Understanding Reasoning in LLMs through Strategic Information Allocation under Uncertainty
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866918524460466176 |
|---|---|
| author | Kim, Jeonghye Luo, Xufang Kim, Minbeom Lee, Sangmook Li, Dongsheng Yang, Yuqing |
| author_facet | Kim, Jeonghye Luo, Xufang Kim, Minbeom Lee, Sangmook Li, Dongsheng Yang, Yuqing |
| contents | LLMs often exhibit Aha moments such as self-correction after tokens like "Wait," yet the underlying mechanism remains unclear. Standard LLMs collapse mainly through silent divergence, where trajectories drift from the correct answer yet remain locally coherent, so no explicit error triggers reactive self-correction. We introduce an information-theoretic framework that separates reasoning into procedural advancement and epistemic verbalization, the token-level externalization of uncertainty, and prove that sporadic verbalization restores convergence toward the correct answer even without explicit error triggers. Empirically, a minimal doubt cue recovers failed trajectories, and small-scale SFT suffices to instill or suppress this capability, suggesting that strong reasoning hinges less on an extraordinary inner mechanism than on the linguistic habit of externalizing uncertainty. Our framework recasts reasoning as strategic information allocation under uncertainty, offering a new lens for understanding and advancing LLM reasoning. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_15500 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Understanding Reasoning in LLMs through Strategic Information Allocation under Uncertainty Kim, Jeonghye Luo, Xufang Kim, Minbeom Lee, Sangmook Li, Dongsheng Yang, Yuqing Artificial Intelligence Machine Learning LLMs often exhibit Aha moments such as self-correction after tokens like "Wait," yet the underlying mechanism remains unclear. Standard LLMs collapse mainly through silent divergence, where trajectories drift from the correct answer yet remain locally coherent, so no explicit error triggers reactive self-correction. We introduce an information-theoretic framework that separates reasoning into procedural advancement and epistemic verbalization, the token-level externalization of uncertainty, and prove that sporadic verbalization restores convergence toward the correct answer even without explicit error triggers. Empirically, a minimal doubt cue recovers failed trajectories, and small-scale SFT suffices to instill or suppress this capability, suggesting that strong reasoning hinges less on an extraordinary inner mechanism than on the linguistic habit of externalizing uncertainty. Our framework recasts reasoning as strategic information allocation under uncertainty, offering a new lens for understanding and advancing LLM reasoning. |
| title | Understanding Reasoning in LLMs through Strategic Information Allocation under Uncertainty |
| topic | Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2603.15500 |