Explaining Variance in Social Competence in School-Aged Children with Autism Spectrum Disorder

Background: Social competence likely depends on multiple social-cognitive and social-affective processes, yet few studies have assessed the relative importance of a range of constructs across these domains, and fewer still have done so in children with autism spectrum disorder (ASD). Deficits in theory of mind (ToM) are often hypothesized to substantially impair social functioning, yet evidence for this link is mixed. Empathy is also considered essential for social competence, and an important distinction may exist between cognitive empathy (recognizing or understanding others’ emotions) and affective empathy (experiencing others’ emotions in oneself).

Objectives: The present study is the first to our knowledge to directly compare relations between social competence and a range of social factors potentially affected in ASD: ToM, empathy, biological motion perception (BMP), social reward, and social anxiety.

Methods: In school-aged children with ASD (n=51, mean age: 11.48 ± 2.11 years, range: 7.11-14.87), we administered standard ToM tasks (Strange Stories and Reading the Mind in the Eyes), a BMP task featuring walking point-light figures embedded in noise, and the Social Reward Questionnaire (SRQ), from which we selected two subscales for the current analysis: Admiration and Prosocial Interactions. Parents reported on their children’s ToM abilities (Theory of Mind Inventory (ToMI): Early, Basic, and Advanced subscales), empathy (Griffith Empathy Measure: cognitive and affective subscales), and social anxiety (Screen for Child Anxiety Related Disorders). Social competence was assessed using the Social Affect domain of the Autism Diagnostic Observation Schedule (ADOS).

Due to high collinearity between the ToMI subscales and between cognitive empathy and ToMI Early, we created a composite of these measures representing parent-reported ToM. Similarly, we created a composite of the two highly correlated SRQ subscales. Predictors also included verbal IQ, nonverbal IQ, and age.

To determine the subset of variables that best explain social competence, we performed Bayesian model selection, which quantified evidence (Bayes Factors; BF) for adding each predictor to the null (intercept-only) model. Then, to determine the relative importance of each predictor, we performed a dominance analysis, which calculates the change in R² when each variable is added to the model.

Results: Model selection indicated moderate evidence in favor of affective empathy predicting social competence (BF=3.6) but inconclusive evidence for the other predictors. The dominance analysis provided converging evidence, as affective empathy explained the most variance in social competence (11%), followed by verbal IQ (6%) and BMP (4%). The remaining predictors each accounted for 1% or less of the variance.

Conclusions: These findings suggest that for some individuals with ASD, social deficits are partly explained by a lack of emotional resonance with one’s social partner. Our results do not support the hypothesized link between ToM and social functioning in ASD. However, we do not conclude that ToM and the other social domains represented in this study have no relevance for social functioning in ASD. Instead, our results could indicate the limitations of several commonly used behavioral measures, which may not capture the cognitive and affective processes most relevant to real-world social functioning.