Title: Responses to correlated and anticorrelated random-dot stereograms in V3/V3A neurons in the Monkey
Graduate School of Frontier Biosciences, Osaka University
Abstract: Binocular disparity ,which is the difference between left and right retinal images, is an important cue for depth perception. The visual system conducts a matching computation that identifies the correct match between retinal images among numerous false ones. Although in many visual areas such as V1, V2, V4 and MT, disparity tuning is well characterized, the same cannot be said in V3 and V3A. We therefore compared disparity tuning in these two areas with that in V2. We used correlated random-dot stereograms (cRDSs) and anticorrelated random-dot stereograms (aRDSs). When one looks at aRDSs in which the contrast polarity in one eye is inverted, the person cannot perceive a three-dimensional structure, because there are no correct match between retinal images. Despite this, V1 and V2 neurons have been shown to be sensitive to aRDSs (Cumming and Parker 1997; Parker 2007), meaning they code for a false match. Higher visual areas must resolve this problem, resulting in aRDS insensitivity. We recorded single unit activities in V2, V3 and V3A while two monkeys performed a fixation task, finding no apparent differences in the Gabor parameters, which describe the disparity tuning curve, between the three areas. We calculated the sensitivity to aRDSs as an amplitude ratio (the ratio of the Gabor amplitude for aRDSs to that of cRDSs). The amplitude ratio in V3 and V3A were significantly lower than that for V2, meaning V3 and V3A had less sensitivity to aRDSs. In conclusion, the disparity tuning curves for cRDSs are very similar in V2, V3, and V3A, while V3 and V3A discards the sensitivity to aRDSs. V3 and V3A are involved in the computations that eliminate any false-match responses elicited in V1 and V2 and are contribute to the matching computation.