Knowledge in Perception and Illusion
Knowledge in Perception and Illusion
Knowledge in Perception and Illusion
RIC H A R D L. GR EG ORY
Department of Psychology, University of Bristol, 8 Woodland Road, Bristol BS8 1TN, UK
SU M M A RY
Following Hermann von Helmholtz, who described visual perceptions as unconscious inferences from
sensory data and knowledge derived from the past, perceptions are regarded as similar to predictive
hypotheses of science, but are psychologically projected into external space and accepted as our most
immediate reality. There are increasing discrepancies between perceptions and conceptions with science's
advances, which makes it hard to dene `illusion'. Visual illusions can provide evidence of object knowledge
and working rules for vision, but only when the phenomena are explained and classied. A tentative clas-
sication is presented, in terms of appearances and kinds of causes.
The large contribution of knowledge from the past for vision raises the issue: how do we recognize the
present, without confusion from the past. This danger is generally avoided as the present is signalled by
real-time sensory inputsperhaps agged by qualia of consciousness.
Phil. Trans. R. Soc. Lond. B (1997) 352, 1121^1128 1121 & 1997 The Royal Society
Printed in Great Britain
1122 R. L. Gregory Knowledge in perception and illusion
mask is rotated, or the observer moves, it appears to such as perspective, and is normally applied to the
rotate in the opposite to normal direction, at twice the world of objects but is activated by the patterns of paint.
speed; because distances are reversed motion parallax
becomes eectively reversed. This also happens with a
3. W H AT A R E I L LU S ION S ?
depth-reversed wire cube.)
It is signicant that this, and very many other illu- It is extraordinarily hard to give a satisfactory deni-
sions, are experienced perceptually though the tion of an `illusion'. It may be the departure from
observer knows conceptually that they are illusory reality, or from truth; but how are these to be dened ?
even to the point of appreciating the causes of the As science's accounts of reality get ever more dierent
phenomena. This does not, however, show that know- from appearances, to say that this separation is `illu-
ledge has no part to play in vision. Rather, it shows sion' would have the absurd consequence of implying
that conceptual and perceptual knowledge are largely that almost all perceptions are illusory. It seems better
separate. This is not altogether surprising because to limit `illusion' to systematic visual and other sensed
perception must work extremely fast (in a fraction of a discrepancies from simple measurements with rulers,
second) to be useful for survival, though conceptual photometers, clocks and so on.
decisions may take minutes, or even years. Further, There are two clearly very dierent kinds of illu-
perceptions are of particulars, rather than the general- sions: those with a physical cause and cognitive
ities of conceptions. (We perceive a triangle, but only illusions due to misapplication of knowledge. Although
conceptually can we appreciate triangularity.) Also, if they have extremely dierent kinds of causes, they can
knowledge or belief determined perception we would produce some surprisingly similar phenomena (such as
be blind to the unusual, or the seemingly impossible, distortions of length or curvature), so there are dicul-
which would be dangerous in unusual situations, and ties of classication that require experimental evidence.
would limit perceptual learning. Illusions due to the disturbance of light, between
The distinguished biologist J. Z. Young was a pioneer objects and the eyes, are dierent from illusions due to
who stressed the importance of handling knowledge for the disturbance of sensory signals of eye, though both
understanding brain function, and that there may be a might be classied as `physical'. Extremely dierent
`brain language' preceding spoken or written language. from both of these are cognitive illusions, due to misap-
Thus (Young 1978, p. 56): `If the essential feature of the plied knowledge employed by the brain to interpret or
brain is that it contains information then the task is to read sensory signals. For cognitive illusions, it is useful
learn to translate the language that it uses. But of course to distinguish specic knowledge of objects, from general
this is not the method that is generally used in the knowledge embodied as rules. Either can be mislead in
attempt to understand the brain. Physiologists do not unusual conditions, and so can be revealed by observa-
go around saying that they are trying to translate tion and experiment. An example of misleading specic
brain language. They would rather think that they are knowledge is how a grainy texture is seen as wood,
trying to understand it in the ``ordinary scientic terms though it is a plastic imitation or a picture. More
of physics and chemistry''.' Cognitive illusions reveal dramatic is how a hollow face or mask is seen as
knowledge and assumptions for vision, and perhaps convex (gure 1), because faces are very rarely hollow.
take us close to `brain language', but they must be (Evidently the perceptual hypothesis of a face carries
understood and also classied. Classifying is important the, not always appropriate, knowledge that it is
for the natural sciences: it should be equally important convex.) Examples of misleading rules are the Gestalt
for the `unnatural science' of illusions. laws of `closure', `proximity', `continuity' and the
Classifying must be important for learning and `common fate' of movements of parts of objects
perception, for it is impossible to make inductive gener- (Wertheimer 1923, 1938). When these do not apply illu-
alizations without at least implicit classes. It is also sion can result, because not all objects are closed in
impossible to make deductive inferences, as deductions form, with close-together parts and continuous edges,
are not from facts or events, but from descriptions (in or with parts moving together as leaves of a tree in the
words or mathematics) of real or imaginary members wind. Exceptional objects are mis-seen when Gestalt
of classes. Von Helmholtz's `unconscious inference' for laws are applied, and when perspective rules are
vision was inductive; for example inferring distances applied for atypical objects, such as the Ames window
from perspective and shapes from shading. As there and at projections of pictures.
are frequent exceptions certainty is not attainable.
Thus atypical shapes give systematic errors, when
general rules or specic knowledge are inappropriate 4 . ` I N S -A N D - OU T S '
for these unusual objects or scenes, as shown most To the usual terms `bottom-up' signals and `top-
dramatically by the Ames demonstrations such as the down' knowledge, we add what might be called `side-
Ames window (Ittelson 1952). (This is a slowly rotating ways' rules. Both top-down and sideways are
trapezoid, the shape of a rectangle as viewed from an knowledge; the rst specic (such as faces being
oblique angle. It changes bizarrely in size and form as convex), the second being general rules applied to all
it does not go through the usual perspective transfor- objects and scenes (such as the Gestalt laws and
mations of a familiar rectangle, such as a normal perspective). These are `ins-and-outs' of vision, which
window.) Much the same applies to seeing familiar it might be useful to consider, before attempting to
objects in the very dierent brush strokes of paintings; explain how the visual brain works, with the scheme
this is evidently seen by object knowledge and rules, presented in gure 2.
physics knowledge
(xii) Kanizsa triangle and many other illusory contours A `downwards' loop is also shown, from the
and surfaces. Some are due to `postulating' a nearer prevailing perceptual hypothesis, aecting bottom-up
occluding surface, to `explain' surprising gaps (Gregory signal processing. This may be demonstrated by the
1972; Petry et al. 1987). change of apparent brightness with depth-reversal of
(xiii) Hollow face. This illustrates the power of prob- the Mach's corner illusion (gure 5). Though as
abilities (and so knowledge) for object perception Barlow points out (personal communication, 1997) this
(gure 1). is not necessarily the explanation; it requires experi-
(xiv)Size ^ weightillusion. Small objects feel heavier than ments.
larger objects of the same scale weight; muscles are set by
knowledge-based expectation that the larger will be hea-
6 . QUA L I A
vier, which is generally, though not always true.
(xv) Magritte mirror. Rene Magritte's painting La repro- Most mysterious of all brain phenomena is
duction interdite (1937) shows a man facing a mirror, but the consciousness, especially how sensations, qualia, are
back of his head appears in the glass.This looks impossible produced and their possible uses.
from ourknowledge of mirrors (Gregory1997). In the account given here, perception depends very
(xvi) Faces-in-the-re, ink blots, galleons in the clouds largely on knowledge (specic `top-down' and general
and so on, show the dynamics of perception. Hypotheses `sideways' rules), derived from past experience of the
are generatedthat go fancifully beyondthe evidence. individual and from ancestral, sometimes even pre-
The Cafe wall distortion, due to disturbed neural human experience. So perceptions are largely based on
signals, is shown in gure 3a, for comparison with the the past, but recognizing the present is essential for
knowledge rules-distortion of the Muller^ Lyer distor- survival in the here and now.
tion (gure 3b) and the specic-knowledge distortion The present moment must not be confused with the
of the size ^ weight illusion (gure 3c). They may past, or with imagination, i.e. as indeed one appreciates
appear similar (all being distortions) but their causes when crossing a busy road. So, although knowledge
are fundamentally dierent. from the past is so important, it must not obtrude into
We may develop the `at box' of ins-and-outs (gure the present. Primitive non-cognitive animals have no
2) to a fuller `black box' (gure 4), These diagrams do such danger of confusion, as their present is simply
not attempt to show anatomical paths or brain regions, signalled by real-time aerent inputs. Time-confusion
but rather, functional ins-and-outs of vision. is likely only for `higher' animals, especially humans,
conceptual knowledge
feedback learning
from failures
he
and successes
i n gt
gg
perceptual knowledge fla t?
sen
pre
a
top-down ali
qu
r
u
l sideways output to behaviour
hypothesis
e
generator
s
objects explored
task
for
set
signal processing top-down loop
(many channels)
bottom-up
from the eyes
sensed reality
Figure 4. Ins-and-outs: black box of vision. The scheme of gure 2 with additions: set, for selecting needed knowledge;
qualia, perhaps for signalling the present.
R E F E R E NC E S
Barlow, H. B. 1997 The knowledge used in vision: and where
it comes from. Phil. Trans. R. Soc. Lond. B. 352, 1143^1149.
(This volume.)
Boring, E. G. 1950 A history of experimental psychology, 2nd edn.
New York: Appleton Century Crofts.
Gregory, R. L. l963 Distortion of visual space as inap-
propriate constancy scaling. Nature 199, 678^691.
Gregory, R. L. l968a Perceptual illusions and brain models.
Proc. R. Soc. Lond. B 171, 179^196.
Gregory, R. L. l968b On how so little information controls so
much behaviour. In Towards a theoretical biology 2 (ed. C. H.
Waddington). Edinburgh: University of Edinburgh Press.
Gregory, R. L. 1970aThe intelligent eye. London: Weidenfeld &
Nicolson.
Gregory, R. L. 1970b The grammar of vision. Listener 83, 242.
Reprinted in R. L. Gregory 1974 Concepts of vision, pp. 622^
629. London: Duckworth.
Gregory, R. L. l972 Cognitive contours. Nature 238, 51^52.
Gregory, R. L. l980 Perceptions as hypotheses. Phil. Trans. R.
Soc. Lond. B 290, 181^197.
Gregory, R. L. 1997 Mirrors in mind. Oxford: Spektrum/New
York: W. H. Freeman.
Gregory, R. L. & Harris, J. P. l975 Illusion-destruction by
appropriate scaling. Perception 4, 203^220.
Gregory, R. L. & Heard, P. 1979 Border-locking and the Cafe
Wall illusion. Perception 4, 203^220.
Figure 5. Mach's corner. The dark region changes Helmholtz, H. von 1866 Concerning the perceptions in
apparent brightness when the corner ips in or out; it is general. In Treatise on physiological optics, vol. III, 3rd edn
brighter when in, and so a likely shadow, although there is (translated by J. P. C. Southall 1925 Opt. Soc. Am. Section
no physical change (Mach 1897). 26, reprinted New York: Dover, 1962).
Humphries, G. W. & Riddock, M. J. 1987a The fractionation
of visual agnosia. In Visual object processing: a cognitive neuro-
where knowledge derived from the past dominates psychological approach (ed. G. W. Humphries & M. J.
present perception. As for primitive (reex and Riddock), pp. 281^306. London: Lawrence Erlbaum.
Humphries, G. W. & Riddock, M. J. 1987bTo see but not to see: a
tropism-controlled) animals our present is also case study of visual agnosia. London: Lawrence Erlbaum.
signalled by real-time aerent inputs, but as input Ittelson, W. H. 1952 The Ames demonstrations in perception.
signals have a smaller part to play than knowledge Princeton University Press.
from the past, for cognitive perception, they must be Luria, A. R. 1969 The mind of a mnemonist. London: Penguin.
very clearly distinguished. (Exceptions are qualia in Mach, E. 1897 The analysis of sensations (English translation
dreams and in schizophrenic hallucinations. There are 1959, 5th edn). New York: Dover.
rare cases (Luria 1969) of individuals having such vivid Marr, D. 1982 Vision. New York: W. H. Freeman.
Nijhawan, R. 1997 Visual decomposition of colour through
memory that their present is dangerously confused with motion extrapolation. Nature 386, 66^69.
their past and with imagination. Memories of emotion Petry, S. & Meyer, G. E. (eds) 1987 The perception of illusory
such as embarrassment can evoke qualia, perhaps from contours. New York: Springer-Verlag.
real-time signals from visceral changes or blushing Pinker, S. 1994 The language instinct. London: Allen Lane/
evoked by memory.) As a speculation: are real-time Penguin.
sensory signalsand so the presentagged by the Sacks, O. 1985 The man who mistook his wife for a hat. London:
Duckworth.
vividness of qualia?
Sillito, A. 1995 Chemical soup: where and how drugs may
It is interesting to compare the qualia of seeing, with inuence visual perception. In The artful eye (ed. R. L.
memory of a scene immediately the eyes are closed. Gregory, J. Harris, P. Heard & D. Rose), pp. 294^306.
Surely the visual qualia almost if not entirely disappear Oxford University Press.
when the sensory inputs cease. Reversing this simple Wertheimer, M. 1923 Untersuchungen zur Lehre von Gestalt
experiment by opening the eyes following immediate II. Psychol. Forsch 4, 301^350. Transl. 1938 Organization of
memory, the onset of the visual qualia is so striking perceptual forms. In A source book of Gestalt psychology (ed.
W. D. Ellis), pp. 71^88. London: Routledge and Kegan
that they make the memory pale by comparison. So Paul.
perhaps consciousness serves to avoid confusion with Young, J. Z. 1978 Programs of the brain. Oxford University
the remembered past, by agging the present with the Press.
unique vividness of sensory qualia. Zeki, S. 1993 A vision of the brain. Oxford: Blackwell.