想测 AI 到底懂不懂人心？这个基准用阿瓦隆桌游把心智理论拆成细粒度任务，28 个模型都栽了。
Avalon-ToM-Bench 将心智理论拆分为 2×2 分类法，涵盖认知与动机推理、推断与行动。对 28 个大语言模型的测试显示，模型对游戏规则理解良好，但心智理论能力明显偏弱。线性探测从隐藏状态恢复心智判断的准确率达 77-82%，高于模型自身思维链的 62-70%。专门的推理训练平均带来 +11.0 分提升，而测试时思维链仅 +1.1 分。
Avalon-ToM-Bench: Evaluating Fine-Grained Theory of Mind via Asymmetric Game Mechanics
Theory of Mind (ToM) is essential for agent interactions, yet existing evaluations either rely on static scenarios that oversimplify mental-state reasoning or interactive settings that provide limited diagnostic insight. We present Avalon-ToM-Bench, a fine-grained benchmark that operationalizes ToM through the asymmetric-information mechanics of The Resistance: Avalon. Rather than evaluating end-to-end gameplay, it decomposes ToM into a 2$\times$2 taxonomy -- epistemic versus motivational reasoning crossed with inference versus action -- using human-crafted, perspective-constrained queries. Benchmarking 28 LLMs reveals three insights: 1) Reasoning, not knowledge. Models show strong game-rule comprehension but markedly weaker ToM abilities, isolating failures to social reasoning rather than missing domain knowledge. 2) Expression, not representation. Mechanistic analyses via linear probing and activation steering show that models frequently represent correct mental-state inferences in their hidden states but fail to express them during generation -- linear probes recover 77-82% accuracy versus 62-70% from the models' own chain-of-thought. 3) Policy, not deliberation. Dedicated reasoning training yields substantial improvements whereas test-time chain-of-thought provides only marginal gains (+11.0 versus +1.1 points on average), suggesting that robust ToM depends on a learned reasoning policy rather than increased inference-time deliberation.