The Interaction of Autism Traits and Psychosis Traits on Mentalizing in a Clinical-Community Sample

Background: Previous research suggests dimensional traits that characterize autism spectrum disorder (ASD) and psychotic spectrum disorders (PSDs) are useful for understanding variations in social cognition. Notably, in individuals without clinical diagnoses, concurrent high levels of both sets of traits are associated with better performance on social cognitive tasks relative to discordant levels of traits (i.e., high on one set of traits, low on the second set of traits; Abu-Akel et al., 2015). The current study extends previous work on this “normalizing effect” by examining effects of autism traits, psychosis traits, and their interaction on mentalizing in a combined clinical-community sample of individuals with and without ASD.

Objectives: The current study examines 1) main effects of autism traits, 2) main effects of psychosis traits, and 3) interaction effects of autism and psychosis traits on mentalizing accuracy (forced-choice task) and spontaneous mentalizing levels (open-ended task).

Methods: Seventy-six adolescents and adults (M_age=22.18 years, SD=4.51 years; 19 females) completed the Wechsler Abbreviated Scale of Intelligence, 2^nd Edition (WASI-II), Autism Spectrum Quotient (AQ; autism traits), Schizotypal Personality Questionnaire, Factor 1 Cognitive-Perceptual (SPQ-Cognitive; psychosis traits), Social Attribution Task-Multiple Choice (SAT-MC; mentalizing accuracy), and the Dynamic Interactive Shape Clips Task (DISC; mentalizing levels). Thirty-two individuals (eight females) had ASD diagnoses confirmed by record review. We conducted multiple linear regression analyses to evaluate main effects and interaction effects of mean-centered autism traits and psychosis traits on SAT-MC mentalizing accuracy and DISC mentalizing levels with WASI-II Vocabulary scores as a covariate. When data distributions violated assumptions of analytic procedures, we evaluated the significance of results by generating p-values and confidence intervals using bootstrapping based on 5,000 samples. We used the PROCESS macro for SPSS and, specifically, the Johnson-Neyman Technique to probe significant interactions.

Results: Bivariate correlations showed that AQ scores had significant relationships with SPQ-Cognitive (ρ(76)=.26, p=.02) and SAT-MC scores (ρ(76)=-.26, p=.02). SAT-MC and DISC scores were positively related (ρ(75)=.37, p=.001). In the regression models, AQ scores were significantly negatively related to SAT-MC scores, F(2,73)=14.67, p<.001, R²=.29, ΔR²=.07, β=-.26,p=.01, but not DISC (ps>.05). SPQ-Cognitive was not related to SAT-MC or DISC scores (ps>.05). The interaction term, AQ x SPQ-Cognitive, was significantly negatively related to SAT-MC scores (R²=.33, F(4,71)=8.65, p<.001, ΔR²=.04, β=-.20,p=.03) and DISC scores (R²=.18, F(4,70)=3.72, p=.008, ΔR²=.05,β=-.23,p=.04), such that higher levels of both autism and psychosis traits related to the poorest performance on mentalizing tasks.

Conclusions: We found an enhancing negative effect of autism traits and psychosis traits on mentalizing accuracy and levels in a combined sample of individuals with and without ASD. Specifically, individuals with concurrent high levels of these traits performed less accurately on a forced-choice mentalizing task and made the fewest, least complex mental state attributions during an open-ended mentalizing task compared to individuals with discordant levels of traits. Our results suggest that relative levels of these traits exert an enhancing negative effect on some forms of social cognition and behavior. The normalizing effect documented in a previous study may not be present for all social cognitive skills or when traits reach clinically significant levels.