# MEDLEY-BENCH: Benchmarking Behavioural Metacognition and Belief Revision Under Social Pressure in Large Language Models

arXiv: https://arxiv.org/abs/2604.16009 · HTML: https://mlsanity.com/paper/2604.16009/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 19 of its 32 references that are in the map.
- First posted on arXiv 17 Apr 2026.

## Builds on (most cited first, 19 of 19)

- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [TruthfulQA: Measuring How Models Mimic Human Falsehoods](https://mlsanity.com/paper/2109.07958/) (2021)
- [Language Models (Mostly) Know What They Know](https://mlsanity.com/paper/2207.05221/) (2022)
- [Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Prompting](https://mlsanity.com/paper/2305.04388/) (2023)
- [Towards Understanding Sycophancy in Language Models](https://mlsanity.com/paper/2310.13548/) (2023)
- [Measuring Faithfulness in Chain-of-Thought Reasoning](https://mlsanity.com/paper/2307.13702/) (2023)
- [AmbigQA: Answering Ambiguous Open-domain Questions](https://mlsanity.com/paper/2004.10645/) (2020)
- [ASQA: Factoid Questions Meet Long-Form Answers](https://mlsanity.com/paper/2204.06092/) (2022)
- [SycEval: Evaluating LLM Sycophancy](https://mlsanity.com/paper/2502.08177/) (2025)
- [Do as We Do, Not as You Think: the Conformity of Large Language Models](https://mlsanity.com/paper/2501.13381/) (2025)
- [BeliefShift: Benchmarking Temporal Belief Consistency and Opinion Drift in LLM Agents](https://mlsanity.com/paper/2603.23848/) (2026)
- [When Two LLMs Debate, Both Think They'll Win](https://mlsanity.com/paper/2505.19184/) (2025)
- [Are LLM Belief Updates Consistent with Bayes' Theorem?](https://mlsanity.com/paper/2507.17951/) (2025)
- [The Illusion of Insight in Reasoning Models](https://mlsanity.com/paper/2601.00514/) (2026)
- [Do LLMs Know What They Know? Measuring Metacognitive Efficiency with Signal Detection Theory](https://mlsanity.com/paper/2603.25112/) (2026)
- [Leveraging Imperfection with MEDLEY A Multi-Model Approach Harnessing Bias in Medical AI](https://mlsanity.com/paper/2508.21648/) (2025)
- [Measuring (some aspects of) the metacognition of AI](https://mlsanity.com/paper/2603.29693/) (2026)
- [Beyond Social Pressure: Benchmarking Epistemic Attack in Large Language Models](https://mlsanity.com/paper/2604.07749/) (2026)

## Abstract (arXiv)

Most large language model benchmarks evaluate final-answer quality but reveal little about how models revise beliefs under disagreement or conflicting evidence. We introduce MEDLEY-BENCH, an open benchmark comparing structured private self-review and analyst-conditioned social revision from a common solo baseline. We evaluated 35 models from 12 families on 130 instances using the Medley Metacognition Score (MMS) and four taxonomy-aligned, rubric-derived composites. MMS point estimates were not consistently ordered in the available within-family size or generation comparisons. Under the prespecified ipsative procedure, the Evaluation-mapped composite had the lowest relative rubric score in 30 of 35 models; Self-regulation was lowest in four models and Control in one. This rubric- and centering-dependent pattern may reflect model behavior, judge severity, data-pipeline effects, or their combination; it is not evidence of an absolute Evaluation deficit. In an exploratory adversarial analysis of 11 purposively selected models, sensitivity to manipulated consensus labels ranged from near zero to larger response shifts. A preliminary human rubric-application study of 24 paired vignettes found a mean composite difference of 0.727 (95% CI: 0.500-0.942). Across 24 cross-rated response items from 12 shared vignettes, quadratic-weighted inter-reviewer agreement was kappa = 0.389, and response-profile human-LLM convergence was rho = 0.637. This preprint reports MEDLEY-BENCH v1.0, the audited proof-of-concept release. A separately versioned v1.5 will rerun the full protocol with corrected social-summary rendering and stronger scoring reproducibility, experimental control, and uncertainty analysis. MEDLEY-BENCH complements accuracy-based evaluation by characterizing prompted belief revision under ambiguity and social disagreement.