Abstract
Visual information is represented in multiple areas in the human brain. The primary visual cortex (V1) is typically associated with representing low-level image properties of the visual stimuli, whereas higher-level areas encode more abstract information, such as object category. What is not well understood is how the visual information is represented in the intermediate-level visual areas, such as visual areas V2 and V4. Here we used representational similarity analysis (RSA; Kriegeskorte et al. 2008) to characterize population-code representations of natural images across hierarchy of visual areas (for details on the fMRI data, see Kay et al. 2008; Naselaris et al. 2009). We found a gradual change in the representational similarity structure across the visual areas V1, V2, V3, V4 and LO (lateral occipital area). The representations in V3A and V3B were most similar to that in V3 and were more distinct from the representations in V4 and LO. We aim to characterize the visual features that drive these differences in the representations across the hierarchy of visual areas and to relate the results to computational models of visual processing.
Meeting abstract presented at VSS 2013