STRUCTURE-SENSITIVE INTEGRATION

IN HUMAN MOTION PERCEPTION

We live in a world of objects that move. To perceive these motions correctly, the human visual system cannot simply register motion signals at different positions on an object contour. In fact, all local motions are ambiguous (a property of vision that may be called kinematic indeterminacy). To overcome this ambiguity, local motion signals must be integrated into a coherent global interpretation. In some cases, the integration may consist of a simple form of vector averaging (a structure-blind integration process). Most often, however, to recover the correct global motion of an object the visual system must take into account the spatial structure of the optic array. This tutorial discusses several demonstrations of the role of spatial organization in motion perception. The barberpole effect illustrates the role of contour terminators. The chopstick illusion and the sliding effect illustrate the role of figure/ground segregation. Motion stereograms illustrate the role of binocular disparity. Occluded terminators, Petter plaids, apparent rest and paradoxical rest illustrate the role of occlusion, modal, and amodal completion. Invisible corners illustrate the role of higher-level expectations. Finally, the breathing square and illusory blob effects illustrate that motion integration does not necessarily obey a rigidity constraint.


created by Fauzia Mosca and Nicola Bruno