Abstract
Visual perception depends not only on local stimulus features but also on their relationship to the surrounding stimulus context. Many known visual figure-ground cues are based on local edge features such as edge convexity, adjacent region brightness, or previous experience. Recently though, contextual influences from remote, non-adjacent edges have also been shown to affect figure-ground organization (Zhang & von der Heydt, 2010; Brooks & Driver, 2010; Peterson & Salvagio, 2008). Intermediate visual areas may play a role in such contextual influences. We tested this by examining a rare case (LG) of developmental visual agnosia. A previous study showed that LG had no evident abnormality of brain structure and functional neuroimaging showed relatively normal V1 function. In contrast, his intermediate visual areas (V2/V3) were deactivated in response to visual stimulation. To determine whether this abnormal pattern of visual cortical function may selectively affect contextual processing, we tested LG on a set of local figure-ground cues as well as contextual figure-ground influences. We showed LG dynamic displays with two bipartite sections. Figure-ground assignment of the dividing edge within one bipartite section was locally-biased in one direction by an edge-region grouping cue (Palmer & Brooks, 2008). The dividing edge of the other section was locally-ambiguous with regard to figure-ground assignment. In control participants figure-ground assignment of the locally-ambiguous edge is affected by the locally-biased edge in the context. We found that although LG’s figure-ground assignment in the locally-biased section was equivalent to controls, contextual influence on the locally-ambiguous section was significantly reduced. Our results suggest dissociable mechanisms for contextual and local influences on figure-ground assignment. These results suggest that computational and neural models of figure-ground organization should contain dissociable mechanisms for local and contextual influences.
Meeting abstract presented at VSS 2012