A study published in Autism Research suggests that autistic people possess a brain configured with a highly interconnected reward centre and social processing networks. Led by Chen Yang and Ai-Ping Sun, the team used resting-state brain imaging to analyse data from 646 participants, including 298 autistic individuals and 348 typically developing controls, to map how these networks communicate when the brain is at rest.
The human brain handles social interactions through a distributed system known as the social brain, comprising four subnetworks that govern different facets of connection. The reward system encodes motivation and the pleasure attached to social stimuli, while the face perception network supports encoding and recognition of facial expressions.
The remaining two subnetworks manage more abstract social processes: the theory of mind network enables the attribution of mental states and intentions, and the mirror neuron system activates both during action and observation of the same action, aiding imitation and learning.
For social cognition to function smoothly, these subnetworks must maintain a careful balance. They need to operate in a modular fashion to process tasks efficiently, a concept known as modular segregation, while also sharing information across the broader brain network to produce coherent social behaviours, in a process called modular integration.
Previous research suggests this balance is altered in autism spectrum disorder. Autistic people often show differences in social motivation, facial recognition and mimicking behaviours. Researchers Chen Yang, Ai-Ping Sun and colleagues sought to map how these social subnetworks integrate and whether any imbalances relate to clinical social symptoms, while also probing the chemical and genetic profiles of these brain regions.
To examine these network dynamics, the researchers analysed resting-state functional magnetic resonance imaging data from 646 participants—298 autistic individuals and 348 typically developing controls. Resting-state fMRI tracks blood flow to infer brain activity while participants are awake but not engaged in a task, revealing which areas activate together over time.
Using an automated process, the team isolated 102 brain regions that make up the four social subnetworks and calculated a metric called the participation coefficient for each participant. This figure quantifies how much a given brain node communicates with outside modules compared with its own internal connections; a low score signals high segregation, while a high score signals greater cross-network integration.
Autistic participants showed increased modular integration of the reward system. Rather than remaining relatively isolated, the reward network formed a greater number of connections with the face perception network and with the mirror neuron system. The face perception network itself also exhibited higher integration, though the changes were most pronounced for the reward system.
To link these network patterns with observable social behaviour, the researchers compared connectivity metrics with standard clinical assessments, including parental reports of social communication and motivation. They found that greater reward-system integration correlated with higher scores on the Social Responsiveness Scale, indicating more pronounced challenges in social communication, social awareness and social cognition. The team noted that this correlation was driven specifically by social functioning rather than restricted and repetitive behaviours.
Turning to neurochemistry, the researchers overlaid their connectivity maps onto established atlases of neurotransmitter distribution. A leading theory of autism posits an imbalance between excitatory and inhibitory signals, often linked to how neurotransmitters bind to receptors. The regions of the reward system that were highly integrated in autistic participants overlapped with brain areas known to have lower densities of certain serotonin and gamma-aminobutyric acid (GABA) receptors. A reduction in these receptors implies diminished inhibitory control, supporting a view that a hyper-excitable reward network may underpin the observed social differences.
In search of a genetic basis, the team compared their brain maps with a public database of human brain tissue gene expression. They identified gene transcripts whose spatial distribution aligned with the areas of heightened reward-network integration. The implicated genes are predominantly involved in developmental processes, including cellular proliferation, positive regulation of cell migration and tissue formation, suggesting that the hyper-integration seen in the autistic brain may stem from foundational developmental changes.
The researchers then examined how this reward-network integration evolves with age by testing a second, independent dataset. Using linear regression models, they tracked age-related changes in both autistic and typically developing participants and found that integration increased over time in both groups, following parallel trajectories. However, autistic participants maintained higher levels of integration across age groups. In the second dataset, seven of nine sites showed a similar pattern, though the overall pooled effect was small and not statistically significant.
While the study links brain activity, genetics and behaviour, it relies on chemical and genetic maps derived from neurotypical brains and overlays these onto autistic participants’ scans. As a result, the researchers caution that the results do not prove that the proposed neurochemical deficits cause the altered connectivity, only that there is a correspondence worth further investigation.
Future work should incorporate molecular imaging directly within autistic cohorts and adopt longitudinal designs to follow individuals over time, enabling a clearer view of how the social brain develops across the lifespan.
The study, titled Disrupted Modular Integration of the Reward System Is Associated With Social Deficits in Autism Spectrum Disorder, is authored by Chen Yang, Ai-Ping Sun, Sheng-Zhi Ma, Wen-Qiang Dong, Xiao Chen, Shuai-Yu Chen, Yuqi You, Yu-Feng Zang and Li-Xia Yuan.
