Overcoming Multi-step Complexity in Multimodal Theory-of-Mind Reasoning: A Scalable Bayesian Planner

Chunhui Zhang, Zhongyu Ouyang, Kwonjoon Lee, Nakul Agarwal, Sean Dae Houlihan, Soroush Vosoughi, Shao-Yuan Lo
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:75878-75900, 2025.

Abstract

Theory-of-mind (ToM) enables humans to infer mental states—such as beliefs, desires, and intentions—forming the foundation of social cognition. Existing computational ToM methods rely on structured workflows with ToM-specific priors or deep model fine-tuning but struggle with scalability in multimodal environments. They remain trapped within the gravitational pull of multi-step planning complexity, failing to generalize as task demands increase. To overcome these limitations, we propose a scalable Bayesian ToM planner. It breaks down ToM complexity into stepwise Bayesian updates. Meanwhile, weak-to-strong control specializes smaller LMs to refine ToM-specific likelihood estimation, transferring their ToM reasoning behavior to larger LMs (7B to 405B) for social and world knowledge integration. This synergistic approach enables scalability, aligning large-model inference with human mental states with Bayesian principles. Extensive experiments demonstrate a 4.6% improvement in accuracy over state-of-the-art methods on multimodal ToM benchmarks, including unseen scenarios, establishing a new standard for modeling human mental states in complex environments.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-zhang25bk, title = {Overcoming Multi-step Complexity in Multimodal Theory-of-Mind Reasoning: A Scalable {B}ayesian Planner}, author = {Zhang, Chunhui and Ouyang, Zhongyu and Lee, Kwonjoon and Agarwal, Nakul and Houlihan, Sean Dae and Vosoughi, Soroush and Lo, Shao-Yuan}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {75878--75900}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/zhang25bk/zhang25bk.pdf}, url = {https://proceedings.mlr.press/v267/zhang25bk.html}, abstract = {Theory-of-mind (ToM) enables humans to infer mental states—such as beliefs, desires, and intentions—forming the foundation of social cognition. Existing computational ToM methods rely on structured workflows with ToM-specific priors or deep model fine-tuning but struggle with scalability in multimodal environments. They remain trapped within the gravitational pull of multi-step planning complexity, failing to generalize as task demands increase. To overcome these limitations, we propose a scalable Bayesian ToM planner. It breaks down ToM complexity into stepwise Bayesian updates. Meanwhile, weak-to-strong control specializes smaller LMs to refine ToM-specific likelihood estimation, transferring their ToM reasoning behavior to larger LMs (7B to 405B) for social and world knowledge integration. This synergistic approach enables scalability, aligning large-model inference with human mental states with Bayesian principles. Extensive experiments demonstrate a 4.6% improvement in accuracy over state-of-the-art methods on multimodal ToM benchmarks, including unseen scenarios, establishing a new standard for modeling human mental states in complex environments.} }
Endnote
%0 Conference Paper %T Overcoming Multi-step Complexity in Multimodal Theory-of-Mind Reasoning: A Scalable Bayesian Planner %A Chunhui Zhang %A Zhongyu Ouyang %A Kwonjoon Lee %A Nakul Agarwal %A Sean Dae Houlihan %A Soroush Vosoughi %A Shao-Yuan Lo %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-zhang25bk %I PMLR %P 75878--75900 %U https://proceedings.mlr.press/v267/zhang25bk.html %V 267 %X Theory-of-mind (ToM) enables humans to infer mental states—such as beliefs, desires, and intentions—forming the foundation of social cognition. Existing computational ToM methods rely on structured workflows with ToM-specific priors or deep model fine-tuning but struggle with scalability in multimodal environments. They remain trapped within the gravitational pull of multi-step planning complexity, failing to generalize as task demands increase. To overcome these limitations, we propose a scalable Bayesian ToM planner. It breaks down ToM complexity into stepwise Bayesian updates. Meanwhile, weak-to-strong control specializes smaller LMs to refine ToM-specific likelihood estimation, transferring their ToM reasoning behavior to larger LMs (7B to 405B) for social and world knowledge integration. This synergistic approach enables scalability, aligning large-model inference with human mental states with Bayesian principles. Extensive experiments demonstrate a 4.6% improvement in accuracy over state-of-the-art methods on multimodal ToM benchmarks, including unseen scenarios, establishing a new standard for modeling human mental states in complex environments.
APA
Zhang, C., Ouyang, Z., Lee, K., Agarwal, N., Houlihan, S.D., Vosoughi, S. & Lo, S.. (2025). Overcoming Multi-step Complexity in Multimodal Theory-of-Mind Reasoning: A Scalable Bayesian Planner. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:75878-75900 Available from https://proceedings.mlr.press/v267/zhang25bk.html.

Related Material