Direct gaze facilitates rapid orienting to faces: Evidence from express saccades and saccadic potentials

Highlights • We investigated the role of direct gaze for rapid orienting to faces.• Faster express saccades towards faces occurred only when faces were with direct gaze.• Saccade-locked ERPs discriminated between gaze directions from the saccade onset.


Data analysis method of target locked ERPs
Target locked ERPs were analysed in a similar way as saccade locked ERPs for the key components relevant to the main findings. The epochs used for this analysis were centred around stimulus onset (-100 ms to 500 ms). For comparability, the trials used for this analysis were the same as those used for the analysis of all saccades, which excluded trials with saccades beginning before 80 ms and after 500 ms. Incorrect trials or trials which included a switch of gaze direction occurring before 500 ms, were also excluded from the subsequent ERP analysis due to the associated double saccade response. As with the main analysis global field power (GFP, Skrandies, 1990) was used to identify the time windows to analyse the selected components, P100 and N170 ( Figure S1). The beginning and end of each time window was set around half the maximum GFP value of each component.
Consequently, P100 and N170 mean amplitudes were respectively analysed between 99 and 130 ms and 172 and 205 ms. All components were analysed in the same parieto-occipital regions as the main analysis (P7/P8, P5/P6, PO7/PO8) to allow for comparison ( Figure S2 for topographical maps).
Statistical analyses of the mean amplitudes in each time window were carried out using a three-way ANOVA with hemisphere (right and left), visual hemifield of target (right and left visual presentation) and condition (buildings, averted and direct gaze) as within-participant factors. When appropriate, post hoc planned comparisons were performed using two-tailed paired t-tests. Violations of sphericity were corrected with the Greenhouse-Geisser correction.

P100
A three-way interaction between hemisphere, hemifield and condition was significant (F(2,28)= 9.13, p=.001), which was driven by an interaction between hemifield and condition on the left hemisphere (F(2,28)= 8.78, p=.001, see Figure S3 for ERP waveforms). Differences between conditions occurred only in the contralateral hemifield (right hemifield; F(2,28)= 6.58, p=.005), where faces with averted gaze had a smaller (M=1.60) amplitude than faces GFP ms Supplementary Material with direct gaze (M=2.55; p=.009) or buildings (M=2.76; p=.010). This difference was unexpected, and different from the combination of hemifield and hemisphere which yielded the main results.

N170
Main effects of hemifield (F(1,14)= 4.75, p=.047) and condition (F(1.11,28)= 55.33, p<.001) were significant, which was driven by larger amplitudes in the right hemifield (M=-1.92) than the left (M=-1.20; p=.047), and to faces with direct (M=-2.84, p<.001) and averted gaze (M=-2.61, p<.001) than buildings (M=-.78). No difference was significant between gaze directions (p=.102). A three way interaction between hemisphere, hemifield and condition was not significant either (F(1.27,17.79)=.72, p=.439). This overall face specific effect, which occurred independently of hemisphere and hemifield, could partially account for the differences observed in saccade-locked components. Nonetheless in express saccades where the spike potential precedes the N170, a significant difference between faces with direct gaze and buildings was still observed, showing that these effects cannot be fully justified by target locked differences.

Right hemifield presentation Presaccadic positivity (PSP).
Differences in PSP between conditions were significant for right hemifield presentation on the left hemisphere (F(2, 28) = 7.61, p=.002) but not on the right (F(2, 28) = 1.26, p=.299). A significantly larger PSP amplitude was found for