Abstract
Discrimination of the direction of motion depends on (1) the difference in direction between the discriminanda, (2) the signal to noise ratio of the motion signal, and (3) the viewing time. An increase in any of these factors should improve performance. Here, we examined the dependence of subjects' performance on these factors by asking them to judge the direction of motion of a peripherally-located dynamic random dot field in a single-interval two-alternative forced choice experiment. The signal strength was controlled by the fraction of dots that moved coherently in the target direction; the remaining dots changed position randomly and served as an external source of noise. We tested 5 different angular differences (θ: 12, 22.5, 45, 90 and 180 degrees) and 5 different viewing times (T: 100, 200, 400, 600 and 800 ms). In a single block, we measured coherence threshold for all 5 viewing times for a given θ. For the range of values tested, coherence threshold is a monotonically decreasing function of both θ and T. The drop in performance at shorter viewing times is not compensated by a proportional increase in the signal strength, suggesting imperfect temporal integration at all values of θ. The decrease in threshold with angular difference is most pronounced for short viewing times, and suggests a loss of signal for smaller angular differences, presumably because of the overlap of the tuning curves of relevant channels. We interpret our results in the framework of a computational model of direction selective neurons, with which we account quantitatively for the roles of tuning bandwidth, spontaneous firing rate and inter-neuronal correlations in determining perceptual performance.