Abstract
A controversy remains in the literature on the visual mechanisms underlying the detection of a plaid pattern. One theory suggested the detection was mediated by a plaid selective mechanism while the other, a combination of two oriented channels with contrast gain control. Here, we used a pattern masking paradigm to investigate this issue. The target was either a spiral plaid or a spiral grating on a spiral plaid or a spiral grating (in either parallel or orthogonal orientation to the grating target) pedestal. A temporal 2AFC with staircase method was used to measure the discrimination thresholds of the target at 7 pedestal contrasts. All conditions, except the one with orthogonal target and pedestal gratings, showed a typical dipper shape target threshold vs. pedestal contrast (TvC) functions. That is, the threshold first decreased (facilitation) and then increased (suppression) with pedestal contrast. TvC function for grating-on-grating and plaid-on-plaid conditions overlaped at high pedestal contrasts. The TvC function shifted upward for the grating-on-plaid condition. The plaid-on-grating condition showed a weaker facilitative and suppressive effects than the plaid-on-plaid conditions. We fitted the data with versions of the divisive inhibition model, where the facilitative input was divided by the sum of broadband inhibitory inputs, to test whether our result can be accounted for by a filter specific for plaid or by a combination of two orthogonally oriented filters. The latter model described the data better with fewer free parameters. We conclude that plaid detection is mediated by a combination of two contrast gain control mechanisms with receptive fields whose orientation are orthogonal to each other.
Meeting abstract presented at VSS 2014