Size perception from an egocentric point of view.

Abstract

It was hypothesized that the apparent height of a visual stimulus is affected by the amount of space which it is seen to occupy within a surrounding frame of reference. It was further speculated that the lower limit of the surround is established by the ground plane and the upper limit by the apparent eye-level position. In other words, it was proposed that visual and egocentric cues to spatial location combine to create a frame of reference which can be used relationally to determine apparent visual size. It was felt that srimulus size is judged in comparison to the observer's own body and that the height of the body is represented in external space by the distance which separates the ground from the apparent eye-level position. If this is true, then an object which stands upon the ground and extends up to the level of our eyes should be judged to be of equal height to ourselves. One which falls below eye level should be seen as short and one which extends over our heads should appear to be tall. Several predictions stem from this point of view. Consider first the case of a small object. Such an object would be seen to rise with increasing distance from the observer, with the bottom of the object rising with the ground and the top approaching eye level. Since the boaom of the object is tied to the ground and must follow the laws of natural perspective, the decrement in angular size produced by increasing distance would be attributed to the perceived incline of the ground plane and not to the object itself. Thus, a stimulus of progressively decreasing angular size would be seen to occupy more and more space between the ground and eye level and, therefore, should appear to grow. Second, an object as tall as ourselves should seem to remain fairly constant in size with increasing distance, as it would continually occupy the full amount of space between the ground and eye level. Again, decreases in the size of the projected image could naturally be aaributed to perspective. Third, a tall object should be seen to shrink with increasing distance, as the top of such an object is seen to collapse angularly toward the eye level. In this instance, however, the decrement in perceived size should be less than could be expected from the decrease in the projected image as some of the decrease in visual angle would again be attributed to perspective. Finally, a complete regression to visual angle would be expected with the occlusion or omission of a ground plane. All of the above speculations were supported in a series of experiments dealing with the perceived size of moving stimuli, using binocular vision in darkness, open field settings, and well articulated corridors. W e are presently studying the effect of the slope of the ground plane on apparent size. The perception of size seems to involve a dynamic interaction between a stimulus and its surround. The process seems to be attributional, in that changes in the proximal stimulus can be allocated to the surround. Paradoxical phenomenological reports, such as seeing an object remain constant while smaller, can be understood in these terms. While constancy may be based upon the maintenance of an invariant relationship between an object and its surround, the sense of getting smaller may be derived from changes in the proximal stimulus which are attributed to the perspective of the ground plane. The purpose of this paper was to present the reader with another way of thinking about size constancy, one which does not involve the utilization of distance cues.