Self-Reflection in Language Model Agents

Latest papers

36 papers7 papers in the last four weeks, up 75% on the four weeks before. 0.0% of all new papers.

MIRROR: A Hierarchical Benchmark for Metacognitive Calibration in Large Language Models
Apr 15, 2026Jason Z WangMetacognition in Language Models Language Model Self-Assessment
Independent
ReflectRM: Boosting Generative Reward Models via Self-Reflection within a Unified Judgment Framework
Apr 8, 2026Kai Qin, Liangxin Liu, Yu Liang +7Preference Modeling Self-Reflection in Language Model Agents
Shenzhen International Graduate School, Tsinghua University · Baidu Inc., Beijing, China
From Hallucination to Structure Snowballing: The Alignment Tax of Constrained Decoding in LLM Reflection
Apr 7, 2026Hongxu ZhouSelf-Reflection in Language Model Agents LLM Self-Correction
Erasmus Mundus Master’s in Language and Communication Technologies Saarland University Saarbrücken, Germany
Me, Myself, and $π$ : Evaluating and Explaining LLM Introspection
Mar 17, 2026Atharv Naphade, Samarth Bhargav, Sean Lim +1Language Model Introspection Self-Reflection in Language Model Agents
Carnegie Mellon University
GRPO and Reflection Reward for Mathematical Reasoning in Large Language Models
Mar 14, 2026Zhijie WangSelf-Reflection in Language Model Agents LLM Reasoning
School of Computer Science and Engineering, Sun Yat-sen University(SYSU), Guangzhou, China
ParamMem: Augmenting Language Agents with Parametric Reflective Memory
Feb 26, 2026Tianjun Yao, Yongqiang Chen, Yujia Zheng +3Self-Reflection in Language Model Agents Memory-Augmented Language Model Agents
Mohamed bin Zayed University of Artificial Intelligence · Carnegie Mellon University · Georgia Institute of Technology

Apr 15, 2026J/K move · Enter open · S save

MIRROR: A Hierarchical Benchmark for Metacognitive Calibration in Large Language Models

Jason Z Wang

Independent

Metacognition in Language Models Language Model Self-Assessment Self-Reflection in Language Model Agents Language Model Calibration LLM Safety Benchmarks

Open paper AskPDF arXiv

We introduce MIRROR, a benchmark comprising eight experiments across four metacognitive levels that evaluates whether large language models can use self-knowledge to make better decisions. We evaluate 16 models from 8 labs across approximately 250,000 evaluation instances using five independent behavioral measurement channels. Core experiments are run across the full model roster; experiments with specialized infrastructure requirements report explicitly marked model subsets. We find two phenomena with direct implications for agentic deployment: (1) compositional self-prediction fails universally -- the Compositional Calibration Error ranges from 0.500 to 0.943 on the original 15-model Exp3-v1 set (and 0.434 to 0.758 on the balanced 16-model Exp3-v2 expansion), indicating that models cannot predict their own performance on multi-domain tasks, and (2) models exhibit above-chance but imperfect domain-specific self-knowledge yet systematically fail to translate even this partial awareness into appropriate agentic action-selection -- external metacognitive control reduces the Confident Failure Rate from 0.600 to 0.143 (76% reduction at temperature 0; mean 70% at temperature 0.7 across 5 models from 4 labs). Providing models with their own calibration scores produces no significant improvement (p > 0.05); only architectural constraint is effective. This suggests that external metacognitive scaffolding -- not improved self-knowledge -- is the path to safer autonomous AI systems. Code, data, and Croissant metadata will be released publicly with the benchmark.

Latest papers

MIRROR: A Hierarchical Benchmark for Metacognitive Calibration in Large Language Models

ReflectRM: Boosting Generative Reward Models via Self-Reflection within a Unified Judgment Framework

From Hallucination to Structure Snowballing: The Alignment Tax of Constrained Decoding in LLM Reflection

Me, Myself, and πππ : Evaluating and Explaining LLM Introspection

GRPO and Reflection Reward for Mathematical Reasoning in Large Language Models

ParamMem: Augmenting Language Agents with Parametric Reflective Memory

Me, Myself, and $π$ : Evaluating and Explaining LLM Introspection