Neural mechanism of figure-ground segregation at occluding contours in monkey prestriate cortex

An important aspect of figure-ground segregation is the detection of occluding contours and the discrimination of figure and ground at such contours. In this paper we investigate the neural processing and the representation of occluding contours defined by occlusion cues. We trained rhesus monkeys on a visual fixation task that reinforced foveal viewing. During the periods of active visual fixation we recorded the responses of single neurons in areas V1 and V2 of the visual cortex. The stimulus conditions that we used mimicked situations of spatial occlusion; usually an opaque, uniform rectangle (or tongue) which appeared to overlay a larger, rectangular grating texture. The neural responses were analyzed with respect to figure-ground direction and contrast polarity. Neurons in area V1 either failed to respond to such stimuli, or they were sensitive to contrast polarity only; others responded unselectively to all types of contour. By contrast, more than one third of the neurons studied in area V2 were sensitive to the direction of figure and ground at such contours. The majority of these neurons were not sensitive to the contrast polarity of these contours. Some rare neurons preferred a certain combination of figure-ground direction and contrast polarity. We explain these results in terms of a model that had been developed to explain neural signals of illusory contours, contours that often coincide with occluding contours. In conclusion, the results of the present paper suggest that in monkey visual cortex occluding contours and mechanisms for the segregation of figure and ground at such contours are first represented at the level of area V2.