Early Visual Cortex Assigns Border Ownership in Natural Scenes According to Image Context

J R Williford1, R von der Heydt2

1School of Medicine, Neuroscience Department, Johns Hopkins University, MD, United States
2Zanvyl Krieger Mind/Brain Institute, Johns Hopkins University, MD, United States

Contact: williford@jhu.edu

Discerning objects from their backgrounds is a fundamental process of vision. A neural correlate is coding of border ownership in early visual cortex [Zhou et al, 2000, J. Neuroscience, 20(17), 6594-6611]. When stimulated with the contour of a figure, neurons with this correlate respond more strongly when the figure is on one side of their receptive field (the "preferred" side) versus the other. So far, border ownership coding has only been shown with simple displays of geometric shapes (e.g., squares). Here we studied border ownership coding with static images of natural scenes, using microelectrodes to record from isolated neurons in V1 and V2 of macaques. We found that subsets of V1 and V2 neurons indeed code for border ownership in complex natural scenes. Decomposition of local and context influences showed that the context-based border ownership signals correlated with those for the (locally ambiguous) edge of a square, but were weaker. We used stimuli with intermediate complexity along several dimensions to measure the relative influences of object shape, occlusion between objects, texture and color contrast to determine how they contribute to the border ownership signal strength. The signal decreased with complexity along all dimensions, especially with shape, occlusion and texture.

Up Home