Neuroimaging of shape similarity in object recognition

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): The goal of the proposed research is to determine the neural substrates associated with visual recognition. An area of the human brain, referred to as ventral temporal cortex (VTC), is implicated in visual recognition based on evidence from electrophysiological and lesion studies with non-human primates and on clinical observations that visual recognition abilities are lost when this region of the brain is damaged. In addition, human lesion studies and recent functional neuroimaging findings suggest that different regions of VTC may preferentially respond to different object categories. A popular interpretation of these results is that specialized neural subsystems are dedicated to recognizing different object classes. Cognitive behavioral studies, however, have demonstrated that object recognition depends critically on processing information about an object's shape, and that shape information drives category organization. Consequently, the thesis of the present proposal is that category-specific patterns of brain activation in VTC may, in fact, reflect varying degrees of shape information processing that is crucial for object identification. For example, manufactured object categories like tools and furniture do not overlap much in shape, whereas faces overlap tremendously in shape. Brain areas that appear to be specific to face processing may, instead, be involved in resolving high degrees of shape similarity. Functional magnetic resonance imaging (fMRI) is used in the present proposal to examine blood-flow changes in VTC during object recognition. The first goal is to determine the role of VTC in processing shape information associated with objects. One hypothesis is that posterior VTC is critically involved in mapping perceptual information onto stored information about object form. Another hypothesis examines the magnitude and spatial distribution of fMRI response in VTC as shape similarity parametrically increases among objects. Posterior VTC regions are likely involved when shape similarity is low (as in the case of manufactured object categories), but anterior VTC regions are likely recruited to resolve high degrees of shape similarity (as in the case of faces). The second goal is to compare the efficacy of this shape similarity framework with two other popular approaches to category specificity - a taxonomic category account and a categorization-level account. Shape similarity may explain both taxonomic category differences and categorization level differences in brain activation patterns in VTC. Isolating the proper cognitive processing components of visual recognition and mapping these components to functional neuroanatomy is invaluable for understanding the nature of visual recognition deficits following brain damage, and for leading to potential therapies for rehabilitation and recovery of function.