mlsanityData to 7 Oct 2026

Doing Experiments and Revising Rules with Natural Language and Probabilistic Reasoning

arXiv 2402.06025 ↗First posted 8 Feb 2024cs.AI cs.CLcited by 23

On the map: LLM reasoning → Test-time compute and efficient thinking

Abstract (arXiv)

We give a model of how to infer natural language rules by doing experiments. The model integrates Large Language Models (LLMs) with Monte Carlo algorithms for probabilistic inference, interleaving online belief updates with experiment design under information-theoretic criteria. We conduct a human-model comparison on a Zendo-style task, finding that a critical ingredient for modeling the human data is to assume that humans also consider fuzzy, probabilistic rules, in addition to assuming that humans perform approximately-Bayesian belief updates. We also compare with recent algorithms for using LLMs to generate and revise hypotheses, finding that our online inference method yields higher accuracy at recovering the true underlying rule, and provides better support for designing optimal experiments.

Key facts. Cited by 23 papers in the map (data through 7 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 28 of its 85 references that are in the map. First posted on arXiv 8 Feb 2024.

Builds on

28 of 85 references.

References · 28

  1. 2023GPT-4 Technical Reportcited by 18,132
  2. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,302
  3. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,355
  4. 2022Large Language Models are Zero-Shot Reasonerscited by 4,323
  5. 2023Teaching Large Language Models to Self-Debugcited by 963
  6. 2021Show Your Work: Scratchpads for Intermediate Computation with Language Modelscited by 871
  7. 2022Using Large Language Models to Simulate Multiple Humans and Replicate Human Subject Studiescited by 506
  8. 2020Bayesian Deep Learning and a Probabilistic Perspective of Generalizationcited by 480
  9. 2022Selection-Inference: Exploiting Large Language Models for Interpretable Logical Reasoningcited by 331
  10. 2023Large Language Models as Commonsense Knowledge for Large-Scale Task Planningcited by 231
  11. 2022Language models show human-like content effects on reasoning taskscited by 171
  12. 2023Modern Bayesian Experimental Designcited by 121
16 more
  1. 2023Hypothesis Search: Inductive Reasoning with Language Modelscited by 119
  2. 2022Language Model Cascadescited by 104
  3. 2023Eliciting Human Preferences with Language Modelscited by 90
  4. 2023Phenomenal Yet Puzzling: Testing Inductive Reasoning Capabilities of Language Models with Hypothesis Refinementcited by 89
  5. 2023Sequential Monte Carlo Steering of Large Language Models using Probabilistic Programscited by 79
  6. 2023Training Chain-of-Thought via Latent-Variable Inferencecited by 46
  7. 2021Human-Level Reinforcement Learning through Theory-Based Modeling, Exploration, and Planningcited by 43
  8. 2023FIND: A Function Description Benchmark for Evaluating Interpretability Methodscited by 32
  9. 2021ACRE: Abstract Causal REasoning Beyond Covariationcited by 30
  10. 2023Active Preference Inference using Language Models and Probabilistic Reasoningcited by 29
  11. 2020Learning abstract structure for drawing by efficient motor program inductioncited by 23
  12. 2023Human-like Few-Shot Learning via Bayesian Reasoning over Natural Languagecited by 21
  13. 2022Using Natural Language and Program Abstractions to Instill Human Inductive Biases in Machinescited by 20
  14. 2015Nested Sequential Monte Carlo Methodscited by 19
  15. 2019Bayesian Synthesis of Probabilistic Programs for Automatic Data Modelingcited by 19
  16. 2023Modeling infant object perception as program inductioncited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 1; 2025: 12; 2026: 10.

23 papers.

Sort:

By area:

  1. 2025Bayesian Teaching Enables Probabilistic Reasoning in Large Language Modelscited by 36
  2. 2024Is Programming by Example solved by LLMs?cited by 27
  3. 2025PoE-World: Compositional World Modeling with Products of Programmatic Expertscited by 25
  4. 2025AutoToM: Scaling Model-based Mental Inference via Automated Agent Modelingcited by 23
  5. 2025WorldLLM: Improving LLMs' world modeling using curiosity-driven theory-makingcited by 10
  6. 2026Failing to Falsify: Evaluating and Mitigating Confirmation Bias in Language Modelscited by 9
  7. 2025Shoot First, Ask Questions Later? Building Rational Agents that Explore and Act Like Peoplecited by 7
  8. 2026OPINE-World: Programmatic World Modeling with Ontology-error-Prioritized Interactive Exploration for ARC-AGI-3cited by 7
  9. 2025The Curious Language Model: Strategic Test-Time Information Acquisitioncited by 5
  10. 2026Wiring the 'Why': A Unified Taxonomy and Survey of Abductive Reasoning in LLMscited by 5
  11. 2025Program Synthesis via Test-Time Transductioncited by 4
  12. 2025Disentangling Exploration of Large Language Models by Optimal Exploitationcited by 1
11 more
  1. 2025Bonsai: Interpretable Tree-Adaptive Grounded Reasoningcited by 1
  2. 2025Beyond Either-Or Reasoning: Transduction and Induction as Cooperative Problem-Solving Paradigmscited by 1
  3. 2025Information Seeking for Robust Decision Making under Partial Observabilitycited by 1
  4. 2026Playing ZendoWorld: Challenging AI Agents on Active Visual Concept Inductioncited by 1
  5. 2025Analysis of Error Sources in LLM-based Hypothesis Search for Few-Shot Rule Inductioncited by 0
  6. 2026Uncertainty as a Planning Signal: Multi-Turn Decision Making for Goal-Oriented Conversationcited by 0
  7. 2026Hypothesis Generation and Inductive Inference in Children and Language Modelscited by 0
  8. 2026Human Adults and LLMs as Scientists: Who Benefits from Active Exploration?cited by 0
  9. 2026Solver-Aware Decompositions for Programming-by-Example: When Dividing Requires Knowing how to Conquercited by 0
  10. 2026Schema: Discovering Unknown Environments via Agentic Program Inductioncited by 0
  11. 2026Bayesian Fine-tuning Yields Language Models that are as Bayesian as their Beliefs Allowcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024QUITE: Quantifying Uncertainty in Natural Language Text in Bayesian Reasoning Scenarios
  2. 2026Diagnosing and Improving Probabilistic Reasoning in Large Language Models
  3. 2025Reasoning Under Uncertainty: Exploring Probabilistic Reasoning Capabilities of LLMs
  4. 2026Benchmarking LLM Competence on Logical Inference over Probability Operators
  5. 2024Towards Logically Consistent Language Models via Probabilistic Reasoning
  6. 2025Always Tell Me The Odds: Fine-grained Conditional Probability Estimation
  7. 2021Towards General Natural Language Understanding with Probabilistic Worldbuilding
  8. 2025Exploring the Potential for Large Language Models to Demonstrate Rational Probabilistic Beliefs