arXiv cs.AI· Hanbo Xie, Yue Zhao, Robert C. Wilson·· 3 天前AI 评分31
超越成功对弈：诊断大语言模型中的对手建模能力
Beyond Successful Play: Diagnosing Opponent Modeling in Large Language Models
AI 导读
研究者提出一个针对重复捉迷藏博弈的基准与诊断框架，用具有明确递归信念更新的算法对手，通过实时交互、固定历史评估和受控信息干预区分自适应对弈、动作预测与信念追踪。在 Qwen2.5-7B-Instruct 上，告知对手类型与参数设置可提升预测，但需从历史推断对手时准确率下降；GPT-5.4 同样受益于正确对手信念，而显式信念报告与自生成记忆未能稳定改善预测。
正文 · 原文
Abstract:Language-model agents are expected to adapt to others whose behavior changes through interaction. Yet adaptation may rely on learned behavioral patterns rather than theory of mind (ToM): the ability to infer others' mental states, including beliefs and intentions. We introduce a benchmark and diagnostic framework for repeated hide-and-seek against algorithmic opponents with explicitly defined recursive belief updates. Their ground-truth beliefs and action probabilities let us distinguish adaptive play, action prediction, and belief tracking through live interaction, fixed-history evaluation, and controlled information interventions. With structured elicitation and behavioral supervised fine-tuning (SFT), we find improved play in several tested settings, but diagnostics reveal limitations in opponent prediction and belief tracking. We test whether supervising opponent action probabilities can address this predictive deficit in Qwen2.5-7B-Instruct. Predictions improve when the model is told the opponent's type and parameter settings, but become less accurate when it must infer the opponent from history without this information. Training on examples that also omit these details reduces the gap, although improvements vary across opponents. Is this limitation specific to smaller open-weight models, or does it also appear in a frontier model? GPT-5.4 also benefits from correct opponent beliefs: supplying them improves its predictions and choices, whereas deeper explicit belief reports and self-generated memory do not consistently improve forecasting from interaction history. This contrast suggests that using useful belief information is easier than inferring it autonomously, even for a frontier model. Together, our findings distinguish improved play and action prediction from reliable belief tracking, providing a framework for evaluating mentalizing and testing targeted improvements.
| Subjects: | Artificial Intelligence (cs.AI); Neurons and Cognition (q-bio.NC) |
| Cite as: | arXiv:2610.17310 [cs.AI] |
| (or arXiv:2610.17310v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.17310 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Hanbo Xie [view email]
[v1]
Mon, 12 Oct 2026 19:08:43 UTC (4,098 KB)
来源：arXiv cs.AI · arxiv.org