Visual cortex consists of low-level regions containing systematic representations of the visual field (e.g. retinotopic maps), as well as high-level regions specialized for processing complex shapes such as faces and limbs (e.g. the fusiform face area and extrastriate body area, respectively). Prevailing views adopt dichotomous organization principles for low- and high-level functional regions and associated neural computations, where the former are thought to be systematically organized and the latter less systematic and highly variable - both anatomically, as well as relative to other high-level visual areas. Using higher-resolution (1.5mm voxels) fMRI scanning methods than past studies (3-5mm voxels), we recently conducted a series of experiments (Weiner & Grill-Spector, 2010, 2011a,b) revealing that the spatial location of high-level visual regions selective for faces and limbs are much more consistent than once thought. Furthermore, these experiments reveal a topographic organization of face- and limb-selective regions extending from lateral occipitotemporal to ventral temporal cortex where each high-level region is defined by a combination of anatomical and functional boundaries separating them from neighboring regions. The anatomical and functional boundaries of this high-level topographic map are reliable across subjects and longitudinally within-subjects over a span of three years. Here, we propose a multi-factor parcellation framework resulting from our empirical measurements using the following criteria: 1) precise anatomical location of functional regions, 2) preserved spatial relationship among functional regions, 3) preserved relationship relative to known visual field maps, and 4) reliable functional differences among regions. The implementation of this framework allows the first consistent parcellation of high-level visual regions outside of visual field maps. Moreover, it illustrates that defining brain areas from one measurement of category selectivity is insufficient. Though we use high-level visual cortex as a model system, this framework can also be applied to other sensory and potentially nonsensory cortical systems.
Meeting abstract presented at VSS 2012