Consciousness and Cognition 60 (2018) 17–24

Contents lists available at ScienceDirect

Consciousness and Cognition

journal homepage: www.elsevier.com/locate/concog

Review article

Explaining the felt location of bodily sensations through body

T
representations
⁎
Luis Alejandro Murillo Lara
Universitaria Agustiniana, Bogotá, Colombia
Avenida Ciudad de Cali, No. 11b-95, 110821 Bogotá, Colombia

AR TI CLE I NF O AB S T R A CT

Keywords: Why are bodily sensations felt on specific body parts? This paper discusses the view according to
Bodily sensations which we need body representations to account for the felt location of bodily sensations. My aim
Body representations will be to consider whether or not some claims linked with that view are substantiated (namely,
Body image that all of our grasp of the spatiality of our bodies must come from bodily sensations, that the
Brian O’Shaughnessy
representation of the body can determine bodily sensations surmounting sensory input, that the
content of body representations cannot be action-oriented). To do this, I first introduce and assess
Brian O’Shaughnessy’s seminal version of the representationalist approach to bodily sensations.
Next, I will inquire into a purported objection to any version of the approach, showing its in-
adequacy. Finally, I will concentrate on Frédérique de Vignemont’s variant of the re-
presentationalist view, trying to pin down a few of her assertions. My conclusion will be that the
scope and strength of the representationalist position in regard to the aforementioned claims is
different from what it is usually thought to be.

1. Introduction: The problem of the location of bodily sensations

If we are bitten by a mosquito, then we feel not just the characteristic itch but we are also able to tell the place of the itching. So,
together with their intrinsic qualitative aspect, bodily sensations apparently bring with them a spatial aspect.1 But, are bodily
sensations literally located on the body parts in which they are felt? It seems that they are not: sensations are psychological entities,
and to that extent they must take place in the brain. What, then, explains the fact that despite being in the brain, bodily sensations
seem to us to be located on specific body parts?
An answer that may come immediately to mind is that the felt location of bodily sensation is the place of its cause. However, some
physical events on certain body parts may cause bodily sensations that are felt somewhere else, which are described as referred
sensations (O’Shaughnessy, 1980). Other attempts at a definition of the felt location of bodily sensations can be equally discarded: the
place where we believe (or are inclined to believe) the cause of the sensation is, the place in objective physical space in which the

⁎
Address: Avenida Ciudad de Cali, No. 11b-95, 110821 Bogotá, Colombia.
E-mail address: luis.murillol@uniagustiniana.edu.co.
1
Sensation is sometimes defined as the process in which a sensory receptor is stimulated, making the brain to interpret it as sound, image, odor, taste, pain, among
other qualities, whereas perception is consequently defined the process of making sense of the information coming from sensation (see Mather, 2016). In this sense, it is
thought that without perception sensation would be raw feeling (Russell, 1912), and perception without sensation would be impossible (see O’Shaughnessy, 1980).
However, in this paper I employ de Vignemont’s (2017, 2016a, 2018) characterization of bodily sensations as items endowed with both qualitative character and
spatial content (typically related to spatial properties of the stimulus and the sensation’s location on the body). To that extent, bodily sensations (sometimes also called,
bodily experiences) go beyond the mere qualitative and phenomenal character customarily ascribed to sensations, including a certain “aboutness” (that is, conveying
information about body parts or body areas) but without being labeled as perceptions (perhaps insofar as what they present is not a discrete perceptual object).

https://doi.org/10.1016/j.concog.2018.01.007
Received 2 October 2017; Received in revised form 26 December 2017; Accepted 30 January 2018
1053-8100/ © 2018 Elsevier Inc. All rights reserved.
L.A. Murillo Lara Consciousness and Cognition 60 (2018) 17–24

relevant body part is, the place in the biological body where it seems to be, and so on (see O’Shaughnessy, 1980, p. 222 et seq., and
Armstrong, 1962, p. 85 et seq.)
Difficulty to define the felt location of bodily sensation may lead one to the thought that the problem is assuming that bodily
sensations have an inherent spatial component. It could be said that bodily sensations exclusively have a qualitative aspect and that
this qualitative aspect is different depending on the stimulated body part. Alternatively, one could contend that the location of bodily
sensations should be specified in terms of motor abilities; that is, as the place we know how to reach when a body part is stimulated.
However, against the first suggestion, we find that there can be a difference in the felt location of sensations that are qualitatively
indistinguishable. Against the second suggestion, neuropsychological cases have shown that it is possible to know how to get to the
location of the stimulation in the absence of any sensation whatsoever, such as numbsense patients (de Vignemont, 2010, 2018).
In this paper I want to delve into a different way to explain why bodily sensations seem to us to be located on specific body parts,
which might be called the representationalist approach. According to this approach, the specification of the location of bodily sen-
sations involves the existence of a certain spatial representation of the body. Brian O’Shaughnessy (1980, 1989, 1995) can be
considered to be the first prominent advocate of such view, which he calls the body-image hypothesis. However, my focus will be
Frédérique de Vignemont’s body-map theory (de Vignemont and Massin, 2015, de Vignemont, 2018), probably the most recent version
of the representationalist approach and a refinement of O’Shaughnessy’s body-image hypothesis. I will argue that, despite being
essentially right, the representationalist approach (both in O’Shaughnessy’s and in de Vignemont’s version) has not really established
some of its claims and it still has a few key points to clarify. Therefore, the actual scope and strength of the representationalist
approach is different than has previously been thought.
I will proceed in three steps. In the first step, I will illustrate how O’Shaughnessy (1980, 1995) argues for his body-image
hypothesis, with the aim of making clear what he achieves to demonstrate (particularly as regards an idea coming from Elizabeth
Anscombe, 1957, 1962). In the second step, I will consider an alleged objection to any version of the representationalist approach:
that a representation of the whole body is not necessary and that a bunch of local representations are enough for the task. The third
step will consist of an analysis of the two main lines of argument that de Vignemont uses to motivate the body-map theory. I intend to
address some obscure elements in de Vignemont’s first argument, and to show that the scope of her second argument is more limited
than it might seem.

2. The ‘body-image’

O’Shaughnessy’s explanation of the spatiality of bodily sensations depends on the notion of a body-image (1980, 1989, 1995): he
contends that an awareness of the body together with some awareness of its general spatial traits is a necessary condition of the
appearance of a bodily sensation on the body. He then claims that this second awareness amounts to the possession of a kind of
‘internal picture’ of the body, the body image. The body image (or, sometimes, long-term body image) is a general sense of the body
endowed with spatial content (1995, pp. 184, 188). In his view, bodily sensations can only appear on a body that is so internalized:
they appear “first and foremost (…) onto the body-image; and thence, and only secondly, albeit immediately, onto the body itself” (1980, p.
211). In this regard, he asserts that the content of proprioceptive awareness at a given moment is caused by the conjunction of
somatosensory information and the body image.
The way O’Shaughnessy (1980, 1995) argues for the body image as an account of the spatiality of bodily sensations can be seen as
an argument to the best explanation where other candidates have been proved to be inadequate. Let us think, for example, that
instead of a body image it is the body itself that explains the spatial content of bodily sensations. According to O’Shaughnessy (1995),
the explanandum requires an explanans already spatially endowed, and the body itself meets this requirement. However, if the
explanans is the body itself, then we can only explain the spatial content of veridical bodily sensations but manifestly not that of
phantom limb sensations. This might illuminate his reply that what we need to explain the spatial content of proprioception is a
representation causally intervening between the body itself and individual bodily sensations with their spatial regularity (and capable
of explaining this spatial regularity by its possession of spatial content).
O’Shaughnessy (1995) considers another option: bodily sensations themselves are endowed with spatial content. It could be said
that they bring to awareness their own shape, which in turn maps that of the body, thus explaining the spatial content of proprio-
ception. But, as Bermúdez (1998) has also argued, in order to possess their spatiality, bodily sensations require a frame of reference.
O’Shaughnessy surmises that “the spatial properties of bodily sensations cannot be the epistemological foundation of the spatial content of
proprioception” (1995, p. 191). In being represented as spatially endowed, bodily sensations presuppose a representation of a spatial
frame of reference, of something already endowed with spatial content.
A third alternative is a sort of three-dimensional image generated by sensory inputs, which he calls I3D. Among other char-
acteristics, I3D constantly mutates as the body moves and its content determines our immediate knowledge of our body. As an
additional feature, the posture of the moment would have a direct cerebral effect in the current I3D, so this knowledge could be had
independently of the existence of a sensation. This third feature is particularly relevant for the purposes of our discussion, insofar as it
is the hallmark of Anscombe’s conception of how non-observational knowledge is acquired (Anscombe, 1957, 1962).
O’Shaughnessy (1995) notes that this would be an anaesthetic knowledge of the body, not proprioceptive experience—which he
characterizes as feeling the body. Indeed, the deliverances of I3D could not be considered bodily sensations, but this does not prove
that this kind of immediate anaesthetic knowledge of the body is impossible. In fact, the line of argument O’Shaughnessy uses to
reject any role for such knowledge is not tenable: that without bodily sensations the will would not have access to the body. As a
matter of empirical fact, we know that the bodily will can have its immediate object in the absence of any proprioceptive experience
(as in deafferentation; see Gallagher and Cole, 1995). Bodily sensation cannot be the only way for the will to have access to the body.

18
L.A. Murillo Lara Consciousness and Cognition 60 (2018) 17–24

Of course, O’Shaughnessy’s (1995) failure to exclude the possibility of such anaesthetic knowledge does not jeopardize the hypothesis
of the body-image: it is only incompatible with the claim that this representation necessarily always involves a conscious grasp of the
body.
Interestingly, O’Shaughnessy seems to concede that the spatial content of bodily sensations has a non-conceptual action-oriented
structure; for example, he comments that “the content seems somehow to be ‘subconceptual,’ as one might say, not ineffable but forged out
of something other than the concepts enshrined in our language” (1995, p. 186), and that “the content of proprioception seems best expressed
in a practical medium rather than in conceptual terms” (1995, emphasis added) and finally that the content must be so qua content of the
action (cf. 1995, p. 181). This might constitute a significant difference with de Vignemont’s body-map theory, which allows a rather
narrow space (if any) to such action-oriented contents.

3. Is a global body map dispensable?

Having introduced the classic version of the representationalist view of the location of bodily sensations, I want to consider a
potential threat that may lurk in the assumption that there is a representation of the whole body. Perhaps a representation of the
whole body is not necessary and a bunch of local representations suffice to do the job. If so, the claim that proprioceptive content is
something like “right-index-finger-attached-to-hand-attached-to-arm-attached-to-etc.” (O’Shaughnessy, 1989, p. 42) would be simply
wrong.
The view that a representation of the whole body is not necessary has been maintained by Adrian Alsmith (2009). His argument
begins with a criticism of Bermúdez’s conception of bodily space.2 He contends that Bermúdez’s (1998) view according to which
bodily locations are specified by hinges is unable to account for some features of our experience of bodily locations; namely, the facts
that sometimes we do not experience bodily sensations in clearly bounded areas (“vague spatial phenomenology”) and that some-
times the place of such sensations is felt as more precisely definite than others (“grades of spatial determinacy”).3 Alsmith (2009)
points out that to account for these features we need something that Bermúdez’s (1998) account does not provide.
What is this missing thing? Alsmith’s (2009) opinion is that what Bermúdez’s (1998) view lacks is a conception of bodily space in
which mereological relations (part-to-part and part-to-whole) play a crucial role. These relations, he holds, are crucial both in the
representation of the spatial orientation of parts and in the experience of the body in its spatial aspects. Moreover, Alsmith’s (2009)
position is not only that mereological relations are necessary to account for bodily space but that they are sufficient for the task.
According to Alsmith (2009), the hypothesis of a global representation of the spatial properties of the body as a whole that enables
globally controlling bodily posture and movement has not been demonstrated. Given that, there is room for an alternative ex-
planation in which instead of a global representation of the body, the existence of multiple body-part centered representations is
posited.4 But then, how does this plethora of local representations come to be articulated in a single, coherent behavior? Alsmith
(2009) answers that the very unity and stability of the body itself provides the requisite coherence, without needing an internal
representation of the said unity and stability.
Nonetheless, Alsmith (2009) acknowledges that, although most of the time we can only focus on particular body parts and not on
our whole body, it does make sense to claim that the body is experienced as a whole. This allows us to understand attentional focus as
a relation in which an experience of my body as a whole functions as a sort of background within which is located what I attend to.
Yet, he asserts that “such an insight, however, need not lead us to postulate an internal representation of the body as a whole” (2009, p. 96)
and insists that the role of any purported global representation can be better fulfilled by the unity of the actual body.
Surprisingly, although he recognizes that a good reason for introducing a representation of the whole body comes from the
phenomenon of attentional focus, he does not offer an alternative explanation. How else do we account for attentional focus if not by
appealing to a background sense of the body as a whole? There seems to be no answer to this question in Alsmith’s (2009) approach.
The absence of such an account negatively impacts other aspects of this approach. One consequence of his view that Alsmith
(2009) considers is that we would not have a single bodily space, but multiple spatial points of origin. He replies that this multiplicity
would only be a problem if each of those spaces gave rise to a separate behavioral space. But why does it not happen? His answer is
that the structure of the actual body entails that agents have a unified behavioral space. The idea seems to be something like: one
body, one behavioral space. But this answer is rather unsatisfactory. Think of patients who suffer alien hand syndrome (Marchetti &
Della Salla, 1998): it is incontestable that they have one physical body, but it is clear that they do not have a unified behavioral space.
Intuitively, in everyday life what seems to impede agents having more than one behavioral space at a given time is attentional
focus (which might be altered in alien hand syndrome patients). However, since Alsmith (2009) does not account for attentional focus
and our only comprehension of it involves representing of the whole body, he cannot appeal to it to explain what happens in alien
hand syndrome. What is more, this condition also compromises a key assertion of his approach: that coherent behavior exists due to
the very unity and stability of the body itself. Unless this issue is addressed, it seems more advantageous to maintain that the body as

2
Alsmith develops two different lines of reasoning to show that no global representation of the body is necessary to account for bodily awareness: the first has to do
with the personal level of explanation and the second with the subpersonal level. Insofar as we are moving at the experiential sphere, I will focus exclusively on what
happens at the personal level.
3
To the extent that is mentioned for the purpose of illustrating Alsmith’s (2009) starting point, I will not dwell on whether this criticism really affects Bermúdez’s
(1998) account. I will only remark that it has not been proved that in Bermúdez view the specification of a sensation’s location relative to some hinges cannot be vague
or more or less determinate.
4
The spatial vagueness of some bodily sensations and the existence of grades of spatial determinacy in the location of such sensations would be accounted for in
terms of occurrent points being tied to particular actions.

19
L.A. Murillo Lara Consciousness and Cognition 60 (2018) 17–24

a whole is somehow represented because it provides us with a way to understand attentional focus (as in O’Shaughnessy, 1995) and
allows us to hypothesize a fragmentation of this body map to explain alien hands.

4. The body-map theory

So far we have examined O’Shaughnessy’s (1980, 1995) argument for the body-image hypothesis and questioned Alsmith’s (2009)
contention that a representation of the whole body is not necessary and that a bunch of local representations are enough. Now I want
to introduce Frédérique de Vignemont’s version of the thesis that our grasp of the spatial aspects of bodily sensations is best explained
by means of representations of the body; it is her view of this thesis that she calls the body map theory (de Vignemont, 2017, 2014a,
2018; see also de Vignemont and Massin, 2015). In order to do so, I will proceed in two steps. The first is to consider the main
argument that de Vignemont presents for the existence of a body map, let us label this line of reasoning “the argument from the
spatial poverty of sensory input”. The second step will be looking at one of her arguments against what she calls the “hot conception”
of bodily space, in which a relevant distinction within the body map theory enters the scene.

4.1. The spatial poverty of sensory input

The view that the content of somatosensory input does not include most of the spatial properties of our bodies we are aware of is
characteristic of O’Shaughnessy’s (1980, 1989, 1995) thought. For example, he speaks of “spatially impoverished respects” (1995, p.
196) and of “poverty of spatial determinations” (1989, p. 43). In his opinion, in order to be aware of complex spatial properties, more
than this bare content is required: “this (complex) object [the shapes of objects we are aware of] came to attention (…) across time, and
assisted by attending-activity (…)” (1989, p. 45). In line with O’Shaughnessy’s (1995) remarks, de Vignemont maintains that the spatial
information carried by somatosensory input is insufficient to account for the spatiality of bodily sensations. In her view, the spatial
information carried by somatosensory input is like a cross on a blank map: “there is no reference point, no orientation, no border” (2017,
p. 466). One needs the tracing paper of the body map behind the blank map of somatosensory information.
Granting Bermúdez’s (1998) distinction between A- and B-locations, de Vignemont underscores that the spatial content of bodily
sensations is rich enough to include both of those locations, whereas information carried by somatosensory input consists of “raw
signals about muscle stretch, tendon tension and joint angle” so that “the scope of information that the body senses directly carry is limited”
(2014a, pp. 991-992). According to de Vignemont, the spatial accuracy of proprioceptive input is weak not only when one is still but
also when one moves. Therefore, the spatiality of bodily experience involves spatial information that “cannot be easily derived from the
body senses” (2014a, p. 992).
In de Vignemont’s view, the spatial poverty of somatosensory input together with the spatial richness of bodily experience is what
justifies a hypothesis of a representation of the long-term spatial properties of the body–the body map. It is thanks to the body map
that bodily sensations are experienced at more than “an isolated body point” (de Vignemont and Massin, 2015, p. 305; see also de
Vignemont, 2017); it is the body map that provides them with a spatial frame of reference and constitutes the background on which
they are experienced.
Nonetheless, some elements of the argument from the spatial poverty of sensory input seem to require clarification. First, what
does it mean that without the body map bodily sensations would be experienced as being at an isolated body point? And, relatedly,
does somatosensory input by itself carry any spatial information whatsoever? Second, what does it mean that bodily sensations are
intrinsically spatial even when they need a body map to have spatiality? Third, does somatosensory information play any role
whatsoever in shaping the body image?

4.2. Spaceless bodily sensations or somatosensory input?

If somatosensory information is as a blank map with an isolated cross in it, and if bodily sensation without a body map is
experienced as this very isolated point in a blank, then bodily sensations without a body map would amount to somatosensory input.
The body map theory would seem close to the idea that bodily sensations are devoid of intrinsic spatial content in absence of a body
map (a view known as the local sign theory, which was rejected by de Vignemont, 2015). It is worth noting that in criticizing the view
that sensations are raw feelings, Michael Martin (1993, p. 209) argues that the only way to articulate the idea that sensations do
inform one about one’s body is by showing that their spatial content was somehow additional to its purely sensational core–and
observed that the way local sign theory tried to meet this challenge was simply implausible. But if bodily sensation exist devoid of
spatiality without a body map, and it is the body map that endows spaceless bodily sensations with spatial content, then it might seem
that the body map theory meets Martin’s (1993) challenge: the spatial content of sensations is an additional contribution of the
already spatial body map.
Still, there are two reasons to think that the body map theory does not imply that bodily sensations are spaceless without the body
map. The first is that the body map theory seems to grant to somatosensory information some spatial content. The second is that,
according to body map theory, bodily sensation is not somatosensory information.
Without a body map, somatosensory information would be reduced to spaceless, raw feelings? De Vignemont concedes that the
somatosensory input one receives as feedback in acting “can carry information about the size of the limbs” (2014a, p. 992), and mentions
that there is “spatial information conveyed by the body senses” (de Vignemont, 2018). Of course, that information is poorer and more
limited than that of regular bodily experience, but it is still spatial information. It seems to be likewise even for O’Shaughnessy: “The
objects of body-sense are such things as: ‘presence of arm’, ‘flexedness of arm’, ‘hand near chin’, ‘body presence’. But they do not include the

20
L.A. Murillo Lara Consciousness and Cognition 60 (2018) 17–24

shape of the limb whose presence and posture one perceives. Now body-sense depends upon the possession of a long-term body-image” (1989,
p. 54, emphasis added).
What does the spatial content in somatosensory information amount to? What kind of spatial information does not entail the body
map? O’Shaughnessy’s (1989) quote from the previous paragraph may shed light in this regard: it would be unarticulated information
about local spatial properties (partial information about body-part sizes, current relative distances between body parts, and current
spatial disposition of body parts). Perhaps, the spatial information somatosensory input would carry independently of the body map
would amount to that unarticulated spatiality.
Even if somatosensory information was spaceless in the body map theory, it would not follow that bodily sensation is spaceless in
the absence of the body map. This comes from the fact that the body map theory entails that somatosensory information is different
from bodily sensation. We saw that the rich spatiality that we acknowledge in bodily sensation can only come from a body map. We
have also just seen that if somatosensory information and bodily sensation were the same, then this rich spatiality should be extrinsic
to bodily sensation. But there is no way that bodily sensation could intrinsically possess spatial richness—as the body map theory
claims— were it a separate existence from the body map. Thence, if the spatial richness of bodily sensations is intrinsic, and this
spatial richness implies a body map, then what body map theory must be claiming is that bodily sensations must be intrinsically
representational (that is, intrinsically involving a body map). Thus, whereas somatosensory information might exist independently
from the body map (the cross in a blank map), bodily sensation would not.
Therefore, there is no risk that bodily sensations end up being spaceless without the body map because, strictly speaking, there are
no bodily sensations without a body map. The confusion would come from de Vignemont (2015, 2014a, 2018) inadvertently calling
the experience of somatosensory information devoid of a body map bodily sensation.
This also explains why bodily sensations are intrinsically spatial, even when they need a body map to have spatiality. As in
O’Shaughnessy (1980, 1995), bodily sensations would be the result of conjunction of somatosensory information and a body map.
Bodily sensations would have both somatosensory information and the body map as independently necessary but jointly sufficient
conditions.5

4.3. The role of somatosensory information and body image in bodily sensations

Is somatosensory information actually a necessary condition for bodily sensation in the body map theory? There are two reasons
why one could think that such information might end up being explanatorily idle here.
The first comes from the way that body map theory accounts for phantom limb sensations and supernumerary phantom limb
sensations, which seem to be explained in terms of the sole activity of the body map.6 In this vein, de Vignemont writes: “phantom
limbs are the clearest pathological expression of the existence of the representation of the body in the mind. Patients can feel sensations in a
limb that no longer exists thanks to the body representation that still includes the amputated limb” (2014b, p. 289). This way of explaining
phantom limb sensations entails that bodily sensations are possible in the absence of somatosensory information. We can see the
expression of basically the same idea in O’Shaughnessy (1980), when he characterizes the body image as the cause (the ‘deepest’
cause) of bodily sensations. If it may cause bodily sensations by itself, in the absence of peripheral input, then body image would be a
sufficient cause of bodily sensations.
It could be replied that there is still a form of abnormal somatosensory information playing a role in phantom limb sensations: that
coming from the nerve endings in the stump, which can even activate spontaneously. Yet, this consideration does not apply to the
case of supernumerary limb sensations. Alternatively, it could be said that the fact that somatosensory information is not really a
necessary condition for bodily sensations in the body map theory does not imply that they are explanatorily idle: somatosensory
information usually happens to serve as input for the body map, so that in normal cases—as a matter of empirical fact—somato-
sensory information joins the body map in the production of bodily sensations.
Does the possibility of having experiences of a certain sort (commonly had by virtue of having some sensory input) without the
need of their usual input make the input unnecessary for having those experiences? In some sense it does because, strictly speaking,
this possibility makes the following statement false: every instance of that type of experience is such that it can only be had if a certain
input takes place. But is this sense relevant?
It could be objected that, despite the fact that I can imagine colors (and so to a certain extent I am able to have some experience of
colors without having thereby visuosensory input at the time of the imagining), visual input is still necessary for visual experience.
The relevant sense in which somatosensory input is necessary for bodily experience would be analogous to the sense in which visual
input is necessary for visual experience in spite of the existence of visual imagination.
Nonetheless, visual imagination has visual input as a pre-condition in a way that bodily experience might not have somatosensory
input. I can only imagine a certain color if I ever had the relevant visual input, but supernumerary phantom limbs seem to suggest

5
However, where does the body-map come from? de Vignemont’s (2016b) answer is that whereas the representation of short-term bodily properties (for instance,
limb posture) are constantly updated on the basis of afferent information, the representation of long-term bodily properties would include “innate components” that
convey information about the structure of the human body (“two-arm/two-leg”) and would not be based purely on afferent information. In addition, she has asserted
(2010, p. 11) that all the sources of sensory information about our own body interact with each other at the computational level to update body representations. They
participate in body representations formation and updating through sensory combination (between non-redundant sensory signals) or through sensory integration
(between redundant signals).
6
Supernumerary limb sensations refer to cases where, following brain lesions, some patients experience both the presence of more than two legs or arms and various
sensations in the “supernumerary” limb (see Miyazawa, Hayashi, Komiya, & Akiyama, 2004).

21
L.A. Murillo Lara Consciousness and Cognition 60 (2018) 17–24

that I can have certain bodily experiences without ever having had the relevant somatosensory input. If so, the analogy with visual
imagination does not help us to argue that somatosensory input is still necessary for us to have bodily experiences.7
There is another way in which somatosensory information could turn out to be explanatorily idle, namely, if body representations
shape bodily experiences regardless of the ongoing somatosensory feedback. The source of this latter idea is O’Shaughnessy’s claim
that if one had an ‘octopus-shaped’ body image, then one could not experience having a flexed arm despite having a human shape:
“(…) if in general one took oneself to be (say) octopus-shaped instead, then despite having a human shape and despite the presence of
posture-caused phenomena like sensations of posture, one could not have the experience of seeming to be in the presence of a flexed (very
roughly) arm-shaped thing.” (1995, p. 184) These lines are sometimes interpreted as claiming that in such a situation one would feel
like one was an octopus. But, is that claim intuitive? Could one actually feel like an octopus in this situation?8
My contention is that the implication drawn from O’Shaughnessy’s example (that in this situation one would feel like one was an
octopus) is misleading. Consider the following Kafkaesque imaginary situation: one morning one wakes from troubled dreams and
finds oneself transformed into a horrible vermin; not enough with this, one is blind and—of course—one keeps one’s human body
representation. As in its literary version, the intuition in this situation is not that even if one had a human-shaped body image, one
could experience having a flexed bug-leg when having a bug-shaped body. The intuition is rather that if one had a bug-shaped body,
one could not simply experience having a flexed human arm even if one had a human-shaped body image.
Let x and y be two different anatomical shapes (human and octopus/bug, respectively). O’Shaughnessy’s (1995) original claim is
that:

(1) If one had a y-shaped body representation, then one could not have an x-shaped bodily experience despite having an x-shaped
body.

In other words, that the kind of somatosensory information one receives due to one’s body shape does not determine one’s bodily
experiences, because one’s body representation plays a role in the determination of one’s bodily experience. For its part, what the
Gregor-Samsa-situation illustrates is that:

(2) If one had a y-shaped body, one could not just have an x-shaped bodily experience even if one had an x-shaped body re-
presentation.

Put another way, the kind of body representation that one has is not enough for determining bodily experience. Somatosensory
information due to one’s body shape also plays a role in the determination of one’s bodily experiences.
As can be seen, neither (1) nor (2) say anything about the kind of bodily experience that one could have, they only point out what
kind of bodily experience one could not have. Thus, it does follow neither from (2) nor from (1) that:

(3) One would have y-shaped bodily experience.

The implication from (1) to (3) presupposes that one must have either x-bodily-experiences or y-bodily-experiences. But this
assumption is not justified because, as far as we know, one could have bodily experiences that are none. A discrepancy between one’s
body shape and one’s body representation would probably deliver rather messy bodily experiences (determined neither by one’s body
shape nor by one’s body representation). It is hard to know how one could feel if one had an octopus-shaped body representation and
a human-shaped body (or a human-shaped body representation and a bug-shaped body). But given the conflict between the de-
liverances of somatosensory information and the internalized body, it is not clear that one would simply feel as an octopus (or bug) or
as a human.
Therefore, that O’Shaughnessy maintains (3) through (1) is a misreading because (3) does not follow from (1). Furthermore, as (2)
illustrates, ongoing somatosensory information influences both body representation and bodily experiences, and it cannot be claimed
that body representation determines the spatiality of bodily experiences regardless of any incoming sensory information.

4.4. The rough-grainedness of action-oriented body content

After having seen the scope and implications of the argument from the spatial poverty of somatosensory input, we can move to
another argument of de Vignemont (2016a) through which we can come to know more about the body map theory. This argument

7
It could be replied that even in supernumerary phantom limb experiences, patients can be said to have had the relevant somatosensory input: despite not having
had it coming from that specific limb, they have had it coming from other limbs. I take this reply to mean something like that: someone experiencing a supernumerary
phantom limb might feel a warm sensation on it, due to the fact that he has felt warm sensations elsewhere in the past. Besides, without those past sensations, he could
not have had the warm sensation on his supernumerary phantom limb. Therefore, somatosensory input is necessary even in the case of supernumerary phantom limb
experiences. As a consequence, this reply accounts for supernumerary phantom limb experiences as something close to what happens in imagination: I can somehow
experience qualities in the absence of an ongoing appropriate input, because I have experienced those qualities in the past. It is not clear whether such an account
makes justice to what patients experience, inasmuch as they seem to describe them as ongoing bodily experiences rather than as imagined bodily experiences. The
challenge for the advocate of the depicted explanation would be accounting for supernumerary phantom limbs as ongoing bodily experiences rather than as imagined
bodily experiences, despite appealing to somatosensory information from the past. It is hard for me to see how it could be done.
8
Likewise, how can we make these claims compatible with O’Shaughnessy’s (1980) and de Vignemont’s (2018) acknowledgement of the role of somatosensory
feedback in re-shaping the body map after significant bodily change (as in growth or amputation)?

22
L.A. Murillo Lara Consciousness and Cognition 60 (2018) 17–24

addresses the idea that the spatial content of the body map is action-oriented, or what she calls the hot conception. I am going to call
this argument “the argument from the rough-grainedness of action-oriented body-content”.9 According to this argument, whereas
bodily sensations can be highly focal (involving sometimes quite specific skin areas), action-oriented content about the body (which
de Vignemont, 2016a, calls bodily pushmi-pullyu representations) involves rather groups of segments of the body moving together
(bodily synergies).
On the one hand, as a result from the fact that particular actions bring together body parts (typically effectors), action-oriented
body-content is characteristically related with representations that bring together groups of effectors. On the other hand, the loca-
tions of bodily sensations are specified in terms of well-segmented body parts. As action assembles body parts into functional units,
the spatial structure of action-oriented body-content is rough-grained but the body map that provides spatiality to bodily sensations
has a fine-grained spatial structure. Thence, the rough-grained action-oriented body-content is inadequate for the location of bodily
sensations; the spatiality of the latter cannot be hot.
In a similar vein, de Vignemont (2016a, 2018) uses some empirical results to call into question the relation Bermúdez (1998)
posits between action and the specification of the A-location of sensations by means of the concept of “hinge”. According to Bermúdez
(1998, p. 158), the link between action and the specification of sensations’ A-locations derives from the fact that hinges both afford
possibilities for movement and allow the specification of those locations. De Vignemont observes that the idea that body parts are
individuated by joints may indeed be supported by the fact that there is more sensitivity to stimulation around joints—to the point
that cross-joint distances are overestimated. As Le Cornu Knight, Longo, and Bremner (2014) as well as de Vignemont, Majid, Jola,
and Haggard (2009) have noted, when stimuli are applied across the wrist, the perceived tactile distance is larger than when they
were applied within a single body part (forearm or hand). Consistent with Bermúdez segmentation, Le Cornu Knight, et al. (2014)
conclude that this effect is indicative of a mental representation of the body segmented into discrete parts delineated by joints
modulating tactile spatial perception. However, de Vignemont casts doubt on the idea that this higher sensitivity effect comes from
the role joints play for action. What de Vignemont, et al. (2009) found is that action actually reduces the overestimation of cross-joint
distances. Their idea seems to be that even if joints play a role in structuring bodily sensations, it does not follow that this role is
dependent on the function that joints play in action.
Nonetheless, in order to establish that idea, what must be shown is that action reduces sensitivity to stimulation around joints and
not the overestimation of cross-joint distances. It could still be said that in reducing overestimation, action makes the estimation of
cross-joint distances more accurate, thus specifying the precise availability of body parts for action.
Let us concede for the sake of the argument that the rough-grainedness of action-oriented body-content achieves to show that the
spatiality of bodily sensations cannot be hot. But, does this argument prevent other aspects of our grasp of the spatiality of our own
bodies from being hot? We might distinguish between (a) the spatiality of bodily sensations (that is, their location) and (b) our
immediate grasp of spatial properties of body parts. This distinction is based on the simple fact that whereas in (a) there is no question
about whether sensations play a role in our knowledge of the location of sensations, in (b) there are reasonable doubts about whether
sensations play a role in the acquisition of knowledge of spatial properties of body parts (or at least it seems conceptually coherent;
see Anscombe, 1957, 1962, Bermúdez, 1998).
Were we to grant the distinction between (a) and (b), the first thing that we should note is that the argument from the rough-
grainedness of action-oriented body-content is devised to prove that the spatiality of bodily sensations (i.e., (a)) is not hot, and
argument is needed to show that what applies to (a) applies to (b) as well (that is, to the immediate spatial content about properties of
body parts).
Furthermore, de Vignemont (2016a, 2018) herself grants the existence of a hot body map, a body representation whose content is
used to guide action. Its content, needless to say, must be hot. So we have at least two body-maps, and the content of at least one of
them is hot. The argument from the rough-grainedness of action-oriented body-content would show that the spatiality of bodily
sensations cannot be hot, not that the content of a body map cannot be hot. Besides, if we take into account the indispensability of our
grasp of spatial properties of body parts for guiding action, then (b) seems to be a fitting candidate for content of the hot body map.
Consequently, we have to distinguish within the body map theory between a hot body map—whose content could be our im-
mediate grasp of spatial properties of body parts—and a ‘cold’ body map, whose content would be used for the location of bodily
sensations. De Vignemont’s argument from the rough-grainedness of action-oriented body-content would amount to the claim that
there is more than one type of bodily spatial representation/content and that only one of them is actions-oriented (or hot).

5. Conclusions

After introducing the problem of the location of bodily sensations, I started by examining O’Shaughnessy’s (1980, 1995) notion of
body image, one representationalist solution to the problem. According to O’Shaughnessy (1995), the body image is the best ex-
planation for the spatiality bodily sensations; bodily sensations would be the result of the conjunction of somatosensory information
and the already spatially endowed body image. Even so, we realized that O’Shaughnessy’s (1995) explanation does not discard an
anaesthetic knowledge of the spatial properties of one’s body.

9
The poverty of somatosensory input and the coarseness of action-oriented content are clearly different claims: whereas in the former the possession of content is at
issue, the second is by definition contentful –what is at issue is the capacity of such content to explain the detail of bodily sensation. What they have in common is the
idea that the spatiality of bodily sensation is rich. However, the role played by that premise is different in each argument: in the first, it leads to the postulation of a
body representation, in the second leads to a conclusion about the content of this very representation.

23
L.A. Murillo Lara Consciousness and Cognition 60 (2018) 17–24

In the second section, we dismissed an alleged objection to the representationalist approach to the location of bodily sensations.
We appreciated that a global body map is necessary at least to the extent that a bunch of local representations are unable to account
neither for the background experience of the body as a whole nor for the unity of behavior.
I discussed two of de Vignemont’s arguments to the effect that the spatial aspects of our bodies are best explained by means of a
body representation that she calls the body map (de Vignemont, 2017, 2016a, 2015, 2014a, 2018). According to the first and central
argument, sensory input displays a spatial poverty that makes it insufficient for explaining our knowledge of the location of bodily
sensations, making pertinent the postulation of the body map. This poverty, however, does not necessarily mean that the deliverances
of somatosensory input are raw feelings (although a proper account of their spatial content has yet to be provided); and even if it were
the case, somatosensory input is not bodily sensation. It has to do with the fact that the existence of a body map does not affect the
claim that the spatiality of bodily sensations is an intrinsic (non-relational) property, because the body map is a constitutive element
of bodily sensations.
Nonetheless, I have also underscored that the argument from the spatial poverty of sensory input might imply that somatosensory
information, besides not being enough for having bodily sensations, is not indispensable for bodily sensation. Moreover, I showed
that the argument from the spatial poverty of sensory input is sometimes misread as implying that body representation shapes bodily
experiences regardless of the ongoing somatosensory feedback. Indeed, somatosensory feedback influences both bodily sensations
and the body map.
The second of de Vignemont’s arguments states that hot body content’s rough-grainedness makes it inadequate to explain the fine-
grainedness of the felt location of bodily sensations. However, we saw that this argument does not impede hot (action-oriented) body
contents from capturing our knowledge of the spatial properties of our body and body parts. As a result, we could agree with de
Vignemont in distinguishing a cold body map—which is used for the location of bodily sensations—from a hot body map—which is
used to guide action.
In summary, there seems to be good reason to grant a global representation of the body in order to account for the location of
bodily sensations. However, in spite of what O’Shaughnessy (1995) and some of his readers believe, the representationalist approach
is compatible with an anscombian anaesthetic knowledge of the spatial properties of one’s body. As regards, de Vignemont’s version of
the representationalist approach, it is desirable to dispel any ambiguity regarding the spatial content of somatosensory input.
Moreover, in her version of the approach the tension as regards the role of somatosensory input is rather clear: whether it is a
necessary condition for bodily sensation or not and whether body representation may shape bodily experiences regardless of so-
matosensory input. The answer to the latter question, I think, is no.

References

Alsmith, A. J. (2009). Acting on (bodily) experience. Psyche, 1(15), 82–99.

Anscombe, G. E. (1957). Intention. Cambridge, MA: Harvard University Press.
Anscombe, G. E. (1962). On sensations of position. Analysis, 55–58.
Armstrong, D. (1962). Bodily Sensations. London: Routledge and Paul.
Bermúdez, J. L. (1998). The paradox of self-consciousness. Cambridge, MA: MIT Press.
de Vignemont, F. (2017). Pain and touch. The Monist, 100(4), 465–477.
de Vignemont, F. (2016a). Bodily experiences and bodily affordances. In Y. Coello, & M. Fischer (Eds.), Foundations of Embodied Cognition (Vol. 1: Perceptual and
Emotional Embodiment, pp. 149-164). Routledge.
de Vignemont, F. (2016b). Bodily awareness. In E.N. Zalta (Ed.). Retrieved December 21, 2017, from The Stanford Encyclopedia of Philosophy (Summer 2016
Edition). < https://plato.stanford.edu/archives/sum2016/entries/bodily-awareness/ > .
de Vignemont, F. (2014a). A multimodal conception of bodily awareness. Mind, 123(492), 989–1020.
de Vignemont, F. (2014b). Acting for bodily awareness. In: L. Shapiro (Ed.), The Routledge Handbook of Embodied Cognition (pp.287-295). Routledge.
de Vignemont, F. (2010). Body schema and body image: Pros and cons. Neuropsychologia, 3(48), 669–680.
de Vignemont, F. (2018). Mind the body. Oxford: Oxford University Press.
de Vignemont, F., & Massin, O. (2015). Touch. In M. Matthen (Ed.). The Oxford handbook of philosophy of perception (pp. 294–311). Oxford University Press.
de Vignemont, F., Majid, A., Jola, C., & Haggard, P. (2009). Segmenting the body into parts: Evidence from biases in tactile perception. The Quarterly Journal of
Experimental Psychology, 62(3), 500–512.
Gallagher, S., & Cole, J. (1995). Body image and body schema in a deafferented subject. Journal of mind and behavior, 16, 369–390.
Le Cornu Knight, F., Longo, M. R., & Bremner, A. J. (2014). Categorical perception of tactile distance. Cognition, 131(2), 254–262.
Marchetti, C., & Della Salla, S. (1998). Disentangling the alien and the anarchic hand. Cognitive Neuropsychiatry, 3(3), 191–207.
Martin, M. (1993). Sense modalities and spatial properties. In N. Eilan, R. Mc-Carthy, & B. Brewer (Eds.). spatial representation (pp. 206–218). Oxford: Basil Blackwell.
Mather, G. (2016). Foundations of sensation and perception. New York: Psychology Press.
Miyazawa, N., Hayashi, M., Komiya, K., & Akiyama, I. (2004). Supernumerary phantom limbs associated with left hemispheric stroke: Case report and review of the
literature. Neurosurgery, 54(1), 228–231.
O'Shaughnessy, B. (1980). The will, Vol. I. Cambridge: Cambridge University Press.
O’Shaughnessy, B. (1989). The sense of touch. Australasian Journal Journal of Philosophyphilosophy, 1(67), 37–58.
O’Shaughnessy, B. (1995). Proprioception and the body image. In J. L. Bermundez, A. Marcel, & N. Eilan (Eds.). the body and the self (pp. 175–203). Cambridge, MA:
MIT Press.
Russell, B. (1912). The problems of philosophy. Oxford: Oxford University Press.

24