Neural Correlates of Learning From Social Engagement in Children with ASD

Background:  Children with autism spectrum disorder (ASD) show less social engagement and, therefore, are less likely to learn from social cues.  Social learning represents a challenging cognitive domain to characterize given the somewhat restricted behavioral repertoire of children with ASD. One approach to address this challenge is to inform behavior with quantifiable biomarkers (Gandal, 2010). Electroencephalography (EEG) and event related potentials (ERPs) hold particular promise in this population because they are non-invasive and more feasible than neuroimaging in challenging populations, and they also provide time sensitive information about brain processing (Jeste, 2009). No prior studies have investigated biomarkers for social engagement in children with autism using EEG.

Objectives:  To develop an ERP paradigm that captures social learning, with the goal of differentiating social engagement in children with ASD from typically developing children.

Methods:  The paradigm consists of a behavioral exposure phase and an ERP test phase. The examiner presents children with 4 visually-matched pairs of toys. Half of the toys are presented in a “non-social” setting, with the examiner wearing a neutral expression and refraining from engaging with toy or child, while the other half of the toys are presented in a “social” setting, with the examiner engaging with the child around the toy. The behavioral component is videotaped and later coded by two independent and reliable raters for variables including: eye contact with examiner, child looking time to toy, spontaneous vocalizations, and degree of social engagement during each trial. The ERP portion immediately follows the exposure phase. Children view a continuous sequence of photographs of the toys in random order. The component of interest is the frontocentral Nc and, specifically, the differentiation of the social and non-social conditions based on Nc mean amplitude.

Results:  Four 3-5 year old children with ASD and six typically developing, age matched, children have completed the paradigm thus far. There are significant group differences in the level of social engagement during the behavioral exposure phase, with typically developing children demonstrating more social engagement (t=3.98, p=.004). Preliminary ERP data show that typically developing children differentiate the social and non-social stimuli more than the ASD kids, as quantified by the difference in Nc mean amplitude (4.44 for the typical group, 2.68 for the autism group). Furthermore, mean Nc amplitude shows moderate correlations with behavioral data.

Conclusions:  This is the one of the first studies to use EEG as a biomarker for learning from social engagement. Our preliminary data show that neural correlates of learning from social engagement are quantifiable using EEG. Behaviorally, preschoolers with ASD demonstrate less social engagement, and less differentiation between social conditions in the EEG. Future work will include using this paradigm to better characterize individual variability in social engagement in children with ASD and to investigate this variability as predictor of treatment response.