Integration of local features into visual shapes in the human visual cortex

The perception of visual shapes entails that local image features are integrated into global configurations that represent visual forms. The lateral occipital complex (LOC) in the human brain has been proposed to be primarily involved in the visual analysis of shape. The goal of the present study was to investigate the role of the LOC in figure-ground segmentation and contour integration of simple geometric shapes by using event-related functional magnetic resonance imaging in human subjects. The stimuli consisted of arrays of Gabor elements. Two types of stimuli were used: a) random patterns that consisted of randomly oriented and aligned Gabor elements and b) contours that consisted of a set of Gabor elements that were aligned to a closed contour and embedded in a background of randomly oriented Gabors. We independently localized the LOC in each subject and tested fMRI responses in this region of interest. Our first experiment showed stronger activation in the LOC for contours than for random patterns. Moreover, we found stronger activation in the LOC when the detection of contours was facilitated by additional visual cues, namely motion or disparity. In a second set of experiments, we degraded the contours and found decreased activation in the LOC when the contours were difficult to detect from their background. Our findings suggest that neural populations in the LOC are involved in the integration of local image features and the visual perception of shapes