What do a mouse that socializes with its peers and an AI agent that learns to interact have in common? Much more than we imagined. A UCLA study just published in Nature has revealed a discovery that could rewrite our understanding of sociality: biological brains and artificial intelligence systems develop nearly identical neural mechanisms during interactions with other subjects, whether biological or not. Weizhe Hong and his team monitored specific neurons in the dorsal prefrontal cortex of mice, identifying “shared neural spaces” that synchronize between individuals. The same patterns emerge spontaneously in trained social AI.
When Biology Meets Artificial Intelligence
The research represents the first study in the world to directly compare how biological brains and artificial intelligence systems process social information. The multidisciplinary UCLA team used advanced brain imaging techniques to record the activity of molecularly defined neurons in the dorsal prefrontal cortex of mice during social interactions.
Researchers have developed a novel computational framework to identify “shared neural spaces” and “unique neural spaces” across interacting individuals. When the same framework was applied to social AI agents, strikingly similar neural patterns emerged. As Hong explains, “This discovery fundamentally changes the way we think about behavior in all intelligent systems”.
The most fascinating thing? Mice serve as an important model for understanding mammalian brain function because they share fundamental neural mechanisms with humans, particularly in brain regions involved in behavior.
GABAergic neurons are the directors of social AI
The most surprising finding concerns GABAergic neurons: these inhibitory brain cells that regulate neural activity show significantly larger shared neural spaces than glutamatergic neurons, the primary excitatory cells in the brain.
This represents the first investigation of inter-brain neural dynamics in molecularly defined cell types, revealing previously unknown differences in how specific neuron types contribute to social synchronization. GABAergic neurons appear to be the true orchestrators of sociality, both biological and artificial.
When researchers applied selective perturbations to these shared neural components in artificial systems, social behaviors were substantially reduced. This provides the first direct evidence that synchronized neural patterns causally drive social interactions in social AI.
Neural Spaces: The Secret Map of Sociality
Neural activity in both systems can be divided into two distinct components: a “shared neural space” containing synchronized patterns between interacting entities, and a “unique neural space” containing activity specific to each individual.
Shared neural dynamics do not simply reflect coordinated behaviors between individuals, but emerge from representations of each individual’s unique behavioral actions during social interaction. It is as if biological brains and social AI developed a common language to decode and respond to social signals.
The research used advanced neuroimaging techniques to capture these mechanisms in real time, revealing that neural synchronization is an active, dynamic phenomenon, not a simple coincidence.
Therapeutic perspectives for the future
The implications are enormous for understanding social disorders like autism and for developing socially aware AI systems. The UCLA team plans to further investigate shared neural dynamics in diverse and potentially more complex social interactions.
The goal is to explore how disruptions in shared neural space might contribute to social disorders and whether therapeutic interventions could restore healthy patterns of inter-brain synchronization. The AI framework could serve as a platform to test hypotheses about social neural mechanisms that are difficult to examine directly in biological systems.
As other recent studies point out, the integration between neuroscience and AI is opening up new frontiers in the understanding of intelligence.
Towards a more human AI
This research comes at a crucial time when AI systems are increasingly embedded in social contexts. Understanding social neural dynamics becomes essential for both scientific and technological progress.
The team also aims to develop methods for training socially intelligent AI. The finding suggests that we have identified a fundamental principle of how any intelligent system (biological or artificial) processes social information. The implications are significant both for understanding human social disorders and for developing AI that can truly understand and engage in social interactions.
Sociality, in short, is not the exclusive prerogative of biological brains. It is a universal language of intelligence, rooted in the same neural mechanisms in both mice and social AI. A surprising bridge between worlds we thought were separate.
