Abstract
We examined the ability of disparity-selective neurons (recorded from V1 of awake fixating macaques) to support depth judgements over the range of visible disparities. A selected population of 180 neurons with strong disparity selectivity (eccentricities 1 to 5 degrees) was used. To allow comparison between neurons that respond over different ranges of disparity, the responses were fit by Gabor functions, and the disparity that caused the largest difference from the response to uncorrelation was measured. The distribution of this maximum interaction position was centred close to zero disparity (mean −0.04 degrees) with an S.D. of 0.21 degrees. The disparity at which the rate of change of firing was greatest had a similar mean, with an S.D. of 0.19. By both of these measures, 97% of neurons encode disparities < 0.5 degrees. To summarize modulation over the whole tuning curve, we first took the square root of spike counts, to remove the dependence of variance on rate. The rate of change was then calculated as a function of disparity at each point on a tuning curve. The average across 180 neurons indicates the extent to which the population is able to signal changes in disparity at different pedestal disparities. The peak sensitivity occurred at zero disparity, while at +−0.5 degrees the sensitivity was reduced to less than 10% of this peak value. At +−1 degree there was almost no modulation. These data suggest that the maximum disparity that could be reliably signaled by V1 neurons (for stimuli at eccentricities < 5 degrees) is between 0.5 and 1 degrees. Using matched RDS stimuli, the largest disparity at which the animals were able to report correctly the stimulus depth on 75% of trials was less than 1 degree. Although these judgements can be supported by the responses of disparity selective V1 neurons, the ability to judge depth from larger disparities with spots (or other non-RDS stimuli) is presumably derived from a different population of neurons.