Abstract
The case of holes presents a unique problem for shape perception and figure-ground organization. Normally in bipartite figure-ground displays, the figural side is remembered while the ground is not, even though they share the same contour. When a surrounded region is perceived as a hole through another surface, however, the shape of the hole itself is remembered. This presents a case in which two factors which are normally coupled in figure-ground perception — depth assignment and contour ownership — become decoupled. In the present experiments, a distinction is made between holes which are an intrinsic part of the surface to which they belong and holes which only arise from an accidental relationship to their bounding surfaces. In the first experiment, it is demonstrated that memory for intrinsic holes is superior to that for both accidental holes and accidental objects, whose boundaries and shape change according to viewing perspective. A second control experiment revealed a confound in the structure of these stimuli, however, indicating that larger shapes are remembered better than their constituent parts. A third experiment using stereo viewing with shutter glasses and a different set of stimuli without this problem, however, validated the original conclusion that intrinsic holes are remembered, whereas accidental holes are not. It is concluded that certain types of holes in surfaces are perceived as having shape (in the way that figures have shape), whereas others are not.