Computational Human-AI Experimental Neuroscience (CHEN) Lab
Understanding Interacting Minds Across Humans and AI
Human-Human Interaction and Social Neuroscience
Human minds are shaped by the people and social environments around them. We study how individuals understand and influence one another across dyads, groups, and social networks, with a particular interest in how these processes are represented in the brain. Our research examines how social and cultural experiences shape neural representations of the self and others, how beliefs and emotions are transmitted between individuals, and how shared experiences bring behavior and brain responses into alignment. From acculturation and friendship development to trust, social influence, and shared affect, we seek to understand how interacting brains shape our relationships, experiences, and understanding of the social world.
Computational and Multimodal Approaches
To capture the complexity of interacting minds, we integrate multimodal data, naturalistic paradigms, and computational approaches. By combining neural activity, behavior, facial expressions, language, and social network information, we examine psychological processes across different levels of analysis. Naturalistic experiments allow us to study how people respond to shared experiences and social contexts as they unfold over time. We then apply computational modeling, machine learning, ISC, ISRSA, and RSA to identify patterns within and across individuals. These approaches help us move beyond isolated observations to understand the dynamic relationships among brain representations, psychological experiences, and social behavior, providing a common methodological approach for studying interactions among humans and AI.
Human-AI Interaction and Alignment
Building on our understanding of human interaction, we investigate how biological and artificial minds understand and influence one another. We study how people perceive and respond to AI, how these interactions may reshape neural representations in the human brain, and how AI models represent human psychological states. A central question is whether interaction brings human and artificial representations into alignment or reveals systematic differences between them. By comparing human behavior and brain representations with the internal representations of large language models (LLM), we aim to understand how alignment develops across different cognitive systems and what it reveals about the processes underlying social interaction.
AI-AI Interaction and Interpretability
Extending these questions to artificial minds, we investigate how AI agents communicate, cooperate, and influence one another, and how their internal representations support emerging social behavior. Multi-agent systems allow us to manipulate goals, information, and social contexts to examine the development of shared representations, coordination, and social norms. Through AI interpretability, we seek to understand the computational mechanisms underlying these interactions and how they relate to human psychological representations. This work may inform the development of AI systems that remain aligned with human goals and values. More broadly, we use interacting artificial minds as computational models to uncover principles that help explain how human minds understand, influence, and align with one another.