mlsanityData to 7 Oct 2026

Small transformers track Bayesian evidence for latent common causes via a context-invariant mechanism

arXiv 2609.35161 ↗First posted 28 Sep 2026cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

We present an in-depth investigation of how a form of Bayesian reasoning about common causes can emerge as a cross-contextual generalization in small, tractable transformers. Incrementing on recent work, our set-up (i) disentangles causal mechanisms in the model from the causal structure of the true data-generating process, (ii) orients more towards natural language prediction by considering inference of latent common causes, and (iii) considers whether and how Bayesian evidence accumulation for latent common causes can be implemented in representations and mechanisms that allow for cross-context generalization to novel test cases.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 34 of its 45 references that are in the map. First posted on arXiv 28 Sep 2026.

Builds on

34 of 45 references.

References · 34

  1. 2021An Explanation of In-context Learning as Implicit Bayesian Inferencecited by 915
  2. 2022Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Taskcited by 417
  3. 2024Mechanistic Interpretability for AI Safety -- A Reviewcited by 415
  4. 2021Causal Abstractions of Neural Networkscited by 401
  5. 2022Using cognitive psychology to understand GPT-3cited by 389
  6. 2023Causal Reasoning and Large Language Models: Opening a New Frontier for Causalitycited by 348
  7. 2021Transformers Can Do Bayesian Inferencecited by 338
  8. 2021Implicit Representations of Meaning in Neural Language Modelscited by 179
  9. 2023Pretraining task diversity and the emergence of non-Bayesian in-context learning for regressioncited by 138
  10. 2017Causal Consistency of Structural Equation Modelscited by 114
  11. 2024How Transformers Learn Causal Structure with Gradient Descentcited by 114
  12. 2024Evaluating the World Model Implicit in a Generative Modelcited by 107
22 more
  1. 2018Abstracting Causal Modelscited by 104
  2. 2023Amortizing intractable inference in large language modelscited by 94
  3. 2024Transformers represent belief state geometry in their residual streamcited by 73
  4. 2021Chess as a Testbed for Language Model State Trackingcited by 69
  5. 2024Is In-Context Learning in Large Language Models Bayesian? A Martingale Perspectivecited by 51
  6. 2022Event knowledge in large language models: the gap between the impossible and the unlikelycited by 44
  7. 2023A Latent Space Theory for Emergent Abilities in Large Language Modelscited by 42
  8. 2025Bayesian Teaching Enables Probabilistic Reasoning in Large Language Modelscited by 36
  9. 2025Unveiling Causal Reasoning in Large Language Models: Reality or Mirage?cited by 36
  10. 2023Causal Interpretation of Self-Attention in Pre-Trained Transformerscited by 30
  11. 2024Incoherent Probability Judgments in Large Language Modelscited by 29
  12. 2024Neural Causal Abstractionscited by 20
  13. 2025Enough Coin Flips Can Make LLMs Act Bayesiancited by 19
  14. 2025Constrained belief updates explain geometric structures in transformer representationscited by 14
  15. 2025The Bayesian Geometry of Transformer Attentioncited by 13
  16. 2025Revisiting the Othello World Model Hypothesiscited by 11
  17. 2023Deep de Finetti: Recovering Topic Distributions from Large Language Modelscited by 7
  18. 2025What Does it Mean for a Neural Network to Learn a "World Model"?cited by 7
  19. 2025Transformers Simulate MLE for Sequence Generation in Bayesian Networkscited by 5
  20. 2025A mathematical theory for understanding when abstract representations emerge in neural networkscited by 3
  21. 2026Finding Belief Geometries with Sparse Autoencoderscited by 2
  22. 2024Dual Traits in Probabilistic Reasoning of Large Language Modelscited by 1

Built on by

Not cited yet.