Learning Perception to Action Rules through Reinforcement Learning for Theory-of-Mind Reasoning
In Theory-of-Mind (ToM) reasoning, large language models may correctly infer the agent's mental state but fail to use that inference when predicting or judging the agent's subsequent actions.
ProofPaper ↗
Key points
- In this work, we investigate whether explicitly connecting perception, belief, and action with rules improves ToM reasoning.
- We generate concise if--then rules that express relations between perception to action and use Group Relative Policy Optimization (GRPO) to train Qwen2.5 and Llama3 models across four ToM benchmarks.
- GRPO with rule generation outperforms prompting, supervised fine-tuning, and think reasoning.
- These findings suggest that explicit rules help models bridge the reasoning gap in ToM and support knowledge transfer across models.
Sources (1)
- [1]Learning Perception to Action Rules through Reinforcement Learning for Theory-of-Mind ReasoningarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 14, 08:40 AM
In Theory-of-Mind (ToM) reasoning, large language models may correctly infer the agent's mental state but fail to use that inference when predicting or judging the agent's subsequent actions.
In this work, we investigate whether explicitly connecting perception, belief, and action with rules improves ToM reasoning.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 14, 2026ForkGRPO: Efficient Exploration and Fine-Grained Credit Assignment for Diffusion Language Models
- Oct 14, 2026The Design Space of Tri-Modal Masked Diffusion Models
- Oct 13, 2026Run fast, private agents on-device with LFM2.5-24B-A2B
- Oct 13, 2026Fish Audio Drama 3 Technical Report
- Oct 13, 2026Agent-Reflex: Teaching Language Agents to Act Reflectively
- Oct 13, 2026Cowork-Gym: Synthesizing Interactive Work Environments at Scale