Metacognition in Language Models

Follow this topic Ask about this topic

Momentum

10 papers in the last four weeks, up 67% on the four weeks before. 0.1% of all new papers.

Jul 13Week of Sep 28

Latest papers 74

All topics

LLMs Know When They Know, but Do Not Act on It: A Metacognitive Harness for Test-time Scaling
May 13, 2026Qi Cao, Yufan Wang, Peijia Qin +2LLM Inference Test-Time Scaling
University of California, San Diego
TRIAGE: Evaluating Prospective Metacognitive Control in LLMs under Resource Constraints
May 13, 2026Zabir Al Nazi, Shubhashis Roy DiptaLLM Agent Evaluation AI Agent Benchmarks
The Metacognitive Probe: Five Behavioural Calibration Diagnostics for LLMs
May 11, 2026Rafael C. T. OliveiraLanguage Model Calibration LLM Uncertainty Estimation
Independent Researcher
Decomposing and Steering Functional Metacognition in Large Language Models
May 9, 2026Yanshi Li, Xueru Bai, Shuman Liu +2LLM Evaluation LLM Interpretability
Shopee Singapore, Singapore
Hallucinations Undermine Trust; Metacognition is a Way Forward
May 2, 2026Gal Yona, Mor Geva, Yossi MatiasLLM Hallucination Mitigation Hallucination in Language Models
Google Research · Tel Aviv University
How LLMs Detect and Correct Their Own Errors: The Role of Internal Confidence Signals
Apr 24, 2026Dharshan Kumaran, Viorica Patraucean, Simon Osindero +2LLM Self-Correction Language Model Error Detection
Google DeepMind · Princeton University
Before You Interpret the Profile: Validity Scaling for LLM Metacognitive Self-Report
Apr 20, 2026Jon-Paul CacioliPsychometric Validation LLM Uncertainty Estimation
Independent Researcher, Melbourne, Australia
Beyond Meta-Reasoning: Metacognitive Consolidation for Self-Improving LLM Reasoning
Apr 19, 2026Ziqing Zhuang, Linhai Zhang, Jiasheng Si +2Continual Learning for LLMs Meta-Learning
Southeast University · King’s College London · Qilu University of Technology (Shandong Academy of Sciences) +1
MEDLEY-BENCH: Scale Buys Evaluation but Not Control in AI Metacognition
Apr 17, 2026Farhad Abtahi, Abdolamir Karbalaie, Eduardo Illueca-Fernandez +1Belief Revision LLM Evaluation
Department of Clinical Science, Intervention and Technology (CLINTEC), Karolinska Institutet, 17177 Stockholm, Sweden · Department of Clinical Physiology, Karolinska University Hospital, 17176 Stockholm, Sweden · Department of Biomedical Engineering and Health Systems, KTH Royal Institute of Technology, 14157 Huddinge, Sweden +2
The Metacognitive Monitoring Battery: A Cross-Domain Benchmark for LLM Self-Monitoring
Apr 17, 2026Jon-Paul CacioliLLM Evaluation Language Model Calibration
Independent Researcher Melbourne, Australia
Do LLMs Know What They Know? Measuring Metacognitive Efficiency with Signal Detection Theory
Mar 26, 2026Jon-Paul CacioliConfidence Estimation in Language Models Language Model Calibration
Independent Researcher Melbourne, Australia
Knowing When to Critique: Task-Adaptive Metacognitive Regulation for Reliable LLM Reasoning
Jul 20, 2025Xinmeng Hou, Ziting Chang, Zhouquan Lu +5LLM Self-Correction Multi-Agent LLMs
Nanyang Technological University · Unicorn Verse · Shanghai Jiao Tong University +3
Evidence for Limited Metacognition in LLMs
Date pendingChristopher AckermanLLM Evaluation LLM Uncertainty Estimation
Rescaling Confidence: What Scale Design Reveals About LLM Metacognition
Date pendingYuyang Dai, Yuxia WangConfidence Estimation in Language Models Language Model Calibration
INSAIT

May 13, 2026, cs.LGJ/K move · Enter open · S save

LLMs Know When They Know, but Do Not Act on It: A Metacognitive Harness for Test-time Scaling

Qi Cao, Yufan Wang, Peijia Qin, Shuhao Zhang+1

University of California, San Diego

LLM Inference Test-Time Scaling Metacognition in Language Models LLM Reasoning

Open paper AskPDF arXiv

Large language models (LLMs) often expose useful signals of self-monitoring: before solving a problem, they can estimate whether they are likely to succeed, and after solving it, they can judge whether their answer is likely to be correct. However, these signals are typically measured or elicited in isolation, rather than used to control inference. In this work, we ask whether LLMs possess latent metacognitive ability that can be turned into effective test-time control. Inspired by the Nelson--Narens theory from cognitive psychology, we propose a metacognitive harness that separates monitoring from reasoning. For each problem, the model first reports a pre-solve feeling-of-knowing (FOK) signal; after each solve attempt, it reports a post-solve judgment-of-learning (JOL) signal. Rather than treating these signals as passive confidence estimates, the harness turns them into an explicit control interface for reasoning: it decides when to trust the current solution, when to retry with compact metacognitive feedback, and when to pass multiple attempts to a final aggregator. Across text, code, and multimodal reasoning benchmarks, our harness substantially improves a fixed Claude Sonnet-4.6 base model without parameter updates or benchmark-specific fine-tuning. On the evaluated public benchmark snapshots, it raises pooled accuracy from 48.3 to 56.9 and exceeds the strongest listed leaderboard entries on the three primary evaluation settings: HLE-Verified, LiveCodeBench v6, and R-Bench-V. These results suggest that strong LLMs may already possess useful metacognitive ability, but require an explicit control harness to act on it during reasoning.

Momentum

Related topics

Latest papers 74