NEURAL VISUAL CODING: IMAGE TO OBJECT REPRESENTATION

Johns Hopkins University, Baltimore MD

Investigators

Linked publications & trials

Abstract

DESCRIPTION (Adapted From The Applicant's Abstract): This proposal will continue neurophysiological studies on the visual processes that lead from representations of images to representations of objects. Based on the concept of the receptive field, the first stages of the visual cortex have generally been interpreted as representations of local image features, with progressive addition of details and increasing sensitivity to the context around the receptive field, but still based on local neighborhood operations. The PI's laboratory has discovered that neurons of areas V1, V2 and V4 respond differently to the same local contrast border depending on whether the border belongs to a figure on one or the other side of the border (border ownership coding), indicating an object-related feature representation and processes that require global form processing. These findings challenge the traditional concept. In order to interpret the new findings in terms of psychological and computational theories of vision, we need (1) to understand the nature of object-related feature representation at those cortical levels, and (2) to understand how object-related feature coding relates to high-level processes such as top-down attentional selection. The proposed experiments will explore object-related feature coding in V1, V2 and V4 with various paradigms of border ownership and perceptual completion and they will determine how this coding changes between the conditions of passive viewing and active visual selection in a shape discrimination task. The results will distinguish between two hypotheses: (a) that object-related signals at lower cortical levels are the byproduct of attention processes, and (b) that object-related feature representations result from independent processes which may be used by attentional mechanisms to select visual information efficiently.