How do different visual cortical regions mediate human object recognition? Functional MRI (fMRI) can shed light on this question, measuring neural activity in awake, behaving humans, but interpreting these signals poses several challenges. fMRI responses may average across different neural populations, and may reflect several distinct aspects of visual stimuli and recognition behavior. This dissertation attempts to distinguish between the multiple factors which contribute to fMRI signals in human extrastriate visual cortex. First, I examined functional MR adaptation in object-selective cortex (fMR-A), which has been used to probe whether neural populations within a voxel are homogenous or heterogeneous in their tuning. I found that fMR-A reflects independent components of performance improvement and stimulus repetition, with a significant effect of repetition even when performance variation is controlled for. Second, I examined the effects of object identity on response patterns at high resolution (1mm3). I found that object identity can be readily inferred, using even a small number of voxels; and describe two metrics to rate the informativeness of voxels. Finally, I examined the effects of object category and retinotopic position on visual cortex. I found a strong sensitivity to retinotopic position across all visual cortical regions examined, even relatively high-order ones like the fusiform face area (FFA). fMRI responses in the FFA show an interaction between category and position, with a higher retinotopic sensitivity for faces compared with other categories, although behavioral data do not show a higher sensitivity for face identification. These results demonstrate that fMRI responses reflect multiple aspects of both stimulus properties and subject behavior, and proper interpretation of these signals requires an understanding of the multiple factors involved.
展开▼