Social inference brain networks in autistic adults during movie-viewing: functional specialization and heterogeneity

Difficulty in social inferences is a core feature in autism spectrum disorders (ASD). On the behavioral level, it remains unclear whether reasoning about others' mental states (Theory of Mind, ToM) and empathic responses to others' physical states may be similarly or differentially affected in autism. On the neural level, these inferences typically engage distinct brain networks (ToM versus Pain networks), but their functional specialization remains not well understood in autism. This study aimed to investigate the functional specialization, heterogeneity, and brain-behavior relationships of the ToM and Pain networks in autistic compared to neurotypical (NT) participants. We hypothesized differential functional network specialization (i.e., functional connectivity), increased heterogeneity, and less typical network responses specifically in the ToM network, with relatively similar responses in the Pain network in ASD.

Using functional magnetic resonance imaging (fMRI), we investigated neural responses in 107 adults (autistic: 34 (female = 11), NT: 73 (female = 23); matched for age, intellectual functioning, sex, motion) while they passively watched a short, animated movie including events that evoke reasoning about characters' mental states and bodily sensations. Preregistered analyses included regression models to assess inter-region correlation of within- and across-network connectivity, inter-subject correlation to quantify similarity to the average neurotypical, as well as to within- and across-group timecourse responses, and brain-behavior relationships relevant for social inferences.

Functional specialization of ToM and Pain networks were overall intact, with distinct network responses in both groups. The autistic group showed differential ToM network responses and reduced similarity to the average typical response for both networks. Network responses were more idiosyncratic and heterogenous in the autistic group. Brain-behavior relationships differed between groups for ToM behavior only.

Effects between groups were overall small. Samples were acquired across two sites, yet the sample size restricts subgroup analyses that may further inform autistic heterogeneity and limits generalizability.

We found weak evidence for greater differential responses in brain networks underlying ToM inferences than those involved in empathic responses in autism, consistent with a prior empathy imbalance hypothesis. We outline suggestions for replicating, generalizing and extending these results in future research.