RESEARCH AREAS
Social interactions between individuals and among groups are a hallmark of human society and are critical to the physical and mental health of a wide variety of species including humans. The central goal of our lab is to investigate the fundamental principles underlying the regulation of social behavior in both biological and artificial intelligence systems, with a focus on empathy and prosociality. We study how neural circuits and the underlying computation regulate social behavioral decisions within a single brain, as well as how shared neural dynamics arise across individuals. We take a multi-disciplinary approach and use a variety of experimental and computational technologies across molecular, circuit, and behavioral levels.
Review: Chen & Hong. Neural circuit mechanisms of social behavior. Neuron 2018.
Review: Chen & Hong. Neural circuit mechanisms of social behavior. Neuron 2018.
1. Neural mechanisms of social information processing and social decision-making
|
Understanding how neural circuits control social behavior has long been a major challenge in neuroscience. Social interactions involve the active detection of various social cues and the selection of appropriate behavioral decisions. We are interested in studying (1) how social sensory information is processed and integrated in the brain, (2) how different social behavioral decisions are selected and modulated by neural circuits, and (3) how social interactions are motivated and can be rewarding for individuals.
|
|
Social reward and social motivation
Hu et al. An amygdala-to-hypothalamus circuit for social reward Nature Neuroscience 2021
Representation of social information in the brain
Kingsbury et al. Cortical representations of conspecific sex shape social behavior Neuron 2020
Sexually dimorphic control of parental behavior
Chen et al. Sexually dimorphic control of parenting behavior by the medial amygdala. Cell 2019
Neural mechanisms of aggressive behavior
Hong et al. Antagonistic control of social versus repetitive self-grooming behaviors by separable amygdala neuronal subsets. Cell 2014
Hu et al. An amygdala-to-hypothalamus circuit for social reward Nature Neuroscience 2021
Representation of social information in the brain
Kingsbury et al. Cortical representations of conspecific sex shape social behavior Neuron 2020
Sexually dimorphic control of parental behavior
Chen et al. Sexually dimorphic control of parenting behavior by the medial amygdala. Cell 2019
Neural mechanisms of aggressive behavior
Hong et al. Antagonistic control of social versus repetitive self-grooming behaviors by separable amygdala neuronal subsets. Cell 2014
2. The neuroscience of empathy, compassion, and prosocial behavior
Prosocial behavior—actions that benefit others—represents one of the most evolutionarily puzzling yet critically important aspects of social life. Unlike other social behaviors that primarily serve direct self-interest, prosocial behaviors appear to contradict basic evolutionary principles by directing effort and resources toward helping others, often at one’s own cost. This paradox makes prosocial behavior both fascinating from a scientific perspective and profoundly understudied compared to self-serving social behaviors. We pioneered the systematic study of prosocial behaviors in rodents, demonstrating several forms of prosocial behavior and their underlying neural mechanisms, which have profoundly transformed our understanding of these behaviors. We demonstrated that mice display comforting behavior to reduce other’s emotional distress (Wu et al. Nature 2021), targeted helping behavior to attend to others' injuries or pain (Zhang et al. Nature 2024), and rescue-like behavior to facilitate the recovery of an unconscious animal (Sun et al. Science 2025). Using these paradigms, combined with molecular genetics and computational approaches, we have identified the molecularly and anatomically defined neural pathways in the amygdala and the prefrontal cortex that specifically encode and control these behaviors.
|
Review: Wu & Hong. Neural basis of prosocial behavior. Trends in Neurosciences 2022 Neural mechanism of rescue-like behavior Sun et al. A neural basis for prosocial behavior toward unresponsive individuals. Science 2025 Neural mechanism of helping behavior Zhang et al. Cortical regulation of helping behavior towards others in pain. Nature 2024 Neural mechanism of comforting behavior Wu et al. Neural control of affiliative touch in prosocial interaction. Nature 2021 |
|
|
3. Social cooperation and its neural basis
Cooperation represents a unique form of prosocial behavior requiring real-time coordination between individuals to achieve mutual benefits. In Jiang et al. (Science 2025), we developed novel cooperative tasks where two mice must coordinate their actions to obtain mutualistic rewards. Through comprehensive behavioral analysis, we demonstrated that mice develop sophisticated decision-making strategies, learning to observe their partner’s state and making real-time choices. We also revealed that the anterior cingulate cortex (ACC) represents cooperative behavioral processes and plays a causal role in coordinating joint actions between animals. This work represents the first systematic investigation of mutualistic cooperation in mice at both behavioral and neural levels, establishing a new framework for understanding how individuals coordinate in real-time to achieve shared goals.
|
|
Neural basis of cooperative behavior
Jiang et al. Neural basis of cooperative behavior in biological and artificial intelligence systems. Science 2025
4. Neural dynamics across interacting individuals
Social interaction can be seen as a dynamic feedback loop that couples two or more high-dimension neural networks (i.e. brains). A fuller understanding of the social brain requires a description of how the neural dynamics are coupled across brains and how they coevolve over time. We study social decisions and emergent inter-brain neural properties in a multi-brain framework that considers social interaction as an integrated network of neural systems. In Kingsbury et al. (Cell 2019), we made the discovery of inter-brain neural synchrony in mice—the first demonstration of correlated neural activity across the brains of non-primate animals. Using simultaneous calcium imaging, we revealed that animals exhibit cross-brain correlations in medial prefrontal cortex (mPFC) activity that arise from distinct neuronal populations encoding one's own behaviors versus those of social partners.
While our initial inter-brain work focused on demonstrating the existence of neural correlations between interacting animals, we next sought to understand the detailed mechanisms underlying these phenomena and their relationship to specific behaviors. In Zhang et al. (Nature 2025), we discovered that shared neural dynamics emerge from specific cell types in the mPFC and manifest across multiple dimensions of neural activity space. This study constitutes the first observation of interbrain synchrony in molecularly defined cell types and provides a deeper explanation for the emergence of synchronization from activity in the high-dimensional neural space. The study provides unprecedented insight into how individual brains become coupled during social interaction to form integrated multi-brain networks.
While our initial inter-brain work focused on demonstrating the existence of neural correlations between interacting animals, we next sought to understand the detailed mechanisms underlying these phenomena and their relationship to specific behaviors. In Zhang et al. (Nature 2025), we discovered that shared neural dynamics emerge from specific cell types in the mPFC and manifest across multiple dimensions of neural activity space. This study constitutes the first observation of interbrain synchrony in molecularly defined cell types and provides a deeper explanation for the emergence of synchronization from activity in the high-dimensional neural space. The study provides unprecedented insight into how individual brains become coupled during social interaction to form integrated multi-brain networks.
|
Review: Kingsbury & Hong. A multi-brain framework for social interaction. Trends in Neurosciences 2020 Inter-brain neural dynamics in genetically defined neuronal populations and in artificial intelligence systems Zhang et al. Inter-brain neural dynamics in biological and artificial intelligence systems. Nature 2025 Inter-brain neural synchrony across brains of socially interacting animals Kingsbury et al. Correlated neural activity and encoding of behavior across brains of socially interacting animals. Cell 2019 |
|
5. Emerging social interactions between artificial intelligence (AI) systems
In collaboration with Prof. Jonathan Kao, we investigated the principles that govern social interactions between artificial agents using multi-agent reinforcement learning (MARL). Drawing from our insights about biological inter-brain dynamics, we demonstrated that artificial agents trained to interact socially develop neural network dynamics remarkably similar to those observed in biological brains (Zhang et al. Nature 2025). Strikingly, selectively disrupting the neural components that contribute to shared neural dynamics substantially reduced the social actions of agents. This presents conclusive evidence for a functional role for synchronized neural components in driving social interactions in artificial intelligence systems and suggests that the multi-agent systems may serve as a platform not only for understanding emergent behavioral strategies during social interaction but also for testing the causal role of specific neural components, which could be otherwise challenging to examine in animal models.
We also modeled cooperation between artificial intelligence systems by training two artificial agents to perform a task analogous to the mouse task (Jiang et al. Science 2025). Remarkably, the agents developed behavioral strategies and neural representations reminiscent of those observed in the biological brain, revealing striking parallels between cooperative behavior in biological and artificial systems. This convergent evolution of cooperative mechanisms across biological and artificial systems provides powerful evidence for universal principles underlying social cooperation and offers new approaches for understanding and designing social AI systems.
Inter-brain neural dynamics in genetically defined neuronal populations and in artificial intelligence systems
Zhang et al. Inter-brain neural dynamics in biological and artificial intelligence systems. Nature 2025
Neural basis of cooperative behavior
Jiang et al. Neural basis of cooperative behavior in biological and artificial intelligence systems. Science 2025
We also modeled cooperation between artificial intelligence systems by training two artificial agents to perform a task analogous to the mouse task (Jiang et al. Science 2025). Remarkably, the agents developed behavioral strategies and neural representations reminiscent of those observed in the biological brain, revealing striking parallels between cooperative behavior in biological and artificial systems. This convergent evolution of cooperative mechanisms across biological and artificial systems provides powerful evidence for universal principles underlying social cooperation and offers new approaches for understanding and designing social AI systems.
Inter-brain neural dynamics in genetically defined neuronal populations and in artificial intelligence systems
Zhang et al. Inter-brain neural dynamics in biological and artificial intelligence systems. Nature 2025
Neural basis of cooperative behavior
Jiang et al. Neural basis of cooperative behavior in biological and artificial intelligence systems. Science 2025
6. New molecular and behavioral technologies for studying social behavior
|
To enable our discoveries, we developed Act-seq (Wu et al. Neuron 2017), a new single-cell RNA-seq method that systematically maps molecularly defined neuronal subpopulations activated by specific behaviors or experiences. This technique solved the major technical challenge of artifactual transcriptional changes during cell dissociation, enabling accurate profiling of behavior-activated neurons. Act-seq provided the first molecular taxonomy of amygdala cell types and robustly detected experience-induced gene expression changes across multiple cell types. This methodological advance has become essential for linking specific behaviors to molecularly defined neural populations and continues to drive discoveries across the neuroscience community.
|
|
Act-seq: a single-cell sequencing approach to identify active neuronal populations
Wu et al. Detecting Activated Cell Populations Using Single-Cell RNA-Seq. Neuron 2017
Using depth sensing and machine learning to track and analyze social behavior
Hong et al. Automated Measurement of Mouse Social Behaviors Using Depth Sensing, Video Tracking, and Machine Learning. Proc. Natl. Acad. Sci. USA. 2015
Wu et al. Detecting Activated Cell Populations Using Single-Cell RNA-Seq. Neuron 2017
Using depth sensing and machine learning to track and analyze social behavior
Hong et al. Automated Measurement of Mouse Social Behaviors Using Depth Sensing, Video Tracking, and Machine Learning. Proc. Natl. Acad. Sci. USA. 2015