Abstract
Contrast gain control allows visual neurons to adjust their responsivity based on the total contrast of the visual input. We have recently shown in monkeys that the strength of gain control in primary visual cortex varies widely. We also observed that nearby neurons (recorded on the same electrode track) often showed similar gain control strength, suggesting a functional architecture. To further explore the nature of this architecture we measured gain control in V1 of 8 anesthetized macaques. We used 32-channel linear arrays to record multiunit responses at regularly spaced sites along penetrations at different angles to the cortical surface, ranging from nearly surface-normal to nearly tangential. Our stimuli were pairs of orthogonal sinusoidal gratings presented individually or superimposed as a plaid. We compared responses to each plaid with the sum of responses to its two components; the ratio of the measured plaid response to the sum of the component responses was used to quantify the strength of contrast gain control. For every electrode pair on the same track, we computed the similarity in gain control strength (using the fractional difference between ratios) and analyzed how similarity depended on physical separation. Regardless of approach angle, we found that similarity was highest at neighboring recording sites and decreased with distance between sites. We compared the distance-dependence of gain control to that of orientation tuning. As expected, nearby sites preferred similar orientations, but only for distances of 300 μm or less when averaged over approach angle. Similarity in gain control fell more gradually with distance: sites separated by as much as 500 μm were more similar than those farther apart. We found no correlation between orientation similarity and gain control similarity at any distance, suggesting independent architectures for these functional properties.
Meeting abstract presented at VSS 2018