AI summaryⓘ
The authors present a new method for teaching large language models (LLMs) to better understand and predict how people's changing beliefs influence their actions, especially in uncertain and high-pressure situations like disaster response. Unlike previous methods that treated beliefs as fixed and separate, their approach models beliefs as interconnected and evolving over time. They develop a system that represents these beliefs using a dynamic graph, learns how beliefs depend on each other, and relates belief changes to decisions and information-seeking behavior. Tested on real disaster data, their method improved predictions about people's actions and generated belief trajectories that align well with human reasoning.
Theory of MindLarge Language ModelsDynamic Belief GraphProbabilistic Graphical ModelsEnergy-based ModelsBelief TrajectoriesDisaster ResponseVariational InferenceHuman-in-the-loop AutonomyAction Prediction
Authors
Ruxiao Chen, Xilei Zhao, Thomas J. Cova, Frank A. Drews, Susu Xu
Abstract
Theory of Mind (ToM) reasoning with Large Language Models (LLMs) requires inferring how people's implicit, evolving beliefs shape what they seek and how they act under uncertainty -- especially in high-stakes settings such as disaster response, emergency medicine, and human-in-the-loop autonomy. Prior approaches either prompt LLMs directly or use latent-state models that treat beliefs as static and independent, often producing incoherent mental models over time and weak reasoning in dynamic contexts. We introduce a structured cognitive trajectory model for LLM-based ToM that represents mental state as a dynamic belief graph, jointly inferring latent beliefs, learning their time-varying dependencies, and linking belief evolution to information seeking and decisions. Our model contributes (i) a novel projection from textualized probabilistic statements to consistent probabilistic graphical model updates, (ii) an energy-based factor graph representation of belief interdependencies, and (iii) an ELBO-based objective that captures belief accumulation and delayed decisions. Across multiple real-world disaster evacuation datasets, our model significantly improves action prediction and recovers interpretable belief trajectories consistent with human reasoning, providing a principled module for augmenting LLMs with ToM in high-uncertainty environment. https://anonymous.4open.science/r/ICML_submission-6373/