Abstract
The N170 event-related potential, which is preferentially tuned to faces (see for review Rossion, 2014), has been linked with processing of the eyes (Rousselet, Ince, van Rijsbergen & Schyns, 2014), of diagnostic facial features of emotions (Schyns, Petro & Smith, 2007), and of horizontal facial information (Jacques, Schiltz & Goffaux, 2014). Recent findings have shown that horizontal information is highly diagnostic of the basic facial expressions, and this link is best predicted by utilization of the eyes (Duncan et al., 2017). Given these findings, we were interested in how N170 amplitude relates with spatial orientations in a facial expressions categorization task. Five subjects each completed 7,000 trials (1,000 per expression) while EEG activity was measured at a 256 Hz sampling rate. Faces were randomly filtered with orientation bubbles (Duncan et al., 2017) and presented on screen for 150ms. Performance was maintained at 57.14%, using QUEST (Watson & Pelli, 1983) to modulate stimulus contrast. The signal was referenced to the mastoid electrodes and bandpass filtered (1-30 Hz). It was epoched between -300 and 700 ms relative to stimulus onset, and eye movements were removed using ICA. Single-trial spherical spline current source density (CSD) was computed using the CSD toolbox (Kayser & Tenke, 2006; Tenke & Kayser, 2012). Our main analysis consisted in conducting a multiple linear regression of single-trial orientation filters on PO8 voltages at each time point. The statistical threshold (Zcrit= 3.6, p< .05, two-tailed) was established with the Stat4CI toolbox (Chauvin et al., 2005). We found a negative correlation between horizontal information availability and voltage (Zmin= -5.43, p< .05) in the 50ms leading up to the N170's peak. Consistent with the proposition that the N170 component reflects the integration of diagnostic information (Schyns, Petro & Smith, 2007), the association between horizontal information and amplitude was strongest 25 ms before the peak, and completely disappeared at peak.
Meeting abstract presented at VSS 2018