Abstract
Adaptation is ubiquitous in the visual system, but how perceptual adaptation relates to neural adaptation is unclear. Studies with orientation suggest that a fixed readout of adapted neurons in V1 can explain perceptual adaptation, but this has been difficult to test directly. Here we examine whether a fixed read out of V1 can explain the perceptual shifts produced by adaptation to disparity in random dot patterns. We exploit the systematic relationship between the effects of correlated and anticorrelated adaptors to provide a direct test. Adaptation was measured with 3s stimuli where the first half consisted of the same adapting disparity (correlated or anticorrelated), and the second half contained a random mixture of correlated disparities. This produced systematic changes in spike count during the second half of each trial. In units of dprime, correlated adaptors produced a change ~3.5 times larger than anticorrelated adaptors at the same disparity (n = 126 neurons from two monkeys). A similar trial structure was used in separate experiments to assess the perceptual effect of adaptation in the same two monkeys. Separate 1.5 sec adaptor and test trials alternated, and choices were only reported after test trials. Test trials that had no signal (20%), were identical to the second 1.5sec used for neural recording. While psychometric curves were shifted further by correlated than anticorrelated adaptors, the relative magnitude of the perceptual shift (anticorrelated adaptor/correlated adaptor) was much greater than the relative magnitude observed for neurons. The ratio of the behavioral difference to the neural difference was 2.4 (p< 0.052) and 9.4 (p< 0.002) for each monkey, a six-fold difference on average. The observed relative effect size is hard to reconcile with a fixed readout of V1 neurons, suggesting that the perceptual effects of adaptation require adaptor-induced changes in processing beyond the primary visual cortex.
Meeting abstract presented at VSS 2018