Abstract

Border ownership seems instantaneous and automatic, emerging as rapidly as figure separates from background. But this apparent immediacy hides an extremely subtle problem that lies at an intermediate level of visual processing. Somehow border ownership must integrate upward-flowing image-derived information about boundaries with downward-flowing figural information derived from boundary arrangement. It is related to Gestalt notions of closure. To appreciate border ownership as a process, we contrast it with a more traditional pathway to surface inferences, from boundaries to stereo and, finally, to shape from shading. In the end border ownership emerges as a rather different concept from these more geometric visual computations, thus establishing two poles to perceptual organization. The challenge is integrating the geometric content of borders with the topological content of figures within standard modeling formalisms. We suggest an expanded framework based on mathematical and physiological ideas, perhaps involving local field potentials.