Purchase this article with an account.
Talia L Retter, Caroline Michel, Fang Jiang, Michael A Webster, Bruno Rossion; The speed of individual face recognition. Journal of Vision 2019;19(10):229c. doi: https://doi.org/10.1167/19.10.229c.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Neurotypical human adults can often recognize the identity of a face at a glance. Yet, the minimal and optimal presentation duration at which individual faces can be discriminated from one another, beyond low-level image changes, remains largely unknown. Here, we used a frequency-tagging sweep design with increasing presentation duration to examine firstly the duration at which a face individuation response arises. Responses were recorded with 128-channel EEG from 16 participants with ascending 77-s sequences of 11 presentation durations from 25 to 333 ms (40 to 3 Hz), throughout which the same unfamiliar face was repeated with changes in size and luminance at every presentation; a different unfamiliar facial identity was presented within this sequence every 1 s (1 Hz). In a second behavioral experiment with the same participants, we presented identity changes non-periodically within fixed-rate 30-s sequences while participants performed an explicit individual face discrimination task. A neural individual face recognition response, tagged at 1 Hz and its specific harmonics, emerged over the occipito-temporal cortex at the 50-ms presentation time (25- to 100-ms across individuals), with a maximal response at about 170 ms presentation time. This corresponds to a delay of approximately 20 ms relative to the minimum, and 80 ms relative to the maximum, responses to generic face vs. non-face categorization, measured previously with a similar sweep design. Importantly, behavioral accuracy correlated with individual participants’ weighted neural response amplitude only in a mid-frequency presentation range. These results present a step towards quantifying performance of individual as opposed to generic face categorization in the human brain, and extend the finding that individuals’ weighted EEG amplitudes at mid-frequency ranges predicts behavioral performance.
This PDF is available to Subscribers Only