Causal Abstraction

Follow this topic Ask about this topic

Causal abstraction studies how to map a detailed system, such as a neural network or a large structural causal model, to a smaller model whose variables and mechanisms retain meaningful causal relationships. Approaches include constructive mappings between model variables, low-rank or cluster-based discovery of higher-level variables, and intervention-based methods that test whether effects agree across levels. These abstractions can support neural-network interpretation, causal transfer between settings, and more compact planning models. A central challenge is that observational fit alone does not ensure causal agreement: researchers are developing stronger tests and certificates while addressing when abstractions are identifiable and how to handle approximate mappings.

Latest papers

27 papers4 papers in the last four weeks, against 2 the four weeks before. 0.0% of all new papers.

When Do Learned Blocks Preserve Causal Meaning? Block Identifiability and Constructive Causal Abstractions
Oct 13, 2026Karim Zaghw, Wieland Brendel, Patrik ReizingerCausal Representation Learning Causal Abstraction
University of Tübingen · ELLIS Institute Tübingen · Max Planck Institute for Intelligent Systems, Tübingen, Germany
Causal Effect Identifiability in Causal Abstractions: A Strict Hierarchy of Criteria
Oct 9, 2026Clément Yvernes, Marianne Clausel, Emilie Devijver +1Causal Identifiability Causal Abstraction
Univ Grenoble Alpes, CNRS, Grenoble INP, LIG · Université de Lorraine, CNRS, CRAN, 54000 Nancy
Score-Based Learning of Cluster DAGs from Interventions
Oct 8, 2026Gaetano Tedesco, Alex MarkhamScore-Based Causal Discovery Causal Structure Learning
Informatics Institute, University of Amsterdam · Department of Mathematical Sciences, University of Copenhagen
Trajectory Abstraction for the Science of Language Agent Behavior
Oct 6, 2026Tianqiang YanAgent Trajectory Analysis Causal Abstraction
Faculty of Information Technology, Monash University
Certifying Interventional Agreement Among Observationally Equivalent Causal Models
Sep 26, 2026Sourena Khanzadeh, Daniel Platnick, Marjan Alirezaie +1Causal Inference Causal Intervention
Toronto Metropolitan University · Flybits Labs, Creative AI Hub · MIT Media Lab, Massachusetts Institute of Technology
Efficient Fairness Auditing Across Guidance Scales in Text-to-Image Diffusion Models via Causal Abstraction
Sep 8, 2026Nabila Tasfiha Rahman, Rajatsubhra Chakraborty, Depeng Xu +1Classifier-Free Guidance Algorithmic Fairness
University of Arkansas Fayetteville, AR, USA · University of North Carolina at Charlotte Charlotte, NC, USA
Generalised Transportability via Causal Abstractions
Aug 16, 2026Yorgos Felekis, Paris Giampouras, Fabio Massimo Zennaro +1Causal Inference Causal Abstraction
Department of Computer Science University of Warwick Coventry, CV4 7AL, UK · Department of Informatics University of Bergen Bergen, N-5020, NO · Departments of Computer Science & Statistics University of Warwick Coventry, CV4 7AL, UK
Operationalising Relative Causal Knowledge: Backbone Identifiability from Private Reports on a Shared Outcome
Aug 11, 2026Fabrizio Russo, Mark SomersStructural Causal Models Causal Abstraction
Department of Computing, Imperial College London · London, UK · Fifty One Degrees Ltd
Abstract Event Causal Rules: Induction and Application
Aug 5, 2026Ziwei Zheng, Peiqiong Chen, Bang WangCausal Abstraction Causal Reasoning
School of Electronic Information and Communications, Huazhong University of Science and Technology, Wuhan, China
Hidden APIs in Language Models: Discovering Reusable Causal Interfaces from Forked Futures
Jul 30, 2026SiYuan Ma, Yiqin Luo, Zhangji +8LLM Interpretability Causal Abstraction
1Nanyang Technological University · 2Southern University of Science and Technology · 3Tianjin University +7
Property-driven Causal Abstractions for Markov Decision Processes
Jul 29, 2026Jule Schmidt, Maximilian Weininger, Clemens Dubslaff +2Markov Decision Processes Markov Models
Ruhr-University Bochum, Germany · Eindhoven University of Technology, The Netherlands · Oxford University, UK +1
Validating Causal Abstraction Metrics on Simulated Complex Systems
Jun 30, 2026Maxime Méloux, Tiago Pimentel, François Portet +1Causal Abstraction Benchmark Validity
Université Grenoble Alpes, CNRS, Grenoble INP, LIG · ETH Zürich
Unsupervised Causal Abstractions Discovery
Jun 17, 2026Théo Saulus, Simon Lacoste-Julien, Dhanya SridharStructural Causal Models Causal Discovery
Mila - Quebec AI Institute · Université de Montréal · Canada CIFAR AI Chair
A Simple Hierarchical Causality Primer
Jun 1, 2026Tim GebbieCausal Abstraction
Department of Statistical Science, University of Cape Town, Rondebosch 7701, South Africa
Formalizing and falsifying causal pathways of rare events
May 29, 2026Anahita Haghighat, Dominik JanzingStructural Causal Models Causal Abstraction
Independent · Amazon Research, Tübingen, Germany
Coarsening Linear Non-Gaussian Causal Models with Cycles
May 11, 2026Francisco Madaleno, Francisco C Pereira, Alex MarkhamStructural Causal Models Causal Discovery
Department of Technology, Management and Economics, Technical University of Denmark · Department of Mathematical Sciences,11 University of Copenhagen
PLOT: Progressive Localization via Optimal Transport in Neural Causal Abstraction
May 7, 2026Jonathn Chang, Arya Datla, Ziv GoldfeldMechanistic Interpretability Causal Abstraction
Cornell University
Bucketing the Good Apples: A Method for Diagnosing and Improving Causal Abstraction
May 4, 2026Li Puyin, Jiyuan Tan, Ahmad Jabbar +2Mechanistic Interpretability Causal Abstraction
♢Stanford University · ♠Goodfire
Arithmetic in the Wild: Llama uses Base-10 Addition to Reason About Cyclic Concepts
May 1, 2026Sheridan Feucht, Tal Haklay, Usha Bhalla +9Numerical Reasoning in Language Models LLM Interpretability
aNortheastern University · bTechnion IIT · cHarvard University +1
Causal Foundations of Collective Agency
Apr 30, 2026Frederik Hytting Jørgensen, Sebastian Weichwald, Lewis HammondAI Agent Safety Causal Abstraction
Copenhagen Causality Lab and Pioneer Centre for AI, University of Copenhagen, Denmark · Cooperative AI Foundation, United Kingdom
Grounding Before Generalizing: How AI Differs from Humans in Causal Transfer
Apr 27, 2026Liangru Xiang, Yuxi Ma, Zhihao Cao +2LLM Grounding Causal Structure Learning
Department of Automation, Tsinghua University · State Key Laboratory of General Artificial Intelligence · Institute for Artificial Intelligence, Peking University +2
From Local to Cluster: A Unified Framework for Causal Discovery with Latent Variables
Apr 24, 2026Zongyu LiCausal Effect Estimation Causal Discovery
HCP-DCNet: A Hierarchical Causal Primitive Dynamic Composition Network for Self-Improving Causal Understanding
Mar 12, 2026Ming Lei, Shufan Wu, Christophe BaehrCausal Reasoning Causal Abstraction
Shanghai JiaoTong University, Shanghai · M´et´eo France/CNRS, Universit´e de Toulouse, CNRM UMR 3589, Toulouse, France
Causal Mechanism Reduction: Mechanism Replacement for Neural Network Pruning and Abstraction
Feb 27, 2026Amir AsiaeeStructural Causal Models Causal Abstraction
Vanderbilt University Medical Center Nashville, TN
Multi-Level Causal Embeddings
Feb 25, 2026Willem Schooltink, Fabio Massimo ZennaroStructural Causal Models Causal Abstraction
Department of Informatics, University of Bergen
Causal and Compositional Abstraction
Feb 18, 2026Robin Lorenz, Sean TullCategory Theory Causal Abstraction
Quantinuum, London, United Kingdom
How Causal Abstraction Underpins Computational Explanation
Aug 15, 2025Atticus Geiger, Jacqueline Harding, Thomas IcardComputational Cognitive Modeling Causal Abstraction

Oct 13, 2026J/K move · Enter open · S save

When Do Learned Blocks Preserve Causal Meaning? Block Identifiability and Constructive Causal Abstractions

Karim Zaghw, Wieland Brendel, Patrik Reizinger

University of Tübingen · ELLIS Institute Tübingen · Max Planck Institute for Intelligent Systems, Tübingen, Germany

Causal Representation Learning Causal Abstraction Causal Identifiability

Open paper AskPDF arXiv

Standard identifiability theory assumes one-dimensional latent factors, but learned representations can encode latent information across multidimensional blocks whose internal coordinates are not uniquely determined. When such blocks are used to represent and intervene on higher-level concepts, it is unclear whether their causal meaning survives the transformations left unresolved by block-identifiability. We connect block-identifiability to constructive causal abstraction. We show that block-diagonal invertible transformations followed by compatible coordinate permutations form the maximal class that preserves every constructive causal abstraction built on those blocks. This guarantee assumes that the correspondence between blocks and concepts is updated with the representation. A small synthetic experiment with additive autoencoders shows that high-level interventions can be transferred between independently learned representations while approximately preserving their intended effects.