Electrophysiology of segregation

in collaboration with steve luck, ed vogel, and geoff woodman

In everyday visual scenes, objects overlap and occlude one another. For example, as you look at a book on a cluttered desk, the boundaries between the book and other objects seem obvious, even when the book is partially covered by some objects and partially covers others. How does the visual system segregate foreground objects from background objects?

The gestalt psychologists proposed a set of rules that describe figure-ground segregation. For example, a symmetrical region is more likely to be perceived as a figure than as the background. However, this descriptive approach does not specify the mechanisms of figure-ground segregation--the precise computations that allow these rules to be instantiated. Neurophysiological studies in nonhuman primates, in contrast, show promise for elucidating the mechanisms of figure-ground segregation, but the results of these studies are difficult to link with studies of human perception.

The goal of this project is to bridge the gap between human perception and monkey neurophysiology in figure-ground segregation by using noninvasive electrophysiological recordings (event-related potentials [ERPs]) from human subjects. These recordings will allow us to assess the time course and neuroanatomical substrates of figure-ground segregation, linking the human psychophysical and monkey neurophysiological literatures.

Our preliminary results indicate that the neural representation of a figure is more activated than the neural representation of the ground. The ERPs to task-irrelevant probes on 'figures' have higher voltage amplitudes than the ERPs to probes on 'grounds.'

This research is supported by the National Science Foundation (BCS 99-10727)