首页 > AI前沿 > Rewarding Efficient Reasoning Improves Abstention on Underspecified Tasks in Reasoning Models

Rewarding Efficient Reasoning Improves Abstention on Underspecified Tasks in Reasoning Models

arXiv自然语言 2026-08-06 04:40 6 阅读 查看原文

While modern large reasoning models (LRMs) excel at providing correct answers in many tasks, we provide additional evidence for the observation that they often struggle with a critical capability: knowing when to abstain from answering.

We analyze this gap by comparing LRM behavior to results from a human study, revealing that human reasoning effort on unanswerable tasks is upper-bounded by answerable tasks, whereas LRMs waste computational resources by generating longer Chains of Thought (CoTs) on unanswerable than on answerable prompts.

To overcome this inefficiency, we take inspiration from a resource-rational perspective on human cognition and introduce a novel GRPO reward that encourages efficient reasoning about whether the task contains all the information needed to solve it.

Fine-tuning several 4B LRMs with this reward leads to human-like abstention performance gains (+12.8% on average) while retaining answering capabilities and boosting the models' efficiency (44% shorter CoTs on average).