Learn Like Humans: Use Meta-cognitive Reflection for Efficient Self-Improvement
Xinmeng Hou, Peiliang Gong, Bohao Qu, Wuqi Wang, Qing Guo, Yang Liu · 2026
Read Learn Like Humans: Use Meta-cognitive Reflection for Efficient Self-Improvement by Xinmeng Hou, Peiliang Gong, Bohao Qu, Wuqi Wang, Qing Guo, Yang Liu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-11974.txt.
Description
Xinmeng Hou, Peiliang Gong, Bohao Qu, and their co-authors give language model agents a way to improve with less waste. Agents today rest on fixed prompts made by humans. That caps how well they can adapt to new tasks.
Past self-improve tools use loops that run many turns and cost much compute. The team take a cue from human meta-cognition, our skill at thinking about our own thought. Their method uses brief self-review to guide change. A leaner loop could make agents adapt at lower cost. Brief self checks keep the cost of each task in bounds.
Catalog record
| CREATOR | Xinmeng Hou, Peiliang Gong, Bohao Qu, Wuqi Wang, Qing Guo, Yang Liu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 56 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-11974