Abstract
At an early stage, visual motion is processed by a bank of local detectors, each tuned to a narrow range of motion direction. To investigate how the brain encodes the spatiotemporal relationship of the detector bank outputs, past studies have mainly focused on relatively simple optical flows, i.e., translation, rotation and expansion/contraction. However, we can effortlessly perceive many natural scenes containing complex non-uniform motion patterns produced by multiple rigid objects or non-rigid substances. In an attempt to understand the mechanism underlying these percepts, we examined whether and how the visual system adapts to non-uniform motion flows. Our stimulus was an array of Gabor plaids, each moving in some direction within a stationary Gaussian window. In the non-uniform adaptation stimuli, two local motion directions spatially alternated every two rows and columns in a checkerboard fashion, with the spatial phase of the checkerboard boundaries being randomly updated every 1s. In the uniform adaptation stimuli, one local direction was presented at a time, and switched to the other direction every 1s. The test pattern was a non-uniform checkerboard pattern defined by a motion direction change, and the minimum direction difference with which observers could detect the checkerboard structure was measured by the method of adjustment. We found a significant increase in the threshold direction difference after adaptation to non-uniform motion stimuli in comparison to adaptation to uniform motion stimuli, although the state of local motion adaptation was expected to be similar between the two conditions. That is, a judgment of non-uniformity became difficult after prolonged viewing of a non-uniform motion pattern. Furthermore, the aftereffect was observed even when the local motion directions for the adaptation and test stimuli were widely separated. These results suggest the existence of specialized mechanism encoding non-uniform local motion flow changes.
Meeting abstract presented at VSS 2014