Neural Principles of Visual Cortical Processing: Scene Segmentation Investigated with Microelectrodes and Models

Abstract

Stimuli composed of coherent features are integrated by our sensory systems into perceptual entities, even if the features are dispersed among different sensory modalities. We can perceive a sensory object as a perceptual whole even if various aspects of the object are occluded, obscured by the background, or are not present at all. The visual system can easily detect coherencies in an object’s stimulus features, and is able to link, intensify and isolate them. From theoretical considerations it seems clear that early processing of feature relations is highly important because the infinite number of possibilities in which even simple “toy scenes” can be arranged has to be reduced substantially before interactions with visual object memories. While the rules governing such capabilities of grouping, mutual facilitation, and figure/ground separation have been analyzed in detail at the beginning of this century by “Gestalt” psychologists (e.g., Wertheimer, 1938 the underlying neural mechanisms are still largely unknown.