Abstract

Title: Decoding object categories from fMRI data in parietal cortex
Object categorization is one of the key functions of the visual system. According to the traditional views, object category information is mainly processed in the occipito-temporal cortex whereas parietal cortex mainly represents object location information. This view however, has been recently challenged by both monkey physiological and human functional magnetic resonance imaging (fMRI) studies indicating the direct involvement of the parietal cortex in object representation. In a series of experiment using fMRI, we investigated the contribution of the human parietal cortex to various types of object categorization. We used multi-voxel pattern analysis to decode object categories in sub-regions of parietal cortex. Decoding accuracy of a support vector machine classifier was measured in two types of object categorization tasks: “perceptual categorization” that involves grouping objects based on their visual similarities and “conceptual categorization” that requires classifying objects based on their conceptual properties and independent of their visual features. We were able to decode both the perceptual and the conceptual categories of objects in a sub-region of parietal cortex located in the posterior parietal cortex. These results support a more integrated view of the visual processing in which both parietal and occipito-temporal areas are involved in processing object category information.