Abstract

Background

Previous research reports that increased attention towards social stimuli (i.e., Social Attention, SA) in autistic females may modulate exposure to social contingencies, thereby reducing social interaction difficulties and symptom severity. Therefore, heightened SA could relate to specific manifestations of Autism in females, in relation to symptom intensity and socio-cognitive skills.

Objectives

This research aims to investigate the second-by-second changes of SA in autistic compared to neurotypical individuals and their relation to symptomatology and sex, as measured by eye-tracking during a dynamic video rich in social information.

Methods

388 autistic (M:F ratio=2.68; mean age=16.7, SD=5.8; mean IQ=96.8, SD=19.9) and 271 neurotypical participants (M:F ratio=1.9; mean age=17.2, SD=5.9; mean IQ=103.8, SD=18.9) took part in the study. Participants’ eye-movements were recorded via eye-tracking while watching medium close-up shots of 1-2 people talking to the camera. Each gaze sample was scored 0/1 if it fell within the area of the face, and averaged every 0.5-second since the onset of an individual face, within a 3-seconds time-window. The obtained Proportional Looking Time (PLT) was used in 3 separate Growth-Curve Mixed Models, with 2ndorder polynomials as predictors to capture PLT second-by-second change, and each including the neurotypical, autistic and female participants separately. We report the unstandardised effect sizes (ꞵ) of sex/diagnosis on average and second-by-second change of PLT. Further, we extracted individual ꞵ from the models, and tested the partial correlations with autism symptom intensity (SRS-2), and, exploratively, neural event-related responses during a face processing task, measured by the Global Field Power (GFP) of brain microstates, representing quasi-stable topographies switching across time.

Results

Average PLT was higher in females than males, both in neurotypical (Model 1; ꞵ=0.06, SE=0.01, p-value<0.01) and autistic group (Model 2; ꞵ=0.04, SE.=0.02, p-value=0.02; Fig. 1A). The second-by-second change of PLT configured a parabolic shape, with final net increase, that was sharper in autistic females than autistic males (ꞵ=-0.05, SE.=0.02, p-value=0.002). Average PLT was lower in autistic females compared to neurotypical females (Model 3; ꞵ=-0.09, SE=0.02, p-value<0.01; Fig. 1B). Despite no significant association with symptom intensity, we found a negative correlation with GFP of the early microstate, in autistic (N=46; Intercept: rho=-0.30, p-value=0.04; Slope: -0.35, p-value=0.02) and neurotypical females (N=47; Intercept: rho=-0.33, p-value=0.03; Slope: -0.31, p-value=0.03; Fig. 1C), but not in males.

Conclusions

This task was able to elicit robust sex differences; autistic females showed lower SA compared to neurotypical females, possibly suggesting concurring adaptation/compensation. Further, higher interest in faces associated with smaller GFP in females only, suggesting that the process at work may influence face-processing abilities.