Systems with Switching Causal Relations: A Meta-Causal Perspective
Not on the map. Why?
Abstract (arXiv)
Most work on causality in machine learning assumes that causal relationships are driven by a constant underlying process. However, the flexibility of agents' actions or tipping points in the environmental process can change the qualitative dynamics of the system. As a result, new causal relationships may emerge, while existing ones change or disappear, resulting in an altered causal graph. To analyze these qualitative changes on the causal graph, we propose the concept of meta-causal states, which groups classical causal models into clusters based on equivalent qualitative behavior and consolidates specific mechanism parameterizations. We demonstrate how meta-causal states can be inferred from observed agent behavior, and discuss potential methods for disentangling these states from unlabeled data. Finally, we direct our analysis towards the application of a dynamical system, showing that meta-causal states can also emerge from inherent system dynamics, and thus constitute more than a context-dependent framework in which mechanisms emerge only as a result of external factors.
Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 24 of its 66 references that are in the map. First posted on arXiv 16 Oct 2024.
Builds on
24 of 66 references.
References · 24
- 2015Causal inference using invariant prediction: identification and confidence intervalscited by 536
- 2022Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Taskcited by 431
- 2021Implicit Representations of Meaning in Neural Language Modelscited by 180
- 2023Causal Parrots: Large Language Models May Talk Causality But Are Not Causalcited by 150
- 2017Invariant Causal Prediction for Nonlinear Modelscited by 140
- 2017Causal Consistency of Structural Equation Modelscited by 118
- 2018Abstracting Causal Modelscited by 107
- 2020On the Fairness of Causal Algorithmic Recoursecited by 70
- 2019Causal Reasoning from Meta-reinforcement Learningcited by 53
- 2023A Survey on Causal Reinforcement Learningcited by 45
- 2018Meta Reinforcement Learning with Latent Variable Gaussian Processescited by 44
- 2020Invariant Policy Optimization: Towards Stronger Generalization in Reinforcement Learningcited by 37
12 more
- 2023Foundations of Causal Discovery on Groups of Variablescited by 22
- 2023Passive learning of active causal strategies in agents and language modelscited by 21
- 2023Targeted Reduction of Causal Modelscited by 15
- 2022A Meta-Reinforcement Learning Algorithm for Causal Discoverycited by 11
- 2021Typing assumptions improve identification in causal discoverycited by 10
- 2023Graph Switching Dynamical Systemscited by 10
- 2022Towards a Grounded Theory of Causation for Embodied AIcited by 9
- 2023A Measure-Theoretic Axiomatisation of Causalitycited by 8
- 2022Phenomenological Causalitycited by 7
- 2024Towards the Reusability and Compositionality of Causal Representationscited by 5
- 2024Learning Causal Dynamics Models in Object-Oriented Environmentscited by 3
- 2023Do Not Marginalize Mechanisms, Rather Consolidate!cited by 1
Built on by
Citing papers per year: 2025: 1; 2026: 1.
2 papers.