Abstract
The integration of spatially separated elements into coherent objects is a process that the visual system performs continuously in order to make sense of the world around us. Yet notwithstanding an enormous literature on perceptual grouping, we still lack a clear understanding of the specific representational consequences of grouping disparate visual locations. Our experiment investigated this issue using a stimulus that could induce amodal completion of four elements into two objects behind a central occluding disk. The strength of the amodal completion percept was manipulated by varying the support ratio of the objects. The stimulus was backward-masked at a variable SOA, allowing us to examine changes in the perceptual organization of the visual elements over time. The four elements were curved solid bars, terminating distal to the central occluder in several possible end-shapes, and subjects were asked to indicate the other end matching the one at a pre-cued location (3 alternative forced choice). We compared performance when the matching element belonged to the same apparent object (within-object condition) to when the matching element was at the equidistant location on the other object (between-objects condition). Intriguingly, we generally found superior performance for between-object matches. In a second study where the same stimulus configuration was rotated 45° so that feature comparisons crossed the central vertical or horizontal axes (thus amounting to a symmetry detection task) the effect sometimes reversed, but less consistently. We discuss possible reasons for the differences between the two conditions, the time course of the observed effects, and their significance with respect to object perception. We also relate our findings to the more prevalent finding in the literature of superior performance for within-object comparisons. In conclusion, we suggest that formation of a visual object tends to suppress representation of distinct features at disparate locations within the object.