Abstract
We measured tuning for step size and delay for left-right direction discrimination of moving random pixel arrays (RPAs) by cats. Hereto, four cats were trained to perform a left-right direction discrimination task, indicating the perceived direction of motion with their paws. Performance was measured using a luminance signal-to -noise ratio in a Quest staircase procedure converging at the 80% correct level. The noise consisted of a spatiotemporally incoherent RPA added to the moving RPA on a pixel by pixel basis. Velocity tuning curves show an optimum at about 20 deg/s, corresponding to human peripheral vision. Two different paradigms were used to measure the step sizes and delays that enable cats to detect coherent motion. First, we used a single step pattern lifetime (SSPLT) stimulus, in which patterns on alternate frames make a coherent step and are being refreshed. Since there is no correlation over multiple steps, motion detection is necessarily based on the specific combination of step size and delay. Results show a limited range of step sizes and delays supporting direction discrimination. Second, we measured thresholds for motion seen through a bar pattern orthogonal to the direction of motion. In this case, one expects maximal interference if the width of the bars equals the preferred step size. Results indeed show minimal sensitivity at intermediate bar widths, corresponding to the preferred step size for SSPLT stimuli. The results support a bilocal correlator type of motion detector.