June 1969
Heterogeneous Optical Stimulation and Visual 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 References to these essays must cite them explicitly as unpublished manuscripts.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.
If it is true that the light available to an eye must be heterogeneous in order for the light stimulating the retina to be heterogeneous (that is, in order for a focused retinal image to occur) then the notion of the ambient optic array is central to the theory of visual perception. We must ask what the correspondences are between in homogeneities in the optic array and in homogeneities in the world from which that array is projected.
In the Senses Considered (Ch. 10) the main causes of the structure of ambient light were considered. Three kinds of discontinuity were said to be common in the environment, departures of a surface from uniform planarity, departures from homogeneous composition, and departures from homogeneous illumination. The first is variation in layout, the second is variation in substance or pigmentation (and therefore reflectance), and the third is variation in lighting-shadowing. In these ways the optic array at any point of observation becomes differentiated with respect to intensity and wavelength.
We may note that a surface which is planar, which is the same substance at all points, and which is uniformly illuminated is simply invisible, being a mirror.
I. Consider departures of a surface from planarity. The small convexities and concavities of an otherwise flat surface bring about what we call “highlights” and “shadows” on the surface if the illumination is even partly directional, and these cause highs and lows of intensity in the optic array. There is therefore a correspondence of sorts between the texture of the substance and the texture of the light. This fact is what Metzger and Koffka meant by microstructure. Other similar departures from flatness of a surface are what might be called wrinkles and cracks. Corresponding to these features of small-scale layout are characteristic small discontinuities in the optic array. If the various substances of the environment have characteristic forms of texture, there will be correspondence in the patterning of the array.
Another kind of departure from planarity is a dihedral angle. The apex of a plane-angle may be called an edge when it is convex toward the point of observation and a corner when it is concave toward the point of observation. The apex of a dihedral angle causes an abrupt contrast in the array.
Still another kind of departure from flatness is curvature of the surface which, like angularity, may be either convex or concave, and which may occur along one or both dimensions of the surface. This causes a gradient of intensity in the array (as distinguished from an abrupt contrast).
Finally, perhaps the most important feature of opaque surface-layout is the occluding edge. This may be either a convex dihedral angle or a convex curved surface. The essence of an occluding edge is thatone surface partially hides another but that both surfaces exist. For the optic array at a given point of observation the environment will consist of a set of projected surfaces and another set of occludedsurfaces, but these sets will be gradually interchanged with movement of the point of observation. And when an observer perceives an occluding edge he perceives both the projected and the occluded surface.
Defining an object as a detachable substance in the environment with a topologically closed surface, we can now say that the solid shape of an object is constituted by its dihedral angles and curvatures. We can also say that its projected outline shape in the optic array, its “form,” is caused by its occluding edges. It will also of course have “in lines” as well as outlines. An object in the world is thus “delineated” or “outlined” by its occluding edges, as these are projected in the optic array. But if one sees them as such, one also perceives the occluded surface in a special meaning of the term “perceive.” That is, one “perceives” the back of the object as well as the front, and observes its solid shape. Moreover, in certain circumstances, one perceives the background of the object extending uninterrupted behind it. The latter perception has been described as the “figure-ground phenomenon.”
The depth at an occluding edge, and the direction of this depth are a component of this perception, of course, but not the only or even the most important component. When we take our problem to be the perception of lay out, the problem of the perception of “depth” takes on an entirely new meaning. All these departures of a surface from planarity: texture, edge, corner, convexity, concavity, and the occluding edge, have the effect of structuring the light at possible points of observation.
II. Consider next departures of a surface from homogeneous composition. When a surface is composed of a conglomerate of substances instead of a single substance there will be differences in reflectance over the surface, and differences in intensity (and spectral distribution) over the optic array. The same will occur if the pigmentation of the surface is not homogeneous (as in a painting). When the surface is planar and extended then the optic array is wholly caused in this way; otherwise the optic array is caused by both differences in pigmentation and differences in spatial layout acting together. In short, there are two pure cases; the structuring of light by surface layout but not surface reflectance, and the structuring of light by surface reflectance but not surface layout. The correspondence between discontinuities in pigmentation and discontinuities of the optic array is so simple that it has misled us. A plane surface maps into a cross-section of an optic array with the simple point-to-point correspondence of projective geometry. The more interesting and more important specificities that hold for differential layout and differential shadowing of a surface, specifies that are not simple correspondences, have been neglected and relegated to the vague status of being “cues for depth” or “cues for illumination.”
III. Finally, consider departures of a surface from homogeneous illumination. A plane surface that is not uniformly illuminated, or that does not have a uniform gradient of illumination across it, is said to be shadowed. Shadows on a plane surface are “cast,” as distinguished from the shadows on the faces and facets of a non-plane surface which are “attached.” Shadows depend on the degree of which the illumination has a prevailing direction.
The pure case in which an optic array is caused wholly by cast shadows is provided when a surface is both planar and of uniform composition, and when there is (in fact or in effect) a point source of illumination. The shadow-plays of the ancients and the projection systems of modern times are examples, but normally an optic array in a natural environment is caused by shadows and by pigmentation andby layout, all in combination.
It is assumed that all three of these sources of optical structuring, layout, reflectance, and light-shade are external facts. But the layout of surfaces (or the shapes of objects) are probably more diagnostic of what they afford than are the colors or the shadings of surfaces. The spatial layout of things tends to be persistent and invariant. Shadows are thought to be insubstantial, variable, and ephemeral; colors fade at dusk, but shapes are permanent. Nevertheless, all three facts of the environment are perceived, and it is the task of ecological optics to discover the optical information by means of which they are perceived.