Research map / Probabilistic and scientific modeling / Causal discovery
Counterfactual identification and bounds
Characterize and bound counterfactual quantities from observational and experimental data.
92 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Causal discovery, keyword cluster “counterfactual · causal inference · causal representation”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
16 papers in 2023–24 and 46 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 6.9% to 13.8%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Learning Representations for Counterfactual Inference (ICML 2016 · 185 citations)
- 2018 · Towards Formal Definitions of Blameworthiness, Intention, and Moral Responsibility (AAAI 2018 · 43 citations)
- 2019 · Abstracting Causal Models (AAAI 2019 · 51 citations)
- 2020 · Algorithmic recourse under imperfect causal knowledge: a probabilistic approach (NeurIPS 2020 · 87 citations)
- 2021 · A Causal Lens for Controllable Text Generation (NeurIPS 2021 · 19 citations)
- 2022 · Debiasing NLU Models via Causal Intervention and Counterfactual Reasoning (AAAI 2022 · 35 citations)
- 2023 · Causal Proxy Models for Concept-based Model Explanations (ICML 2023 · 5 citations)
- 2024 · A Compositional Atlas for Algebraic Circuits (NeurIPS 2024 · 1 citation)
- 2025 · Causal Order: The Key to Leveraging Imperfect Experts in Causal Inference (ICLR 2025 · 7 citations)
- 2026 · KTCF: Actionable Recourse in Knowledge Tracing via Counterfactual Explanations for Education (AAAI 2026 · 2 citations)
Earliest papers here
- Learning Representations for Counterfactual Inference (ICML 2016 · 185 citations)
- Towards Formal Definitions of Blameworthiness, Intention, and Moral Responsibility (AAAI 2018 · 43 citations)
- Situation Calculus Semantics for Actual Causality (AAAI 2018 · 30 citations)
What has been tried
51 papers here report 385 measured changes (a component added, removed or swapped, compared with a baseline): helped 70%, no clear effect 9%, hurt 17%, mixed 5%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Replace pairwise base prompting with pairwise Chain-of-Thought prompting | Child · SHD | 177 -> 138 | Helped | Causal Order: The Key to Leveraging Imperfect Experts in Causal Inference (ICLR 2025 · 7 citations) |
| Add zero-mean Gaussian noise with standard deviation 5 to the f3 coefficients | German Credit simulation sensitivity analysis · counterfactual stability and sparsity | No clear effect | A New Approach to Backtracking Counterfactual Explanations: A Unified Causal Framework for Efficient Model Interpretability (ICML 2025 · 0 citations) | |
| use the correct causal order instead of reversing it | ER-Diag-50 · Ctf RMSE | DNME 0.78 ± 0.05 -> 0.53 ± 0.05; TNME 11.24 ± 20.98 -> 0.47 ± 0.05; CMSM 2.64 ± 3.72 -> 0.37 ± 0.05; TVSM 0.79 ± 0.04 -> | Helped | Exogenous Isomorphism for Counterfactual Identifiability (ICML 2025 · 0 citations) |
| Use FRA with coalition caching instead of caching alone | Synthetic graphs with K=100 · execution time | 25m20s -> 21m14s; 4m | Helped | Practical do-Shapley Explanations with Estimand-Agnostic Causal Inference (NeurIPS 2025 · 0 citations) |
| Add latent divergence to SAE | Morpho-MNIST do(th) · Image MSE (· 10⁻²) | 0.67 -> 0.42 | Helped | Counterfactual Generative Modeling with Variational Causal Inference (ICLR 2025 · 0 citations) |
| Increase λ3 to 0.4 for higher validity focus | MNIST add · validity (%) | 99.1 ± 0.2 | Helped | Counterfactual Concept Bottleneck Models (ICLR 2025 · 0 citations) |
| Increase the target intervention size | Simulated dataset · Counterfactual error | Hurt | Towards Characterizing Domain Counterfactuals for Invertible Latent Causal Models (ICLR 2024 · 0 citations) | |
| Compute d-TOE on a subsampled set of indices during training | Synthetic TO inference · topological ordering score | No clear effect | A Fixed-Point Approach for Causal Generative Modeling (ICML 2024 · 0 citations) | |
| replace the RBF kernel with the Physics-Informed Kernel | Virtual Tools · AUCCESS | 42.0±33.0 -> 65.0±25.0 | Helped | Causal-PIK: Causality-based Physical Reasoning with a Physics-Informed Kernel (ICML 2025) |
What reviewers will expect
Datasets and baselines named in the 35 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- German credit · 2 of 35 papers
Baselines
- CausalNF · 2 of 35 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 62 papers; from OpenAlex and the papers' affiliations).
- Columbia University · 6 papers
- Imperial College London · 4 papers
- CNRS · 3 papers
- Stanford University · 3 papers
- Broad Institute · 2 papers
- Massachusetts Institute of Technology · 2 papers
- Technical University of Munich · 2 papers
- TU Darmstadt · 2 papers
- Technical University of Darmstadt · 2 papers
- University of California, Los Angeles · 2 papers
Use it
Download BibTeX of 40 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Latent-variable causal estimation (117 papers)
- Causal effect identification (107 papers)
- Causal representation identifiability (122 papers)
- Causal approaches to fairness (46 papers)
Elsewhere in the topic
- Constraint-based structure learning with latents (90 papers)
- Identifiability with hidden confounders (85 papers)
- Interventional causal discovery (68 papers)
- Differentiable and score-based DAG learning (52 papers)
- Causal discovery from time series (51 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).