Social behavioural adaptation in Autism.

Autism is still diagnosed on the basis of subjective assessments of elusive notions such as interpersonal contact and social reciprocity. We propose to decompose reciprocal social interactions in their basic computational constituents. Specifically, we test the assumption that autistic individuals disregard information regarding the stakes of social interactions when adapting to others. We compared 24 adult autistic participants to 24 neurotypical (NT) participants engaging in a repeated dyadic competitive game against artificial agents with calibrated reciprocal adaptation capabilities. Critically, participants were framed to believe either that they were competing against somebody else or that they were playing a gambling game. Only the NT participants did alter their adaptation strategy when they held information regarding others' competitive incentives, in which case they outperformed the AS group. Computational analyses of trial-by-trial choice sequences show that the behavioural repertoire of autistic people exhibits subnormal flexibility and mentalizing sophistication, especially when information regarding opponents' incentives was available. These two computational phenotypes yield 79% diagnosis classification accuracy and explain 62% of the severity of social symptoms in autistic participants. Such computational decomposition of the autistic social phenotype may prove relevant for drawing novel diagnostic boundaries and guiding individualized clinical interventions in autism. Author summary: Autism or AS is mostly characterized by impairments in a very specific yet intricate skill set, namely: social intelligence. In this work, we focus on "social reciprocity", i.e. the continuous adaptation of one's behaviour that both moulds and appropriately responds to others' behaviour. Our working hypothesis is that social reciprocity deficits in people with AS derive from a basic inability to tune one's adaptation strategy to contextual knowledge about the stakes of social interactions (e.g., others' cooperative or competitive incentives). We ask participants to engage in simple interactive games with AI agents that are endowed with calibrated reciprocal adaptation capabilities. Critically, participants are framed to believe either that they are competing against somebody else (social framing) or that they are playing a gambling game (non-social framing). Only in the social condition do participants know about the (competitive) incentives of their opponents. Computational analyses of action sequences in the games show that, contrary to healthy controls, people with AS do not change their strategy according to whether they hold information regarding their opponents' incentives or not. In addition, these analyses yield 79% diagnosis out-of-sample classification accuracy (AS versus controls) and predict 62% of the severity of social symptoms in people with AS. This demonstrates the feasibility of AI-based quantitative assessments of social cognition and its deficits. [ABSTRACT FROM AUTHOR]

Copyright of PLoS Computational Biology is the property of Public Library of Science and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)