Think Fast and Slow: Step-Level Cognitive Depth Adaptation for LLM Agents
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866918335762923520 |
|---|---|
| author | Yang, Ruihan Ye, Fanghua We, Xiang Zhao, Ruoqing Luo, Kang Xu, Xinbo Zhao, Bo Ma, Ruotian Wang, Shanyi Tu, Zhaopeng Li, Xiaolong Yang, Deqing Linus |
| author_facet | Yang, Ruihan Ye, Fanghua We, Xiang Zhao, Ruoqing Luo, Kang Xu, Xinbo Zhao, Bo Ma, Ruotian Wang, Shanyi Tu, Zhaopeng Li, Xiaolong Yang, Deqing Linus |
| contents | Large language models (LLMs) are increasingly deployed as autonomous agents for multi-turn decision-making tasks. However, current agents typically rely on fixed cognitive patterns: non-thinking models generate immediate responses, while thinking models engage in deep reasoning uniformly. This rigidity is inefficient for long-horizon tasks, where cognitive demands vary significantly from step to step, with some requiring strategic planning and others only routine execution. In this paper, we introduce CogRouter, a framework that trains agents to dynamically adapt cognitive depth at each step. Grounded in ACT-R theory, we design four hierarchical cognitive levels ranging from instinctive responses to strategic planning. Our two-stage training approach includes Cognition-aware Supervised Fine-tuning (CoSFT) to instill stable level-specific patterns, and Cognition-aware Policy Optimization (CoPO) for step-level credit assignment via confidence-aware advantage reweighting. The key insight is that appropriate cognitive depth should maximize the confidence of the resulting action. Experiments on ALFWorld and ScienceWorld demonstrate that CogRouter achieves state-of-the-art performance with superior efficiency. With Qwen2.5-7B, it reaches an 82.3% success rate, outperforming GPT-4o (+40.3%), OpenAI-o3 (+18.3%), and GRPO (+14.0%), while using 62% fewer tokens. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_12662 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Think Fast and Slow: Step-Level Cognitive Depth Adaptation for LLM Agents Yang, Ruihan Ye, Fanghua We, Xiang Zhao, Ruoqing Luo, Kang Xu, Xinbo Zhao, Bo Ma, Ruotian Wang, Shanyi Tu, Zhaopeng Li, Xiaolong Yang, Deqing Linus Artificial Intelligence Computation and Language Large language models (LLMs) are increasingly deployed as autonomous agents for multi-turn decision-making tasks. However, current agents typically rely on fixed cognitive patterns: non-thinking models generate immediate responses, while thinking models engage in deep reasoning uniformly. This rigidity is inefficient for long-horizon tasks, where cognitive demands vary significantly from step to step, with some requiring strategic planning and others only routine execution. In this paper, we introduce CogRouter, a framework that trains agents to dynamically adapt cognitive depth at each step. Grounded in ACT-R theory, we design four hierarchical cognitive levels ranging from instinctive responses to strategic planning. Our two-stage training approach includes Cognition-aware Supervised Fine-tuning (CoSFT) to instill stable level-specific patterns, and Cognition-aware Policy Optimization (CoPO) for step-level credit assignment via confidence-aware advantage reweighting. The key insight is that appropriate cognitive depth should maximize the confidence of the resulting action. Experiments on ALFWorld and ScienceWorld demonstrate that CogRouter achieves state-of-the-art performance with superior efficiency. With Qwen2.5-7B, it reaches an 82.3% success rate, outperforming GPT-4o (+40.3%), OpenAI-o3 (+18.3%), and GRPO (+14.0%), while using 62% fewer tokens. |
| title | Think Fast and Slow: Step-Level Cognitive Depth Adaptation for LLM Agents |
| topic | Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2602.12662 |