July 1974
The Puzzle of Optical Structure
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 References to these essays must cite them explicitly as unpublished manuscripts.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 is an optical array? Should it be treated as a distribution of different luminous intensities (stimuli) or as a topological map of different areas (patches)? It certainly is not a stimulus; is it then a pattern of stimuli, or is it a map? A distribution is analyzed in terms of energies; a map is defined in terms of contours. Intensities vary continuously but contours are discontinuous. Intensities imply corresponding point-sensations of brightness; the map implies only impressions of contour or “contrast.” The lower-level analysis suggests that perception is based on sensations; the higher-level description does not suggest this, since sensations are changing all the time . It suggests that perception is based on invariants.
We need to be able to talk about an optic array at a point of observation whether or not an eye is stationed at that point. If no eye is there, no stimulation occurs; but information is available there (and even more information if the point moves). How do we describe this information? What is the structure of the array, especially its invariant structure? Is it to be treated as a distribution of energies or as a topological map?
The issue becomes evident in the phenomenon of Mach bands. If we ask what this phenomenon means for the perception of the environment, instead of treating it as a problem for visual physiology (Ratliffe, 1965), a deep puzzle must be faced. In the case of an array in which a Mach band appears, there is no step of intensity, no jump in the distribution and thus no contour in the array; and yet a contour is seen! This puzzle has led sensationists to suppose that somehow a contour in the array is “accentuated” by physiological interaction in the nervous system. But this reasoning will not do, for a contour cannot be accentuated if there was none there to begin with. The sensationist ought to conclude that a contour is created by the nervous system where none exists in the array. But this would also be unsatisfactory, for psychophysics teaches that no experience is created without a stimulus correlate. Is there a contour in the array or is there not?
My tentative solution to this puzzle is to suppose that a discontinuity in the array exists and that it is perceived as a discontinuity of the environment. The discontinuity in the array need not be a jump or step of intensity; it can be one of several kinds. As we now know, there can arise “anomalous” contours in perception without there being any ordinary contours at all in the array. The effort to explain the Mach band in terms of a point of inflection on the curve of the intensity distribution, or in terms of a high second derivative of the function, is to be understood as a sort of halfway effort in this direction.
A discontinuity in the array is an “objective” fact, not a creation of the nervous system, or of the mind. It may not be amenable to analytical geometry and calculus, inasmuch as these disciplines presuppose continuity; but that does not prove it is subjective. The environment is sliced up by all sorts of discontinuities (even if the world of physics and mathematics is not) and it is a great mistake to think that these cuts and splits are imposed on the world by the act of perception. We have too long supposed that the physical world is continuous (or at least very fine-grained) and that hence the light to the eye is continuous or fine-grained, concluding from this that the discontinuities in perception are a “subjective contribution.” But ecological physics and ecological optics do not make this mistake.
The deep question is not whether Mach bands are subjective or objective but whether contours are subjective or objective. The problem is whether optical structure, more generally “form,” is a contribution of the mind or a fact of the world, and my solution to this problem is ultimately to assert that it is a false problem. It is false because the objective-subjective dichotomy is false. The basic fact of ecological optics, the ambient optic array at a point, is neither “objective” nor “subjective” since it escapes the objective-subjective dichotomy. It implies both the environment and a potential observer, both the world and the possibility of observing it. It assumes a logical reciprocity of animal and environment, not a dualism of the phenomenal and the physical, nor a correspondence between the phenomenal and the physical. To perceive something is not to have a thing called a “percept” of it, and to be conscious of something is not to have a consciousness-copy of it. To perceive is simply to perceive.
The ambient light rays coming to a “point sink” in space can be represented as a distribution of intensities that make a closed surface, i.e., a sort of “lumpy bag.” The ambient optic array at a point of observation in an illuminated environment can be represented as a set of nested visual solid angles with a common apex having a reflecting surface as its base. Which representation is more appropriate for the study of perception? Are there intermediate representations that would be appropriate for certain problems of vision? I favor the “ecological” representation, but it’s not the only possible one. It does, surely, have the virtue of emphasizing discontinuities, and suggesting problems like the perception of occluding edges. It presupposes a steady state of reverberating light in the medium. It abandons certain classical problems of sensory physiology, but it embraces certain other problems that have either been neglected or treated in terms of a vague and speculative mentalism.