Professional Documents
Culture Documents
by the phenomenon of heat. He reasoned Notice that the distinction privileges pri-
that, in reality, heat was probably the motion mary qualities as uniquely real, by running
of tiny ‘corpuscles’: atoms, as conceived very fast by the fact that there is no brain-
by Democritus (460–370 BC). Galileo was independent or representation-free access to
struck by the fact that heat is not experienced reality. If colour and smell representations are
in this way. The feeling of heat seems to have the brain’s causal response to certain external
nothing to do with the perceived motion of stimuli, then so are spatial representations
anything. To explain why the appearance and motion representations. The brain can-
does not resemble the reality, he suggested not directly compare its representation of the
that particle motion kicks off a causal process external world with the external world itself, as
in the body that results in something quite we might compare the on-stage Wizard of Oz
different: heat as we experience it. Given this with the man behind the curtain.
and other differences, discovered by science, To be sure, instruments can provide the
between appearance and reality, thinkers brain with further data, as Galileo’s clever
such as John Locke (1632–1704) saw wisdom thermal-expansion thermometer provided
in making a general distinction between him with data concerning heat. But instru-
real-world properties (primary qualities) and mental data require human observation and
brain-constructed properties (secondary theory-backed interpretation, both of which
qualities)10. involve filtering through the lens of represen-
Something akin to the primary–secondary tational models. Hence, ‘objective’ instru-
distinction is still commonly wheeled out in ments are not the general solution to making Figure 2 | The refraction illusion. The pencil
appears to have a kink where it enters the water.
discussion of the conceptual framework out- a principled distinction between primary and
This appearance is explained by the refractive
lined above. Despite its prima facie plausibil- secondary qualities. properties of the water medium. Image courtesy
ity, the distinction carries the seeds of its own Here, then, is the dilemma: if the distinc- of D. Stack.
destruction. tions between inner-me and outer-world are
made within the brain’s representational
models, then how does the distinction between and other German Idealists in the nineteenth
brain-constructed properties and real-world century found their intellectual home at the
properties gain objective grounding? And bottom13. Their efforts went into explaining
why should we believe that primary qualities the apparently ‘physical’ world in terms of the
accurately characterize reality whereas sec- allegedly more basic world of mental ideas.
ondary qualities are merely representational Because some contemporary neuroscientists
fabrications that are incident on the brain’s consider the primary–secondary distinction
interactions with reality? to be unavoidable and the slippery slope to
Idealism to be inevitable, we shall briefly dis-
Into Idealism and out again cuss the disadvantages of Idealism before
The philosopher Bishop Berkeley (1685–1753) returning to the question of the fidelity of
realized that the arguments that support the brain models to the world modelled.
secondary qualities as mind creations start us An obvious and fundamental objection to
down a very slippery slope (FIG. 1). Once on the Idealism is that it cannot account for the
slope, we find that we have slipped to the next coherence and regularity of the (idea of the)
stage: primary qualities, just like secondary external world, or even of one’s own mental
qualities, are nothing but mind-created life. Even simple regularities — objects thrown
responses to a real world, the true nature of in the air regularly fall towards the Earth, dry
which we can never know. At the bottom wood regularly burns, water reliably quenches
of the slippery slope is the proposition that fire — are inexplicable. Berkeley’s solution
the so-called external world is, after all, appealed to the supernatural: God keeps the
nothing more than my idea of an external flux of ideas coherent. But what happens in
world. Ideas are the only things there really deep sleep, or in coma, when ideas vanish?
Figure 1 | The slippery slope to Idealism. The are. Furthermore, my apparently physical What of the existence of the Universe before
slippery slope begins with the seemingly plausible
brain also must be nothing but an idea. In this there were minds? Luckily, Berkeley’s God
distinction between primary and secondary
qualities, which was first conceived by Galileo when
case, only non-physical minds — constella- saves the day by having all of that as ideas in
puzzling about the differences between heat as a tions of ideas — genuinely exist. Classically, His ample mind. The supernatural solution is
subjective experience and heat as an objective this view is known as ‘Idealism’, and it is what transparently ad hoc — a pixie-dust, magic-
property in the world. John Locke was moved by awaits us at the bottom of the slope. wand solution — and Berkeley’s contempo-
similar considerations and explored the distinction Berkeley gleefully took the trip to the bot- raries tore it to shreds. However, unfortunately
further, glimpsing but not worrying too much about tom of the slope, and was an uncompromising for Idealism, no one ever contrived a solution
the slope in front of him. Immanuel Kant struggled
to keep from sliding down, Bishop Berkeley was
Idealist to the end11. In the Critique of Pure that was both intelligible and less ad hoc than
convinced that sliding was the only logical recourse, Reason, Immanuel Kant struggled mightily to Berkeley’s.
and Georg Hegel was a thoroughgoing Idealist. stop two-thirds of the way down the slope, but Idealists have also been unable to account
Image courtesy of M. Churchland. hit bottom even so12. Georg Hegel (1770–1831) for the progress of science in falsifying
observe, think, test, don’t be dogmatic but Traditional semantics, by contrast, assumes the 4. Churchland, P. S. Brain-Wise: Studies in Neurophilosophy
(MIT Press, Cambridge, Massachusetts, 2002).
don’t change your mind too easily, don’t get primary representational relationship to hold 5. Churchland, P. M. The Engine of Reason, The Seat of the
in a rut but don’t give up too soon, take between our internal concepts taken one by Soul (MIT Press, Cambridge, Massachusetts, 1996).
6. Suri, R. E. Anticipatory responses of dopamine neurons
advantage of statistical tools but don’t sup- one, and external features taken one by one. and cortical neurons reproduced by internal model. Exp.
pose that good theories will simply waft up According to the domain-portrayal hypothesis, Brain Res. 140, 234–240 (2001).
7. Damasio, A. R. The Feeling of What Happens (Harcourt
from statistical analyses, and so on26–29. single concepts derive their representational Brace, New York, 1999).
Ultimately, we want to understand in significance entirely from the larger neural 8. Gazzaniga, M. S. Cerebral specialization and
interhemispheric communication. Brain 123, 1293–1326
detail the neural mechanisms by which com- model in which they are embedded. Intuitively, (2000).
mon sense and science develop coherent rep- of course, it may seem otherwise, but ‘folk 9. Galileo The Assayer (1623) in The Scientific Background
to Modern Philosophy: Selected Readings (ed. Matthews,
resentational models, and what ‘coherence’ semantics’ is undoubtedly as misconceived as M. R.) 56–60 (Hackett, New York, 1989).
means in neural terms. Meanwhile, we can were folk physics and folk cosmology. 10. Locke, J. An Essay Concerning Human Understanding
(1690). Modern edition (ed. Nidditch, P. H.) 544–547
begin to see what the metaphor of ‘isomor- The proposal to distinguish between ‘pri- (Oxford Univ. Press, Oxford, UK, 1975).
phisms between models and world’ might mary’ and ‘secondary’ properties runs amok 11. Berkeley, G. in Berkeley’s Philosophical Writings (ed.
Armstrong, D.) (Macmillan, New York, 1965).
amount to in non-metaphorical terms. because it attempts to explain successful repre- 12. Kant, I. Critique of Pure Reason (1781). Modern edition
Assume that a representational model can sentation in terms of a resemblance relation- (eds Guyer, P. & Wood, A.) (Cambridge Univ. Press,
Cambridge, UK, 1998).
be characterized in terms of a parameter ship between inner concepts and outer prop- 13. Hegel, G. Phenomenology of Spirit (1807). Translated by
space, the dimensions of which are those neu- erties taken one by one. The slippery slope to Miller, A. V. (Oxford Univ. Press, Oxford, UK, 1977).
14. Kaufman, L. & Rock, I. The moon illusion: I. Science 136,
rons that participate in the model. Distinct Idealism is avoided not by working ever more 953–961 (1962).
representations (for example, the face of Clint feverishly to establish the resemblance between 15. Palmer, S. E. Vision Science (MIT Press, Cambridge,
Massachusetts, 1999).
Eastwood and the face of Sophia Loren) will individual primary qualities and individual 16. Purves, D., Lotto, R. B., Williams, S. M., Nundy, S. &
involve distinct patterns of activation across properties in the real world, but by looking Yang, Z. Why we see things the way we do: evidence for
the participating neurons and will thus indi- to the general function of representational a wholly empirical strategy of vision. Phil. Trans. R. Soc.
Lond. B 356, 285–297 (2001).
viduate distinct points in the neuronal para- domain models in nervous systems. This ush- 17. Ramachandran, V. S. Filling gaps in perception: part II.
meter space. In a neuronal parameter space ers in the recognition that representational Scotomas and phantom limbs. Curr. Dir. Psychol. Sci. 2,
56–65 (1993).
for faces, for example, the activation pattern utility depends on the higher-order, multi- 18. von Helmholtz, H. L. F. Treatise on Physiological Optics
for Queen Elizabeth’s face will be closer to the dimensional ‘resemblances’ that nervous sys- (1866). Translated by Southall, J. P. C. (Dover, New York,
1925).
pattern for Sophia Loren’s face than to that tems produce; that is, on the relative richness 19. Gregory, R. L. The Intelligent Eye (McGraw–Hill, New
for either Clint Eastwood or George Bush. In of the isomorphism between the representa- York, 1970).
20. Shepard, R. N. & Chipman, S. Second-order
the neuronal taste space, the activation pat- tional model and the world modelled, as isomorphism of internal representations: shapes of
terns for apricots, raspberries and honey clus- indexed by the model’s predictive profile. states. Cogn. Psychol. 1, 1–7 (1970).
21. Glymour, C. The Mind’s Arrows: Bayes Nets and
ter together, and are more distant from the Domain-portrayal semantics is, so far, Graphical Causal Models in Psychology (MIT Press,
cluster of patterns for bitter things, such as described only in the most general terms. Cambridge, Massachusetts, 2001).
22. O’Brien, G. & Opie, J. in Representations in Mind: New
choke cherries, quinine and urea, which in Nevertheless, reorienting semantics away Approaches to Mental Representation (eds Clapin, H.,
turn are separated from the cluster of patterns from the one-by-one paradigm and towards a Staines, P. & Slezak, P.) (Greenwood, Westport,
Connecticut, in the press).
for salty things. model-to-domain paradigm that is more con- 23. Sutton, R. S. & Barto, A. G. Reinforcement Learning:
Whence isomorphism? The various dis- silient with current neuroscience motivates an Introduction (MIT Press, Cambridge, Massachusetts,
1998).
tance relationships between the learned clus- continued exploration of specific neuronal 24. Hutchins, E. Cognition in the Wild (MIT Press, Cambridge,
ters (prototype points), within the activation populations to discover what relationship- Massachusetts, 1995).
25. Tomasello, M. The Cultural Origins of Human Cognition
space of a given population of neurons, are preserving mapping constitutes the particular (Harvard Univ. Press, Cambridge, Massachusetts, 1999).
collectively and literally isomorphic with the representational success of particular neuronal 26. Bacon, F. Novum Organum (1660). Modern edition (eds
Jardine, L. & Silverthorne, M.) (Cambridge Univ. Press,
similarity relationships that objectively exist populations. Contrary to both Locke and the Cambridge, UK, 2000).
between the various categories in the external Idealists, the mind–brain does model the real 27. Medawar, P. B. Advice to a Young Scientist (Basic Books,
New York, 1981).
world. The same holds for causal regularities world, including that part of the world that 28. Crick, F. H. C. What Mad Pursuit: a Personal View of
in the world and prototype trajectories in neu- is the mind–brain. The reality–appearance Scientific Discovery (Basic Books, New York, 1988).
ronal-activation space. The greater the degree distinction ultimately rests on comparisons 29. Spirtes, P., Glymour, C. & Scheines, R. Causation,
Prediction, and Search (Springer, New York, 1993).
of isomorphism, the greater the fidelity of the between the predictive merits of distinct repre-
model to the world. As we cannot directly sentational models, and the best explanation Acknowledgements
We are grateful for advice and assistance from F. Crick,
compare model and world modelled, predic- for why one theory out-predicts another is that A. Damasio, L. Goble, E. McAmis and S. Rickless.
tive success is the measure of fidelity and the one theory is closer to the truth than the other.
guide to the need for model revision. Patricia S. Churchland and Paul M. Churchland Online links
Somehow, the various mechanisms of neu- are at the Philosophy Department, FURTHER INFORMATION
ronal plasticity, including dendritic growth, University of California San Diego, Al Seckel’s laboratory: http://neuro.caltech.edu/~seckel/
the emergence of new synapses, changes in the La Jolla, California 92093, USA. Dale Purves’s laboratory: http://www.purveslab.net/
Correspondence to P.S.C. Donald D. Hoffman’s laboratory:
probability of vesicle release and changes in http://aris.ss.uci.edu/cogsci/personnel/hoffman/
e-mail: pschurchland@ucsd.edu
transmitter released per spike, are orchestrated Edward H. Adelson’s laboratory:
doi:10.1038/nrn958 http://www-bcs.mit.edu/people/adelson/
to enhance the fidelity of the basic model. 1. Hoffman, D. D. Visual Intelligence (Norton, New York, Grand Illusions: http://www.grand-illusions.com/
We call this hypothesis ‘domain-portrayal 1998). Illusionworks: http://psylux.psych.tu-dresden.de/i1/kaw/
2. Wolpert, D. M., Ghahramani, Z. & Jordan, M. I. An internal diverses%20Material/www.illusionworks.com/
semantics’, because it proposes that the pri- model for sensorimotor integration. Science 269, MIT Encyclopedia of Cognitive Sciences:
mary representational relationship holds 180–182 (1995). http://cognet.mit.edu/MITECS/
3. Grush, R. Self, world and space: on the meaning and Kant, Immanuel
between the high-dimensional map as a whole, mechanisms of egocentric and allocentric spatial Richard L. Gregory online: http://www.richardgregory.org/
and the categorical/causal domain as a whole. representation. Brain Mind 1, 59–92 (2001). Access to this interactive links box is free online.