Contrastive explanation: a structural-model approach
Open accessAbstract
This paper presents a model of contrastive explanation using structural casual models. The topic of causal explanation in artificial intelligence has gathered interest in recent years as researchers and practitioners aim to increase trust and understanding of intelligent decision-making. While different sub-fields of artificial intelligence have looked into this problem with a sub-field-specific view, there are few models that aim to capture explanation more generally. One general model is based on structural causal models . It defines an explanation as a fact that, if found to be true, would constitute an actual cause of a specific event. However, research in philosophy and social sciences shows that explanations are contrastive : that is, when people ask for an explanation of an event—the fact —they (sometimes implicitly) are asking for an explanation relative to some contrast case ; that is, ‘Why P rather than Q ?’. In this paper, we extend the structural causal model approach to define two complementary notions of contrastive explanation , and demonstrate them on two classical problems in artificial intelligence: classification and planning. We believe that this model can help researchers in subfields of artificial intelligence to better understand contrastive explanation.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Deep Taylor decomposition is leveraged to explain models such as OCSVM and KDE.
“Third, deep Taylor decomposition [144] is leveraged to explain models such as OCSVM, KDE, and so on.”
From A Survey on Explainable Anomaly Detection (2024) · cites this paper as [144]
The Abstract (p. 1) describes a structural-causal-model approach to contrastive explanation, with examples in classification and planning. Searches of our full-text copy for “deep Taylor decomposition” and “OCSVM KDE” found no passage addressing the claimed use of deep Taylor decomposition to explain those models. (LLM, conf=0.75)
Experimental — accuracy not yet measuredContrastive explanations partially support trade-offs when they answer 'Why A instead of B?', where A is the output and B a foil.
“Contrastive explanations [27] partially support trade-offs when they answer 'Why 𝐴 instead of 𝐵?', where 𝐴 is the output and 𝐵 a foil.”
From Explainable AI is Dead, Long Live Explainable AI!: Hypothesis-driven Decision Support using Evaluative AI (2023) · cites this paper as [27]
Many existing tools will play a part in evaluative AI, including contrastive explanation, Weights of Evidence, feature importance, and case-based reasoning, all of which generate evidence.
“Fourth, from the earlier example, it is clear that many existing tools will play a part in evaluative AI: constrastive explanation [27], Weights of Evidence (WoE) [24], feature importance, case-based reasoning techniques [38], etc., all generate evidence.”
From Explainable AI is Dead, Long Live Explainable AI!: Hypothesis-driven Decision Support using Evaluative AI (2023) · cites this paper as [27]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
Grice, H. P. (1975) · Speech Acts · doi:10.1163/9789004368811_003 · Cited by 23,221 · see citing works
"Why Should I Trust You?": Explaining the Predictions of Any Classifier
Ribeiro, M. T., Singh, S., & Guestrin, C. (2016) · Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining · doi:10.1145/2939672.2939778 · Cited by 17,555 · see citing works
Explanation in artificial intelligence: Insights from the social sciences
Miller, T. (2019) · Artificial Intelligence · doi:10.1016/j.artint.2018.07.007 · Cited by 5,302 · see citing works
Self-citationFraassen, B. C. V. (1980) · doi:10.1093/0198244274.001.0001 · Cited by 4,682 · see citing works
Demopoulos, W., & van Fraassen, B. C. (1982) · The Philosophical Review · doi:10.2307/2184804 · Cited by 3,317 · see citing works
Rule-Based Expert Systems: The MYCIN Experiments of the Stanford Heuristic Programming Project
(1986) · Telematics and Informatics · doi:10.1016/s0736-5853(86)80073-9 · Cited by 1,701 · see citing works
Rule-based expert systems: The mycin experiments of the stanford heuristic programming project
Swartout, W. (1985) · Artificial Intelligence · doi:10.1016/0004-3702(85)90067-0 · Cited by 1,248 · see citing works
Explaining machine learning classifiers through diverse counterfactual explanations
Mothilal, R. K., Sharma, A., & Tan, C. (2020) · Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency · doi:10.1145/3351095.3372850 · Cited by 1,233 · see citing works
Designing Theory-Driven User-Centric Explainable AI
Wang, D., Yang, Q., Abdul, A., & Lim, B. Y. (2019) · Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3290605.3300831 · Cited by 906 · see citing works
Causes and Explanations: A Structural-Model Approach. Part I: Causes
Halpern, J. Y., & Pearl, J. (2005) · The British Journal for the Philosophy of Science · doi:10.1093/bjps/axi147 · Cited by 781 · see citing works
Conversational processes and causal explanation.
Hilton, D. J. (1990) · Psychological Bulletin · doi:10.1037/0033-2909.107.1.65 · Cited by 456 · see citing works
Intelligent agent transparency in human–agent teaming for multi-UxV management
Mercado, J. E., Rupp, M. A., Chen, J. Y. C., Barnes, M. J., Barber, D., & Procci, K. (2016) · Human Factors: The Journal of the Human Factors and Ergonomics Society · doi:10.1177/0018720815621206 · Cited by 403 · see citing works
Lipton, P. (1990) · Royal Institute of Philosophy Supplement · doi:10.1017/s1358246100005130 · Cited by 343 · see citing works
Forms of Explanation: Rethinking the Questions of Social Theory.
Hollis, M., Garfinkel, A., & Rosenberg, A. (1982) · The Journal of Philosophy · doi:10.2307/2026064 · Cited by 311 · see citing works
Assessing demand for intelligibility in context-aware applications
Lim, B. Y., & Dey, A. K. (2009) · Proceedings of the 11th international conference on Ubiquitous computing · doi:10.1145/1620545.1620576 · Cited by 246 · see citing works
Causes and Explanations: A Structural-Model Approach. Part II: Explanations
Halpern, J. Y., & Pearl, J. (2005) · The British Journal for the Philosophy of Science · doi:10.1093/bjps/axi148 · Cited by 242 · see citing works
Contemporary science and natural explanation: commonsense conceptions of causality
(1988) · Choice Reviews Online · doi:10.5860/choice.26-1476 · Cited by 185 · see citing works
Autonomy and Common Ground in Human-Robot Interaction: A Field Study
Stubbs, K. N., HINDS, P. J., & Wettergreen, D. S. (2007) · IEEE Intelligent Systems · doi:10.1109/mis.2007.21 · Cited by 162 · see citing works
Explanation in Second Generation Expert Systems
Swartout, W. R., & Moore, J. D. (1993) · Second Generation Expert Systems · doi:10.1007/978-3-642-77927-5_24 · Cited by 160 · see citing works
Explaining control strategies in problem solving
Chandrasekaran, B., Tanner, M. C., & Josephson, J. R. (1989) · IEEE Expert · doi:10.1109/64.21896 · Cited by 137 · see citing works
The problem of causal selection.
Hesslow, G. (1988) · Cited by 86 · see citing works
Attribution in conversational context: Effect of mutual knowledge on explanation‐giving
Slugoski, B. R., Lalljee, M., Lamb, R., & Ginsburg, G. P. (1993) · European Journal of Social Psychology · doi:10.1002/ejsp.2420230302 · Cited by 79 · see citing works
Explaining differences and weighting causes
Hesslow, G. (1983) · Theoria · doi:10.1111/j.1755-2567.1983.tb00209.x · Cited by 76 · see citing works
Linegang, M. P., Stoner, H. A., Patterson, M. J., Seppelt, B., Hoffman, J. D., Crittendon, Z. B., & Lee, J. D. (2006) · Proceedings of the Human Factors and Ergonomics Society Annual Meeting · doi:10.1177/154193120605002304 · Cited by 76 · see citing works
Designs for explaining intelligent agents
Haynes, S. R., Cohen, M. A., & Ritter, F. (2009) · Int. J. Hum. Comput. Stud. · doi:10.1016/j.ijhcs.2008.09.008 · Cited by 75 · see citing works
CoCoX: Generating Conceptual and Counterfactual Explanations via Fault-Lines
Akula, A., Wang, S., & Zhu, S. C. (2020) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v34i03.5643 · Cited by 68 · see citing works
Remote Causes, Bad Explanations?
BOUWEL, J. V., & Weber, E. (2002) · Journal for the Theory of Social Behaviour · doi:10.1111/1468-5914.00197 · Cited by 61 · see citing works
Explainable Reinforcement Learning through a Causal Lens
Madumal, P., Miller, T., Sonenberg, L., & Vetere, F. (2020) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v34i03.5631 · Cited by 58 · see citing works
Self-citationDiscussion: The Contrast Theory of Why-Questions
Temple, D. (1988) · Philosophy of Science · doi:10.1086/289421 · Cited by 54 · see citing works
Hierarchical Expertise Level Modeling for User Specific Contrastive Explanations
Sreedharan, S., Srivastava, S., & Kambhampati, S. (2018) · Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2018/671 · Cited by 50 · see citing works
THE IDEA OF CONTRASTIVE EXPLANANDUM
Ylikoski, P. (2007) · Springer eBooks · doi:10.1007/978-1-4020-5581-2_3 · Cited by 31 · see citing works
Ruben, D.-H. (1987) · Analysis · doi:10.1093/analys/47.1.35 · Cited by 31 · see citing works
Model-based contrastive explanations for explainable planning
Krarup, B., Cashmore, M., Magazzeni, D., & Miller, T. (2019) · Strathprints: The University of Strathclyde institutional repository (University of Strathclyde) · Cited by 29 · see citing works
Self-citationDebugging Agent Programs with Why?: Questions
Winikoff, M. (2017) · Adaptive Agents and Multi-Agents Systems · doi:10.5555/3091125.3091166 · Cited by 19 · see citing works
Contrastive Constraints Guide Explanation‐Based Category Learning
Chin‐Parker, S., & Cantelon, J. (2017) · Cognitive Science · doi:10.1111/cogs.12405 · Cited by 18 · see citing works
A characterization of contrastive explanations computation
Kean, A. (1998) · Lecture notes in computer science · doi:10.1007/bfb0095304 · Cited by 4 · see citing works
Flagged first, then by citations
Research works printed but not linked: 4
- Sylvain Bromberger. Why-questions. In R. G. Colodny, editor, Mind and Cosmos: Essays in Contemporary Science and Philosophy , pages 68-111. Pittsburgh University Press, Pittsburgh, 1966.
- Amit Dhurandhar, Pin-Yu Chen, Ronny Luss, Chun-Chen Tu, Paishun Ting, Karthikeyan Shanmugam, and Payel Das. Explanations based on the missing: Towards contrastive explanations with pertinent negatives. In Advances in Neural Information Processing Systems , pages 592-603, 2018.
- Joseph Y Halpern. A modification of the halpern-pearl definition of causality. In Proceedings of the 24th International Joint Conference on Artificial Intelligence (IJCAI 2015) , pages 3022-3033, 2015.
- J Waa, J van Diggelen, K Bosch, and M Neerincx. Contrastive explanations for reinforcement learning in terms of expected consequences. In Proceedings of the Workshop on Explainable AI at IJCAI , 2018.
Cited by
Sorted by citations
Formalizing Trust in Artificial Intelligence: Prerequisites, Causes and Goals of Human Trust in AI
Jacovi, A., Marasović, A., Miller, T., & Goldberg, Y. (2021) · Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3442188.3445923 · Cited by 545 · see citing works
Self-citationEvaluating XAI: A comparison of rule-based and example-based explanations
van der Waa, J., Nieuwburg, E., Cremers, A., & Neerincx, M. (2021) · Artificial Intelligence · doi:10.1016/j.artint.2020.103404 · Cited by 393 · see citing works
Miller, T. (2023) · 2023 ACM Conference on Fairness Accountability and Transparency · doi:10.1145/3593013.3594001 · Cited by 172 · see citing works
Self-citationA Survey on Explainable Anomaly Detection
Li, Z., Zhu, Y., & Van Leeuwen, M. (2024) · ACM Transactions on Knowledge Discovery from Data · doi:10.1145/3609333 · Cited by 139 · see citing works
“Instead, we refer to recent papers for perspectives from HCI [203] and social science [142, 143], and leave a broader discussion of these aspects to a future article.”
BackgroundKnowing About Knowing: An Illusion of Human Competence Can Hinder Appropriate Reliance on AI Systems
He, G., Kuiper, L., & Gadiraju, U. (2023) · Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3544548.3581025 · Cited by 107 · see citing works
“For example, contrastive explanations may work in the context of human-AI decision making [51, 59].”
BackgroundContrastive Explanations for Model Interpretability
Jacovi, A., Swayamdipta, S., Ravfogel, S., Elazar, Y., Choi, Y., & Goldberg, Y. (2021) · Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2021.emnlp-main.120 · Cited by 83 · see citing works
Random forest explainability using counterfactual sets
Fernández, R. R., Martín de Diego, I., Aceña, V., Fernández-Isabel, A., & Moguerza, J. M. (2020) · Information Fusion · doi:10.1016/j.inffus.2020.07.001 · Cited by 73 · see citing works
Towards relatable explainable AI with the perceptual process
Zhang, W., & Lim, B. Y. (2022) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3491102.3501826 · Cited by 67 · see citing works
Explainable Reinforcement Learning through a Causal Lens
Madumal, P., Miller, T., Sonenberg, L., & Vetere, F. (2020) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v34i03.5631 · Cited by 58 · see citing works
Self-citationA General-Purpose Method for Applying Explainable AI for Anomaly Detection
Sipple, J., & Youssef, A. (2022) · Lecture Notes in Computer Science · doi:10.1007/978-3-031-16564-1_16 · Cited by 18 · see citing works
Do Models Explain Themselves? Counterfactual Simulatability of Natural Language Explanations
Chen, Y., Zhong, R., Ri, N., Zhao, C., He, H., Steinhardt, J., Yu, Z., & McKeown, K. (2023) · arXiv · doi:10.48550/arxiv.2307.08678 · Cited by 5 · see citing works
“One possible solution is contrastive simulation (Jacovi et al., 2021; Miller, 2021; Yin and Neubig, 2022), where a human simulator is shown the model’s output mixed with fake outputs (distractors) and selects which output is from the model based on the explanation.”
Background
- Authors
- Miller, T.
- Affiliations
- The University of Melbourne (AU)
- Corresponding author
- Miller, Tim
- ORCID
- Miller, Tim
- Venue
- The Knowledge Engineering Review
- Publisher
- Maximum Academic Press
- Published
- 2021
- Volume
- 36
- Issue
- 1
- ISSN
- 0269-8889 (ISSN-L)
- 1469-8005
- OpenAlex
- W2899652826
- Cited by
- 186
- References
- 36
- Citations by year
- 2026: 13 · 2025: 44 · 2024: 21 · 2023: 38 · 2022: 20 · 2021: 24 · 2020: 21 · 2019: 5
- FWCI
- 9.10
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- The Knowledge Engineering Review (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Physical Sciences › Computer Science › Artificial Intelligence › Bayesian Modeling and Causal Inference
- Physical Sciences › Computer Science › Artificial Intelligence › Logic, Reasoning, and Knowledge
- Keywords
- contrastive explanations, structural causal models, causal explanation, actual causes, explanatory models, classification, planning, explainable AI
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
