Abstract
Parallel perception of multiple orientations is essential for identification of patterns and 3-D shapes from texture. While the primary visual cortex of primates and carnivores is organized in columns of neurons that respond to the same orientation preferences, neuronal responses to gratings of preferred orientation can be suppressed when another orientation is added to create a plaid. It is currently unknown how the different factors contributing to this cross-orientation suppression vary across the orientation map. To address this question, we performed horizontal penetrations with multielectrode arrays in the primary visual cortex of anesthetized cats and sampled populations of neurons in different cortical orientation domains (32-channel Neuronexus probe, 100 microns inter-electrode distance). Cortical neurons were stimulated with a sequence of static sinusoidal gratings (100 msec duration) containing 8 contrasts, 8 orientations, and 4 phases and a sequence of plaids consisting of superimposed pairs of gratings at 50% contrast, 8 orientations and 4 phases. We first confirmed contrast-invariance of orientation tuning and orientation-invariance of contrast response functions with these sequences of static gratings. We then calculated a Suppression Index as the ratio between the response to the preferred grating at 50% contrast (GR) and the average response to the plaids (PR), as (GR-PR) / (GR+PR). The distribution of Suppression Indices could be accurately fit with a Gaussian function (R2= 0.96, mean= 0.18 ± 0.10). In 95% of neuronal recordings (131 out 138), the added grating suppressed responses below that to the preferred component alone, which clearly rules out contrast saturation. Interestingly, the suppression index was negatively correlated with orientation tuning bandwidth (r = -0.34, p < 0.0001). In addition, preliminary analyses suggest that the response suppression is related to the rate of orientation change in visual cortical domains.
Meeting abstract presented at VSS 2015