1967
The Consequences of the Pictorial Attitude
(First draft; approximately as read. For criticism only)
J. J. Gibson, Cornell University
The World Wide Web distribution of James Gibson’s “Purple Perils” is for scholarly use with the understanding that Gibson did not intend them for publication. References to these essays must cite them explicitly as unpublished manuscripts. Copies may be circulated if this statement is included on each copy.
What I will try to do in this work is to reduce my own confusion about representative art and its relation to nonrepresentative art, hoping that this may help to reduce the confusion that many of you also feel.
Let us begin with the puzzle of representation itself. Yesterday’s speaker, Gombrich, has written a book entitled Art and Illusion. He means by illusion what we will call the illusion of reality. He says (p.8) “the main aim I have set myself in these chapters is to restore our sense of wonder at man’s capacity to conjure up by forms, lines, shades, or colors those mysterious phantoms of visual reality we call pictures.” [Comment: I will suggest that the conjuring up is done not so much by forms, lines, colors, etc. but by what I call stimulus information. And the differences between forms, lines, and colors on the one hand and information on the other is the crux of my argument.]
Consider three cases of the illusion of reality, three cases of the paradox arising from what is loosely called an image. The first is the paradox of Narcissus in the pool of water: the second is the paradox of a mammoth on the wall of a cave (a cave painting), and the third is the paradox of Pygmalion’s beautiful girl found in a chunk of cold marble.
(Slide 1. The optical principle of a mirror.)
1. The mirror. I think the earliest case of the contrast between appearance and reality must have arisen when a man first looked into a still pool and saw the trees above his head and what seemed to be his hand down in the water. The observer saw the trees and his body in the pool and yet, surely, not in the pool. Narcissus, according to the myth, was thoroughly deceived by what he saw in the pool. But, to be hard headed about it, no sensible observer, not even an animal, has ever been permanently deceived by the situation. Discuss the following points: the laws of regular reflection. The virtual object and its cause in the optic array. The fact that the regular eye is exactly twice the distance from the real eye to the water. The apparent movements of the virtual objects in relation to the visual movements of the real objects. The deformation of virtual objects with ripples of the water. The disappearance of virtual objects when the plane surface is destroyed.
Even the simplest experiments by our primitive observer would show him that what he saw in the pool was not like what he saw about the pool. The puzzle must be very old–perhaps a million years. It may well be one of the roots of philosophy. (The polished metal mirror, of course, is recent compared to the pool mirror.) The contradiction is that one sees something there but yet not there.
(Slide 2: The mammoth on the wall of a cave)
2. The picture. The next case is that of the first man made image. Here the observer (and also the artist ) sees a natural object that is on a surface (e.g., a wall) and yet not on the surface. This image, like the next to be considered) is about 20 – 30 thousand years old. [Insert p. 228-9 Senses Considered as marked]
The fundamental plastic act developed in man, along with the fundamental graphic act. Let us consider the sculpture. (Slide 3: Two conceptions of the myth of Pygmalion)
3. The sculpture (statue, scale-model, replica). The third case is that of the solid image as distinguished from the flat image (Reliefs are intermediate cases). The statue projects its perspectives in alldirections; the picture projects only to the front, By the time of the Greeks, the techniques of sculpture had been perfected. Hence Pygmalion could make a statue which he could fall in love with. It was his beloved, and yet surely it was not his beloved. What was Pygmalion’s error? Why was he such a fool?
Consider then, these three cases the mirror, the picture, and the sculpture. They are all examples of a so-called image, but the term is a very slippery one, and the examples are quite different.
The first is an image of optics; the latter two are images of art (i.e., of artifice – they are artificial). The image of optics does not involve the creation of a new object, but only operations on light rays by mirror, prisms, or other devices. The original meaning of the term image is the picture or sculpture, although it was borrowed by optics to our perceptual confusion! Consider the “visual image” and the resulting muddle).
I will always mean by an image a man-made object, the source of an optic array, that yields the experience of a virtual object that is not itself.
And now I think we can be clear about a representation; it is a means by which an artist can enable others to see what he has seen. It is a thing which provides perception at second hand. The experience is essentially a mediated perception (as contrasted with immediate). It is an indirect perception (as contrasted with a direct perception).
Next consider the psychological puzzle or contradiction involved in representation. The paradox consists in having to attend to both the point on the wall and the mammoth on the world at the same time.(To see both the sculptured stone and the naked girl in the same place). This what I mean by pictorial attitude. It is contrasted with what could be called the naive attitude. Even now I will speak of the pictorial attitude, although the concept can be applied to the medium of sculpture as well as that of painting.
I will suggest that the artist has to be able to take the pictorial attitude toward the world without losing the ability to take a naive attitude toward it. The same applies to the viewer of art. Let us see what this implies and presupposes.
Prehistoric man before the discovery of image-making could only take the naive toward the objects of their environment. This means that they had not learned to notice the aspects of things, that is, theirperspectives. They were not aware of people (animals, things) in profile, or in front view or from above or below. Likewise, I am convinced, young children before they are pictorially educated take only the naive attitude towards objects and scenes. In our culture, of course, they are given practice in scribing, and are bombarded with pictures, so that they do not have to discover the pictorial attitude for themselves, as the caveman did, but nevertheless children do not have it as early as we suppose. Note what is implied: Primitive man were not, and children are not, aware of visual sensations! This is a radical assertion. It is completely in contradiction to all traditional theories of perception, sensations being supposedly innate and perceptions learned. But so much the worse for the theories! There is a better alternative, which I will try to explain. It says that naive perception is based on the detection of invariant information – invariants under perspective transformations – and that we learn to see in perspective.
If I am right, one can take either a pictorial attitude or a naive attitude in viewing the environment around him. The former yields the experience of the visual field, the first patchwork of color sensations, while the latter yields a direct awareness of the visual world – the animals, objects, places, events, and the layout of the terrain. Similarly, one can take either a pictorial attitude or a naive attitude in viewing a picture. The former means paying attention to the “medium” – the canvas, paint, form, technique, style, perspective – while the latter means paying no attention to the medium but only to the information for the virtual object. And this, we can now understand was poor Pygmalion’s error.
Can an ordinary observer in fact see both the picture as a thing and the thing pictured? Are there two separate spaces associated with a picture – the space of the picture and the space in the picture? I believe the answer is yes, although we are only starting to make formal experiments on the question. Procedure: we ask a subject to estimate the distance away of the photograph or painting (In fact) and the heightof the surface. Then we ask him to estimate the distance away and the height of a tree (say) that is “in” the picture. There is evidence to suggest that the distances and the sizes will be incommensurate. The tree may be 200 feet away and 50 feet high!
Another question: Can the ordinary observer learn to adopt both attitudes without the one interfering with the other? Are they mutually exclusive, so that one cannot see both the paint on the wall and the mammoth in the world? Gombrich suggests at one place in his book that the demand is impossible. But I wonder. I am not so sure that these modes of perception are incompatible. In a sense, both attitudesmust be taken – almost at the same time. I think the painter has to do it and that we should all learn to do it. There is evidence, moreover, to show that a compromise between the two modes of perception is very common. [Facts of partial constancy.]
It may now be possible to state a clear definition of a picture. Thirteen years ago I offered a definition (in a paper called A Theory of Pictorial Perception) which now seems to me wrong. I said that a picture was a surface so treated that it delivered to a properly placed eye more or less the same light – rays as the original scene would deliver. If it delivered the same light rays it would yield the same virtual sensations, and to that extent (I said) the picture had fidelity to the original (I was thinking of fidelity in sound reproduction). Let me try again. Do you agree that the following is truer? A picture is a surface so treated that it delivers to an eye more or less the same optical information as the original scene would deliver. It need not be faithful to the original in arousing the same visual sensations – only in providing the essential structural information. It need not have the same array of stimulus energy but it must contain the stimulus information. It may or may not stimulate the receptors of the retina in the same way, but it must cause the eye – brain system to resonate in the same way.
I mean by structural information in the light the invariants of structure that specify (not replicate) the sources of the array – the real things from which the light comes. For example, the faces of men have distinctive features, as Gombrich so clearly demonstrated yesterday. At another level, there are the distinctive features of the human face as against ape-faces, dog-faces, or fish-faces. And at still another level, there are distinguishing features of the animal head as against coconuts or rocks. At the most general level, there are the distinguishing features of the earth and the sky. All of these are in the light. They are examples of what I mean by optical stimulus information as distinguished from optical stimulation.
It must be evident that this theory of perception by way of pictorial information depends on a quite unorthodox theory of perception by way of natural information. I am proposing that when we see a solid object in the world we do not infer the solid shape from a series of retinal pictures or momentary perspectives but we detect it by registering the invariants under perspective transformation. The series of retinal pictures has nothing to do with the perception; the non-change of the changing pictures is the basis of it.
Our awareness of an object, thus, is a simultaneous awareness of its back and sides as well as its front, and this depends on the detection of constant relations over time. The French have a neat saying that expresses this point: “the more it changes the more it is the same thing”. The information for the object is not given by any single form of the object – any single front face – but by the formless invariants that emerge when we walk around it, or turn around it. A crowd of observers standing around a table thus get, in a way, the same information about it as a single observer who walks around it, and herein lies the justification for our confidence that they all see the same table.
In a word, sensations depend on frozen stimulus patterns but perceptions depend on invariants of stimulus patterns, and perception do not depend on sensations.
When a man says, “I see a tree,”. information is available in the light for the presence of a tree. When he says, “I see a picture of a tree,” information is available in the light for the presence of a material surface and also for the presence of a tree. Since the texture of the surface, the edges, scratches, dust, pigments, luster, brush-strokes, photographic grain, and stereoscopic flatness is not consistent with the tree, the perception of the latter is of a special sort that we call virtual. (But note that nothing whatever has been said about sensations.)
Of course we can have sensations. The having of them is a consequence of the pictorial attitude – so I think. Philosophers and psychologists and painters are deeply interested in those visual feelings that accompany perceiving. But animals, children, and ordinary persons are not concerned with how it feels to perceive something. Sensations are curious and interesting accompaniments of perceiving, the symptoms of it but not the causes. If you want to notice your visual sensations, you must introspect – try to observe the light as it falls on the retina and the corresponding patches of color in the visual field. To do this is almost to see the world as a painting – not quite but almost. The development of the doctrine of sensations in the history of philosophy runs parallel to the development of the techniques of perspective painting. After Alberti in 1435 and Leonardo in 1500 discovered how a scene could be represented as if in a window, on a picture plane, the viewers of these paintings began themselves to see the world in perspective and to see the perspective in a painting. By 1715 an English mathematician could write: “We must consider that a Picture painted in its utmost degree of Perfection ought to effect the Eye of the Beholder in that he should not be able to judge whether what he saw is only a few colours laid artificially on a cloth or the very Objects these represented, seen through the frame or the Picture as through a Window. To produce this effect, it is plain the light ought to come from the Picture to the spectator’s eye in the very same manner as it would do from the Objects themselves” (Taylor, 1715).
And now, finally we come to non-representative painting. If a picture is more essentially the conveying of stimulus information for perception than it is the conveying of stimulus energy for the painter’s sensations, do we have a new basis for criticism?
In order to perceive something you must get enough information about it to distinguish it from what it is not. But you do not have to get all the information about it to distinguish it from everything that it is not. That would be a waste of time and effort. And in any case it would be impossible to distinguish everything in the world from everything else that it is not. You notice as many distinctive features that you could notice if you had nothing else to do. You see enough for a diagnosis and then go on to something else.
Psychologists describe this tendency of perceiving as selective or schematic. It can be dangerous as well as economical. You may fail to see differences that you should wee – subtle or complex or uncommon features of the world that are nevertheless important. You may make the wrong diagnosis of a man, or an event (or the doctor may mistake a disease!). We say that your perception is a mere stereotype of the real thing, And we are apt to mean by that word a distortion of the object in contrast with a true representation of it, a poor picture of the situation instead of a good picture. But in meaning this we are mislead, for a percept is never a representation. It is a set of features.
Now here is where the artist could help us perceive if he would. He is a man who takes time to look carefully. As Gombrich has suggested, he is in general a caricaturist – not only of persons but of all sorts of things. He deals in features; the features of a face and the features of the earth. He selects and emphasizes of necessity. The same information in light is available to us as to him, but if he can capture the essential properties of things, we may see them too. He can show us things to which attention must be paid.
The graphic artist cannot represent an invariant, as I defined it, but that depends upon change in time. He cannot paint or depict or photograph but he can sometimes capture it. But this capturing is hard to understand or explain. Some artists have been tempted to explain it by saying that the capturing of the invariants in a frozen picture is a matter of conveying abstract concepts of the sort conveyed by words. They suppose that graphic symbols exist, like hieroglyphics, the meanings of which can be discovered by artists and learned by the rest of us. When art is not representative, they say, it must be “abstract”.
But this seems to me to confuse different things. Symbols or words are conventional and arbitrary. They constitute a vocabulary or code. There is no vocabulary of painting, no language of art that can be learned in the way we learn to speak and write. Language can do things that art cannot – for example it can formally predicate by means of a sentence. Art cannot do this. But art can do things that language cannot.
First, art can exploit the illusion of reality – it can “conjure up phantoms of visual reality”. This is no mean accomplishment. It is not to be sneered at just because the technique of photography has made it so easy that anyone can do it without having to go through a long apprenticeship that painters once had to go through. Representation by drawing is still wonderful – and every scribbling child, I think, discovers this wonder for himself. It leads to the pictorial attitude, from which flows all kinds of psychological consequences, including the theory of visual form sensations.
But, second, art can do more. It can sometimes capture the invariant distinctive features of things – capture, not represent them. This is information for perception, not sensation. Paradoxically, these essential features are formless. That is, they are not momentary and they are not frozen. They are not composites of images in memory – of photographs stored in the brain. They exist outside us and are there for anyone to detect. They constitute direct information about reality.