Perceptual Organization in Motion and Lightness Perception
Massachusetts Institute of Technology
It is hard to make progress in any area of vision without studying perceptual organization. For example, there are very simple motion stimuli that foil almost every model of motion analysis, because these models lack the sophisticated machinery needed to combined motion and form information across space. I will illustrate some of problems, and and suggest some solutions, based on our recent research with ambiguous motion stimuli. I will also discuss problems in lightness perception. Here again, perceptual organization turns out to be a critical part of computations, as evidenced by numerous illusions and other phenomena that depend on configural properties of a stimulus. I will also argue that the standard notions of grouping and binding are too simplistic to be very useful in visual computations. Psychophysical and computational considerations suggest that there may be multiple grouping mechanisms operating in parallel, each with rules appropriate to its task.