更多令牌是否理性？语言模型推理时的扩展作为适应性资源理性
Are More Tokens Rational? Inference-Time Scaling in Language Models as Adaptive Resource Rationality
AI总结：
研究通过变量归因任务探讨语言模型在推理时扩展对资源理性的影响，发现模型能根据任务复杂性调整策略，即使无显式成本奖励。
摘要：
人类推理受资源理性影响——在约束下优化性能。最近，推理时的扩展作为一种强大的范式，通过在测试时增加计算量来提高大型语言模型的推理性能。具体而言，指令微调（IT）模型在推理过程中明确生成长推理步骤，而大型推理模型（LRMs）则通过强化学习训练以发现能最大化准确性的推理路径。然而，资源理性是否能从这种扩展中自发产生，而无需与计算成本相关的显式奖励，仍不清楚。我们引入了一个变量归因任务，其中模型根据候选变量、输入输出试验和预定义的逻辑函数推断哪些变量决定结果。通过改变候选变量和试验的数量，我们系统地操纵任务复杂性。两种模型在复杂性增加时都从暴力策略转向分析策略。IT模型在XOR和XNOR函数上表现下降，而LRMs保持稳健。这些发现表明，模型可以根据任务复杂性调整其推理行为，即使没有显式的基于成本的奖励。这提供了强有力的证据，表明资源理性是推理时扩展本身的一种涌现属性。
英文摘要：
Human reasoning is shaped by resource rationality -- optimizing performance under constraints. Recently, inference-time scaling has emerged as a powerful paradigm to improve the reasoning performance of Large Language Models by expanding test-time computation. Specifically, instruction-tuned (IT) models explicitly generate long reasoning steps during inference, whereas Large Reasoning Models (LRMs) are trained by reinforcement learning to discover reasoning paths that maximize accuracy. However, it remains unclear whether resource-rationality can emerge from such scaling without explicit reward related to computational costs. We introduce a Variable Attribution Task in which models infer which variables determine outcomes given candidate variables, input-output trials, and predefined logical functions. By varying the number of candidate variables and trials, we systematically manipulate task complexity. Both models exhibit a transition from brute-force to analytic strategies as complexity increases. IT models degrade on XOR and XNOR functions, whereas LRMs remain robust. These findings suggest that models can adjust their reasoning behavior in response to task complexity, even without explicit cost-based reward. It provides compelling evidence that resource rationality is an emergent property of inference-time scaling itself.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
