Abstract

Orienting to others is crucial for learning about the world. Infants show a marked bias for attending to faces, however, we do not know much about the online changes of their attention to faces during the first 3 years. The definition of temporal profiles of attention might help understanding how infants orient to faces at different stages of development.

The current study aimed to characterise the distribution of attention over time longitudinally, in a large cohort of infants from 5 to 36 months. We recorded the eye-movements of 163 infants (70 females), of which 100 had an older sibling with Autism Spectrum Disorder (ASD+), at 5, 10, 14, 24 and 36 months with a TX300 eye-tracker. The infants watched 8 5-seconds arrays of 1 photograph of a face plus 4 distractors. We excluded trials with missing data >25% (1 participant excluded). We aggregated the data in time bins of 500ms and calculated the proportional looking time (PLT) to the face. We applied Growth Curve Analysis with a multi-level modelling approach, including 3rd degree polynomials, random intercept per visual array and random intercept and slope per participant.

The average PLT was above chance (Intercept = 46.10, CI 95% = 42.29 – 49.92), but tended to decrease over the course of the trial (Slope = -16.21, CI 95% = -20.22 – -12.20) and with age (Age = -0.51, CI 95% = -0.55 – -0.47). Age markedly influenced the change of PLT over time by means of the quadratic and cubic component (quadratic * age = 0.26, CI 95% = 0.11 – 0.42, cubic * age = -0.38, CI 95% = -0.54 – -0.23). Gender modulated the effect of age, with males showing a more pronounced average decrease with age (Age * Gender = -0.21, CI 95% = -0.29 – -0.13). Infants with a sibling with ASD had a more prominent average decrease with age (Age * ASD+ = -0.15, CI 95% = -0.26 – -0.03), but smaller gender differences (Age * Gender * ASD+ = 0.28, CI 95% = 0.12 – 0.43). PLT changed from a flat slope to a curvilinear, asymmetrical profile in increasingly mature infants (Figure 1).

Younger infants were steadily captured, and slowly and linearly disengaged from the face. Older infants instead rapidly disengaged and bounced back by the end of the trial – possibly indicating enhanced social interest and oculomotor control. Furthermore, by 36 months, male infants had significantly dropped their attention level – possibly a sign of emerging gender-conforming looking preferences due to environmental social incentives. The general transition was also evident in the group with an older sibling with ASD, however, these infants showed less interest for the face and a smaller differentiation by gender (Figure 2).

This distinction was marked by male infants showing relatively higher PLT compared to peers – supporting the idea that infants with on the autism spectrum may adjust less to the social environment, which in turn may lead to milder differentiation by gender. These results highlight a developmental transition modulated by gender and familiar likelihood of ASD, with the potential of obtaining individual trajectories of face-orienting tendencies.