Abstract
Purpose: Human adaptation and monkey electrophysiology studies have demonstrated that face sensitive neurons are selective for emotion, gender and ethnicity. But it is not yet clear to what extent face sensitive neurons are singly tuned (e.g. selective for gender, unselective for ethnicity) or jointly tuned (i.e. selective for both ethnicity and gender). We examined the selectivity of face sensitive neurons in humans using an fMRI adaptation paradigm.
Methods: In each scan, subjects were first pre-adapted for 3 minutes to a series of [a] female-Caucasian faces (1 sec/image), [b] male-Asian faces, or [c] female-Caucasian alternating over time with male-Asian faces. This was immediately followed by an uneven block design containing 6×32 sec cycles in which faces from the adapting image set were presented for 24 secs and faces from a non-adapted image set were presented for 8 secs. The non-adapted images differed from the adapted images in either gender, or ethnicity or both. A control condition tested whether subjects were simply adapting to individual faces: subjects adapted to a random set of faces and release from adaptation was measured for a random set of different faces. 4 subjects participated in 8 sessions (7–8 scans/session). Results: If a significant proportion of neurons were selective for both ethnicity and gender, we would expect a complete release of adaptation for non-adapted faces. We did find adaptation effects when non-adapted faces differed from the adapting faces in either race or gender. This was true even under a contingent adaptation paradigm, in which singly tuned neurons would not be expected to show release from adaptation. These adaptation effects occurred selectively in the fusiform gyrus in areas overlapping with face selective areas defined using typical fusiform face area reference scans. Subjects did not show adaptation effects in the control condition or in V1.
Acknowledgments: Supported by NIH EY12925