Abstract
Visual representations of faces vary greatly across individuals, and have been linked to psychopathology (Faghel-Soubeyrand et al., 2020) and recognition ability (Tardif et al., 2019). Classification image (CI) techniques such as Bubbles offer means to measure those representations, but require long and arduous testing. Frequency-tagging has been shown to effectively measure differences in the processing of visual stimuli within seconds of EEG recording, and has been applied to study face processing (Norcia et al., 2015). Here, we evaluated whether EEG frequency-tagging could be a faster and reliable alternative to CI techniques to measure individual use of information on faces. To do so, 20 participants were asked to discriminate the gender and emotion of faces in both a Bubbles paradigm in which the spatial information available was randomly sampled (1000 trials per task, ~1h), and an EEG frequency-tagging paradigm (24x70 seconds per task) in which task relevant facial features flickered at specific frequencies (e.g. left eye: 6.31 Hz, right eye: 3.52 Hz, mouth: 8.00 Hz). We found robust oscillatory signals in occipito-temporal electrodes for all frequency-tagged features, with signal-to-noise ratio reaching >16 in individual participants (p<.0001). This EEG frequency-tagged activity was modulated by tasks (F(1,294)=6.23, p=.0131), matching with findings from our own Bubbles group-results as well as with previous studies (e.g. Faghel-Soubeyrand et al., 2019; Blais et al., 2008), and underlining the relevance of this signal to measure representation-specific visual information. Crucially, a participant’s reliance on the facial features (the average z-scores in the Bubbles’ CIs within the regions of interest) was predictive of his/her EEG signal pertaining to the brain processing of those features in the frequency-tagging paradigm (r=.27, p=.009). Overall, these findings present EEG frequency-tagging not only as a mean to measure task-specific visual representations, but also as a promising method to assess individual differences in face representations.