mlsanityData to 14 Oct 2026

Are LLM Belief Updates Consistent with Bayes' Theorem?

arXiv 2507.17951 ↗First posted 23 Jul 2025cs.CL cs.AIcited by 9

Not on the map. Why?

Abstract (arXiv)

Do larger and more capable language models learn to update their "beliefs" about propositions more consistently with Bayes' theorem when presented with evidence in-context? To test this, we formulate a Bayesian Coherence Coefficient (BCC) metric and generate a dataset with which to measure the BCC. We measure BCC for multiple pre-trained-only language models across five model families, comparing against the number of model parameters, the amount of training data, and model scores on common benchmarks. Our results provide evidence for our hypothesis that larger and more capable pre-trained language models assign credences that are more coherent with Bayes' theorem. These results have important implications for our understanding and governance of LLMs.

Key facts. Cited by 9 papers in the map (data through 14 Oct 2026). Builds on 17 of its 23 references that are in the map. First posted on arXiv 23 Jul 2025.

Builds on

17 of 23 references.

References · 17

  1. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  2. 2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,879
  3. 2024MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmarkcited by 1,848
  4. 2022Challenging BIG-Bench Tasks and Whether Chain-of-Thought Can Solve Themcited by 1,807
  5. 2023Pythia: A Suite for Analyzing Large Language Models Across Training and Scalingcited by 1,786
  6. 2023Instruction-Following Evaluation for Large Language Modelscited by 1,228
  7. 2023The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"cited by 407
  8. 2025The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexitycited by 302
  9. 2019Risks from Learned Optimization in Advanced Machine Learning Systemscited by 258
  10. 2023MuSR: Testing the Limits of Chain-of-thought with Multistep Soft Reasoningcited by 201
  11. 2025Large Language Models Often Know When They Are Being Evaluatedcited by 128
  12. 2025Utility Engineering: Analyzing and Controlling Emergent Value Systems in AIscited by 67
5 more
  1. 2023Evaluating Superhuman Models with Consistency Checkscited by 47
  2. 2024Consistency Checks for Language Model Forecasterscited by 27
  3. 2025Comment on The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexitycited by 23
  4. 2025Constrained belief updates explain geometric structures in transformer representationscited by 15
  5. 2025Evaluation Faking: Unveiling Observer Effects in Safety Evaluation of Frontier AI Systemscited by 14

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 8.

9 papers.

  1. 2026Lying with Truths: Open-Channel Multi-Agent Collusion for Belief Manipulation via Generative Montagecited by 19
  2. 2026BayesBench: Evaluating LLM Belief Trajectories Under Multi-Turn Evidence Accumulationcited by 7
  3. 2026Why Do LLMs Struggle in Strategic Play? Broken Links Between Observations, Beliefs, and Actionscited by 3
  4. 2026Illusions of Confidence? Diagnosing LLM Truthfulness via Neighborhood Consistencycited by 2
  5. 2025Overcoming State Inertia: Minimally Invasive Temporal Alignment for Evolving Contextscited by 1
  6. 2026Assessing mentalization in humans and large language modelscited by 1
  7. 2026Subjective-Graph LLM Agents for Simulating Uncertainty in Classroom Social Perceptioncited by 0
  8. 2026MEDLEY-BENCH: Benchmarking Behavioural Metacognition and Belief Revision Under Social Pressure in Large Language Modelscited by 0
  9. 2026Bayesian Intelligence from the Outsidecited by 0