Social Bayes: Using Bayesian Modeling to Study Autistic Trait–Related Differences in Social Cognition

Abstract

BACKGROUND: The autistic spectrum is characterized by profound impairments of social interaction. The exact subpersonal processes, however, that underlie the observable lack of social reciprocity are still a matter of substantial controversy. Recently, it has been suggested that the autistic spectrum might be characterized by alterations of the brain's inference about the causes of socially relevant sensory signals. METHODS: We used a novel reward-based learning task that required integration of nonsocial and social cues in conjunction with computational modeling. Thirty-six healthy subjects were selected based on their score on the Autism-Spectrum Quotient (AQ), and AQ scores were assessed for correlations with cue-related model parameters and task scores. RESULTS: Individual differences in AQ scores were significantly correlated with participants' total task scores, with high AQ scorers performing more poorly in the task (r = 2.39, 95% confidence interval = 20.68 to 20.13). Computational modeling of the behavioral data unmasked a learning deficit in high AQ scorers, namely, the failure to integrate social context to adapt one's belief precision-the precision afforded to prior beliefs about changing states in the world-particularly in relation to the nonsocial cue. CONCLUSIONS: More pronounced autistic traits in a group of healthy control subjects were related to lower scores associated with misintegration of the social cue. Computational modeling further demonstrated that these traitrelated performance differences are not explained by an inability to process the social stimuli and their causes, but rather by the extent to which participants consider social information to infer the nonsocial cue.

Deconstructing Theory-of-Mind Impairment in High-Functioning Adults with Autism

“…ToM is engaged ubiquitously in our everyday social behavior and features a specific developmental trajectory [4] that is notably delayed in children with autism spectrum disorder (ASD) [5,6]. In healthy individuals, model-based analyses of social learning and decision-making have successfully elucidated specific computational components of ToM processing [7][8][9][10][11]. However, the use of this approach to study ToM impairment in ASD has been extremely limited [10,12].…”

Expecting social punishment facilitates control over a decision under uncertainty by recruiting medial prefrontal cortex

“…Based on recent findings that people became more 'myopic' under the threat (Korn & Bach, 2019), we speculate that directed exploration would also be reduced by blame expectation and it could explain a reduced creativity under the threat. Finally, from the perspective of computational psychiatry, an investigation of the optimization behavior of our task in patients with psychiatric conditions, such as autism and psychopathologies, would be interesting to quantify their lack of an ability to expect social responses and utilize adaptive behaviors (Sevgi, Diaconescu, Tittgemeyer, & Schilbach, 2016).…”

Differential valuation and learning from social and non-social cues in Borderline Personality Disorder

“…In this initial effort to extend the work of Behrens et al, we adhered closely to their approach. However, Diaconescu et al have used the Hierarchical Gaussian Filter (HGF), allowing individual differences and obtaining subject-specific estimates of approximate Bayesian inference (9, 11, 26). Modelling patient behavior using the HGF may also detect subtler between-group differences (11).…”

“…Weighting of social versus non-social cues in the SVT in community samples correlates with self-reported traits. In healthy adults, self-reported autistic traits were directly correlated to poorer overall task performance, and inversely correlated to weighting social over non-social cues (9). Also, subjects with more autistic traits were worse at avoiding the influence of bad advice during the “volatile phase” of the task, when reward for the social cue was most unreliable.…”

“…Weighting of social versus non-social cues in the Social Valuation Task (SVT) in community samples is correlated with self-reported traits. Sevgi et al screened a sample of healthy adults without psychiatric disorders for autistic traits using the Autism Quotient Spectrum Questionnaire (AQ), a self-report measure (8). People with low and high levels of autism traits in a modified SVT with implicit instead of explicit social cues (gaze direction instead of explicit advice).…”