Investigating the relationship between visual object category selectivity measured with functional neuroimaging and electrocorticography in the human ventral temporal cortex

Abstract

Functional neuroimaging (fMRI) is commonly used to investigate the organization of human visual cortex. Previous reports described category-selective (e.g. faces, bodyparts, houses) fMRI activations in consistent locations of the occipitotemporal cortex. Similar selectivity has also been reported using electrocorticography (ECoG). However, the precise relationship between the signals measured with fMRI and ECoG is not well understood. To address this issue, we measured category selectivity to faces, bodyparts, cars and houses using fMRI and ECoG in seven patients undergoing epilepsy surgery evaluation. ECoG signals from electrodes located on ventral temporal cortex were analyzed both by computing event-related potentials (ERP) and broadband power in the gamma range (70-200 Hz). The latter is correlated with mean population firing rate (Manning et. al. 2009) and linked with fMRI signals (Mukamel et al., 2005). Our analyses indicate that (1) category selectivity exhibited in ERP and gamma signals largely overlapped although not completely, suggesting these signals reflect partially distinct neural processes, (2) ECoG selectivity for faces was usually stronger than that measured with fMRI and (3) ECoG bodypart selectivity often spatially overlaps with face selectivity, with the former delayed by 30-50ms relative to the latter. Preliminary co-localization analyses reveal an overlap between the selectivity measured in fMRI and ECoG to faces, bodyparts and houses at the single subject level. Specifically, category-selective ECoG responses were often recorded at electrodes overlapping or near fMRI activations of the same selectivity. Co-localization was best for faces, probably due to the location of face activations on a gyrus (fusiform), relative to activations for houses and bodyparts that are located in a sulcus (collateral and occipito-temporal, respectively), and consequently further away from electrodes recording ECoG signals. Our results thus indicate an overall tight coupling between category selectivity measured with fMRI and electrophysiology, within the same subject in human occipitotemporal cortex.

Meeting abstract presented at VSS 2012