Abstract
It has been shown that humans cannot perceive more than three directions from a multidirectional transparent motion stimulus. However, it remains unknown whether adapting to such imperceptible motion could generate any perceptible motion aftereffects (MAEs). Here, we demonstrate that the visual system integrates local aftereffects over space to produce perceptiple MAE after adapting to the multidirectional motion stimulus. The stimulus contained 260 randomly-oriented, equally-spaced drifting Gabor elements. They were randomly divided into five sets, each of which was assigned one distinct global motion direction. Local velocities of elements within the same set were manipulated so that they were consistent with the set’s global velocity. In Experiment 1, we first confirmed that observers were unable to identify any of the five embedded directions in this multidirectional stimulus. We then had observers adapt to this imperceptible pattern, and then indicate the MAE direction perceived on a static test pattern. We found that, when tested with elements taken from multiple sets (the Mixed condition), reported MAE directions were uniformly distributed. However, when tested with elements taken from the same set (the Single condition), observers reported MAE directions mainly 180° away from the set’s adapting direction. In Experiment 2, we used complex motion patterns (rotations and expansions) and found similar results. These results show that humans can clearly perceive individual MAE directions after adapting to an imperceptible motion pattern. We developed a two-stage computational model that: 1) computes local MAEs based on adaptation-induced changes in tuning properties of local motion detectors, and 2) averages local MAEs across test locations. Prediction from the model matches well with humans’ perceived MAE directions in all experimental conditions. Altogether, our psychophysical and computational results provide evidence for the existence of a mechanism that integrates local aftereffects over space to generate global MAE percepts.
Meeting abstract presented at VSS 2012