Abstract
Neurons in the primate primary visual cortex (V1) are well organized into columns based on their orientation preference. Although this columnar organization has been studied for decades, it is still unclear how neural activity in these columns contributes to orientation perception. As a first step towards addressing this question, we used voltage sensitive dye imaging (VSDI) to measure the columnar signals in V1 of a monkey while it performed a fine orientation discrimination task. To assess neural sensitivity, we developed a linear decoder that pools the single-trial VSDI signals over space using weights proportional to the orientation response map at the columnar scale. The decoder then uses the pooled signals to perform the same task as the monkey. To test the hypothesis that columnar signals provide the main source of information in the orientation discrimination task, we varied the spatial frequency of the stimulus. Under this hypothesis, spatial frequency should have a similar effect on neural and behavioral sensitivities. However, we found systematic differences in the neural and behavioral effects of spatial frequency. The monkey was most sensitive at the middle frequency (2cpd), and performed worse at both low (0.5cpd) and high (8cpd) frequencies. In contrast, neural sensitivity was best at 8cpd, intermediate at 2cpd, and much worse at 0.5cpd. There are at least two possible sources that could contribute to these discrepancies between the neural and behavioral effects. First, additional sources of information beyond the columnar signals may contribute to behavioral performance at intermediate and low spatial frequencies. Second, the efficiency with which V1 signals at the columnar scale are decoded by subsequent processing stages may decrease with increasing spatial frequency. By studying trial-by-trial covariations between V1 signals and behavioral choices we may be able to distinguish between these possibilities.
Meeting abstract presented at VSS 2015