Abstract
The behavioral face inversion effect (Behavioral-FIE) is one of the most established markers for specialized face processing. Functional MRI studies testing for a correspondingly reduced signal for inverted compared to upright faces in face selective fusiform areas (FFA-FIE) have reported mixed results. Two studies (Haxby et al., 1999; Aguirre et al., 1999) reported no face inversion effect in the FFA, whereas one study found a significant but a small effect (Kanwisher et al., 1998). Importantly, the studies that found no FFA-FIE did not obtain or did not test for a Behavioral-FIE in accuracy measures. Here we report two experiments that yielded a significant Behavioral-FIE and a reduced signal for inverted relative to upright faces in the FFA (FFA-FIE). Further, this FFA-FIE was correlated with the Behavioral-FIE across subjects in both experiments. In Experiment 1, 14 subjects performed a same-different matching task on upright and inverted faces presented in separate blocks. Our data show a significant reduced signal to inverted than upright faces and a positive correlation between the FFA-FIE and the Behavioral-FIE (r(12)=.56). To examine whether this effect is specific for faces, in Experiment 2, 15 subjects matched upright or inverted faces and chairs in an event-related design. The fMRI response in the FFA and accuracy were higher for upright than inverted stimuli for both faces and chairs. However, the FFA-FIE was correlated with the Behavioral-FIE (r(13)=.45), whereas the behavioral chair inversion effect was not correlated with the fMRI chair inversion effect (r(13)=.16). Our findings suggest a close link between the behavioral and FFA inversion effects for face stimuli only.