Abstract
Numerous regions along the ventral visual stream have selective responses to particular categories of stimuli. However, it has not been established what types of low- and mid-level visual properties contribute to the emergence of high-level category selectivity during processing. One dimension that has been posited to play a role is the curvilinearity/rectilinearity of shape contours: fMRI studies have found that face-selective areas respond preferentially to curvilinear shapes while place-selective areas respond preferentially to rectilinear shapes. Unknown, however, is whether the dissociation of curved and rectilinear contours precedes or follows category-selective responses. In particular, does the separable processing of faces and scenes rely, at least in part, on differences in shape contours? Alternatively, is shape-contour sensitivity an epiphenomenal consequence of already divergent processing of faces and scenes? To address this question, we recorded event-related potentials during a simple luminance-change detection task while participants viewed a series of images of matched novel curvilinear and rectilinear patterns, as well as faces, scenes, and objects. We observed that curvilinear and rectilinear shape contours give rise to dissociable signals at lateral-occipital electrode sites. However, the signals temporarily diverged after the initial separation of face and scene ERP components. This finding suggests that the presence of differing shape-contours is not the basis for the initial segregation of faces and scenes in visual processing.
Meeting abstract presented at VSS 2016