March 1965

The Comparison of Mediated Perception with Direct Perception

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 References to these essays must cite them explicitly as unpublished manuscripts.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.

Children are said to learn from images and pictures as well as from words, but this is a thoroughly muddled question in education. In order to clarify it we need a theory of the relation between the perception of things and the mediated perception of things. Mediated or indirect perception is of three common types (1) that which depends on the understanding of images and pictures, (2) that which depends on the understanding of speech, and (3) that which depends on the understanding of writing. All these can be described as perception or knowledge at second hand(Gibson 1954). Images, pictures, vocal speech and written language will be called here mediators. They are media of communication among persons. Only men communicate by these means although, of course, animals and men can also communicate by other means like cries, gestures, attitudes and grimaces, that is, by reactions that we loosely call "expressive".

Mediators, it should be noted, are not stimuli as ambient light, impinging sound, and mechanical energy are stimuli, but sources of stimulation as the objects, events, and other animals of the environment are themselves sources of stimulation. The common mediators use light and sound or, rarely, touch. It is assumed (1) that the main function of stimulation is to carry information, and (2) that information does not necessarily depend on the kind of energy that conveys it.

The hypothesis is that direct or first-hand perception is that which comes from environmental sources and that indirect or second-hand knowledge is that which comes from mediators. It is further assumed that the uniquely human media of information-transmissions are of two types, iconic and symbolic. Iconic mediators have been described (e.g., Morris 1946) as "similar" to what they stand for; symbolic mediators are not. But this is not very satisfactory. I have tried to define models and pictures, the iconic mediators, as being specific to what they stand for by proportion or by projection whereas vocal speech and written language, the symbolic mediators, are specific to what they stand for by convention (Gibson 1954). The symbolic object is informative because of the establishment of a social code; the iconic object is informative by non-social laws of stimulus information. A license-plate corresponds to an automobile by virtue of conventional rules but its shadow corresponds to it by optical rules. It is here assumed that perspective geometry derives its validity from laws of space-filling light-rays and that ideal pictorial perspective is therefore not a convention of Renaissance painters but a discovery (Gibson 1960,1961).

The iconic and the symbolic kinds of mediation admittedly must be thought of as pure cases, not as mutually exclusive types, since many images and pictures show the influence of convention, are semi-symbolic, and many forms of writing show the influence of iconic mediation, notably graphs. The two cases are hard to separate in visual art (Grombrich, 1960). Even spoken words may in part represent the sounds of what they stand for. Mixtures of the purely iconic and the purely symbolic relation can be accounted for by the assumption that some qualities of things and events can best be communicated by stimulus representation, by displaying, while other qualities of the world can only be conveyed by symbolic media, such as verbal propositions. Elements of both may be found in the mediator. Both art and language are also, to be sure, mixed with elements of primitive expressive communication but this is a different question. The matter is full of difficulties, but the above assumptions at least have the virtue of being explicit.

Development of Mediated Perception.

Consider the order of development of some perceptual mediators in human evolution as compared with the order in which the child learns to use them. Primitive man had to invent them; the child does not. The conventionalizing of vocal sounds may be supposed to have begun with the emergence of the human species, perhaps a million years ago. Proto-man could make and hear a variety of sounds because they were already highly vocal animals. The making of sculptures and pictures began, it is now fairly certain, some twenty or thirty thousand years ago in the caves of the Ice Age. This invention depended on visually guided manual skills, on making tools, fire, torches, and on shaping of materials. Finally the invention of writing, the second-order conventionalizing of the picture-making skill so as to make the vocal symbols themselves visible and permanent, came only about five thousand years ago.

The child begins to develop the understanding and the production of speech at about the same time, at the age of one year, and does so spontaneously in a family group. Just when he begins to comprehend the iconic mediators is not clear. He is given models in the form of toys and dolls and pictures in the form of drawings and photographs (Hochberg 1962) but not much is known about what he perceives, although there is a great deal of speculation (and muddled theorizing) about the perception of forms vs. the perception of solids. More experimental evidence is needed, taking off from the hypothesis of visual stimulus information instead of the classical dogma of flat visual sensations. What evidence there is suggest that the equivalence of the optical information from an object and from a picture of it is detected early, and there is also some indication that the equivalence of small-scale models and full-scale replicas is effective. Apart from this there are no experiments on what the young child comprehends from moving picture displays or perspective transformations, despite the prevalence of television in the home. Nor is there evidence about working models. What the child learns from unregulated gamut of iconic things presented to him is not known, although it must be important.

It is certain that the young child does not learn to produce iconic mediators, even the simplest pictures or models, until well after he can comprehend them. The necessary manipulative skill seems to lag behind the perceptual skill. However he is usually given much practice in the fundamental graphic act of trace-making or scribbling, that is, making (and controlling) visible marks on a surface. But the perceptual value of this practice has not been studied. Similarly he is often given opportunities for plastic manipulation so as to develop his "creative" capacities but experimenters have not ventured into this field. Iconic information and iconic communication are mixed up in our culture with the concept of play as distinguished from work. Scribbling, however, is not simply play; it provides an opportunity for educating visual perception in a special way.

Finally, the child learn to read around age five or six and to write somewhat later. He has to detect the equivalence of alphabetic combinations (utterances) before he can do this. Sometimes it is easy; sometimes hard. The difficulty may depend on how well he has previously learned to discriminate the variables of graphic displays­the dimensions of variations among surface-tracings.

It is likely that the order of similarity of these mediators to natural information is first­and experience, or knowledge by direct acquaintance is given by the first. This hypothesis should be tested. If so, images could usefully be given the child at an even younger age than is now customary.

Experiments on Mediated Perception.

It is proposed to explore some of the questions that have been raised to see what experiments are feasible.

1. Comparison of full - size replicas and scale models. To what extent does a minified optic array, preserving ratios, carry the same information as an ordinary array for the young child? A systematic study of miniature animals, persons, objects, places, etc., should be made to determine the degree of equivalence of stimulus information.

2. Comparison of solid object, photograph, and line - drawing. Evidence on the perceptions of primates and infants is being obtained and should first be surveyed. Further experiments on the equivalence of pictures, with controlled variation, to what they portray can then be carried out.

3. Comparison of still pictures and motion pictures. There is reason to believe that perspective transformations convey information better that forms. This should be tested with young children. The perception of "animate" motions should also be studied, with a view to systematic experiments. One way to do this is to take a "still"(or series of stills) from a motion picture shot and determine how much more information a child gets, if any, from the latter than he does the former.

4. Analysis of the "fundamental graphic act" (tracing, scribbling) in young children. What is the nature of this special visual feedback and what is learned from seeing a permanent trace being made? At what age do children begin to see Michotte's "tracage" effect (1964)? A separate note has been written on possible experiments to answer these questions.

5. Tests of the hypothesis that environmental information may cut across perceptual systems (cross-modal information). The present approach is based on the idea that stimulation, although necessary, is less relevant for perception than information. The equivalence of information across different media of communication is fairly well accepted. But the equivalence of information across stimulus energies is a new idea that needs verification. It can be tested in various ways, for example by determining the extent to which children can equate the visual appearance of an unfamiliar texture with the tactual feel of it. Cross-modal equivalence can also be tested with solid shapes previously unknown to the child.

References

Gibson, J. J. A Theory of Pictorial Perception. Audio Vis. Communication Rev. 1954, 1, 3-23.

Gibson, J. J. Pictures, Perspectives, and Perception. Daedalus, 1960 (winter issue) 216-227.

Gibson, J. J. Ecological Optics. Vision Research, 1961,1, 253-262.

Gombrich, E. H. Art and Illusion. N.Y. Pantheon, 1960.

Hochberg, J. The Psychophysics of Pictorial Perception. Aud. Vis. Communication Rev. 1962, 10, 22-54.

Morris, C. Signs, Language, and Behavior, N.Y., 1946.