Testing Social Cognition As a Mediator between Neurocognition and Social Outcomes

Background: The specific contributors to poor social functioning outcomes for adults with autism are poorly understood. Neurocognitive abilities (e.g., working memory) and social cognition (e.g., social perception and appraisal) are frequently targeted in intervention studies to improve social outcomes, but to date no studies have empirically tested the contribution of these constructs to social functioning.

Objectives: We examined the relative contribution of neurocognition and social cognitive domains to predict both social skills and functioning. We also tested these constructs for mediation, hypothesizing the effect of neurocognition on outcomes is mediated through a serial pathway between social perception and social appraisal abilities.

Methods: Adults with autism (N=103; age M=24.28; 11 females) completed a battery of tasks. Social perception was measured by computing a composite score of tests measuring ability to recognize emotion in static faces (ER-40; Kohler et al., 2003; Reading the Mind in the Eyes; Baron-Cohen et al., 2001), dynamic videos (Bell Lysaker Emotion Recognition Task; Bryson et al., 1997), and point-light walkers (Emotional Biological Motion; Biological Motion; Kern et al., 2013), as well as ability to recognize faces (Benton Facial Recognition Task; Benton et al., 1968). Social appraisal was measured by computing a composite score of tasks assessing mentalizing abilities (The Awareness of Social Inference Task; MacDonald et al., 2003; Hinting Task; Corcoran et al., 1995) and the Relationships Across Domains task (Sergi et al., 2009), assessing understanding of social relationships.

Participants also completed the MATRICS Consensus Cognitive Battery (MCCB; Nuechterlein et al., 2008), measuring working memory, short term memory, and processing speed through pencil and paper tasks.

Social skill outcomes were assessed through trained coders’ average rating of six skills displayed in a three minute interaction with a researcher role-playing meeting a new neighbor (Social Skills Performance Assessment; Patterson et al., 2001). Social functioning was assessed with the UCSD Performance-based Skills Assessments (UPSA; Mausbach et al., 2007), a performance-based task measuring daily living skills.

Results: Gender, age, neurocognition, social perception, and social appraisal entered as predictors accounted for 48% of variance in social functioning scores and 30% of social skills scores. In both models, older participants and participants with stronger neurocognitive ability scored higher on social functioning and skills (ps<.001), and social appraisal ability predicted better social functioning (p=.003).

Controlling for age, we examined indirect effects and the bootstrapped confidence intervals around the estimate to test for evidence of mediation (Figure 1). The indirect serial pathway was significant in predicting social functioning (b=.08, bootstrapped lower limit=.03, bootstrapped upper limit=.15), suggesting social cognition is a mechanism through which neurocognition influences social functioning. There was no evidence for mediation in predicting social skills (b=.02, bootstrapped lower limit=-0.05, bootstrapped upper limit=.09).

Conclusions: Neurocognition is a significant predictor of social functioning and social skills for adults with autism. The mediation results show neurocognitive ability is transmitted through social perception and social appraisal ability for social functioning, suggesting interventions targeting neurocognition may benefit social cognitive skill and functioning. However, our results suggest social skill is a function of a different mechanism.