CAREER: Modeling the cognitive and neural mechanisms of mental state detection, prediction, and regulation with deep learning

Everyday social interactions – like conversations – play an important role in almost every part of life. Whether negotiating a business deal, teaching a child, or simply having fun with friends, people need certain skills to interact successfully with others. One of the most important of these skills is the ability to understand other people’s mental states: their thoughts, feelings, beliefs, desires, and intentions. Understanding others’ mental states is useful because it helps people to more effectively plan their own actions during interactions (e.g., knowing when to call someone’s bluff, or when a loved one needs cheering up). This project studies how, precisely, people understand each other’s mental states, and how their brains make this understanding possible. This project investigates three key parts of mental state understanding: 1) how people detect which mental state(s) another person is currently experiencing, 2) how people predict what mental states others are likely to experience in the future, and 3) how people intervene to change or regulate other people’s mental states. Using tightly controlled experiments designed to capture real-world mental state dynamics, the project measures how people learn to detect, predict, and regulate mental states. In parallel, deep neural network models learn to complete the same tasks. By testing how well different models explain human behavior and corresponding brain activity, this research provides insight into the mechanisms that support mental state understanding. By characterizing the computational and neural mechanisms of mental state understanding, this research lays the foundation for developing ways to improve a wide variety of human social interactions. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.