Protocol and behavioral data for the fMRI task.

<p>(A) In each trial, subjects saw a male of female target face preceded by a short or long series of noise masks (note that faces used were from the Karolinska database and were equated for luminance, they were not the example faces used in the figure). Subjects were asked to make a fast gender decision on the final target face. At the start of each trial, a word cue (the word <i>short</i> or <i>long</i>) indicated with 75% validity the time at which the target face would appear. 100 ms before the target face appeared, we presented a fearful or a neutral subliminal face (prime) or a noise mask (baseline) for 33 ms, which was both forward and backward masked by the noise masks. Gender of the subliminal face was either congruent or incongruent with the target face, while its expression was either fearful or neutral (orthogonally to other factors). (B) Response times (RTs) show that attention and gender congruence speeded correct responses. This effect did not occur on the unexpected incongruent trials (*p < .05). (C) Mean performance during target-gender decision demonstrated that more errors were made when prime gender was incongruent with target gender in expected trials (**p < .01). We have received informed consent according to Plos guidelines from the 4 individuals portrayed here.