Abstract

This study found that both global and local contrast differences are used to discriminate left-right movement. Movement discrimination was examined by measuring an observer’s ability to discriminate shifts in the position of contrast differences over time for a large range of background contrasts, spatial frequencies, and exposure durations. The contrasts needed to discriminate between two test patterns separated in time by a short interval and shifted in position optimally relative to a multifrequency background were measured using two alternative forced-choice procedures. Multifrequency backgrounds determined the window of visibility used to discriminate movement. A neural network model implemented by the weighted difference and sum of the output of paired odd- and even-symmetric simple and complex cells (the spatial and temporal gradients of paired Gabor filters) pooled across the window of visibility predicts horizontal movement discrimination.

© 1987 Optical Society of America