Perceptual cost for motion transparency depends on the measures of psychophysical performance

Abstract

Transparency perception is recognized as one of the important phenomena for the theory of early vision. This is because transparency perception suggests that a simple theory reconstructing a single-valued field of a visual attribute, such as an optical-flow field, cannot model the neural mechanism in the brain and raises the fundamental issue of how the visual attribute is represented and computed in the visual cortex. Here we present a systematic study that examined the perceptual cost for motion transparency psychophysically. It has been known that the perceptual performance in motion transparency is worse than that predicted by assuming overlapping motions are detected individually. This perceptual ldquocostrdquo would reflect the neural encoding mechanism for transparent motions. The present results showed that the properties of the perceptual costs varied with the measures of psychophysical performance; the perceptual cost evaluated by motion detection thresholds became smaller as the directional difference between overlapping motions decreased, whereas the cost examined with precisions of directional judgments became worse. A computational analysis with a simple population coding model suggests that the perceptual costs evaluated by two measures would arise from different mechanisms.