Neural Circuits of Social Behavior in Autism Spectrum Disorder: From Mechanism to Intervention

Difficulties in social communication and interaction are core diagnostic features of autism spectrum disorder (ASD), yet interventions targeting these difficulties have modest and variable effects. Understanding mechanistically distinct contributions to social behaviors may identify new therapeutic opportunities.

This narrative review integrates human neuroimaging studies, supporting rodent circuit neuroscience, clinical intervention research, and developmental studies. We organize the findings into four overlapping functional-computational domains and relate these domains to the Research Domain Criteria Social Processes framework.

The four domains are: (1) Social Reward and Social Interest, encompassing the incentive and hedonic value of social contact and motivation to engage; (2) Social Memory and Social-Affective Value, encompassing recognition of familiar individuals and learned affective or motivational associations; (3) Social Cue Perception, Action Representation, and Social Prediction/Timing, encompassing the processing and forecasting of dynamic social information; and (4) Higher-Order Social Cognition and Model Updating, encompassing mental-state inference and the updating of person-specific social models. Because these domains are not independent, interactions among reward, memory, cue processing, prediction, and higher-order inference create multiple possible developmental entry points, such that early differences in one process may alter experience-dependent learning in others. Evidence for specific causal cascades remains preliminary and requires longitudinal testing.

A circuit-and-development framework may improve hypothesis generation, domain-specific phenotyping, target-engagement assessment, and mechanism-aligned outcomes. Clinical translation, however, requires caution. Biomarkers should be validated as predictors of treatment response, cascade models should be tested longitudinally, and interventions should combine mechanism-informed targets with behavioral support that promotes generalization to daily life.