Abstract
Holes present an intriguing paradox for figure-ground organization. Although the outside of a hole is seen as the closer figure and the inside as the farther ground, the shape of the interior region appears to be well perceived and remembered, contrary to the usual claim that border assignment is unidirectional and linked for depth and shape. Recent evidence has questioned the claim that the interior shapes of holes are perceived (e.g., Bertamini & Croucher, 2003). We present four experiments that demonstrate circumstances under which the shapes of visual holes are perceived and remembered — or not. In Experiment 1, subjects saw a series of novel shapes presented as real, physical models of 2D surfaces with and without holes. A later shape recognition test showed that they remembered the interior shapes of the holes as well as the shapes of the solid objects. A second study with stereoscopically viewed depth displays demonstrated that memory for the interior shape of holes is limited to intrinsic holes, whose contours are continuous in depth within a single object. Memory for accidental visual holes, arising from the coincidental alignment of discontinuous contours of two or more objects separated in depth, was no better than chance. A control experiment ruled out explanations based on artifacts due to differential size or shape. Further studies used instructional manipulations to investigate the influence of differential attention to intrinsic versus accidental holes to determine whether the observed differences in shape memory are obligatory or strategic. The findings are discussed as supporting an account of hole perception in which the shape of a hole is attended and encoded as an immaterial (or virtual) surface where matter is missing from the otherwise solid surface, perhaps with a “missing sign” (analogous to a minus sign in mathematics) to represent its status as a hole rather than as a material part.