Visual
illusions and effects.
Osvaldo da Pos
In:
L. Sivik (Ed.), AIC 1996 - Colour
and Psychology, Scandinavian Colour Institute AB, Stockholm, 1996, 114-118
About terminology and related
problems.
The following reflections
can be considered a philosophical speculation of little interest to people
working in the field of colour perception. Nevertheless I would like to submit
the reader some observations on the terminology almost all people use speaking
about perception and specifically vision. The awareness of the problems which
such terminology poses is a starting point to become more cautious in defending
a particular theoretical position. It seems that interest in illusions in
perception depends on the theory followed by the researchers. If one considers
illusions as mistakes either of our senses or of our mind, his interest will be
much less lively than that of people who consider illusions striking occasions
to disclose the normal functioning of our mind.
To start with a
psychological approach to illusions, the first distinction which I consider
relevant regards perception as opposed to cognition, i.e. seeing vs knowing. At
first sight one could have the impression that perceiving is in any case a sort
of knowing, and especially in human beings perception cannot be separated by
higher forms of knowledge. One of the most well known scientist of our time who
stresses the role of cognition in perception, and therefore in illusions, is R.
Gregory (1993, see also D.D. Hoffmann, 1983). Perceiving means, for him as for
other researchers, trying to solve some problems, making hypothesis, and on the
basis of some more or less justified assumptions deriving a conclusion. The
main difference between perception and cognition would be only in the
simplicity and velocity of the first. On the other side different 'direct'
approaches to perception (gestalt and ecological psychology, for instance), do
not see perception as involving mediatory cognitive processes. The two positions
seem to be irreconcilable. It seems extremely difficult that people belonging
to one tradition will ever be able to understand and accept the antagonistic
one (Kuhn, 1970).
Let us consider the following
example suggested by Kanizsa's work (da Pos, 1996). What is people usually
perceiving when looking at Fig. 1? Many times patterns are seen like alphabetical
characters (K, V, Y) and arrows ( -->, <-- ) differently oriented. Is it possible, maybe
with an effort of attention and imagination, to see a single, more simple
figure? Some conceptual cues can be added to help the viewer: the figure is
a regular 3D object, with straight and perpendicular lines, regular faces,
and so on. Moreover, we can look at an analogous display as that of fig. 2,
and pose the same question: beyond segments and Xs, which simple geometrical
3D figure can be seen?. It is almost certain that the viewer has already experienced
the intended result, although it does not appear despite many kinds of cognitive
efforts (recourse to geometry, physics, memory and so on). Nevertheless, without
changing a single line, we can add a series of segments as those depicted
in Fig. 3 either to Fig. 1 or to Fig. 2, and the 3D object appears well evident
(see Fig. 4 and Fig. 5). Is in this case the parallelepiped seen or imagined,
maybe deduced as an effect of inference process? We can now go back to Fig.
1 (or to Fig. 2), and try to reach the same perceptive results, i.e. to perceive
the parallelepiped: all the necessary cues are there and nevertheless, despite
knowing the possibility of perceiving that geometrical shape, it does not
appear. It is also very difficult just imagining it. What we have learnt from
looking at Fig. 4 and 5 is that some segments of Fig. 1 and 2 can change their
figural role, and instead of closing the Ks Vs Xs patterns, enter in different
sets of lines which, in combinations with the segments of Fig. 3, induce the
appearance of a series of stripes covering the parallelepipeds. Following
a current style of reasoning, we can point out some T junctions which would
induce the perception of a partially hidden, occluded object. But now that
we know this possibility, is it possible to perceive the parallelepiped looking
again at Fig. 1 (or Fig. 2)? This knowledge, this experience which we could
put in the series of premises from which appropriate inferences can be derived,
are not helping us: the parallelepiped is not visible in Fig. 1. Knowing is
not enough to perceive. What is still amazing is that not only by adding some
sets of lines as done before, but also by subtracting from Fig. 1 and Fig.
2 some segments, the parallelepiped becomes again visible (see Fig. 6 and
7; better if a black paper is put behind the page and the viewer adopts a
global attitude). Here again the parallelepiped appear partially covered,
occluded by white stripes, and different explanations of this result have
been given by visual scientists. But also after this experience, and after
the knowledge we could have so derived, it is still impossible to perceive
the parallelepiped looking at Fig. 1 and 2. The main point of this demonstration
is not how the effect is explained, but the subjective difference between
knowing and perceiving, which should convince also the most reluctant that,
at least sometime, seeing and knowing belong to basically different modalities
by which we interact with the external world (of course this does not exclude
that sometimes seeing can be affected
by knowledge, but this is not the rule).
Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig.7
Moreover, there are some
well documented situations in which what we perceive not only cannot be derived
by abstract knowledge and experience as before shown, but is also in contrast
with what we know very well; and sometimes a firmly consolidated knowledge
cannot change an incompatible perception. Perception is considered by most
scientists an unquestionable stage of our relation with the environment, it
cannot be wrong: only our following reasoning can be subject to errors, for
instance deriving wrong conclusions from limited experience. What could be
considered errors are normal ways of functioning: for instance some old pocket
calculators sometimes give quite odd results which are only the effect of the
particular software implemented. By noticing those unexpected but systematic
results we can understand how the calculator is operating. Correspondingly it
does not seem correct to consider some perceptual results as a consequence of
wrong operations just because they are not in accordance with the physical
reality: if those performances are systematic, they reveal the normal way of
working of perception. This does not imply that we can always rely on our
perception: due to the evolutive limit of perception, a behaviour based on it
is not always guaranteed from unsuccess. On the one side learning, and on the
other side abstract or scientific knowledge, which goes also beyond the simple
learning, are in fact extraordinary tools to overcome the limitations of
perception.
Some people would say
that perception is in any case, also in animals, an intelligent activity, and
therefore perceptive processes, sharing the inferential nature of intelligence,
can be subject to errors. But everything in this world has an intelligent or
rational structure and functioning, for instance the movement of planets is
wisely guided by the gravity forces they are subject to. Nevertheless we would
not use the concept of inference in explaining such mechanical behaviours (the
attainment of an equilibrium state could be in fact be considered as a solution
to a problem, but the use of analogy in such a wide sense does not seem to be
scientifically productive). Therefore, also in perception more specific
concepts seem appropriate than those (like inference) applied to abstract
knowledge.
Secondly there is a terminological
problem which could make people reluctant to treat the argument. One of the
most ambiguous word used in describing our experience is the term 'appearance',
which has at least two completely different meaning: firstly it denotes the
perceptive characteristics that objects show in our direct experience of the
world. In this phenomenological sense (see also Goodman, 1977) there is no
difference in saying that one thing appears or it is simply perceived. On the
other hand the word appearance has a long history both in western and eastern
philosophy: appearance is the contrary of reality. One thing is the appearance
and another thing is the reality: for many thinkers we have no access to the
reality and are confined in a world of appearance only. Most scientists, at
least at the present times, consider themselves as realist, in so far reality
exists independently form the observer and can be grasped by our intellect. As
often perception is not considered as an autonomous function of our brain, independent
from abstract cognition, the same laws of functioning which are applicable to
abstract reasoning are thought to work in perception too, because of the unity
of the human being. This assumption just because it sounds logical, seldom
becomes object of verification.
Another source of
misunderstanding is the ambiguity of the term representation, which is often
used to maintain that content of perception is not the reality but a sign of
it, a 'very similar sign' (Caivano, in press). In this context the discussion
can be developed about the concept of image vs reality. Again the image is not
usually considered reality, but only a tool to grasp the reality. This is why,
according to many people, we should have 'to interpret the image' in order to
arrive at the reality: still more, we should have to interpret the raw data of
our senses to build an image, which already appears somehow structured, but is
still only the appearance of the world. Were this point of view true, we never
perceive the reality, because in our eyes we only have images and nothing else.
Why then looking at bidimensional pictures is considered quite different from
looking at 'real' three-dimensional objects? This question is central when the
discussion is about visual illusions as regard to camera obscura and
photography.
The relationship image
/ reality was treated by R. Evans (1943) under the assumption that good
pictures should have to look like objects. This idea has been left apart not
only by researchers, but also by professional photographers; the goal of
pictures seems now limited to the recognition of objects and their
characteristics, with the wide range of problems related to constancies of many
kinds.
A rather simple
question then arises: do animal perceive as we do? Of course not, according to
most people, because they do not share the abstract knowledge which is
characteristics of human beings. On the other hand animal behaviour shows that
adaptation to the external world is greatly similar to the human behaviour, and
a number of 'mistakes' are performed in the same way by both human and animal
beings (size/distance illusions, movement illusions, and so on). We cannot
ascribe to animals as low as flies or larvae the same kind of reasoning we
perform when we fall in the same perceptual error as they do. Therefore it
seems reasonable to hold that similar behaviours in animals and in humans obey
similar laws, and these laws can, at least sometimes, work independently from
abstract knowledge (formulation of questions, inferences, hypothesis, theories,
and so on).
About
a definition of visual illusion.
Visual illusions are
so common that all people roughly know what they are. Nevertheless a plain
definition of illusions would hardly satisfy all different theoretical
positions. There are at least two different categories of definitions: in one
category illusions are considered as errors of specific operations (for
instance distortions, displacements, overestimations, and so on, belonging to
either physiological or cognitive processes, and which lead to a discrepancy
between the 'external - physical' reality and the content of perception); the
second type of definition has its focal point in the phenomenology of
perception, affirming that what is important is the experienced evidence of a
contradiction (according to my definition, illusion is a perceptual experience
conflicting with another experience about the same object).
R.I. Reynolds (1988)
discusses six different definitions, and expresses his preference for the
following one: "An illusion is a discrepancy between one's perception of
an object or event observed under different conditions" (where
'conditions' take one of three classified forms). According to him this
definition (already suggested by E. Mach, 1900) has the advantage that "it
may help to reconcile the long-standing argument between proponents of the
'indirect' and 'direct' approaches, since the proposed definition does not make
reference to truth or falsity, but only to different percepts occurring under
different conditions". Many authors (Michotte, 1955; Richer, 1978; Bors
& Silberman, 1993; Klymenko & Weisstein, 1987, just to quote some
authors from different schools) along the history of psychology expressed their
opinions in favour of phenomenological definitions (perceptual experiences
sometimes appear illusory sometimes non illusory, i.e. they appear with or
without contrasting characteristics). On the other hand the relationship with
the 'physical reality' is often mentioned by non psychologists (illusion = non
veridical perception, a discrepancy between what is perceived and the physical
reality), although it is almost never specified which concept of physics
(naive, classical, quantistic) should be considered (Gregory, 1996).
Referencing the physical reality, or simply the objective reality, could only
be reasonable under a specific point of view: if the perceptual aspects of a
physical object or event can be deduced either theoretically or directly by
memory from previous informations, then these perceptual aspects can be
compared with the actual percepts we have of the same object or event. If there
is a discrepancy, this is still at the level of perceptive characteristics,
i.e. between one actual and the other
only expected experience. Therefore the mistake could be found not in the
actual perception, but in the erroneous derivation of the expected perceptive
characteristics in the actual context. Illusions in these cases are not
self-evident, like those mentioned in my strict definition of perceptual illusions.
The possible surprise which could emerge necessarily derives from referencing
to something known in a different way, by theory, by memory or by learning.
This way the possible artistic enjoyment which could derive, is not only based
on what we actually perceive but also on what we already know, on our culture.
Context is one (or
probably the only one) critical factor which determines the different
appearances of the objects in our phenomenal world. This topic deserves of
particular attention, because there is no perception at all without a context.
Boynton (1979) in his well known book affirms: "I have decided not to deal
with the very difficult topic of chromatic context in this book". The
statement means that Boynton is dealing only with very simple contexts (one or
two patches of light in complete darkness), which can be reproduced easily, at
least as far as the stimulation is concerned (on the other side, subjective
attitudes, attention, adaptation and so on, although well determined, are not
always so perfectly reproducible). Stating that some percepts depend on context
is clearly redundant: nevertheless, the topic seems today in fashion. As
perception always depends on the context, and therefore different contexts
likely induce different perceptions, it is not a great discovery that visual
illusions depends on context ('A context dependent illusion: ...' is the
beginning formulation of many titles in recent papers). The problem is the
identification of which contexts are relevant, which relationships between
elements of the visual field are determining factors. From this point of view
also the definition of illusion suggested by Boring, 1942 (when we reach the
conclusive explanation of a particular perception, illusions do not exist
anymore) can be understood: illusions as phenomenological aspects of our
experience can remain unmodified, but their relationship with the complex
pattern of physical stimulation and physiological response, gives reason of all
their characteristics, either veridical or not. Referring to some applications
of visual illusions in art, the attractive appearance of stained glasses is
most often due to the rapid changes and instability of their perceptive
characteristics as consequence of easy and frequent changes in the context
(direction of observation, illumination, perspective, and so on); the observer
becomes aware of the discrepancy between the flat surface and the
three-dimensional object which is perceived at the same time. Variations in
surface colour and illumination which are perceived as interdependent and as a
function of subjective attitude, or direction of observation, or some other
factors, are source of pleasure for the observer both when this happens in art,
in architecture and in scientific research.
Bibliography
Boring E.G., Sensation and Perception in the history of experimental psychology,
New York, Appleton-Century-Crofts, 1942
Boynton R.M., Human Color Vision, Holt and Reinhart-Winston, New York, 1979.
Bors D.B., Silberman R.L., The importance of orienting attitudes in the
perception of the Hering and Zollner illusions. J. of Phenomenological
Psychology, 1993, 24, 161-174.
Caivano J.L., Semiotics and reality, Semiotica, Journal of the International
Association for Semiotic Studies, in
press.
da Pos O.,
Zambianchi E., Illusioni ed effetti
visivi. Una raccolta - Visual illusions and effects. A collection. A.Guerini &
Assoc., Milano, 1996.
Evans R., Visual processes and color photography. J. of the Optical Society
of America, 1943, 33, 579.
Goodman N., The structure of appearance. 3rd Edition: D. Reidel Publ. Comp.,
Dordrecht - Holland, 1977.
Gregory R.L., Seeing and thinking. Giornale Italiano di Psicologia, 1993, 20, 749-769.
Gregory R.L., Editorial: What are illusions? Perception, 1996, 25 (5), 503-504.
Hoffmann D.D. The interpretation of visual illusions. Scientific American, 1983, vol. 249 (6) 154-162.
Kanizsa G., Grammatica del Vedere. Saggi su Percezione e Gestalt. Il mulino, Bologna, 1980.
Klymenko V., Weisstein N., The resonance theory of kinetic shape
perception and the motion-induced contour. In: Petry & Meyer Eds., The Perception of Illusory Contours,
Springer Verlag, Berlin, 1987, pp. 143-148
Khun T., The structure of Scientific Revolutions, The University of Chicago,
1970.
Mach E., Die
Analyse der Empfindungen und das Verhältniss des Physischen zum Psychischen. Jena: Gustav Fischer, 1900 (Engl. Trans.: Williams C.M. Ed., The analysis of sensations and the relation
of the physical to the psychical. New York, Dover Publication, 1959)
Michotte van der Berck A., Perception et Cognition. Acta Psychologica, 1955, vol. XI, 70-91
Reynolds R.I., A psychological definition of illusion. Philosophical Psychology,
1988, 1, 217-223.
Richer P., A phenomenological analysis of the perception of geometric illusions.
J. of Phenomenological Psychology, 1978, 9,
132-135.