 177-187.
(2004). Review of Brian OShaughnessys Consciousness and the
World, in: British Journal for the Philosophy of Science, vol. 55, pp.
163-173.
Routley, Richard S. (1982). On What There Isnt, in: Philosophy and
Phenomenological Research, vol. 43, pp. 151-78.
Ryle, Gilbert (1949/1963). The Concept of Mind. London: Penguin.
Russell, Bertrand (1905). On Denoting, in: Mind, vol. 14, pp. 479-493.
Russow, Lilly-Marlene (1978). Some Recent Work on Imagination, in:
American Philosophical Quarterly, vol. 15 (1), pp. 57-66.
Salmon, Nathan (1998). Nonexistence, in: Nous, vol. 32, pp. 277-319.
Sartre, Jean-Paul (1936/1997). Die Imagination. In his: Die Transzendenz
des Ego. Philosophische Essays 1931-1939. Reinbek: Rowohlt.
(1940/2004). The Imaginary: A Phenomenological Psychology of the
Imagination. London: Routledge.
Scruton, Roger (1974). Art and Imagination. London: Methuen & Co.
Searle, John (1983). Intentionality. Cambridge: Cambridge University
Press.
Sellars, Wilfrid (1978). The Role of the Imagination in Kants Theory of
Experience, in: H. Johnstone (ed.), Categories: A Colloquium. University Park: Pennsylvania State University Press.
Shorter, J. M. (1952). Imagination, in: O. Wood & G. Pitcher (eds.)
(1970), Ryle. London: Macmillan.

452

Bibliography

Siegel, Susanna (2010). The Contents of Visual Experience. Oxford: Oxford University Press.
Siewert, Charles (2000). The Significance of Consciousness. Princeton:
Princeton University Press.
Soames, Scott (1999). Understanding Truth. New York: Oxford University
Press.
Soldati Gianfranco (1994). Bedeutung und psychischer Gehalt. Zur sprachanalytischen Kritik von Husserls frher Phnomenologie. Paderborn:
Schningh.
(2006). Review of Currie & Ravenscrofts The Recreative Mind, in:
European Journal of Philosophy, vol. 14, pp. 448-452.
Soldati, Gianfranco & Dorsch, Fabian (2005) Conceptual Qualia and
Communication, in: Interaction Design Institute Ivrea (ed.), The
Foundations of Interaction Design, pp. 1-14.
(2011). Intentionalism, Phenomenal Error and Experiential Error. Manuscript.
Sorensen, Roy (1992). Thought Experiments. Oxford: Oxford University
Press.
Stock, Kathleen (2008). The Role of Imagining in Seeing-In, in: Journal
of Aesthetics and Art Criticism, vol. 66, pp. 365-380.
Strawson, Galen (1994). Mental Reality. Cambridge (Mass.): MIT Press.
(2003). Mental Ballistics or The Involuntariness of Spontaneity, in:
Proceedings of the Aristotelian Society, vol. 103, pp. 227-256.
Strawson, Peter (1970). Imagination and Perception, in: L. Foster and J.
W. Swanson (eds.), Experience and Theory. London: Duckworth.
Stroud, Barry (1977). Hume. London: Routledge.
Thomas, Nigel J. T. (1999). Are Theories of Imagery Theories of Imagination?, in: Cognitive Science, vol. 23, pp. 207-245.
(2010). Mental Imagery, in: E. Zalta (ed.), The Stanford Encyclopedia
of Philosophy (Winter 2010 Edition), http://plato.stanford.edu/archives/
win2011/entries/mental-imagery/.
Travis, Charles (2004). The Silence of the Senses,in: Mind, vol. 113, pp.
57-94.
Tye, Michael (1991). The Imagery Debate. Cambridge (Mass.): MIT Press.
(2000). Consciousness, Color, and Content. Cambridge (Mass.): MIT
Press.
Velleman, David (2000). The Possibility of Practical Reason. Oxford: Oxford University Press.
(2006). Self to Self, in his: Self to Self: Selected Essays. Cambridge:

Bibliography

453

Cambridge University Press.


Voltolini, Alberto (2006). How Ficta Follow Fiction. Dordrecht: Springer.
Walton, Kendall (1990). Mimesis as Make-Believe. Cambridge (Mass.):
Harvard University Press.
(1997). Spelunking, Simulation, and Slime: On Being Moved by Fiction, in: Hjort & Laver (1997).
(2002). Depiction, Perception, and Imagination: Responses to Richard
Wollheim, in: The Journal of Aesthetics and Art Criticism, vol. 60, pp.
27-35.
Warnock, Mary (1976). Imagination. London: Faber and Faber.
Weatherson, Brian (2004). Morality, Fiction, and Possibility, in: Philosophers Imprint, vol. 4, pp. 1-27.
Weinberg, Jonathan & Meskin, Aaron (2006). Puzzling Over the Imagination: Philosophical Problems, Architectural Solutions, in: S. Nichols
(ed.), The Architecture of the Imagination: New Essays on Pretence,
Possibility, and Fiction. Oxford: Oxford University Press.
White, Alan (1990). The Language of Imagination. Oxford: Blackwell.
Williams, Bernard (1966). Imagination and the Self, in: B. Williams
(1973), Problems of the Self. Cambridge: Cambridge University Press.
(1970) Deciding to Believe, in: B. Williams (1973), Problems of the
Self. Cambridge: Cambridge University Press.
Williamson, Timothy (2000). Knowledge and its Limits. Oxford: Oxford
University Press.
(2008). The Philosophy of Philosophy. Oxford: Blackwell.
Wilson, George (1989). The Intentionality of Human Action. Stanford:
Stanford University Press.
Wittgenstein, Ludwig (1984a). Philosophische Untersuchungen. Werkausgabe Band 1. Frankfurt: Suhrkamp.
(1984b). Das Blaue Buch. Werkausgabe Band 5. Frankfurt: Suhrkamp.
(1984c). Bemerkungen ber die Philosophie der Psychologie. Werkausgabe Band 7. Frankfurt: Suhrkamp.
(1984d). Zettel. Werkausgabe Band 8. Frankfurt: Suhrkamp.
Wollheim, Richard (1973). Imagination and Identification, in his: On Art
and the Mind. Cambridge (Mass.): Harvard University Press.
(1984). The Thread of Life. New Haven: Yale University Press.
(1987). Painting as an Art. Princeton: Princeton University Press.
(2003). What Makes Representational Painting Truly Visual?, in: Proceedings of the Aristotelian Society, suppl. vol. 78, pp. 131-147.
Yablo, Stephen (1993). Is Conceivability a Guide to Possibility?, in:

454

Bibliography

Philosophy and Phenomenological Research, vol. 53. pp. 1-42.


Index

Index

acquaintance 19, 361, 363; see also: mental representation


(ACT) 390-1, 395-7, 409-10, 413-5, 417-8, 420, 423, 433, 441; see also:
Agency Account
action see: agency
actualities 12, 161-2
actual world/situation 61-2, 64, 66-7, 75-6, 83-4, 115-6, 161-70, 172, 174,
246, 260, 297, 317, 341, 348, 372, 434, 438; see also: external
world; reality
aesthetic appreciation 2, 14, 46, 49-50, 52-3, 64, 270, 338, 340-1, 356,
387, 440-1; see also: art
affective imagining 4, 17, 27, 36, 41-2, 81, 256, 259-61, 263, 265, 279,
292-3, 295, 298, 301, 306, 308, 337-64 passim, 370, 373, 382, 413,
420-3, 435; see also: bodily sensation; emotional imagining
agency 101-3, 138-9, 180-1, 184-8, 381-7, 396-407; see also: intention;
motivation; volition
bodily 121, 374-6, 386-7, 399-401
involving control 13-4, 19, 63, 119, 140, 142-6, 157, 180-1, 200, 243,
249, 264, 269, 272, 357, 364, 382, 386, 388-9, 404-7, 414, 432,
437, 439
direct 10, 19, 52, 112, 185, 188, 264, 357-8, 385, 389-90, 392-8, 4017, 411-5, 417-8, 420-1, 423, 426, 431-3, 436, 439, 441
mental 2, 8, 10-1, 17, 19, 31, 93, 105, 110, 116, 133, 176, 179-181,
183-8, 192, 249-250, 262, 264, 271-2, 357, 378, 381-407 passim,
410-1, 413-7, 426, 431-4, 438, 441; see also: Agency Account

456

Index

agency (contd.)
non-mental 179, 386-7
sub-intentional 93, 111, 143, 249-50, 392, 401, 424, 440
voluntary 2, 8, 10-1, 13-4, 63, 101-2, 112, 138-44, 146, 157, 178-82,
184-7, 199-200, 210, 247, 357-8, 364, 372, 381-2, 385, 388-90,
392, 394, 397, 399, 404-6, 413, 431-3, 437-40
Agency Account 2, 8, 10-1, 13, 17, 19-20, 24, 30-1, 55, 83, 85, 104-6,
111-2, 139, 143, 146, 149-50, 176, 188, 192, 247, 262, 271-2, 279,
364, 370, 378-9, 381-429 passim, 431-4, 436-7, 439-41; see also:
(ACT); Constitutive View of voluntary imaginings
Alain 140
anaesthesia see: states of consciousness
anticipation 51, 55-7, 354, 374, 440
anxiety 260, 421, 432
Argument from Attention 214, 218, 221, 247
Argument from Origin 200-1, 214, 216, 219, 221-2, 242, 247, 251, 257,
268, 271, 378
Armstrong, D. 54, 394
art 2, 46, 53, 328, 339-41, 343, 345-7, 349-51, 355-6, 360, 363-4, 440-1;
see also: aesthetic appreciation
assertion 35, 63, 72, 375
association 11, 93, 104, 110-1, 140, 145, 179-80, 184, 227, 249, 266, 281,
393, 405, 419, 424, 428, 432, 439
assuming 42-3, 64, 68
attention 20, 39-40, 47, 52-3, 72, 91-3, 124-5, 138, 140-1, 174, 182, 195,
197, 214, 218, 221-5, 228, 247-51, 318, 323, 349, 376-7, 390, 392,
404, 425, 433, 439-40; see also: Argument from Attention
object-directed 92, 178, 222
attitude/commitment 12-3, 18, 23, 32-4, 59-68, 73, 75, 83-4, 87, 129, 143,
149-52, 154, 156-8, 161-78 passim, 184-5, 187, 206, 224, 243, 245,
254, 296-7, 303, 307, 333-4, 351, 368, 377, 407, 412, 416, 422,
437-8, 440
conative see: attitude, practical
cognitive 13, 17, 54, 59, 62-4, 72-3, 80, 83, 128-9, 144, 163, 166,
169-70, 172-4, 238, 245, 258, 260, 267-8, 271, 290, 294, 297, 305,

Index

457

309, 332, 375, 396, 410, 412-3, 424-5, 428, 436


attitude/commitment (contd.)
denying actuality 13, 51, 67, 161, 163, 170-5
denying existence 13, 170-5, 428
denying presence 13, 65, 171-5, 184, 186
imaginative 13, 64-5, 74-5, 159, 163-4, 166, 169-175, 303, 306, 377,
438
neutralising/suspending 171-5
practical 35, 67-8, 71, 111, 118-9, 122, 184-5, 187, 338, 385, 388, 415
theoretical 10, 12-3, 61-70, 75, 84, 90, 93, 99, 111-2, 116, 144, 155,
157, 162-4, 175-6, 185, 187, 258, 310, 338, 388, 416
Attitude Account see: imagining, accounts of
Audi, R. 395, 400-401
auditory awareness see: sense modalities
autism 15, 331
awareness see: consciousness; mental representation
awareness of the outer world 234-42; see also: perceptual contact with
reality; rational integration
Bayne, T. 55
belief 5, 8-10, 15, 24, 30-1, 33-5, 42-3, 46, 52-3, 55-56, 60-1, 63-6, 6973, 76, 89-91, 93, 115-8, 121-4, 126, 128-30, 136, 144, 149, 162,
170, 172-4, 194-5, 199, 202, 206-7, 210, 214-9, 221, 223, 228-34,
236-47, 256-8, 260, 262, 267-71, 275-7, 282, 284, 289-91, 293-5,
297-8, 300, 302-10, 316, 318, 337, 339, 341, 343-7, 356, 358-9,
366, 369-74, 376, 378, 386-9, 410-3, 415-6, 419, 422, 428, 435-6,
439-40
Berg, A. 73, 75, 229
Berkeley, G. 37, 82
blind people 367
Block, N. 47
blur 85, 318
bodily sensation 10, 37, 41, 47, 81, 85, 180, 186, 194, 224-8, 249, 259-60,
266, 275-6, 281-2, 293, 297, 314, 322, 337, 346, 353, 361, 438; see
also: sense modalities

458

Index

real vs. imagined itches 4, 41, 50, 61, 306, 315, 318, 322-3, 361, 4201, 423
bodily sensation (contd.)
real vs. imagined pain 36, 41, 82, 85, 150, 292, 315, 322, 338, 361-2,
386, 421-3, 432
Bodrozic, D. 154, 156-157
Braddon-Mitchell, D. 71
brain see: neural/neurofunctional
Budd, M. 5, 10, 45, 48, 59, 73, 82, 143, 145, 180, 270, 277, 382, 394
Burge, T. 48, 69, 71, 294, 325-6, 329-30, 366
Campbell, J. 316, 324
Carroll, N. 392
Carruthers, P. 41, 47, 370
Casey, E. 1, 3-4, 12-13, 20, 37, 42, 46, 73, 80, 85-7, 139, 152, 155-6, 164,
172-3, 177-8, 180-3, 185-6, 188, 269, 405, 424, 426-8
Causal Account see: imagining, accounts of
Causal View of voluntary imaginings 398-403
chess 261, 387
children 44, 330-1, 374
cognising episodes/states 3, 12, 18, 24, 60, 71, 73, 81, 115-8, 122-5, 1279, 131-4, 136-7, 141, 144, 146, 154-5, 195, 199-200, 202, 205-6,
211-6, 221-2, 226, 228-32, 234, 236, 239-41, 244-5, 257-65, 267-8,
270-2, 293, 305, 311, 378, 381-2, 396, 410-4, 428, 432, 434-6, 438;
see also: cognition; knowledge; perception; recognition
cognition 8-9, 32, 56, 59, 62, 70-2, 84, 103, 115, 128-9, 149, 154, 187,
195, 199, 203, 207, 245, 259-62, 296, 298, 300-1, 428, 433, 437;
see also: cognising episodes/states; perception; knowledge; recognition
cognitions 8-10, 12-3, 18-20, 23-4, 27, 30, 34-5, 56, 59-64, 68-73, 80, 845, 115, 124, 128-9, 134, 143, 146, 149, 151-7, 159-63, 165-7, 16970, 172, 175, 177, 187-8, 192-3, 195, 199, 201, 208, 211, 213-5,
218-20, 222-3, 231, 241-5, 247, 254-5, 258-9, 262, 265-8, 272,
275-9, 288, 293-302, 306, 308, 310, 365-73, 375-7, 381, 389, 407,
410-3, 424, 426-9, 433-5, 437-8; see also: belief; judgement; per-

Index

459

ceptual experience; memory


cognitive commitment see: attitude
cognitive constraints 8, 40, 44, 99, 115-6, 138, 141-6, 187, 195, 199-201,
214, 217, 219, 221-42 passim, 245-6, 256, 258, 267-8, 272, 328-9,
332, 392, 416; see also: awareness of the outer world; mental
representation; perceptual contact with reality; rational integration;
reliability; veridicality
cognitive episodes/states see: cognitions
cognitive ideal see: cognitive prototype
cognitive penetration 105, 110
cognitive projects 3, 35, 59-60, 76, 145, 192, 261-3, 266, 269-70, 272,
310, 377, 385, 414, 416-7
cognitive prototypes 127-8, 198-209, 211-9, 221-3, 226, 228-9, 232-3,
240, 242, 244, 246-7, 257-8, 267-8, 300, 304, 308
cognitive role see: epistemic function
coherence see: rational integration
Collingwood, R. 1, 3, 81-3, 272, 382-3, 405, 423
colour 15-7, 62, 67-8, 82, 90, 95, 102, 110, 124, 134, 136, 159, 176, 2258, 248-9, 285-6, 289, 293, 299, 309, 346, 366-7, 388-9, 414
commitment see: attitude
Conan-Doyle, A. 52, 68
conceivability 3, 162
conceptualisation see: mental representation
consciousness 16, 19, 30, 35, 41, 44, 47, 49, 51-2, 54, 60, 63, 69, 71, 88,
92-3, 100-2, 111, 113, 125, 140-1, 151, 155, 165, 172-3, 177-9,
183-7, 197-8, 200, 207, 218, 223, 225, 230, 233-5, 237-9, 241, 247,
249-50, 260, 283-4, 290, 302-3, 308, 386-7, 389, 394, 399, 407,
410, 425; see also: phenomenal character/aspects; phenomenological salience
consistency see: rational integration
constituting knowledge see: cognising episodes/states
Constitutive View of voluntary imaginings 398-403
content 4, 10, 12, 19, 33-4, 37-42, 44, 48, 52, 59, 61-3, 74, 84, 87, 93, 95,
97-8, 103-4, 106-7, 111, 113, 118, 120-1, 134-5, 139-42, 144, 146,
157, 161-2, 171, 173, 175, 180, 182, 206-8, 225-6, 235, 240, 245,

460

Index

248-9, 264, 275, 277, 285, 287, 293, 295-300, 303-7, 309, 334-5,
339-40, 343, 347, 351, 353-4, 356-61, 365, 368, 370-1, 388-90,
393, 395, 397-8, 404-5, 411-2, 414-8, 420, 422, 425-7, 432, 439,
441; see also: mental representation
conventions 7, 328, 346-7, 349
copies see: dependency of imaginings on cognitions
counterfactuals 3, 66, 310, 369, 440
counterparts see: echo claims
Crane, T. 33, 62, 96, 168, 260, 421
creativity 3, 24, 32, 46-7, 64, 177-8, 184, 307, 338, 383; see also: imaginativeness
Crowther, T. 182
Cruz, J. 69-70, 240, 256, 305
Currie, G. 3-4, 9, 15, 20, 37, 41-4, 46, 59, 84, 276-7, 304, 326, 330, 3345, 337, 353, 366, 368-71, 374-5, 395
Dancy, J. 270, 399
Danto, A. 395
Davidson, D. 240
daydreaming see: imagining, forms of
delusions 15, 54-6, 216, 378, 440
Dennett, D. 47
dependency of imaginings on cognitions; see also: echo claims
asymmetric 9, 370-2
causal (EC) 284-5, 288, 292-4, 296, 298, 365, 368; see also: imagining, accounts of
conceptual 369-71, 373
constitutive 272, 301, 369-71, 373
copies/reproductions 80, 84, 87, 159, 206, 208, 279, 282, 285, 287-9,
291-3, 298, 300, 302, 410; see also: Humes Copy Principle; pictures
imitational (EI) 9, 19-20, 205, 275-7, 280, 365, 368-9, 373, 375-7,
436; see also: imagining, accounts of
representational (ER) 84, 193, 197, 201-2, 205-6, 208-10, 212-3, 216,
220, 247, 279-80, 288, 292, 294-311, 313-6, 322, 325-8, 330-1,

Index

461

334-5, 337, 362, 364-5, 370, 377; see also: imagining, accounts of
dependency of imaginings on cognitions (contd.)
semantic (ES) 9, 146, 275-6, 280, 366-71, 393, 411; see also: imagining, accounts of
types vs. tokens 277-8, 286-7, 298-301
Dependency Account 9-11, 18-9, 30, 84, 149, 159, 188, 192-3, 197, 2002, 207, 213, 220, 255, 272-3, 275-80, 282, 295, 377-8, 381, 431,
433-7; see also: dependency of imaginings on cognitions; echo
claims
depiction see: pictures
Descartes, R. 240
desire 5-6, 9-10, 33-35, 41, 44, 52, 56, 67, 71, 73-4, 133, 146, 163, 209,
227, 260-1, 282, 293-294, 297, 303-4, 307-308, 337, 359, 370-372,
389, 405, 411, 415-6, 420, 425; see also: intention; mental representation; volition
Devil 168
direct agency see: agency
direct object/reference see: mental representation
disjunctivism 24, 33, 47-9, 89-90, 153, 258, 263, 305, 314, 334-5, 399,
435
disorders, mental see: states of consciousness
dispassionate imagining 266, 339, 351, 353, 356, 358-9, 361
dramatic imagining see: imagining, forms of
dreams see: states of consciousness
Dretske, F. 33, 71, 82, 95, 120-121, 128-129, 134, 153, 223, 233, 248
drugs see: intoxication
(EC) see: dependency of imaginings on cognitions
echo claims 34, 77, 84, 159, 171, 192-3, 197-8, 202, 205, 208-9, 212-3,
216-20, 255, 275-80, 288, 292-5, 298-9, 301-4, 307-10, 365-6, 3689, 371, 377, 422; see also: Dependency Account; dependency of
imaginings on cognitions
idea of counterparts 4, 9, 12, 19, 27, 37, 41, 59, 73, 80-1, 133, 149,
160, 173, 193, 195, 200, 204, 216, 243-4, 257, 260-1, 265, 272,
275-9, 288, 291, 293, 295, 300, 337, 365, 367, 369-71, 373, 377,

462

Index

413-4, 429, 432, 435


echo claims (contd.)
idea of dependency 276-7, 369
idea of inheritance 84, 277, 289, 295, 297, 333-4
OShaughnessys echo claim 201-2, 205-9, 212-3, 216-20, 255, 276
Eco, U. 43, 392
egocentric orientation see: perspectivalness
(EI) see: dependency of imaginings on cognitions
emotional imagining 19, 301, 337-64 passim, 419, 421;
see also: affective imagining
emotion 2-3, 6, 10, 15, 33, 35, 41-2, 46, 52, 63, 74, 187, 227, 260, 266,
270, 275, 282, 284, 289, 294, 297, 301, 309, 337-46, 349-64, 371,
374, 405, 415, 418, 434, 436, 441; see also: emotional imagining;
mental representation; moods; quasi-emotion
empathy 5-6, 15, 46-7, 271, 338, 353-5, 440
empirical evidence 14-7, 41, 43, 66, 131-2, 151, 179, 242, 310, 331, 338,
353, 367, 406-7, 439
empiricism 82, 241, 281, 298
entertaining a proposition 34, 37, 43, 167, 171, 270, 276, 295, 303, 305,
307-9, 353, 374, 376-7, 415-6, 425
entitlement see: justification
episodic memory see: memory
epistemic function/role 17-8, 23, 60, 62, 68-74, 81, 115, 129-30, 149-50,
152, 169, 187, 194, 228, 269, 272, 290-1, 296-7, 411-2, 424, 434,
436-7
Epistemological Account 8, 18-9, 30, 116, 149, 176, 187, 189, 191-4,
196-8, 200, 202, 204, 206, 208, 210, 212-4, 216, 218, 220, 222,
224, 226, 228, 230, 232, 234, 236, 238, 240, 242, 244, 246-8, 2501, 253-4, 256, 258, 260, 262, 264, 266, 268, 270-2, 276, 378, 396,
434-7
(ER) see: dependency of imaginings on cognitions
(ES) see: dependency of imaginings on cognitions
Escher, M. 168
evaluations 2, 27, 44-5, 51, 89, 144, 146, 149, 167, 181, 191, 195, 202,
248-9, 338-41, 354, 394, 441

Index
Evans, G. 134, 146, 258
evolution 71-2, 120, 129, 406, 411
examples
calculating a sum in ones head 130, 261, 386
chiliagon 406
Columbus 419
counting windows 3, 124, 393
Devils tuning fork 168
elephant in the city 64-6
field of daffodils 247-8
frogs and their lips 3, 125, 137, 141-2
Hamlet 341
hearing voices 54
Hercules 328
Hertford College 94-9, 103-4, 106-7, 411
imposter 95, 374
Jean of Arc 328
Macbeth 6, 155, 216, 244-5, 425
missing shade of blue 290, 299, 366
Mller-Lyer illusion 110
Napoleon 44, 46-7, 50, 287, 332, 417
Odysseus 154, 162
old friend 116-33 passim, 144-5, 267, 269
Pantheon 140-1, 249-50
Pegasus 176
Perkys experiments 151
Pierre 102, 179, 186
pink elephant 224-5, 232
Rambutan 367
Rome in snow 35-6, 41-2, 44, 46, 141, 391
Saint John 328
Sherlock Holmes 52-3, 68, 156
sofa in the shop 116-33 passim, 144-5, 267, 269, 414
suitcase with a cat behind it 97, 99, 328
Twin Earth 366

463

464

Index

examples (contd.)
Ulysses 107
unicorns 154, 299, 386, 417
Waterfall illusion 168
Wellington 395
wax replicas/fakes 90-1, 108-11, 254, 277
Wozzeck in the Opera Bastille 73-5, 229
expectations 6, 42, 55-7, 163-4, 173, 440
experiential imagining 4, 17, 37, 44-7, 49-50, 84, 97-100, 207, 295, 298,
308, 313-364 passim, 419-20, 422, 432; see also: imagining seeing;
S-imagining
Experiential Rationalism 412
experiments see: empirical evidence
explanatory power 17, 27-9, 106, 108, 210, 253-6, 259, 261, 264-5, 268,
334, 396, 409-10, 432-3, 435, 437
fundamentality 2, 11-3, 15-6, 29-30, 33, 48, 59, 77, 101, 133, 149,
152-4, 156-7, 161, 165, 169, 176, 187, 192-4, 197-8, 200, 202, 204,
207, 210-3, 220, 254-7, 264-5, 268, 271, 275, 293, 295, 302, 304,
345, 365-6, 372, 378, 382, 390-1, 394, 396, 403, 405, 409-10, 4134, 417, 432, 437-8, 441
informativity 20, 28, 254-7, 371, 409-12
extensional adequacy 2, 17, 27-28, 33, 51, 210, 213, 224, 253, 261, 2667, 310, 396, 409, 414-29 passim, 432-3, 435
external world 3, 13, 18, 69, 80, 115-116, 134-5, 141-2, 144-6, 158, 169,
184, 194, 219, 224, 233-5, 237-9, 241, 248, 256, 258-9, 267, 416,
433-4; see also: actual world; reality
factivity see: mental representation
faculty of imagination 31-2, 281
faith 338
fancy/fantasy see: imagining, forms of
fear 33, 42, 55, 164, 342-6, 349-50, 355, 357, 361, 363-4, 374
ficta/fictional objects 41, 154, 340, 342, 344-345, 349-352, 356, 363-364,
437
fiction 6, 15, 36, 41-3, 46, 68, 71, 97, 146, 154, 156-7, 170, 283-4, 338-

Index

465

52, 355-6, 363-4, 410, 416, 437, 441; see also: paradox of fiction
fictional worlds/situations 97, 338, 340-1, 343, 345-9, 352, 356
filling-in 140-3
first-order thought see: thinking
first-personal see: perspectivalness
Fodor, J. 71
foundationalism 240
Frege, G. 162, 416
function, biological/evolutionary 71-2, 120, 411
Furlong, J. 42
games of make-believe see: making-believe
Gendler, T. 3, 15, 42-3, 162, 168, 392
generic memories 93, 137, 141, 145, 393; see also: memory
Gettier, P. 232, 241
Giaquinto, M. 3
Gill, A. 77
Ginet, C. 104, 399
God 416
Goldie, P. 2, 46, 163, 260, 345, 354
Goldman, A. 15, 368
Gordon, R. 309
Goya, F. 395
grounding knowledge see: cognising episodes/states
grouping phenomena see: noticing an aspect
guessing 11, 187, 327, 386, 388
hallucinations see: perceptual experience
Harris, P. 15
Hawley, K. 67
Hawthorne, J. 3, 162
Heal, J. 15, 368
hearing see: sense modalities
higher-order thought see: thinking; introspection
Hjort, M. 42, 343

466

Index

Hobbes, T. 110
hope 33, 51, 55-6, 163-4, 340, 343, 371, 440
Hopkins, R. 1, 3, 5, 12, 37, 39-40, 46-7, 65, 84, 92, 101, 111, 151, 158,
175-6, 201, 250, 269, 277, 287, 305, 307, 309, 320, 348-9, 382,
419, 423
Hornsby, J. 104, 399-400
Hume, D. 4, 8-9, 12, 18, 20, 32-3, 62, 79-82, 150-1, 159, 184, 194, 201,
208, 269, 275, 279, 281-94, 296, 298-302, 333, 436
Humes conception of imagining see: imagining, accounts of
Humes copy principle 281-94 passim, 299, 333
Husserl, E. 1, 13, 20, 33, 47, 62, 64, 79-81, 83-4, 152, 158-9, 169-71,
173-5, 281
hypnosis 231, 234, 236, 240, 389, 397, 415, 417
hypothetical imagining 353; see also: intellectual imagining
illusions see: perceptual experiences
images 3, 6, 11, 19, 34, 38, 44, 54, 64, 74, 77, 79, 82, 86, 88-9, 92, 94, 98102, 112-3, 117-9, 121-5, 128, 131-2, 134, 139-41, 145-6, 172,
175-6, 178-9, 181, 183, 186, 224, 232, 249, 282, 285, 287, 296,
301, 309, 320, 329, 333, 357, 367, 369, 376, 378, 386, 393-4, 3978, 401-2, 405-6, 409, 423-9, 432-3, 437-8
images and thoughts 423-9; see also: images; imagining; mental episodes/states; thinking
difficult to banish 55, 393, 425
fleeting/transient 113, 140, 182, 357, 424-6
hypnagogic 140, 249, 378, 424
instantaneous 35, 140, 181-2, 424
passive/involuntary 13, 82, 102-3, 138-43, 177-82, 184-6, 199-200,
244, 247, 272, 383, 394, 397, 399, 402, 415, 424-5, 427, 431, 438
persisting 140, 175, 243, 246, 429
spontaneous 423-9; see also: spontaneity
surprising/unexpected 113, 181-2, 249, 419, 424
sustainment 103, 113, 177, 244, 357, 425-7, 433
transformation 113, 425-7, 433
imagination as faculty see: faculty of imagination

Index

467

imaginative commitment see: attitude


imaginativeness of persons/idea 24-5, 32, 355; see also: creativity
imaginative perception see: pictorial experience
imaginative projects 3-4, 11, 17-8, 27, 36, 46-7, 64, 66, 68, 77, 98-100,
116, 119, 128, 135-7, 142, 144-5, 150, 165, 192, 250, 259, 261-3,
265, 269, 279, 301, 309-11, 313, 320, 328-9, 353-4, 357, 365, 3778, 382, 396, 410, 413-20, 432-3, 435-6
imagined worlds/situations 61, 64-8, 75-6, 116, 250, 319-21, 325, 328,
330-1, 335, 347; see also: fiction
imagining
active/voluntary 10-1, 112-3, 139, 146, 178, 184-8, 199-200, 211,
246-7, 265, 381-2, 392-4, 396, 402-3, 413, 420, 432-3; see also:
Agency Account; Causal View of voluntary imaginings; Constitutive View of voluntary imaginings
akratic 143, 393-4
borderline cases 6, 27, 31, 49, 51, 401, 440
central/paradigm cases 2, 6-7, 17, 23, 27-9, 31-2, 34-8, 41, 50-1, 53,
55-6, 149, 210, 245, 255, 259, 266-7, 276, 280, 365, 378, 382, 385,
390, 409, 414-5, 417-21, 435
cognising 115-133 passim; see also: cognising episodes/states
compulsive 55, 143, 394, 405, 424, 440
conative 337
concept/definition 31, 202-13 passim, 217-8, 247, 308, 378
concreteness 388-91; see also: mental representation, general subject
matter
difference from cognitions 59-73, see also: mental episodes, difference in attitude/epistemic function
directness see: agency, direct; imagining, immediacy
dispositional 52-3, 376
effort 38, 393
embedded in wider mental projects 43, 64, 116, 145, 182, 269, 295,
298, 392, 414-5
essence/nature 2, 4, 7-8, 11-2, 16, 24, 27-8, 30, 116, 150, 192-4, 197,
199, 200-2, 205-7, 211, 213, 220, 245, 254-8, 261, 263, 265, 269,
275-6, 278-80, 294-5, 304, 306, 368, 371-2, 390, 396-403 passim,

468

Index

406, 409-10, 426, 432-3, 438, 440


imagining (contd.)
immunity to error through misidentification see: immunity to error
through misidentification
immediacy 76, 79-81, 83-5, 87, 127, 156-7, 160-1, 170-1, 208, 251,
287, 291, 295-6, 299, 322-5, 421; see also: mental representation
indeterminacy 68, 85-7, 375
informativity 18, 24, 80-81, 116, 126, 133-47 passim, 194-5, 381; see
also: learning
intellectual element 37, 77, 80, 92, 95-113 passim, 135-6, 139, 157;
see also: intention; S-imagining
limitations 9-10, 46, 160, 328-9, 366-8, 392-3, 434
mental kind see: mental episodes/states, mental kinds
non-central cases 51-7, 378
non-cognitivity 24, 54, 64, 163, 173, 175, 191-4, 213, 253, 260, 262,
265, 271-2, 276, 375, 378, 381, 413, 415, 424, 427-8, 434-6
non-representational 19, 62, 82, 85, 260, 265, 340, 383, 388, 409, 418,
420-3, 432
obsessive 54-55, 143, 393-4, 425, 432-3
object-directed 303, 314, 331, 334, 362-3, 419
objects of 23, 29, 32, 60, 71, 73-7, 94-6, 107, 121, 133, 145-6, 15461, 367, 410, 412; see also: imagining, particularity; imagining,
immediacy
origin in the mind 133, 176-9, 187-8, 193, 200, 222, 243, 428, 438;
see also: Argument from Origin; spontaneity
particularity 77, 95-6, 106-8, 112, 154
passive/involuntary 13, 178, 181, 199-200, 244, 246, 272, 383, 399,
424, 431; see also: images and thoughts
purpose 50, 390, 414-5, 441
quasi-observationality see: quasi-observationality
rationality 43-4, 65-6; see also: attitude, imaginative/theoretical; inference
reference to reality see: imagining, objects of
reliable see: reliability
repleteness 81, 334

Index

469

imagining (contd.)
simplest instances 47, 49-50, 99, 400
simultaneity with perception 38-41, 320-1
specificity in content see: imagining, concreteness
sub-intentional see: agency
surprising 131-2, 249-50
transparent see: transparency
unity 1, 5, 7, 11, 14, 16, 20, 57, 59, 105, 149-50, 152, 154, 167, 174,
176, 185, 294, 310, 359, 382, 431-3, 436, 441; see also: unified accounts of imagining
vagueness 81-2, 86, 406
veridical/correct see: veridicality
imagining, accounts of
Agency Account see: Agency Account
Attitude Account 13, 164, 167-70, 173, 175, 438, 440
Causal Account 18, 279; see also: dependency of imaginings on cognitions
Dependency Account see: Dependency Account; dependency of imaginings on cognitions
Epistemological Account see: Epistemological Account
Humes conception of imagining 18, 81, 279, 281-92, 294
Imitation Account 275-276, 365, 368-77; see also: dependency of
imaginings on cognitions
Modal Account 12, 24, 161-70, 176, 372, 438-9
Ontological Account 12, 24, 76, 154-6, 158-61, 167, 437-8
Origin Account 14
Phenomenal Account 12, 152-5, 169, 437
pictorial model 158-9, 170-1
Quantitative Account 11, 24, 150-1, 437
Representational Account 18, 30, 76, 159-60, 208, 275-313 passim,
370, 436; see also: dependency of imaginings on cognitions
Semantic Account 275, 366-8; see also: dependency of imaginings on
cognitions
Spontaneity Account 13, 24, 176-88, 438
Subjection to the Will Account 13, 18-9, 24, 143, 149, 383, 404-7,

470

Index

439
imagining, accounts of (contd.)
Subpersonal Account 14-8
unified accounts see: unified accounts
imagining, forms of
affective imagining see: affective imagining; emotional imagining
daydreaming 1-2, 4-6, 11, 30, 46, 53, 141, 269, 310, 337, 353, 357,
377, 385, 417-9
dramatic imagining 352-356
experiential imagining see: experiential imagining
fancy 43, 51, 55-6, 283
fantasy 4, 46, 175, 393
imaginative projects see: imaginative projects
imagined pain see: bodily sensation
imagined itches see: bodily sensation
intellectual imagining see: intellectual imagining
reverie 46, 72
sensory imagining see: sensory imagining; visualising
imagining seeing 5, 37, 45, 47, 49-50, 64, 97-8, 100, 207, 266, 298, 300,
313-35 passim, 348, 419-20, 436
imagining that see: intellectual imagining
imitation see: dependency of imaginings on cognitions
Imitation Account see: imagining, accounts of
immediacy see: imagining; mental representation
immunity to error through misidentification 79-81, 88, 94-6, 103, 106-8,
112-3, 131-3, 136, 139, 306, 396, 411-2
impossibilities 68, 168, 170
impressions vs. ideas see: Hume's Copy Principle
inclination see: motivation
indefeasibility see: justification
indeterminacy see: imagining
inference 15, 43, 66, 89, 103, 120, 123, 130, 185, 187, 246, 263, 270-1,
305, 316, 321, 353, 368-71, 376-7, 402, 407, 416, 418; see also: attitude, practical/theoretical
informativity see: explanatory power; new information

Index

471

inheritance see: echo claims


insanity see: states of consciousness
inserted thoughts see: thought insertion
intellectual imagining 3-4, 15, 17-8, 27, 36, 42-5, 52-3, 61, 64-6, 74, 77,
81, 96, 109, 159-60, 162, 165-8, 174, 199, 202, 205, 212, 215, 229,
246, 256-7, 271, 279, 293-5, 298, 301-10, 313, 347-54, 357, 35961, 364-6, 370, 373-4, 376-8, 392, 413, 415-416, 419, 421 435-6;
see also: assuming; dispassionate imagining; hypothetical imagining; make-believing; propositional imagining; S-imagining; supposing
intention 37, 71, 81, 86, 93, 100-7, 109, 111-3, 119, 121, 136-7, 139, 141,
144, 157, 171, 181, 243, 371-2, 383, 385-6, 388-91, 395-6, 399,
401-2, 404-5, 410-1, 419-20
intention-in-action 37, 81, 100, 103-6, 109, 111-2, 139, 396, 399, 410
prior intention 103-5, 137, 139
intentionalism 49, 208, 299-300, 314-5, 320, 335
intentionality see: mental representation
internal pictures see: pictures
intoxication 54-5, 187, 236, 244, 378, 397, 417, 424-5, 428
introspection 47-9, 54, 63, 65, 69, 93, 119, 151, 160, 178, 183, 187, 243,
260, 271, 402, 407, 433-4; see also: higher-order thought; immunity to error through misidentification; perspectivalness; subjective access
intuition 4, 407
inventiveness see: creativity; imaginativeness of people/ideas
Ishiguro, H. 73, 103, 373
itches see: bodily sensation
Jackson, F. 71
Johnson, M. 2
judgements 2, 6, 10, 12, 16, 24, 33-5, 43, 48, 55, 61-3, 65, 67, 69, 74, 76,
83, 88, 103-4, 115, 117-8, 121-8, 130, 150, 155, 158, 161-2, 165,
167, 173-4, 178-9, 183-7, 199, 206, 210, 215, 228, 231-3, 236, 241,
245, 247, 256-7, 260-2, 268, 270-1, 275-7, 282-4, 290-2, 293, 2958, 302-5, 307-8, 316, 337, 340, 353, 366, 370, 376, 385-6, 388-9,

472

Index

394, 412, 414, 416, 418, 427, 435, 438-9


justification 18, 31, 43, 59-60, 69-73, 89-90, 128-9, 137, 149, 197, 201,
233, 241, 267, 270, 279, 293, 316, 318, 340, 411, 425, 428, 436
defeasible 69-70, 72, 89-93, 108-9, 111-2, 136, 240
indefeasible 89-93, 108-9, 111-2, 136
prima facie 69, 128-9, 258, 260, 262, 267-8, 330, 410-1, 413, 436
Kant, I. 13, 32-3, 179-80, 184
kinaesthesia see: sense modalities
Kind, A. 3, 10, 42, 382, 392
knowledge 3, 9, 15, 18-19, 23-4, 31, 43, 55-6, 60, 66, 68-9, 71, 73-4, 76,
79, 81, 88-94, 100-3, 107-13, 115-38, 141-2, 144-6, 151, 186, 188,
195, 199-206, 209-10, 212-9, 222-3, 226, 228-37, 239-44, 246,
256-8, 261-2, 264-5, 267-8, 270-2, 276, 283, 287, 300, 304-5, 307,
316, 333-4, 338, 347, 357-8, 367, 374, 381-2, 385, 389, 393, 395-6,
402, 404, 410, 412-4, 416-7, 424, 433-6, 440-1; see also: cognising
episodes/states; cognition; justification; perception; recognition
Kosslyn, S. 3, 15, 124
learning 76, 88-9, 92, 134-6, 138, 142-3, 145, 270, 389; see also: imagining, informativity
new evidence 92, 130, 133, 146, 434
new information 120, 125, 130, 133, 136-8, 141-2, 144, 146, 195, 434
Levinson, J. 10, 42, 343, 382, 395
Lewis, D. 54, 73, 75, 120, 228
Locke, J. 33
Lowe, E. 395
Macpherson, F. 67, 88, 168
madness see: states of consciousness
making-believe 42-3, 46, 53, 71, 207, 243, 300, 302, 346-51, 363-4, 3734, 387, 410, 416, 440
Marbach, E. 159, 171
Martin, M. 1, 9, 20, 24, 33, 37-8, 41, 48-9, 52, 54, 62-6, 68-9, 73, 75, 77,
80-1, 83-4, 90, 100, 129, 137, 146, 153, 156, 159-60, 187, 201,

Index

473

240, 245, 248, 250, 254, 256, 258, 267, 276-8, 288, 297-9, 302,
306, 313-6, 318-23, 325-7, 333-4, 355, 361, 423
McCann, H. 104, 395, 401
McDowell, J. 90, 126, 146, 180, 184, 240, 258, 267
McGinn, C. 1, 3-4, 8, 10, 46, 52, 54, 155, 194, 269, 337, 357-8, 382
Meinong, A. 154
memory 2-3, 8, 10, 12, 24, 30, 32-5, 39-40, 48, 50, 54-5, 61, 63, 69, 73-4,
76-7, 79, 81-4, 87-8, 93-5, 98, 107, 111, 113, 117-9, 121-30, 136-7,
140-1, 144-5, 151, 153, 158-9, 164-5, 174-5, 185, 194-5, 199-201,
207, 214-5, 234, 241-2, 249, 257-8, 261, 267-9, 278, 280, 282-3,
285, 289-94, 296-9, 301, 309, 319, 321, 334, 353, 361-4, 367, 371,
383, 385-6, 388-90, 393, 395, 413-5, 417-9, 421-2, 427-428, 434-5,
437-9; see also: generic memory
mental agency see: agency
mental episodes 47-49; see also: imagining; mental representation
as indivisible 50, 400
as self-contained 49, 99, 176, 400, 416
complexes thereof 46-51, 68, 74, 100, 105, 262, 357-8, 396-403, 410,
417-420, 433, 440; see also: mental projects
difference in attitude 23, 59-68, 87, 152, 157-8, 162-76, 407
difference in degree/quantity 4, 11, 80, 143, 150, 290-1, 296, 418,
422, 437
difference in epistemic function 59-60, 68-73, 152, 169; see also: epistemic function/role
difference in kind/quality 4, 11, 59, 80, 108-9, 127, 143, 150-2, 251,
290-2, 296, 362, 422, 437
difference in nature 143, 177, 181, 188, 218-9, 222, 277, 399, 432
mental/natural kinds 5, 30-1, 150, 181, 201, 211, 340-1, 391, 398, 432
non-representational 19, 62, 82, 85, 260, 265-6, 340, 383, 388, 409,
418, 420-3, 429, 432
phenomenal difference 12, 48, 63, 73, 150-5, 157-9, 169, 177-9, 1845, 334-5, 437; see also: mental episodes, difference in attitude
introspective error 143, 151, 334-5
mental projects 46-50, 105, 115-118, 133, 180, 261-5, 355, 382, 415-6,
432-6

474

Index

mental projects (contd.)


cognitive see: cognitive projects
imaginative see: imaginative projects
practical 35
mental pictures see: pictures
mental representation 14, 16, 25, 32-4, 37, 76, 81, 96, 112, 120, 145, 153,
229, 232-3, 245, 260-1, 280, 295, 302, 313, 315, 341, 412-3, 421,
423, 432; see also: content
affective 33-4, 41-2, 260, 420; see also: affective imagining; bodily
sensation; desire; emotion; emotional imagining; moods
conative 41, 67, 100-1, 103, 112, 138-9, 158, 163, 169, 260, 265-6,
337, 354, 359, 436; see also: desire; intention; motivation; volition
conceptual 9-10, 32-3, 96, 123, 160, 181, 216, 236, 240, 293, 298-9,
328, 353, 359, 392, 418, 434
de dicto 109
de re 96-7, 101, 105
determinacy 12, 38, 68, 80-2, 85-7, 140, 150, 155, 249-50, 322, 334,
361, 364, 393, 410, 422
direct/immediate 37, 44, 66, 76, 83-4, 87, 126-7, 158, 160, 205-6,
208, 217, 224, 244, 259, 276, 278, 295, 300, 308, 361-2, 364; see
also: imagining
factive 74, 127, 259, 309, 320
general content 95-6, 160-1, 208, 300; see also: imagining, particularity
general subject matter 69, 388-9, 395
intensity 82, 150, 322, 334, 346, 361, 364, 406, 422
intentional/non-relational 33, 41, 49, 121, 128, 159, 164, 172, 205-9,
216, 223-6, 231-3, 238, 242, 258-9, 268, 282, 284, 298, 300-2, 311,
314-5, 320, 334, 383, 421-2
making a claim see: attitude
mode/manner 91, 96, 100-1, 157, 168, 176, 207, 215, 230, 307, 316,
339, 341, 351, 354, 357-63, 365, 370, 391
non-propositional/object-directed 32-3, 45, 101, 144, 156, 260, 302-3,
314, 334, 337, 339, 349, 361-4, 419
propositional 32-3, 41-2, 45, 62, 98, 144, 156, 205, 207, 212, 215,

Index

475

222, 224, 229, 276, 302-3, 307, 337, 339, 352-4, 356, 358-64, 4156; see also: propositions
mental representation (contd.)
relational/non-intentional 19, 49, 178, 187, 202, 206-9, 216-7, 258,
267, 276, 282, 284, 286, 298, 300-2, 361-3
representational element 34, 39, 93, 125, 159, 165, 171, 175, 293,
359-60, 377, 403, 416, 426-7; see also: images and thoughts; propositions
sensory vs. intellectual 32-3
teleological 71, 262
truth-evaluable 144-6, 195
vehicle 65
vivacity see: vivacity
mental rotation 66, 80, 88-9, 117-8, 130-2, 138, 145, 414
Meskin, A. 15, 168
Millar, A. 67, 90
Millikan, R. 71
Modal Account see: imagining, accounts of
modality 12-3, 61, 66, 161-4, 167-9, 175, 204-5, 327, 367, 391, 440-1
Mondrian, P. 270
moods 35, 255, 260, 266, 385, 388, 390, 402-3, 415, 421, 423
morality 2, 43-4, 338-9, 356, 392-3, 441
Moran, R. 20, 42, 337-339, 341, 345, 350-363, 441
motivation 9-10, 35, 42, 50, 59, 69-71, 100, 103, 142, 145-6, 175, 180-1,
185, 232, 238, 283-4, 318, 337, 339, 344, 346, 359, 363, 391, 3934, 410-1, 417, 424
Mulligan, K. 39
multiple use thesis 37, 98-100
negation claims 133, 193-5, 197-201, 205, 209-14, 216-20, 247, 251, 25369, 271, 279, 306, 308, 310, 410, 428; see also: Epistemological
Account
(NC) 194-5, 197, 200-2, 206, 212, 221, 253, 264-66, 268-71, 305
(NC*) 194-195, 197-202, 206, 210-2, 221, 247, 251, 253, 256-9, 2678, 271, 304, 306, 308, 378

476

Index

negation claims (contd.)


(NI) 133-8, 142, 146, 194-5, 253, 258-59, 266
(Ny) 258-65, 267-8
neural/neurofunctional processes 14-7, 40, 153, 160, 179, 248-9, 424, 428
new evidence/information see: learning
Nichols, S. 15, 41, 53, 374
Nida-Rmelin, M. 88
non-accidental veridicality 115, 118-9, 121, 145, 199-200, 229, 235-6,
239-40, 367, 434
non-actual objects 13, 51, 67-8, 161, 163, 170, 173, 440; see also: fiction;
possibilities; possible worlds/situations
non-imaginative phenomena 2, 6-7, 9, 11-2, 14, 27-9, 34-5, 43, 46, 51-2,
56, 149-50, 152-4, 163, 174-5, 185, 210, 212, 215, 255, 259-1, 2656, 270, 294, 297-8, 301, 309-10, 355, 358, 372, 378, 382, 385, 388,
395, 409-10, 413-7, 420-1, 426-7, 429, 432-40; see also: cognitions; cognitive projects
non-neutrality see: attitude, theoretical
normativity 305, 412
Noordhof, P. 37-8, 45-6, 84, 98-100, 207, 269, 326-33, 400, 419
noticing an aspect 52, 171, 440
object awareness see: mental representation
OBrien, L. 30, 180, 250
Ontological Account see: imagining, accounts of
Origin Account see: imagining, accounts of
OShaughnessy, B. 1, 3-4, 8-9, 18-20, 30, 32-3, 42, 46-7, 52, 54-5, 59, 62,
69, 73, 75, 84, 104, 116, 126-7, 129, 134, 138-9, 159, 178, 181,
187, 191-251 passim, 253, 255-9, 266-9, 271-2, 276, 278-9, 298302, 304, 306, 308, 313, 337, 349, 357-8, 378, 381, 383, 392, 3956, 399, 423, 425, 427-8, 435
Owens, D. 388
Pacherie, E. 55
pain see: bodily sensation
Papineau, D. 63, 71

Index

477

paradox of fiction 42, 340-5, 351-2; see also: fiction


particularity see: imagining
passions see: emotion
Peacocke, C. 1, 3, 9, 20, 33, 37-38, 42, 44-6, 69, 73, 75, 84-5, 94, 96-101,
103-5, 112-3, 139, 141, 250, 269, 276, 298, 302, 306-7, 313, 318,
328, 419, 427
perception 5, 16, 24, 33-4, 37, 39-40, 44, 48-9, 62-3, 65, 67, 70-2, 77, 812, 84-6, 88-93, 95-6, 102, 104, 107, 116-8, 123, 126-30, 132, 141,
144, 159-60, 171-3, 175-6, 178, 180, 182-3, 197-206, 208-9, 212-8,
222-30, 233-43, 245, 247-51, 257-8, 261, 266, 268, 281, 284-5,
290-1, 297, 299-300, 304, 309, 314-21, 323-7, 329-35, 337, 339,
341, 348-9, 362-3, 366, 369, 393, 407, 418-20, 427, 436; see also:
perceptual experience
perceptual contact with reality 126, 235-42, 258-62; see also: awareness
of the outer world; cognitive constraints
perceptual experience 2, 8-10, 12-3, 18, 23-4, 32-5, 37-9, 47-9, 54-5, 613, 69-71, 74, 76-7, 79-85, 87-9, 91-2, 94-8, 101, 113, 115-6, 11824, 126-9, 132, 134, 136, 139-41, 144, 150-3, 158-61, 165, 168,
171-3, 175-87, 194-5, 198-202, 205-6, 210, 214-9, 221, 223-6, 228,
230-42, 245-51, 254, 256-9, 261, 267-9, 275-8, 281-2, 284-5, 28893, 296-7, 299, 302, 305, 309-10, 314-7, 319-22, 326-7, 329, 3312, 334-5, 343, 352-3, 363, 366, 369-71, 376, 378, 383, 410, 412-3,
418-9, 424, 427-8, 438; see also: perception
conceptual 96, 240; see also: mental representation, conceptual
hallucinatory 17, 24, 44, 49, 54-5, 64, 75, 89, 97-9, 153, 155, 175,
178, 185, 187, 216-7, 225-6, 243-6, 254, 258, 315, 324, 328, 332,
378, 383, 401, 425, 428-9, 438, 440
illusory 89, 95, 110, 168, 217, 225, 224, 258, 324, 383
non-veridical 47-9, 153, 171, 224-5, 258-9; see also: veridicality
transparent see: transparency
veridical 47-9, 70, 153, 216, 226, 228, 230-1, 235-6; see also: veridicality
Perky, C. 151
perspectivalness
emotional 350, 354

478

Index

perspectivalness (contd.)
first-personal 48, 63, 113, 151, 160, 178, 186-7, 283, 314, 323, 348,
361, 406-7
hidden aspects of objects 80, 89, 99, 131-2, 138, 340
point of view 16, 40-1, 44-5, 80, 88-91, 131-2, 136, 317-8, 320, 325,
333, 342, 348, 354-5, 392
quasi-observational see: quasi-observationality
egocentric orientation 92, 316-8, 320, 322-7, 329-30, 332, 388
third-personal 47, 314, 348
visual 45, 88-9, 93, 108, 131-2, 136, 277, 306, 314-22, 325-7, 330,
332-3, 347-8, 355, 361
Phenomenal Account see: imagining, accounts of
phenomenal character/aspects 45, 47-9, 60, 62-3, 73, 81-2, 96, 152-4,
157, 169, 177-8, 183, 186-7, 201, 254, 283, 287, 289, 303, 310,
314-20, 322, 327, 329-30, 332, 337, 339, 361-3, 402-3, 412, 437
of judgements/beliefs 48, 303, 307
phenomenological salience 47, 59-61, 63, 65, 73, 81, 83-5, 129, 151, 157,
169, 175, 177, 179-80, 183-5, 188, 238, 245, 254, 272, 289-90,
297, 306, 339, 403, 407, 412, 433, 437-9
phenomenology 13, 49, 79, 143, 152-3, 157, 159, 169-70, 176, 186, 437
physics 15-7
physiology 249, 342-4, 425
pictorial experience 5, 11, 39, 45, 51, 64-5, 76, 101, 151, 159, 171, 175-6,
205, 210, 266-7, 309, 347-50, 358, 378, 423, 432, 438-9; see also:
pictures
imagination-based accounts 39, 347-50
as imaginative perception 205, 349
twofoldness 349-50
pictures 5, 34, 39, 64-5, 67, 69, 74, 76-7, 80, 91, 102, 111, 140, 158-60,
168, 171, 175-6, 210, 266-7, 270, 286-7, 298, 309, 318, 321, 328,
333, 335, 338, 347-50, 358, 395, 423, 440; see also: pictorial experience
ambiguous 358, 440
content 110-1, 175, 327-8, 441
internal/mental 34, 76, 88, 158-60

Index

479

pictures (contd.)
paintings 39, 111, 159, 176, 270, 287, 298, 328, 333-5, 347-8, 389
photocopies/reproductions reproductions 287-9, 333-5
photographs 102, 121, 130, 287-8, 333
portraits 158, 298, 328, 338
representing types vs. tokens 286-7
Pietroski, P. 400
Pink, T. 104, 139, 388, 415
pleasure 36, 41, 50, 63, 283, 342, 344, 385
Pollock, J. 69-70, 240, 256, 305
positing see: attitude
possibilities 12, 56, 66, 108, 161-4, 309, 352
possible worlds/situations 12, 67-8, 73, 155, 161-70, 319, 321, 324, 366,
405, 438
practical attitude/reasoning see: attitude
pretense 4, 9, 15, 89, 280, 365, 373-7, 440
propositions 34, 43, 45, 68, 162-3, 166-7, 205, 207, 270-1, 276, 285, 300,
303, 305, 307-9, 343, 347, 350, 353-4, 359, 370, 376-7, 388, 415-6,
425
Priest, G. 167
priming 110
proprioception see: sense modalities
propositional imagining 207, 212, 215, 229, 352-4, 359-61, 364; see also:
intellectual imagining
Pryor, J. 240
psychology 14
Putnam, H. 48, 294, 366
Pylyshyn, Z. 3, 15, 124
Quantitative Account see: imagining, accounts of
quasi-emotion 342-4, 346, 349-52, 356-8, 360, 363-4
quasi-observationality 18, 79-81, 86-94, 106, 108-9, 111-2, 131-3, 136-9,
156, 396, 410, 431
rational integration 18, 126-9, 172, 183, 187, 231-42, 246, 256-8, 260-1,

480

Index

267, 270, 345, 348, 368, 428, 436


Radford, C. 345
rationality 6, 18, 43, 60, 62, 65, 69-70, 73, 103, 122, 126, 128-9, 138,
153, 168, 174, 177, 179-80, 184-5, 187, 199-200, 219, 230-3, 23643, 246, 256-8, 260-1, 267-8, 284, 290, 305, 345, 368-9, 371, 376,
389, 392, 412, 418, 422, 424, 436, 438; see also: attitude,
practical/theoretical; inference; rational integration
Ravenscroft, I. 3-4, 9, 15, 20, 37, 41-4, 46, 59, 84, 276-7, 304, 326, 330,
334-5, 337, 353, 366, 368-71, 374-5, 395
reality 7-9, 13, 18, 23-4, 29, 32, 38, 41-2, 54-6, 60-3, 67-9, 71, 73-6, 81,
83, 87, 90-1, 94, 97-8, 106-7, 111, 115-6, 118-21, 124, 126-9, 1325, 141-6, 154-60, 171-3, 176, 178-9, 184, 187, 195, 200, 203, 205,
210-3, 215, 219, 221, 223-6, 228, 230, 232-47, 254, 256, 258-21,
264-5, 267-72, 275, 277, 284, 287, 289, 297, 321-22, 338-46, 34857, 360-4, 367-8, 375-7, 381, 396, 410-3, 420-3, 425, 428, 432,
434-6; see also: actual world; external world
reasoning see: attitude, practical/theoretical; inference
receptivity 13, 179-80
recognition 10, 118, 122-7, 171, 250, 269-70, 328, 331, 343, 347, 349,
355, 414; see also: cognition; knowledge; perception
recollection see: memory
reflexes 184, 401
regret 163, 340, 354
Reicher, M. 157
reliability 18, 71-2, 115-116, 118-9, 121-2, 125-126, 128, 130, 146, 199,
217, 219, 221, 223, 225-230, 232-3, 235-6, 239-40, 242, 246, 256258, 260-3, 267-269, 272, 411-3, 434, 436-7
religion 338
remembering see: memory
Representational Account see: imagining, accounts of
representational element see: mental representation
representationality see: mental representation
reproductions see: dependency of imaginings on cognitions; pictures
retina 224, 227
Roessler, J. 55, 394, 425

Index

481

Routley, R. 154
Russell, B. 157
Ryle, G. 1, 4, 9, 20, 34, 194, 276, 281, 373-6
sadness 125, 338, 354, 356, 389
Salmon, N. 154, 156
Sartre, J.-P. 1, 3, 8, 12-3, 20, 24, 34, 38, 62, 64, 73, 79-80, 86-92, 94, 96,
100-3, 116, 134-5, 138-40, 142-4, 151-2, 154-6, 158-9, 170-5, 17781, 183-6, 188, 193-5, 253, 258, 269, 272, 281, 320, 382-3, 394,
427-8
science 14-7, 33, 137, 248-9, 338; see also: empirical evidence; experiments
Scruton, R. 1, 3-4, 6, 10, 42, 46, 51-2, 54, 61, 151, 180, 304, 337, 357-8,
367, 369, 381-2, 394, 423
sculpture 85-6, 171
Searle, J. 33, 68, 103, 395
Sellars, W. 32
Semantic Account 287
seeing an aspect see: noticing an aspect
seeing-as 52
sense-data 49, 65, 160, 198, 206-7, 223-7, 259, 302
sense modalities 37-8; see also: bodily sensation
hearing 33, 35, 84, 207-8, 307, 348, 353-4, 359, 391
kinaesthesia 38
proprioception 38
smell 367
taste 71, 270, 286, 289, 293, 327, 367
touch 38, 88, 176, 353, 359, 367, 391
vision see: perception; perceptual experience; perspectivalness; visualising
sense of reality 243-4, 286, 289, 425
sensory imagining 4, 9-10, 12, 15, 17-8, 23, 27, 36-9, 41, 50, 54, 62, 65,
67, 74, 76-7, 79-98, 100-8, 111-3, 115-6, 118-9, 121-2, 127-31,
133-46, 151-2, 155-6, 158-9, 161, 165-6, 170-5, 177-80, 184-5,
199-202, 205, 210, 214, 217-8, 221-2, 246-51, 256-7, 263, 267,

482

Index

276-8, 292, 297, 299, 301-2, 305, 307, 313-4, 316, 325-7, 348, 3524, 357-8, 366-7, 369, 376-7, 382, 391, 393, 396, 410, 428, 431,
436; see also: affective imagining; visualising
Shakespeare, W. 341
Shorter, J. 373-374
Siegel, S. 144
Siewert, C. 47, 303
S-imagining 96-8, 100, 103, 105
simulation 4, 9, 19, 271, 275, 280, 365, 368-73, 377, 440; see also: imitation; Imitation Account
simulationalism 15, 371
sleep see: states of consciousness
smell see: sense modalities
Soames, S. 162
Socrates 287
Soldati, G. 41, 48-9, 73, 75, 85, 96, 169, 178, 187, 245, 259, 303, 307,
318, 370, 403
somnambulism see: states of consciousness
Sorensen, R. 3
Soteriou, M. 180
space 11-2, 39-40, 91-2, 99, 117, 145, 154-6, 173, 235, 316-7, 319-24,
330; see also: perspectivalness
spontaneity 6, 11, 13-4, 19, 24, 51, 100-4, 138-9, 175-88, 296, 301, 309,
357, 382-3, 393-4, 397, 402, 409, 420, 423-9, 432-3, 437-8
Spontaneity Account see: imagining, accounts of
Stich, S. 15, 374
Stock, K. 51, 358, 423
Strawson, G. 303
Strawson, P. 32
Stroud, B. 150
subjection to the will 82, 142-3, 146, 181, 243, 357-8, 404-5, 426, 428
Subjection to the Will Account see: imagining, accounts of
subjective access/indistinguishability 12, 16, 47-9, 54, 63, 87, 89, 127,
151-2, 183, 216, 245, 254, 283, 289-91, 297, 299, 339-40, 361-3,
378, 402; see also: introspection

Index

483

subjectivity 13, 60, 139-40, 159, 177, 179-80, 183, 185-6, 222, 242, 254,
283-4, 289, 315-6, 320-7, 329-30, 332, 335, 338, 342, 348, 350,
361, 438; see also: perspectivalness; phenomenal character; subjective access
subpersonal phenomena 14-8, 93, 104, 110, 152, 160, 179, 185, 428; see
also: neuronal/neurofunctional processes
Subpersonal Account see: imagining, accounts of
supposing 1-6, 12, 15, 30, 34, 42-3, 48, 61, 63, 69, 73, 96, 137, 150, 155,
164, 229, 270-1, 292-3, 295, 298, 307, 328, 351, 353, 359, 370,
416-7, 420, 433
states of consciousness/mind
anaesthesia/coma 230, 234
disorders/pathologies 27, 51, 54-5, 187, 231, 425-6, 440
dreams 51, 53-4, 216, 230-1, 234, 236, 238, 240, 245-6, 378
insanity/madness 282, 378
sleep 198, 230, 237, 241, 249, 282
somnambulism 236, 240
trance 230, 236
unconsciousness 52, 179, 230, 234, 238, 241, 424-5
wakefulness 54, 197-8, 230, 232-7, 239-42
stream of consciousness 47, 49, 140, 198, 233, 238, 284, 399, 425; see
also: consciousness; phenomenal character
taste see: sense modalities
theology 338
theoretical attitude/reasoning see: attitude
thinking 4, 38-9, 64, 93, 104-5, 110, 125, 150, 164-6, 251, 282, 295-6,
331, 347, 385, 400-1, 416, 419, 421, 425; see also: mental episodes; images and thoughts
first-order 68-71, 73, 149, 151, 207, 260, 262, 411, 422, 436
higher-order 209, 216, 260, 270, 295, 303, 307, 419
Thomas, N. 15, 367
thought experiments 3, 43, 310, 338, 353, 416, 440
thought insertion 54, 425, 440
touch see: sense modalities

484

Index

trance see: states of consciousness


transparency 65, 160, 250-1, 300, 313
Travis, C. 33
tryings see: volition
twofoldness see: pictorial experience
Tye, M. 15, 82, 153
unconsciousness see: states of consciousness
unified accounts/theories 2-9, 11, 14, 16-20, 23-5, 27-57 passim, 60, 79,
83, 106, 146, 149, 152-4, 159, 161, 164, 166, 170, 175, 178, 181,
187, 191-3, 195, 197-9, 202, 210-1, 213, 220, 247, 253-7, 261, 263,
265-6, 268-9, 271, 275, 278-81, 292-4, 298, 301, 311, 365-6, 371,
377-8, 381-3, 396, 399, 406-7, 409, 414, 431-9, 441; see also: imagining, accounts of; imagining, unity
desiderata 17, 19, 23-4, 27-9, 56, 60, 146, 161, 210-1, 213, 253, 383,
396, 409-29 passim, 431, 434
exhaustiveness/purity 192-3, 197, 201-2, 254-5, 266, 278-9, 294, 29799, 410, 435-6
explanatory power see: explanatory power
extensional adequacy see: extensional adequacy
unreal/non-existing objects 73, 154-8, 160-1, 170, 172-4, 203, 236, 244;
see also: fiction
values see: evaluations
Velleman, D. 41, 43, 46, 64, 68-9, 260
veridicality 31-2, 44, 47-9, 54, 62, 70, 75, 97-8, 115, 117-21, 125-6, 128,
144-6, 153, 158, 162, 178, 195, 199-201, 216-7, 219, 221, 223-33,
235-6, 239-40, 242, 246, 258, 268, 280, 300, 315, 322, 326, 367,
413, 420, 434; see also: non-accidental veridicality
visualising 1-6, 12, 15, 28, 30, 34-35, 37-40, 44, 48-50, 60-7, 69-70, 73-7,
83, 86-9, 91-6, 99-100, 103-4, 106-11, 117-20, 124-6, 130-3, 135-8,
140-2, 144-6, 150-1, 153, 157-60, 162, 164-6, 168, 171, 176-7,
207, 232, 243, 249-51, 269-70, 277-8, 293-5, 298-301, 313-35
passim, 345, 348, 353, 356, 358-9, 363, 366-7, 374, 387, 391, 393,
395-6, 399, 401-2, 405-6, 411, 417, 419-21, 426, 428, 433, 436; see

Index

485

also: imagining seeing


vivacity 4, 11-2, 56, 79-85, 87, 140, 143, 150-1, 282-5, 287-92, 294, 296,
299, 361, 364, 410, 421-2
volition 10, 35, 93, 104-5, 111, 117, 121, 123, 135, 139, 184, 250, 290,
310, 374, 399-401, 410, 415
Voltolini, A. 154
voluntariness see: agency
wakefulness see: awareness of the outer world; states of consciousness
Walton, K. 1-2, 4-5, 20, 41-2, 44-6, 51, 53, 71, 73, 295, 337-9, 341-343,
345-52, 356, 358, 360, 363, 374, 387, 405, 410, 415, 423, 441
Warnock, M. 32
warrant see: justification
Weatherson, B. 42
Weinberg, J. 15, 168
White, A. 1, 3-4, 12, 20, 37, 42, 164-6, 269, 337, 373-6
Williams, B. 37, 46, 63
Williamson, T. 3, 24, 66, 69, 76, 126, 129, 256, 258, 267, 441
willings see: volition
wishes 51, 55-6, 191, 440
Wittgenstein, L. 3, 8, 10, 20, 24, 31, 34, 38-9, 48, 73, 80, 87, 94, 97, 116,
134, 137-9, 142-6, 151, 155-6, 193-5, 200, 253, 258, 264, 269, 281,
320, 328, 381-2, 394, 429
Wollheim, R. 5, 10, 37, 43, 45-6, 77, 111, 171, 269, 298, 320, 349, 382,
392
wondering 36, 48, 61, 66, 125, 163, 169, 173, 303, 308, 415-6

You might also like