Abstract
We studied the effect of high-contrast Gabor flankers on the contrast response function of a centrally placed Gabor patch (target). First, the lateral facilitation effect was reproduced - a two-fold decrease in target detection thresholds for the collinear configuration (target and flankers are aligned). Target and flanking Gabor patches were separated by 3 standard deviations of the Gaussian window. These results show that the ratio between the slope (gain) of the contrast response and internal noise is increased in the presence of collinear flankers. Thus the flankers effect can be explained by either increase in gain due to lateral interactions or by noise reduction. Here, we employed an identification task to obtain independent estimates of the actual response function and noise [Katkov, Tsodyks, Sagi, Vision Research (2007) p2855]. On each trial, the observer was randomly presented with either a collinear configuration or an orthogonal one (target is orthogonal to flankers). Observer reported which contrast out of four (randomly selected on each trial) the central grating had. To analyze the data we assumed a signal detection theory (SDT) based model where each stimulus evokes internal response normally distributed across trials, with category boundaries (criteria) set at the decision stage. Observer reports a particular category when the internal response falls between the corresponding boundaries. We further assumed that criteria are independent of the stimulus presented on a particular trial. The models fitted to the data show that the internal response is higher in the presence of collinear flankers as compared with the orthogonal ones (separated by at least one noise amplitude), whereas there are very small changes in the noise amplitudes. Since the noise amplitude in SDT represents uncertainty in the internal representation of stimulus, our results support the lateral interaction hypothesis, with a slight reduction of uncertainty for collinear configuration.