Parametric decomposition of complex motion by humans

BARRAZA J.F.; GRZYWACZ N.M.

Sarasota, Florida, USA

Congreso; 3rd Annual Meeting of the Vision Sciences Society; 2003

Vision Sciences Society

Purpose: Previous studies showed that humans can discriminate angular velocities (Barraza & Grzywacz, Vis Res, 42 (2002), p. 2457) and rates of expansion (Wurfel et al. VSS, 2002) finely. However, pure rotations and pure expansions are rare in natural motions. Nevertheless, one can almost always decompose visual motions of small surface patches in terms of few elementary components including rotation and expansion. Hence, we wondered whether the brain can decompose complex motions to estimate elementary motion parameters, such as angular velocity and rate of expansion. Methods: We performed a psychophysical matching velocity experiment by using a masking technique, to estimate the component parameters of complex motions. The stimulus was a spiral motion in which one of the components (e.g. Rotation) was the parameter to be measured, while the other (e.g. Expansion) was the mask. Results: We found that human subjects decompose the spiral motion and estimate correctly the angular velocity and rate of expansion. Although these estimates are correct, the sensitivity for discrimination is maximal for pure motions, decreasing with the increase of the mask component. Conclusions: Our results are consistent with the hypothesis that the brain parametrically decomposes complex optic flows into few elementary components. We interpret these results with a population-code model based on cells selective to rotation, expansion and spiral motions.