After preprocessing steps (see the online data supplement), linear contrasts using canonical hemodynamic response functions estimated a faces (match faces task) > shapes (match shapes task) contrast image for each individual. These contrast images were entered into a second-level random-effects model (one-sample t test) to determine mean task-related responses using a combined voxel-level threshold of p<0.05, family-wise error-corrected whole-brain, and cluster threshold of ≥10 contiguous voxels. BOLD contrast estimates were extracted from functional clusters meeting significance criteria in the right and left anatomical amygdala regions of interest, which were defined using the Brodmann template in the Wake Forest University PickAtlas. By extracting amygdala BOLD parameter estimates from the functional clusters activated by our paradigm rather than clusters specifically correlated with our independent variables of interest, we precluded the possibility of any regression coefficient inflation that may have resulted from capitalizing on the same data twice (33). We have successfully used this conservative strategy in previous studies (27, 34).
