June 1954
Ordinal Stimulation and the Possibility of a Global Psychophysics
J. J. Gibson, Cornell University
(Symposium of Recent Trends in Perceptual theory, Montreal)
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 References to these essays must cite them explicitly as unpublished manuscripts.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.
The Innsbruck experiments seem to prove what Stratton many years ago was looking for but failed to demonstrate conclusively. Kohler has shown that a persistent abnormality of optical stimulation leads in the end to a reduction of the phenomenal abnormality of the world, and that a return to normal optical stimulation then yields a new phenomenal abnormality of the world of exactly the opposite kind. Kohler has disturbed, or distorted, or biased the optical stimulation not merely on one but in several different ways. He has produced abnormalities of color and of reversed direction; both right-left and up-down, and, what is even more interesting, abnormalities of curvature and density and proportion. The habituation that results is not simply a matter of appropriate behavior, of manipulation, and locomotion; the actual appearance of the world changes in the course of time, and then it changes all over again in the opposite direction when the optical device is removed from in front of the eyes.
These results strongly suggest that our perception of the surrounding world is modifiable. Perception is subject to habituation. It might follow that the environment looks stable and rigid, rectilinear and proper, only because we have made some sort of adjustment to it. The assumption that we somehow learn to perceive the spatial order of things is made even more plausible by such evidence. The notion that perceptions spring into existence full blown by way of native intuition is rendered less plausible than ever.
It seems to me however that the results also support another assumption that our perception of the surrounding world is an exact function of stimulation. It is a bias or skew in the optical stimulation over the course of time that brings about a change in perception, and the change is a shift in the correspondence between stimulation and experience, not a disruption of the correspondence. The proof is the fact that when stimulation returns to normal perception becomes abnormal in precisely the way demanded by the kind of change that occurred. Habituation in perception carries with it the necessity of rehabituation. If the appearance if the visual world is said to depend on learning, it must equally be said to depend upon present stimulation in the sense that, at any given stage of learning, the qualities of the world are parallel to certain variables of the optical stimulus.
The conviction that perception depends on learning (past experience) is an ancient one in perceptual theory. So also is the conviction that it depends on stimulation. But it has never been clear how the two could be reconciled. The classical explanation of how learning makes its contribution to perception (e.g. – by the association of memories to sensations) and the classical formula of how stimulation affects perception (by dictating the sensations at its core) are not any longer acceptable. We need a new theory of perceptual learning. To accompany it, we need a new theory of the relation between stimulation and perception one which does not confine its role to the determining of innate, elementary sensations. In short, we need a psychophysics of perception.
One conclusion from the Innsbruck experiments, it seems to me, is perfectly clear. The learning which occurs (if it can be called that) is surely not a matter of learning new “local signs”, or new localizing responses, for the receptors of the retina. And the correspondence which is revealed is certainly not a correspondence between the points of the image and the elements of phenomenal space. Instead, it is a correspondence between certain mathematical properties of the retinal image and certain phenomenal variables of the visual world not only the qualities of color and so-called location but also qualities of the curvature and density and proportions and motions of things. These mathematical properties of the stimulus, if they can be specified, are the key to the deepest problems of perception, and I propose for them the term ordinal stimulation.
A psychophysics of perception is possible if we can isolate and control the ordinal stimulation at receptors. Before psychologists will attempt this, however, they have to be convinced of the possibility that this kind of stimulation exists. For many of us the physiologists¹ conception that a stimulus consists of the energy absorbed by a single receptor-cell still carries great weight. For others among us the common sense notion that the stimuli are simply the objects and events of the environment is the alternative. Still others take the position that stimuli are nothing but accidental and inconsistent hieroglyphics, of minor importance for perception. This position, in fact, seems to be a modern theoretical trend. Another more promising trend is to conceive stimuli as units of “information” and to apply to them the mathematics of communications systems. In view of this confusion about the concept of stimulus and in order to convince you that ordinal stimulation does exist, I should like to give some examples of what I mean by it.
A first example may be taken from the field of tactual perception. Thirty-five years ago Hoisington became interested in the problem of how in the absence of all vision we perceive the length of a stick or rod held at one end by the hand in a horizontal position. For rods of uniform diameter, the lengths could be distinguished with surprising accuracy and this was true both for heavy rods (e.g. – metal) and light rods (e.g. – wood). He discovered that the impression of length is based on two pressures, one on the forefingers and the other on the heel of the hand, but not on the absolute intensities of these pressures, only the relative intensities. It can be asserted that the observers were discriminating a ratio of two forces the ratio between what Hoisington called the “kick” and the “pressure”. This ratio is what I would call an ordinal stimulus (or if you prefer, a case of ordinal stimulation) and the results suggest that variation of this ratio is in a strict psychophysical correspondence with variation of the impression of length. I would classify this experiment as a contribution to perceptual psychophysics, although the term would undoubtedly horrify the Cornell psychologists of its day. Many other examples could be given of responses which are specific to ratios of energy rather than to merely type or amount of energy. Receptors are often arranged in a mosaic, as in the skin and the retina. A gradient of energy, that is, a difference between here and there is consequently a stimulus for a receptor-surface. The slope and direction of such gradients, as the two-point threshold and the acuity experiments show, are the critical variables of the effective stimulus. To speak accurately, then, it is not the energy as such which constitutes stimulation for the skin or the retina but typically the changes of energy. The retinal image, according to this conception does not consist of points or spots of light but of transitions. The phenomenal impression corresponding to a transition is a visual margin, and an array of them yields qualities such as texture and forms the basis for other impressions which characterize visual surfaces and in the end visual objects. A gradient of energy is an ordinal stimulus. The order is one of neighborhood or adjacency. A change in energy with time is also an ordinal stimulus, and the order is one of succession. The hypothesis that a stimulus is always a change in the latter sense, has been paid a great deal of lip-service by physiologists and psychologists because the facts of sensory adaptation require it, but it has not been given any theoretical recognition. There is no question, however, but that variations in the successive order of energies constitute stimuli. The auditory phoneme or speech-sound is a good example. From it are compounded words and sentences, which are stimuli of still higher order. Changes of the margins and shapes of the retinal image with time are also stimuli. They induce impressions of motion with a lawfulness which has not been recognized in the past. A psychophysics of visual motion perspective is possible and is long overdue.
When stimulation on the retina or the skin is conceived in this way, that is to say physically and mathematically, the problem of the “transposable gestalt” appears in a new light. Stimulation as such is transposable by definition. The retina and the important skin areas are movable sense organs with the function of exploring the light-patterns and the pressure-patterns which a generally stable environment provides. The retinal image is transposed over the retina whenever the eye moves, and the evidence indicates that it moves continually. The only problem is one of neuropsychology, that is, how does the nervous system register and preserve the order of the image? It cannot possibly be preserved by the anatomical pattern of the particular neurons that are firing, as we have so long assumed. The notion that the stimulus pattern must be replicated in the optic nerve and transmitted to the brain by point-to-point projection must be nothing but a plausible fallacy. The order must be preserved by some relational physiology of which we now have little conception. Another important variable of ordinal stimulation, at least for the eye, is probably the density of the energy transitions or gradients that compose the retinal image. It has long been suspected that what was called the “microstructure” of the retinal image has something to do with the impression of objectivity or surface-quality. The surface-quality of a visual extent seems to be a promising avenue of research on the way to a general theory of visual perception more promising perhaps than the figure-ground impression and the laws of its organization, which seems to me to have reached a dead end. If we can account for surface-perception we shall have solved the abstract problems of form-perception and of space-perception with one stroke. The question is can a variable of stimulation be isolated for the quality of “surfaceness”? We have been trying for two years at Cornell to manipulate light so as to induce a synthetic perception of a surface. The aim is make an artificial surface, as it were, by optical means the percept of a surface which is phenomenally there but not physically there. We have succeeded, I think, and the stimulus-variable which is critical for the perception of a solid and substantial object in three dimensions is the density of the abrupt transitions between light and dark in the sheaf of light rays to the eye. When the transitions are close together the observer sees a surface; when they are fewer and farther apart (below a certain limit) the experience becomes increasingly filmy, foggy, or insubstantial.
The object induced is a large concavity in the field of view, a thing which the observers identify as the interior of a tunnel or tube. The elements of stimulation are concentric-rings of light and dark in the sheaf of light. The apparatus for producing them consists of circular holes or edges in alternating sheets of black and white reflectors, hung one behind another in a twenty foot rack. It must be set up with some care. The control experiment which enables us to assert that we have created a surface out of light and not merely out of other surfaces is to substitute all-black or all-white reflectors for the alternating series and note whether the phenomenal object then disappears into air or fog. To a sufficient approximation, this is actually what occurs. The experience is somewhat like that of the film-color described which appears in apertures of small visual angle.
A stimulus-variable of still higher order in this “optical tunnel” is the gradient of the density of the transitions, or what I have loosely called the density of texture. This, we believe, is a variable which yields depth and which, in combination with gradients of disparity and motion (at another higher order) produce the qualities of slant and recession in space which characterize the surfaces and objects of the visual world.
One more example of an ordinal stimulus must suffice this one only to show that novel hypotheses can be generated by the theory. What is the relation between concomitant or simultaneous stimuli from different modalities of sense? The stability and constancy of visual perception probably depends in a large part on such a relation. Consider the rotation of the retinal image which occurs when a man tilts his head to one side and the fact that his phenomenal world nevertheless remains upright. The significant fact may be the vestibular and kinesthetic stimulation which accompanies the retinal stimulation and which enters into the whole as a component. When two component changes vary reciprocally an invariant product exists. Can such an invariant itself be considered a stimulus? If so the constant uprightness of the visual world despite the tilting of the retinal image is a matter of correspondence with the stimulus, not a lack of correspondence. The relevant stimulus is only more complex than we had realized.
The proposal that we look to stimulation for an explanation of perception before resorting to subjective modes of explanation is perhaps a strategy of research rather than a theory. It assumes that the energies at sense-organs are rich in variables which have not yet analyzed. The so-called “poverty” of the retinal image as a stimulus for vision has been overstated. Except when experimentally improvised by the darkroom or the tachistoscope it is an astonishingly good basis for perception and behavior. This approach makes it possible to suppose that subtle and intricate variables of stimulation are effective even when the percept is question is demonstrably learned. And it suggests that the learning process in perception is one which will account for the growth and increase of our sensitivity to the world around us a world so complex that we need all the discriminative capacity we can acquire.