Abstract
Eye movement recordings allow precise measurement of the time course of perceptual processing. Previously, we measured increased latencies for equiluminant chromatic targets (ARVO '98) and flicker-defined “non-Fourier” motion (ECVP '98). In the present study we attempted to support the hypothesis that the differences are due to visual processing, by comparing the eye movement response with a psychophysical measure of latency. The stimulus was a dark Gaussian disk presented on a white (20 cd/m^2) or gray (5 cd/m^2) background, presentated binocularly using monitors viewed through a mirror stereoscope. In experiment 1, the disk moved horizontally for 3 seconds, with a 1Hz sinusoidal time course. Subjects judged the direction of 3 dimensional rotation, and a staircase procedure adjusted the temporal phase between stimuli delivered to the two eyes so that the “pendulum” appeared to move in a plane. When the two eyes viewed the targets on backgrounds of unequal luminances, the motion of the target on the darker background had to be advanced by 5–8 milliseconds to appear subjectively flat. In experiment 2, the same targets moved with random velocity in 2 dimensions for 8 seconds, and subjects were instructed to track. Binocular eye movements were measured with a pair of video cameras synchronized with the stimulus display. Eye movement correlograms were averaged for 40 different random motions. The resulting signals were fit well by a Gaussian convolved with an exponential. In spite of the fact that the delays measured perceptually correspond to a difference (at the 60 Hz video rate), temporal differences of less than half a video sample were resolved in the fits to the average waveforms. Results from experiments 1 and 2 were in good agreement. These results demonstrate that the eye movement correlogram is capable of resolving fine differences in the time course of processing, and that these differences are likely to arise from early sensory processes.