Purchase this article with an account.
Chris Baker, Mark Williams, Leila Reddy, Lawrence Wald, Graham Wiggins, Brad Dickerson, Christina Triantafyllou, Nancy Kanwisher; Fine-grained analysis of functional selectivity in human occipitotemporal cortex. Journal of Vision 2007;7(9):1029. doi: 10.1167/7.9.1029.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
FMRI investigations of extrastriate cortex have reported regions responding selectively to faces (FFA), scenes (PPA), and body-parts (EBA, FBA), but not other visual object categories (Downing et al., 2006). However, individual voxels in these studies typically contain millions of neurons and partial voluming could obscure fine-grained patterns of neuronal selectivity. Here we used a custom-built 32-channel phased-array coil to investigate selectivity across occipitotemporal cortex at 1mm isotropic resolution. Four subjects viewed faces, body-parts, cars, chairs and vases in a blocked design. Split-half analyses of unsmoothed data i) identified voxels that responded significantly more to a given category than to the average of the other four (p [[lt]] 0.001), and ii) tested whether the selectivity of these voxels replicated in the other half of the data. An average of 237 face and 663 body-part selective voxels per subject were found. These voxels included clusters corresponding to the FFA, EBA and FBA. Critically, face and body-part selective voxels replicated their strong selectivity in an independent test in the other half of the data. In contrast, an average of only 40 car and 65 vase selective voxels were found; the selectivity in these voxels was not replicated in the other half of the data. For chairs, an average of 226 voxels were identified, mostly in the region of posterior LOC; some weak selectivity was preserved in the other half of the data. Thus despite clear replication of face and body-part selectivity, we find few voxels strongly and reliably selective for other visual categories even at 1mm isotropic resolution. While increased resolution may yet reveal selectivity for categories other than faces and body-parts, our findings suggest that at least for cars, chairs and vases, any such selectivity is likely to be weaker than selectivity for faces and body-parts (in terms of clustering or magnitude).
This PDF is available to Subscribers Only