Pith. sign in

REVIEW 4 major objections 4 minor 34 references

ProToM: Promoting Prosocial Behaviour via Theory of Mind-Informed Feedback

T0 review · 4 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read An AI that observes, rather than joins, a group can nudge people toward prosocial actions by sending rare, goal-grounded feedback.

desk verdict A genuinely new facilitator paradigm with a real human study, but the simulation results are largely internal to the method's own assumptions. read the letter →

arxiv 2509.05091 v1 pith:3GT7YZ4Y submitted 2025-09-05 cs.AI cs.MA

Matteo Bortoletto , Yichao Zhou , Lance Ying , Tianmin Shu , Andreas Bulling This is my paper · ORCID
classification cs.AIcs.MA
keywords prosocialbehaviourTheoryofMindBayesianinverseplanningmulti-agentsystemsfacilitatoragentfeedbackgenerationhuman-AIinteractionexpectedutility
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper asks whether an AI can stand outside a group of humans who have different goals and, simply by sending a few words of feedback, get them to help one another. It proposes ProToM, which reads each agent's goals from observed behaviour with Bayesian inverse planning, then scores each possible message by how many steps it would save if obeyed, and sends it only if it would change the agent's plan and is expected to help. In simulation on two multi-agent environments, ProToM reached perfect success and clearly faster completion with much less chatter than GPT-4o and other large vision-language and reasoning models. An 18-person Overcooked study found the same pattern: humans succeeded more, finished faster, and rated ProToM's feedback as more helpful and better explained. The claim, if right, is that prosocial behaviour can be promoted by a purely observational AI, and that Bayesian goal inference plus expected-utility message selection does this better than end-to-end language reasoning.

What carries the argument

The load-bearing machinery is Bayesian inverse planning over a particle-filtered belief of each agent, feeding an expected-utility message selector. Inverse planning means working backwards from observed actions to a distribution over goals; the particle filter lets the facilitator track what each agent believes while only seeing their observations. The selector evaluates each candidate message by simulating the agent's plan with and without it under sampled goals and scoring the reduction in steps-to-goal; a divergence gate suppresses messages that would not alter the agent's behaviour, which is why ProToM communicates rarely. The formal wrapper is a two-level POMDP: the inner level is the

What would settle it

Run the same Overcooked comparison with human participants told truthfully that the facilitator's suggestions are sometimes wrong, and track how often they follow them; if the perfect success rate and speedup over no-feedback disappear once users start ignoring messages, the compliance assumption is the load-bearing part. A cheaper check: replace the simulated agent's planner with another planner of similar competence and see whether ProToM's utility scores still rank feedback correctly.

Watch

Extended reading notes

Core claim

ProToM claims that an AI facilitator needs no shared goal with the people it helps. It treats the observed agents as the hidden state of a POMDP, infers each agent's goal as a belief distribution via Bayesian inverse planning, and scores candidate feedback messages by the expected reduction in the number of steps to the agents' goals, assuming the recipient follows the message. It communicates only when the best message clears a utility threshold and the agent's predicted trajectory with feedback diverges from the one without. In the paper's evaluations, this produces perfect success rates in both mDKG and Overcooked, substantially faster completion than no feedback, and fewer messages than

Load-bearing premise

The load-bearing premise is that a person who receives feedback follows it and then plans exactly like the simulator ProToM uses to score the feedback.

Editorial extensions

If this is right

  • An AI that never takes actions in the environment can still improve multi-agent success and speed by sending a small number of goal-grounded messages.
  • Feedback should be withheld unless it changes predicted behaviour; over-communication, the failure mode of the tested large vision-language and reasoning models, reduces or reverses the benefit.
  • Explicit goal inference plus expected-utility selection is sufficient to beat those models on these two grid-world benchmarks, suggesting that modular Theory of Mind is a viable alternative to end-to-end language reasoning for real-time assistance.
  • Pairing each message with an explanation of the other agent's inferred goal matters: human participants rated ProToM higher on clarity and goal understanding and ignored fewer of its messages.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same expected-utility gate could decide when to interrupt in non-text assistive systems, such as highlighting a relevant object in an interface, because the feedback set is discrete and grounded in the environment state.
  • A natural next experiment is to make the facilitator adaptive: learn each user's compliance rate and the planner they actually use, then re-rank messages under that empirical model; ProToM's current utility computation assumes obedience and the paper's own planner.
  • Because the human study used 18 participants and one LLM baseline, the sharpest testable claim is whether the advantage survives with users who have reason to distrust the facilitator; that would isolate the compliance assumption from message quality.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. This paper introduces ProToM, a facilitator that observes two agents pursuing independent goals, maintains a Bayesian belief over their goals via inverse planning, evaluates candidate feedback messages by expected utility (defined as step-count reduction under the agents' planners), and communicates only when utility and divergence thresholds are met. ProToM is evaluated in mDKG and Overcooked with simulated agents and compared with VLMs/RMs, and in an Overcooked human study with 18 participants (9 pairs) against GPT-4o and no-facilitator controls. The paper reports higher success rates, larger speedups, lower communication overhead, and higher subjective ratings for ProToM, and concludes that ProToM provides targeted, helpful feedback.

Significance. If the claim is established, the paper makes a useful contribution: it formalizes a third role for AI beyond human-AI coordination and team intervention, and it offers a modular theory-of-mind method that is more transparent than LLM-only feedback. Among the paper's strengths are the public code, the explicit two-level POMDP formulation, and a human study that goes beyond simulation. The core limitation is that the simulation evidence is close to circular; because the utility model and the simulated agents use the same planners, the simulated speedups are largely built into the objective. The human study is the only independent evidence, and it is small and partially confounded by the explanation templates. The conclusion as written overstates the support.

major comments (4)
  1. [§3.2, Eq. (12)–(14); §4.4] The load-bearing simulation evidence is self-consistency rather than independent validation. U(f) in Eqs. (12)–(13) is the expected reduction in completion cost C(f,g)=ℓ(π|f,g) (Eq. 14) under the same planner π that is used to simulate the agents in §4.4 (A* in mDKG and the heuristic planner in Overcooked). Section 3.2 further assumes the agent follows the feedback. Thus a positive utility almost mechanically implies that the simulated 'human' agent, which is π executing the selected feedback, completes the task faster. The simulation therefore cannot establish that ProToM's goal inference or feedback selection transfers to people. I recommend either evaluating with agents whose policies differ from the planner used in U(·) (e.g., noisy, model-mismatch, or independently trained agents), or explicitly presenting the simulation as a sanity check and placing the burden on the human study.
  2. [§4.2 vs. §B.1] Threshold tuning is reported inconsistently, which undermines reproducibility. Section 4.2 states that for Overcooked ProToM uses φ=2 (and ε=0.3), while §B.1 says φ and ε were chosen by a small search on a held-out set with values from 0 to 1.0 in increments of 0.1. A threshold of 2 is outside the stated search range. Please correct the reported value or the search range; otherwise the reader cannot reconstruct the experiment, and the selection of φ=2 appears post hoc.
  3. [§4.5; Eq. (17), §B.2] The only independent evidence for the central claim comes from the human study, but that study conflates the feedback-selection mechanism with the explanation templates. ProToM's communicated feedback is always accompanied by an explanation that explicitly states the inferred goal of the other agent (Eq. 17, templates in §B.2). Participants' high ratings on 'understood my goals' and 'helpfulness' could therefore be driven by the goal-revealing explanation rather than by the expected-utility selection. Since the LLM baseline is also asked for explanations, a cleaner comparison would require an ablation that delivers the same explanations with random or oracle-selected feedback, or that removes explanations from both arms.
  4. [§4.5] The human study is small: 18 participants, 9 pairs, one environment, one LLM baseline, and two trials per condition. With these numbers, the reported speedup differences (0.75±0.17 vs −0.28±0.11) may be real but are fragile; the study is also the only evidence that real humans comply with feedback well enough for the utility model to transfer. The conclusions in §6 that ProToM 'provides targeted and helpful feedback' and is 'consistently preferred by human users' go beyond what this sample supports. Please either expand the study, report effect sizes and power, or temper the scope of the claims.
minor comments (4)
  1. [Abstract; §1] The phrase 'higher communication overhead and task speedup' appears to be a typo; presumably 'lower task speedup' or 'reduced speedup' is intended.
  2. [§4.3] Please clarify in the text or figures which exact statistical test and per-condition sample sizes correspond to the significance markers, especially in Figure 6 where the perceived-rating comparisons are reported.
  3. [Algorithm 1] The initialisation of the belief particles is not described in the text. A brief sentence on the initial proposal distribution would help reproducibility.
  4. [§4.4] Consider using 'simulated agents' rather than 'human agents' for the A* and heuristic planners, to avoid the impression that the simulation involved actual human participants.

Circularity Check

1 steps flagged · score 6.0 of 10

Simulation is a closed loop: feedback is selected by expected step-count reduction under planners π, and the simulated agents are those same π planners that follow the feedback, so the reported speedup/success largely reduce to the maximized utility; only the 18-participant human study provides external support.

  1. self definitional [Section 3.2, Eqs. (12)-(14); Section 4.3 (Speedup metric); Section 4.4 (Simulation Experiments)]
    "U(f) =E_{g∼ˆbG}[∆C(f,g)] (12) ... ∆C(f,g)=C(∅,g)−C(f,g) (13) ... C(f,g)=ℓ(π|f,g) (14). In our case, ℓ(·) corresponds to the number of steps to achieve g ... Speedup = L∅/Lf −1 ... We assume that agent i will follow the feedback upon receiving it ... The two human agents are simulated using the A* planner from (Zhi-Xuan et al. 2024). In Overcooked, ... Human agents are simulated using a stochastic heuristic-based planner."

    Feedback selection maximizes U(f)=E[ΔC(f,g)], where ΔC=C(∅)−C(f) and C(f,g)=ℓ(π|f,g) is the step-count reduction predicted by the planners π. The simulation's headline metric Speedup=L∅/Lf−1 is exactly ΔC/C(f) for those same planners. In §4.4 the simulated 'human' agents are the same planner class (A* in mDKG, stochastic heuristic in Overcooked), and §3.2 assumes the agent follows the feedback. Therefore a positive-ΔC message executed by π produces the observed speedup/success by construction; the simulation results are the optimized objective re-measured, not an independent test of whether the model matches human planning or compliance. Only the human study (where participants could ignore feedback) provides external evidence.

full rationale

The principal circularity is confined to the simulation evaluation. The feedback-selection objective (Eqs. 12–14) is the step-count reduction estimated with the same planner models π that later act as the simulated human agents (§4.4) and that are assumed to follow the feedback (§3.2). Thus the reported simulation speedup and perfect success rate are largely the optimization objective re-measured—a self-consistency check, not evidence of general human-agent transfer. This warrants a 6 rather than 0. The paper does contain independent content: the comparison with ProToM-Oracle tests goal inference (though on data generated from the assumed model), and the human study with 18 participants, in which participants were free to ignore feedback, grounds the central claim externally. The human study is small and uses one LLM baseline, but those are correctness/limitation concerns, not circularity. The paper's own limitations paragraph (Section 6) states that ProToM has not been tested in real-world settings, which further limits the strength of the simulation-based claims but does not change the circularity analysis. Citations to co-authored prior work (Zhi-Xuan et al. 2024; Ying et al. 2024) supply standard environments, planners, and ToM machinery; no load-bearing claim is justified solely by a self-citation or an imported uniqueness theorem, so they do not raise the score. Separately, a non-circular reproducibility issue: §4.2 sets φ=2 for Overcooked while §B.1 reports the threshold search range as 0–1.0; this does not affect the circularity verdict.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

No new physical or ontological entities are introduced; the facilitator, feedback templates, and explanation templates are components of the system, not independently evidenced postulates.

free parameters (3)
  • Utility threshold phi = 0 (mDKG), 2 (Overcooked)
    Minimum expected utility for a message to be considered; tuned via held-out search (Appendix B.1), controls how often the facilitator speaks.
  • Divergence threshold epsilon = 0.1 (mDKG), 0.3 (Overcooked)
    Minimum divergence between simulated behavior with and without feedback; tuned on held-out set; controls timing of feedback.
  • Belief particle count N = 1 (mDKG), 5 (Overcooked)
    Number of particles per agent; N=5 chosen from inference-speed considerations for real-time use, not accuracy (Appendix B.1).
assumptions (6)
  • domain assumption Agents are approximately rational planners with known policies pi_i and known goal sets G_i.
    Section 2 Eqs. 2-4 and Section 3.1; inverse planning and utility evaluation both require the facilitator to know the agents' planning model.
  • domain assumption Agents follow ProToM's feedback until it is completed or becomes infeasible.
    Section 3.2 feedback selection; utility computation assumes compliance, and simulation agents comply by construction.
  • domain assumption The set of possible feedback messages is finite, hand-authored, environment-specific, and constructed only from state-dependent templates.
    Section 3.2 and Appendix A.2; this bounds the communication space and is not learned or evaluated against free-form language.
  • domain assumption The goal prior P(g_i) and observation model for each agent are known inputs.
    Problem formulation Eqs. 2-3; no estimation of priors from data is described.
  • domain assumption Simulated human agents (A* planner in mDKG, stochastic heuristic planner in Overcooked) approximate real human behavior closely enough for simulation conclusions.
    Section 4.4; all simulation claims about 'humans' depend on this.
  • standard math Bayesian inverse planning and particle filtering provide adequate posterior approximations with the chosen number of particles.
    Section 3.1; standard techniques, but the N=5 choice for Overcooked is heuristic.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ProToM: Promoting Prosocial Behaviour via Theory of Mind-Informed Feedback." pith.science (2026). https://pith.science/paper/3GT7YZ4Y

@misc{pith2026250905091,
  author       = {Pith},
  title        = {Pith review of: ProToM: Promoting Prosocial Behaviour via Theory of Mind-Informed Feedback},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3GT7YZ4Y}},
  note         = {Machine review of arXiv:2509.05091}
}
read the original abstract

While humans are inherently social creatures, the challenge of identifying when and how to assist and collaborate with others - particularly when pursuing independent goals - can hinder cooperation. To address this challenge, we aim to develop an AI system that provides useful feedback to promote prosocial behaviour - actions that benefit others, even when not directly aligned with one's own goals. We introduce ProToM, a Theory of Mind-informed facilitator that promotes prosocial actions in multi-agent systems by providing targeted, context-sensitive feedback to individual agents. ProToM first infers agents' goals using Bayesian inverse planning, then selects feedback to communicate by maximising expected utility, conditioned on the inferred goal distribution. We evaluate our approach against baselines in two multi-agent environments: Doors, Keys, and Gems, as well as Overcooked. Our results suggest that state-of-the-art large language and reasoning models fall short of communicating feedback that is both contextually grounded and well-timed - leading to higher communication overhead and task speedup. In contrast, ProToM provides targeted and helpful feedback, achieving a higher success rate, shorter task completion times, and is consistently preferred by human users.

Figures

Figures reproduced from arXiv: 2509.05091 by the authors.

Figure 1
Figure 1. In contrast to the dominant paradigms of human [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of ProToM, a Theory of Mind-informed facilitator that promotes prosocial behaviour by communicating [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Simulation results comparing ProToM to baseline models on mDKG and Overcooked. ProToM achieves perfect [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Example comparing ProToM and o3 on mDKG. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Example comparing ProToM, Gemini 2.5 Pro, and [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 26 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Anthropic. 2025. Introducing Claude 4. https://www.anthropic.com/news/claude-4

    2025

  4. [4]

    Bai, S.; Chen, K.; Liu, X.; Wang, J.; Ge, W.; Song, S.; Dang, K.; Wang, P.; Wang, S.; Tang, J.; et al. 2025. Qwen2.5-VL technical report. arXiv preprint arXiv:2502.13923

    arXiv 2025

  5. [5]

    Bara, C.-P.; Ma, Z.; Yu, Y.; Shah, J.; and Chai, J. 2023. Towards collaborative plan acquisition through theory of mind modeling in situated dialogue. In Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence, 2958--2966

    work page 2023

  6. [6]

    Binns, R. 2018. Fairness in machine learning: Lessons from political philosophy. In Conference on fairness, accountability and transparency, 149--159. PMLR

    work page 2018

  7. [7]

    Bortoletto, M.; Ruhdorfer, C.; Abdessaied, A.; Shi, L.; and Bulling, A. 2024 a . Limits of Theory of Mind Modelling in Dialogue-Based Collaborative Plan Acquisition. In Proc. 62nd Annual Meeting of the Association for Computational Linguistics (ACL), 1--16

    work page 2024

  8. [8]

    Bortoletto, M.; Ruhdorfer, C.; Shi, L.; and Bulling, A. 2024 b . Explicit Modelling of Theory of Mind for Belief Prediction in Nonverbal Social Interactions. In Proc. 27th European Conference on Artificial Intelligence (ECAI), 866--873

    work page 2024

Show all 34 references
  1. [9]

    C.; and Weisswange, T

    Buehler, M. C.; and Weisswange, T. H. 2020. Theory of mind based communication for human agent cooperation. In 2020 IEEE international conference on human-machine systems (ICHMS), 1--6. IEEE

    2020

  2. [10]

    Carattini, S.; and Roesti, M. 2023. Trust, happiness, and pro-social behavior. Review of Economics and Statistics, 1--45

    2023

  3. [11]

    K.; Griffiths, T.; Seshia, S.; Abbeel, P.; and Dragan, A

    Carroll, M.; Shah, R.; Ho, M. K.; Griffiths, T.; Seshia, S.; Abbeel, P.; and Dragan, A. 2019. On the utility of learning about humans for human-ai coordination. Advances in neural information processing systems, 32

    2019

  4. [12]

    D.; Desai, R.; Hlavac, M.; Karashchuk, V.; Krantz, J.; Mottaghi, R.; Parashar, P.; et al

    Chang, M.; Chhablani, G.; Clegg, A.; Cote, M. D.; Desai, R.; Hlavac, M.; Karashchuk, V.; Krantz, J.; Mottaghi, R.; Parashar, P.; et al. 2024. PARTNR: A Benchmark for Planning and Reasoning in Embodied Multi-agent Tasks. arXiv preprint arXiv:2411.00081

    2024 arXiv

  5. [13]

    Comanici, G.; Bieber, E.; Schaekermann, M.; Pasupat, I.; Sachdeva, N.; Dhillon, I.; Blistein, M.; Ram, O.; Zhang, D.; Rosen, E.; et al. 2025. Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities. arXiv ...

    2025 arXiv

  6. [14]

    L.; and Haber, N

    Cross, L.; Xiang, V.; Bhatia, A.; Yamins, D. L.; and Haber, N. 2025. Hypothetical Minds: Scaffolding Theory of Mind for Multi-Agent Tasks with Large Language Models. In The Thirteenth International Conference on Learning Representations

    2025

  7. [15]

    F.; Saad, F

    Cusumano-Towner, M. F.; Saad, F. A.; Lew, A. K.; and Mansinghka, V. K. 2019. Gen: A General-purpose Probabilistic Programming System with Programmable Inference. In Proceedings of the 40th ACM SIGPLAN Conference on Programming Language Design and Implementation, PLDI 2019, 221...

    2019

  8. [16]

    Kakulte, A.; and Shaikh, S. 2023. Prosocial behavior, psychological well-being, positive and negative affect among young adults: A cross-sectional study. Industrial Psychiatry Journal, 32(Suppl 1): S127--S130

    2023

  9. [17]

    P.; and Murthy, A

    Kambhampati, S.; Valmeekam, K.; Guan, L.; Verma, M.; Stechly, K.; Bhambri, S.; Saldyt, L. P.; and Murthy, A. B. 2024. Position: LLM s Can t Plan, But Can Help Planning in LLM -Modulo Frameworks. In Forty-first International Conference on Machine Learning

    2024

  10. [18]

    OpenAI. 2024. GPT-4o System Card. https://openai.com/index/gpt-4o-system-card/

    2024

  11. [19]

    OpenAI. 2025. Introducing OpenAI o3 and o4-mini. https://openai.com/index/introducing-o3-and-o4-mini/

    2025

  12. [20]

    Premack, D.; and Woodruff, G. 1978. Does the chimpanzee have a theory of mind? Behavioral and brain sciences, 1(4): 515--526

    1978

  13. [21]

    B.; Fidler, S.; and Torralba, A

    Puig, X.; Shu, T.; Li, S.; Wang, Z.; Liao, Y.-H.; Tenenbaum, J. B.; Fidler, S.; and Torralba, A. 2021. Watch-And-Help: A Challenge for Social Perception and Human- \ AI \ Collaboration. In International Conference on Learning Representations

    2021

  14. [22]

    B.; and Torralba, A

    Puig, X.; Shu, T.; Tenenbaum, J. B.; and Torralba, A. 2023. Nopa: Neurally-guided online probabilistic assistance for building socially intelligent home assistants. In 2023 IEEE International Conference on Robotics and Automation (ICRA), 7628--7634. IEEE

    2023

  15. [23]

    Ruhdorfer, C.; Bortoletto, M.; and Bulling, A. 2025. The Yokai Learning Environment: Tracking Beliefs Over Space and Time. arXiv preprint arXiv:2508.12480

    2025

  16. [24]

    Seo, S.; Han, B.; and Unhelkar, V. 2023. Automated Task-Time Interventions to Improve Teamwork using Imitation Learning. In Proceedings of the 2023 International Conference on Autonomous Agents and Multiagent Systems, 335--344

    2023

  17. [25]

    H.; Zhou, X.; Choi, Y.; Goldberg, Y.; Sap, M.; and Shwartz, V

    Shapira, N.; Levy, M.; Alavi, S. H.; Zhou, X.; Choi, Y.; Goldberg, Y.; Sap, M.; and Shwartz, V. 2024. Clever Hans or Neural Theory of Mind? Stress Testing Social Reasoning in Large Language Models. In 18th Conference of the European Chapter of the Association for Computational...

    2024

  18. [26]

    Ullman, T.; Baker, C.; Macindoe, O.; Evans, O.; Goodman, N.; and Tenenbaum, J. 2009. Help or hinder: Bayesian models of social goal inference. Advances in neural information processing systems, 22

    2009

  19. [27]

    A.; Wang, R

    Wu, S. A.; Wang, R. E.; Evans, J. A.; Tenenbaum, J. B.; Parkes, D. C.; and Kleiman-Weiner, M. 2021. Too many cooks: Bayesian inference for coordinating multi-agent collaboration. Topics in Cognitive Science, 13(2): 414--432

    2021

  20. [28]

    B.; Torralba, A.; and Shu, T

    Ying, L.; Jha, K.; Aarya, S.; Tenenbaum, J. B.; Torralba, A.; and Shu, T. 2024. Goma: Proactive embodied cooperative communication via goal-oriented mental alignment. In 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 7099--7106. IEEE

    2024

  21. [29]

    Yu, C.; Serhan, B.; and Cangelosi, A. 2024. Top-tom: Trust-aware robot policy with theory of mind. In 2024 IEEE International Conference on Robotics and Automation (ICRA), 7888--7894. IEEE

    2024

  22. [30]

    Zhang, C.; Yang, K.; Hu, S.; Wang, Z.; Li, G.; Sun, Y.; Zhang, C.; Zhang, Z.; Liu, A.; Zhu, S.-C.; et al. 2024 a . Proagent: building proactive cooperative agents with large language models. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, 17591--17599

    2024

  23. [31]

    Zhang, Y.; Robertson, P.; Shu, T.; Hong, S.; and Williams, B. C. 2024 b . Risk-Bounded Online Team Interventions via Theory of Mind. In 2024 IEEE International Conference on Robotics and Automation (ICRA), 12964--12970. IEEE

    2024

  24. [32]

    Y.; and Shu, T

    Zhang, Z.; Jin, C.; Jia, M. Y.; and Shu, T. 2025. Autotom: Automated bayesian inverse planning and model discovery for open-ended theory of mind. arXiv preprint arXiv:2502.15676

    2025

  25. [33]

    Zhi-Xuan, T. 2022. Pddl.jl: An extensible interpreter and compiler interface for fast and flexible ai planning. Ph.D. thesis, Massachusetts Institute of Technology

    2022

  26. [34]

    Zhi-Xuan, T.; Ying, L.; Mansinghka, V.; and Tenenbaum, J. B. 2024. Pragmatic Instruction Following and Goal Assistance via Cooperative Language-Guided Inverse Planning. In AAMAS

    2024

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.