Note on Structure in the Optic Array
J. J. Gibson
Center for Advanced study in the Behavioral Sciences Stanford, California
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 References to these essays must cite them explicitly as unpublished manuscripts.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 following tentative definitions are extensions of the paper Ecological Optics, Vision Research, 1961, 1, 253-262. Criticisms are invited.)
The optic array, the stimulus for an ocular system, consists of ambient (not radiant) light, involving a convergent (not divergent) pencil of rays. It presupposes a medium, a convergence-point, and its environment of light-sources, i.e., emitting, diffusing, or reflecting bodies. During the evolution of eyes the principal emitting sources have been high-temperature substances (the sun by day, and fire of stars by night); the principal diffusing source has been the atmosphere, or sky; and the principal reflecting sources have been the complex layout of terrestrial surfaces. In the recent era, these have been supplemented by artificial emitting sources such as torches or lamps and artificial reflecting sources such as prepared surfaces and pictures.
An optic array may be described, however without reference to its sources. It can be defined simply and generally as having different intensities of light in different directions. It can thus be said to contain or carry information. (When we take into account the sources it can be said to carry information about the world the concern of ecological optics but this is a further step.) It is instructive to consider the ways in which ambient light may fail to carry information, or fail to be an optic array. There seem to be three.
1. Ambient darkness, This is the case of the same intensity of light in all directions, the intensity being zero. There is an absence of stimulus information because of the lack of stimulus energy. In this situation the photoreceptive elements of an eye are not excited and the responses of accommodation, fixation, and convergence are frustrated.
2. The uniform optic array. This is the case of the same intensity of light in all directions, at an intensity above zero. There is stimulus energy but no stimulus information. The photoreceptive elements of the retina are excited, but the higher receptive units are not, and the exploratory responses of the ocular system fail. The situation can only be achieved experimentally by applying the eye of an observer to the aperture of a large optical integrating sphere having an extremely smooth interior surface (Cf. W. Cohen, AJP., 1957, 70, 403-410).
3. The homogeneous optic array. This is the case where, although intensity is not identical in all directions, the changes of intensity with changes in direction are not sufficiently abrupt or sufficiently extended (or both) to arouse the full range of activity of the ocular system. Loosely, there is said to be not enough differentiation, contrast, detail, form, or structure in the array. The situation is encountered naturally in dense fog, with the field of an overcast sky, an in arctic “whiteout”; it can be obtained in the laboratory with a panoramic illuminated screen (“Ganzfeld”); or with diffusing plastic eye-caps.
Some Types of Structure A. Shadowy. A homogeneous optic array may begin to have a vague structure of the sort that can be called “shadowy.” This would consist of very gradual or “penumbral” transitions of intensity. Experiments show that this type of structure is not sufficient to yield any impression of surfaceness, not any clear impression of depth, such as depth-at-an-edge, and no clear impression of distance-from-here. There is evidence to suggest, however, that such very gradual transitions of intensity are sufficient, when changing in time, to yield a primitive impression of motion. But this is a formless motion since there are as yet no abrupt transitions or contours to define a form or figure. Presumably motion is a primitive type of information not dependent on form as ordinarily conceived.
B. Spotty. A homogeneous array can become inhomogeneous by having in it a group of points (more accurately, spots) of high relative to low intensity, or vice versa. This constitutes a type of structure that has been much investigated. The light-on-dark case is exemplified by the stars in the night sky. This structure is sufficient to permit clear impressions of relative displacement – motion to occur. It yields what I will call pattern (constellation) but not form. It does not yield the quality of a continuous surface or of the edge of a surface. Some of Werheimer’s laws refer to a kind of phenomenal unity of spot-patterns, but this is not yet surface-perception.
It has been assumed that if the frequency or density of a constellation of spots is sufficiently increased, it will become a textured array, but I now believe that this is a serious misconception.
C. Textured. A shadowy array will gain in structure if the penumbral transitions between areas of relative light and dark are steepened, i.e., become “sharper”, more “abrupt”, or more “discontinuous.” (It is necessary to distinguish, of course, between blur in an array and blur in an imperfectly focused image of that array. Experiments suggest that a sufficiently textured array will yield such qualities as hardness, impenetrability, and opaqueness, which characterize the experiences of surface. Along with surfaciness go other qualities such as optical slant, and surface-color-with illumination. These are probably determined by higher-order variables of optical texture, but the experiments on this question are not yet decisive.
A texture is not the same as a form in this terminology. A texture can undergo transformations of many sorts in the absence of what are ordinarily called geometrical forms, although the word “transform” admittedly does not suggest this fact.
D. Contoured. When a single transition between relative dark and light is considered, the areas being large, there exists a contour. If a contour is closed, there is a form. If many contours intersect exists what I will call a reticulated structure, composed of pairs of adjacent forms sharing the same contour.
As Rubin and Wertheimer noted many years ago, a single closed contour in an array is sufficient to arouse some of the properties of object-perception, i.e., of an object in “space”, or against a background. The figure-ground phenomenon, in fact, was taken to be the prototype of all perception, although this assumption was surely mistaken.
The study of a single contour in a field of view is the essential feature of all experiments on so-called brightness discrimination. A very small difference of intensity between two areas is visible. But this in only true when the transition is abrupt. When it is gradual, as Mach proved, even a large difference of intensity may be invisible. It may be, as Mach suggested, that the crucial variable is not the first but the second derivative of the intensity-transition. In any event it is intensity transitions, not intensities as such, that are the basis of structure of an optic array and that embody stimulus information. The impression of contour depends on a given ratio of intensities for a given degree of an ocular system to this invariant, and the sensitivity of a photoreceptor to amount of energy the “sensation” of “brightness”.
Implications. The study of the registerable structure in a optic array has been neglected because we thought we had to explain how structure could be imposed on the millions of stimuli (or the sense-data) that we took to compose the array. A terminology of this structure, therefore, does not exist. The foregoing is only a tentative beginning at such a terminology.
We can borrow from the terminology of the artist (such as it is) only with great caution since he is describing pictures and we are describing light. But we can borrow from the terminology of the physicist only with greater caution since he is describing energy and we are describing information.
The four types of structure listed are surely not exhaustive. Their virtue is that they are intended to be capable of mathematical treatment but not a blind analytical treatment. They also may serve to clear up misconceptions such as the notion that visual perception rests in the seeing of forms made up of lines or outlines. Animals see all kinds of things besides forms.
The structure of an optic array depends on its intensity transitions, not on the differences in spectral composition of “color.” All animals with eyes (including compound eyes) can register some of the information in the immensely rich variation of structure in an array, but by no means all seeing animals can register the spectral information. It is not even clear, as yet, what should be meant by “color” in an array of ambient reflected light, as distinguished from a beam of radiant light. The seeing of colors is not necessary for vision, nor is the seeing of brightnesses for that matter, but the detection of structure is.