Removing the right inferior occipital gyrus does not disrupt face-selective responses in human ventral temporal cortex: Evidence against a strict hierarchical model of face perception

Abstract

Neurofunctional models of face perception consider the "occipital face area" (IOG-faces/OFA) the input node to a hierarchy of face processing regions. It is presently unknown how removing this node affects downstream face-selective regions and the functional organization of human ventral temporal cortex (VTC) more generally. Here, we report a series of investigations using functional magnetic resonance imaging (fMRI) and intracerebral recordings with depth electrodes (sEEG) in a rare patient with intractable epilepsy. Behaviorally, the patient performed well above chance level at face perception and recognition tests although she performed significantly lower than normal controls. Pre-resection, a block design fMRI experiment using images of faces, body parts, places, and objects, showed a typical topology of functional regions in VTC (Figure 1). SEEG recordings revealed the highest face-selective response in the high gamma frequency range from 150 ms post-stimulus onset in an electrode located in the posterior fusiform face area (pFus-faces/FFA-1), indicative of normal latency of face selectivity. The resection included all of IOG-faces/OFA and the posterior aspect of pFus-faces/FFA-1. However, the right calcarine sulcus was intact and the patient did not suffer from a left visual field hemianopsia. fMRI conducted a month post-resection revealed that the topology and selectivity of face-, body part-, and place-selective regions anterior to the resection were preserved. Quantifying the topology of face-selective responses with multivoxel pattern analyses in VTC revealed that the correlation between pre- and post-resection scanning sessions was highly significant (r=.62±.04; p<10-3) and comparable to the correlation between two pre-resection scanning sessions (r=.51±.05; p<10-3), indicating the stability of face-selective responses post-resection. Interestingly, the patient's face perception and recognition remained stable after resection while her response times decreased two-fold. Altogether, these observations pose important constraints on the hierarchical neurofunctional model of face-selective responses in the human brain.

Meeting abstract presented at VSS 2014