Abstract
Over the past years, fast periodic visual stimulation (FPVS) has been used extensively as an objective measure of neural face discrimination. Xu et al. (2017) reported that the pattern of individual differences of this measure was observed consistently across four trials assessed in the same testing session. Independently, we reported that the amplitude and topography of this neural response is differentially modulated across observers as a function of the foveally presented facial information (i.e., viewing position; VP) (Stacchi et al., 2017). Nevertheless, whether the reliability of neural face discrimination responses varies across VPs between and within subjects has yet to be clarified. We recorded high-density electrophysiological signals in 14 subjects during FPVS with a test-retest design (6-month inter-session interval). On each trial, fixations were enforced on the center of the screen and faces were realigned to one of 10 VPs, covering all inner facial features. The FPVS face discrimination response reliability was computed for 44 posterior electrodes using Cronbach's alpha on the group amplitude and individual topographies. At the group level, reliability of the FPVS response amplitude aggregated across posterior electrodes was generally stable for all VPs. Interestingly, intra- and inter-subject differences emerged for the reliability of the topographies, which varied at the individual level. While some subjects showed highly reliable FPVS face discrimination responses across all VPs, others were extremely unstable; the majority showed reliable responses only for specific VPs. Crucially, the reliability of the FPVS face discrimination response topography was directly correlated with the overall amplitude. Our findings invite to careful consideration of individual differences according to their reliability, as those that are expressed more consistently could be functionally more meaningful. Moreover, our observations suggest that the magnitude of the neural face discrimination responses is an effective indicator of response reliability.
Meeting abstract presented at VSS 2018