Figure 5.

Reproducibility indices (means and standard errors across all subjects). First row: the reproducibility curves with respect to the numbers of selected voxels for the AV‐fami‐cong‐unfami, AV‐fami‐incong‐unfami, V‐fami‐unfami, and A‐fami‐unfami runs. Second row: the reproducibility results obtained with 1,600 selected voxels. The asterisks indicate significant differences (P < 0.005, corrected). Left: familiar stimuli (facial images in the V‐fami‐unfami, names in the A‐fami‐unfami, congruent face‐name pairs in the AV‐fami‐cong‐unfami, and incongruent face‐name pairs in the AV‐fami‐incong‐unfami runs). Right: unfamiliar stimuli (facial images in the V‐fami‐unfami, names in the A‐fami‐unfami, face‐name pairs in the AV‐fami‐cong‐unfami, and face‐name pairs in the AV‐fami‐incong‐unfami runs). We used as a reproducibility index to assess the similarity of two fMRI activity patterns elicited by the stimuli, where $\theta$ is the angle between the two pattern vectors. [Color figure can be viewed at http://wileyonlinelibrary.com]