AI 中文总结
研究大语言模型中因过早自我怀疑导致的上下文焦虑，通过系统研究发现其部分源于对完成任务所需tokens估计不准，会致效率损失，还表明模型可学替代策略，性能提升或可通过提高评估与适应自身局限能力实现。
AI 中文摘要
传统观点认为，当问题超出其能力范围时，推理模型就会失败。然而，我们发现前沿推理模型有时具备解决问题的必要能力，但却因过早的自我怀疑而失败——这种现象被非正式地称为上下文焦虑。我们首次对上下文焦虑进行了系统研究，表明它部分源于模型无法准确估计完成一项任务所需的 tokens。我们还表明，当模型在感知到的约束下运行时，上下文焦虑会导致显著的效率损失。基于此分析，我们进一步表明，模型可以学习解决长周期问题的替代策略而不表现出上下文焦虑，这表明性能提升可能不是通过扩展模型能力，而是通过提高模型准确评估和适应自身局限性的能力来实现。
英文摘要
Conventional wisdom suggests that reasoning models fail when problems exceed their capabilities. However, we find that frontier reasoning models sometimes possess the necessary capabilities to solve problems but fail due to premature self-doubt -- a phenomenon informally known as context anxiety. We provide the first systematic study of context anxiety, demonstrating that it arises, in part, from a model's inability to accurately estimate the tokens required to complete a task. We also show that context anxiety leads to material efficiency losses when models operate under perceived constraints. Building on this analysis, we further show that models can learn alternative strategies for solving long-horizon problems without exhibiting context anxiety, suggesting that performance improvements may be achievable not through scaling model capabilities, but by improving models' ability to accurately assess and adapt to their own limitations.
CommentsAccepted at ICML 2026. 17 pages
