October 1965
Four Related Problems in the Visual Perception Of Environmental Layout
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 References to these essays must cite them explicitly as unpublished manuscripts.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.
Four kinds of visual phenomena will be described which are not explained by traditional theories of perception. The first two have recently been defined and studied by Michotte and his co-workers(1964). When the other two considered, a common problem emerges:
1. The appreciation of the completeness of an object of which a part is occluded by another object. This is what Michotte calls “amodal completion”, on which grounds that no visual sense-data exist for the hidden part of the object. There seem to be two sub-cases:
a. In line-drawings there is the phenomenon (often called the “cue”) of superposition, one form appearing on top of another (e.g. Chapanis and McCleary, 1953). An especially good example is Michotte’s phenomenal completion of a triangle having a cutaway section, when this section is covered by a pencil. This phenomena should be tried out with solid objects.
b. For both line-drawings and objects, there is the phenomenon of the apparent continuity of a whole background surface that is occluded by a surface-form. This is the classical figure-ground phenomenon (“the ground seems to continue uninterrupted behind the figure”). Koffka (1935) described the same fact for the book that lay upon his table-top. But we note that the figure-ground phenomenon is not, as claimed, the most general case that subsumes all the others.
2. The appreciation of the rear surface of an object when only the front surface is “given” in stimulation. If we agree that there are no sense-data (with modal qualities of hue, brightness, extensity, etc.) for the rear surface, then this perception is also a case of “amodal completion”. Ostensibly, only the front surface is specified in the optic array, but this can be questioned. There may be information for the continuity of the front and rear surfaces. An example, from Michotte (Translation, p.9 f.) is apparent completeness of a textured hemisphere seen from the front, in contrast with the apparent incompleteness of a semisphere sufficiently cut away in back so that edges are detected. The detection of the continuity of “front” and “rear” occurs also for a cylindrical surface(see below).
3. The appreciation of the extendedness of a surface or layout seen through a window (aperture, hole) in an otherwise continuous occluding surface).
a. This case might be called the window-phenomenon to distinguish it from the figure-ground phenomenon. The window must be sufficiently large; otherwise a “film-color” appears in it. Note that in the window case the contour-enclosed surface in behind the outlying surface, not in front of it. The edge-depth is reversed as compared with the figure-ground phenomenon. Only in a textureless outline drawing (which is not ecologically representative) does the apparent surface within a closed contour regularly appear to be in front. Note that the phenomenal extendedness of the “background” seen through an aperture cannot plausibly be explained by a hypothesis of completion if this means Gestalt-completion as usually conceived.
b. A case related to the “window” is that of the “door” beyond which extends a phenomenal (locomotor) environment. There is also the awareness of space “around the corner” of a passage way, and is related to the acquisition of “cognitive maps”. Similarly, there is the perception of a “visual cliff”, of depth downward at an edge (Gibson and Walk, 1960), but the unseen space here does not afford locomotion. The experience of the continuation of a road “over the brow” of a hill, when driving, seems to be different, and to fall under Case 2 above.
4. The appreciation of the unbounded environment outside the field of view of the eyes. This is the “visual world” as contrasted with the “visual field” (Gibson, 1950, Ch. 3). It refers to Koffka’s claim that he could perceive, in a sense, the world behind his head. This is not a problem for animals with panoramic vision.
Transitions between Case 3 and Case 4 are provided by the experiment of viewing a surface (or layout) through a tube, or through an aperture that is close to the eye. Certain special features of perception through small apertures are described by Katz (1935) and the effect of different angular sizes of aperture is described in a preliminary way by Gibson and Dibble (1952). When the aperture, or tube, or the window is large enough, and the enclosed field is sufficiently structured to permit accommodations for its optic array, a continuous surface is perceived that extends outward behind the edge.
DiscussionThese four cases of perception are all amodal in the usage of Michotte (1964) or “sensationless” as I would put it, but there is some question as to whether or not they are all cases of completion. The latter term suggests that a contour-form must be involved, but this is found only in Case 1. It seems to me, rather, that they are all cases of the detection of the continuity or coherence of a subject that is partly occluded in accordance with the law of simultaneous rectilinear optical projection to a point of view. That is to say, they all imply the fact of occlusion and necessarily include the simultaneous detection of what I will call an invariant material edge.
There are three types of material edges, it would appear from the above examples: the the abrupt edge (or the edge of the field of view). But only when the first two are components of environmental layout, the perception of which I take to be the perception of “space”.
The main components or variables of frontal surface layout tentatively seem to be slant, curvature, and dihedral angle. Edge seems to be a component of layout, of “space perception”, but to involve quite different stimulus information. The optical information for and the psychophysical detection of slant-depth has been repeatedly studied (e.g., Flock, 1964). The perception of curvature-depth (for the convex but not for the concave case) has been studied only by Smith and Smith (1961). The perception of the depth of a plane angle or “corner” (illustrated in Gibson, 1950, p.93) has not yet been formerly investigated. However, the information for and the perception of edge-depth has scarcely been formulated. It is only now beginning to be studied. This last is perhaps the most crucial feature of environmental layout for, unlike the others, it involves optical occlusion and yields the seeming paradox of perceiving one thing behind another, as noted above. The slants, curves, and plane-angles of the world are specified by not the complex variables or gradients of adjacent and successive order in the optic array (e.g., density of optical texture),but the edges of the world seem to be specified, if there is available information at all, by mathematically subtle properties of the optic array. Continuous gradients of density or parallax do not hold edges.
The four puzzles of sensationless perception were all said to be cases of the continuity or coherence of a surface that is partly occluded. This problem, however, is inseparable from that of the perception of a surface that is temporarily occluded. An optic array is “temporal” as well as “spatial”, and the frozen array is atypical. Partial occlusions are always temporary in life. When a surface loses its sensory visibility it may always regain it later(or vice versa) by the fact that progressive occlusion is reversible by progressive disocclusion. Whatever “goes behind” may later “come from behind,” and those optical transformations are simply temporal reversals of one another.
To put it another way, the continuity of a layout extended in space implies the continuity of existence of the layout in time. Rigid extendedness and rigid persistence go together. Space perception alreadyentails consistency of size and shape (e.g., Wertheimer’s law of proximity for simultaneous coherence with his “law of common fate” for successive coherence).
Consider the following kinds of appreciation in parallel to the four listed above: (1) The phenomenal persistence of an object (or surface) after it has been progressively occluded. (2) The phenomenal persistence of the “front” surface of an object after it has been turned into the back surface (either by rotation of the object or an orbit of the observer’s viewpoint). (3) The phenomenal persistence of a room after its array has been wiped out (by walking into the next room). (4) The phenomenal persistence of the room-behind-one’s-back after one has turned away.
The outcome of reversible transformations for Case 1 has been studied by Michotte in the tunnel phenomenon. The outcome of reversible transformations for the cases has not yet been studied. What is the apprehension that results from such pairs of occlusion-transformations when they are separated by short intervals and by long intervals? This is the question.
Michotte assumes, under the influence of Gestalt theory, that a percept “completes itself”. But is this the only possible explanation? The puzzle of sensationless perception can be dismissed if we accept information instead of sensation as the basis of perception. We could then assume that the detection of invariant material edges the invariant component in what seems to be a flux of changing stimulationis what explains the perception of a persisting and constant layout of the extended environment (Gibson, 1963, 1958). The “cognitive map” thus has parts without “sensory” visibility but not without informational visibility. The animal who has “been around” can “see” things and places behind the modality sensed things and places (Gibson, 1966, Ch. 9).
Developmentally, the implication would be that the child does not learn to fill in the gaps of the intervals between sensory impressions with memories (e. g. Piaget, 1954), but learns to attend to the invariant information for material edges, and to span the reversible transformations over longer and longer intervalsthe pairs that specify going behind and coming from behind, turning and turning back, wiping out and wiping in again, or looking away and looking back. The explanation of phenomenal permanence may well prove to be found here.
References (in sequential order)
Michotte, A., Thinès, G., and Crabbé, G. (1964) Compléments Amodaux des Structure Perspectives, Studia Psychologica, Louvain. (Cornell translation, 1964)
Chapanis,A., and McCleary, R. A. (1953) Interposition as a cue for the perception of relative distance., J. Gen. Psychol, 48, 113-132.
Gibson, E. J., and Walk, R. (1960) The Visual Cliff. Sci. Amer., 202, 64-71.
Gibson, J. J. (1950) The Perception of the Visual World, Boston
Katz, D. (1935) The World of Colour.
Gibson, J. J., and Dibble, F. (1952) Exploratory experiments on the stimulus conditions for the perception of a visual surface. J. Exp. Psychol., 43, 414-419.
Flock, H. (1964) A possible optical basis for monocular slant perception. Psychol. Rev., 71, 380-391.
Smith, P., and Smith, O. W. (1961) Veridical perception of cylindricality: a problem of depth discrimination and object identification. J. exp. Psychol., 62, 145-152.
Gibson, J. J.(1958) Visually controlled locomotion and visual orientation in animals. Brit. J. Psychol., 49, 182-194.
Gibson, J. J. (1963) The useful dimensions of sensitivity. Amer. Psychol., 18,1-15.
Gibson, J. J. (1966) The Senses Considered as Perceptual Systems, Boston.
Piaget, J. (1954) The Construction of Reality in the Child. N.Y.