Abstract
Human faces are readily and automatically categorized for gender across a wide range of variable cues, a critical visual function for social interactions. To identify an implicit measure of rapid face gender categorization, we recorded scalp electroencephalogram (EEG) from 32 participants (16 females). In a first experiment, highly variable face images from one gender alternated at a rapid rate of 6 Hz (i.e., 6 images per second) with images of the other gender inserted every 6th stimuli, objectively isolating a gender categorization response at a 1 Hz rate in the EEG spectrum. In a second experiment, images from only one gender (i.e., male or female face images) were inserted at the 1 Hz categorization rate in a 6 Hz sequence of non-face objects. In the first experiment, a significant categorization response was identified for both face genders over the right occipito-temporal cortex, but the response was larger for female faces presented among males than the reverse. This asymmetrical pattern suggests either greater generalization across female than male exemplars, or a more inclusive female category. Results from the second experiment provide an answer: a larger generic face categorization response is recorded for male faces, indicating higher generalizability across male than female faces, and thus supporting the second interpretation. Importantly, these effects disappear for upside-down faces, ruling out any contribution of low-level physical variability across images. Moreover, no own-gender bias was found. Altogether, these findings reveal that rapid visual gender categorization from natural face images can be objectively isolated and quantified in the human brain in a few minutes of recording. They also suggest that male faces are highly generalizable within a well-defined category that excludes female faces, while female face category boundaries are less demarcated, female exemplars sharing some male characteristics. Keywords: face gender, categorization, EEG, fast periodic visual stimulation, frequency-tagging
Meeting abstract presented at VSS 2018