Cortical Processing of Visual Parts and Wholes

TR Number
Date
2014-08-25
Journal Title
Journal ISSN
Volume Title
Publisher
Virginia Tech
Abstract

Visual perception theory distinguishes between two distinct levels of object processing: holistic, based on global shape, and configural, based on local features and/or component parts. Empirical evidence suggests that different cortical regions may support these levels; holistic processing correlates with activation in the lateral occipital-temporal cortex (LOC), whereas configural processes correspond to activation in the parietal lobe, particularly the intraparietal sulcus (IPS). This study combined theories of visual part structure with neuroimaging methods to investigate the relative contribution of holistic and configural processing in an ecologically valid object recognition task. Rather than rely on stimuli specifically designed to evoke holistic or configural processing, this study used photographs of objects selected without a priori assumptions concerning physical part structure. Twenty participants viewed objects at fixation while undergoing fMRI, followed by a behavioral object identification task involving the same objects presented in peripheral vision. Behavioral data were analyzed according to theories of visual crowding to yield an objective estimate of the number of parts perceived within each object. Neuroimaging results revealed increased activation for holistic objects containing fewer parts in the right parietal lobe and superior temporal gyrus and bilaterally in the fusiform gyrus, suggesting a relation between holistic processing areas and object perception. Configural objects with many parts elicited increased activation in the left angular gyrus. This study, to our knowledge, is the first to investigate the cortical visual regions involved when observers engage in holistic and configural processing as a natural part of visual recognition.

Description
Keywords
hemispheric lateralization, fMRI, holism, visual part structure
Citation
Collections