Abstract
Single cell recordings from monkey visual cortex show that many neurons, especially in area V2, are selective for border ownership. These neurons are edge selective and have ordinary classical receptive fields, but in addition, their responses are modulated (enhanced or suppressed) depending on the location of a ‘figure’ relative to the edge in their receptive field. This selectivity is derived from the image context far beyond the classical receptive field. The large context sensitivity might be interpreted as indicating that border ownership assignment involves object memory. To test this hypothesis, we measured neural border ownership selectivity for silhouettes of face profiles and matched control shapes. The hypothesis predicts that face profiles produce stronger border ownership selectivity than the control shapes. As reported (Ko & von der Heydt, Soc. Neurosci. Abstr. 464.11, 2012), recordings did not show this superiority of faces. Here we report results of two behavioral tests with the same stimuli. (1) One monkey was trained in a match to sample shape discrimination task. Discrimination performance (proportion correct minus chance) was 50% higher for the face profiles than for the control shapes (p = 0.015). (2) In another, ‘naïve’ monkey that had not seen our test shapes before, we examined whether the face profiles would attract attention more often than the control shapes. The task required one second of fixation during which a face and a control shape were presented simultaneously at symmetrical eccentric positions. The monkey occasionally broke fixation after the onset of the stimuli. In these trials, saccades went to the face profile significantly more often than to the control shape (p=0.008). Thus, while face profiles are clearly distinguished behaviorally, recordings fail to show a corresponding influence on border ownership signals. We tentatively conclude that the visual cortex assigns border ownership without recourse to object memory.
Meeting abstract presented at VSS 2013