Abstract
Binding by temporal synchrony of firing has been proposed as a means by which 3-D structural representations of objects by their parts can be constructed from simple features such as bars and edges (Hummel & Biederman, 1992; Hummel & Stankiewicz, 1998; Hummel, 2001). Gray et al. (1989) showed that spatially separate neurons in cat V1 fire in temporal synchrony according to properties of the global visual stimulus. In a computational model of Visual Area V1 that operates on full-color images, we demonstrate a mechanism by which spatially separate but collinear bar and edge detecting neurons can achieve temporal synchronicity of firing, consistent with the findings of Gray and colleagues. The model also shows mutual desynchronization of different contours, which has been proposed to be a function of visual attention (Stankiewicz & Hummel, 1996).