Functional selectivity of human extrastriate visual cortex at high resolution

Abstract

Previous functional imaging studies have isolated a number of regions within human extrastriate visual cortex that appear to be specialized for the processing of different visual stimuli, including the FFA (faces), PPA (scenes), LOC (objects), and EBA (body parts). However, it is unclear whether these regions reflect the processing of specific types of visual stimuli in separate and discrete regions of cortex, or whether they are simply part of a more distributed representation. To address this issue, we investigated the functional selectivity within human extrastriate visual cortex at high resolution using an 8-channel phased-array surface coil. Functional images were collected with 1.4 × 1.4 mm within-plane resolution, and 2 mm slice thickness, giving a 10-fold increase in spatial resolution compared with many previous fMRI studies. Comparisons of the BOLD response to visual stimuli from different visual categories revealed patchy distributions of selective voxels, clustered in distinct regions corresponding to previously described areas. In some of these selective regions, comparison of the response to different visual stimuli showed that the activation was often highly correlated across visual categories. Furthermore much of the activation in these regions coincided with the location of large veins, visible both in high-resolution anatomical scans (0.4 × 0.4 mm in-plane resolution) and in the functional images themselves, which may be pooling blood from a large area of cortex. In other regions, however, there was little correlation across different stimulus categories and these activations did not overlay large veins. These regions may correspond to the cortical focus for selectivity. Importantly, the selectivity in these regions was often very high with very little response, if any, to non-preferred categories. These findings suggest that in local regions of cortex there is very strong and focal selectivity arguing against a more distributed representation.