Concept-Based Approaches (Concept Bottleneck Models)

Assumptions and requirements

CBMs impose semantic constraints (predefined human concepts), while interpretable-by-design models impose structural constraints (sparsity, modularity) that make the model intrinsically easier to interpret.

Intended applications

In general, these methods do not provide causal evidence about the role of those concepts in model computation, although some probing variants aim to support stronger causal claims, such as amnesic probing (Elazar et al., 2021). Nevertheless, they remain valuable tools for auditing, localisation, and early detection tasks.

CBMs can be more appropriate when the set of concepts of interest is known. Interpretability-by-design models may be preferable when the goal is to discover what the model has learned—including features that were not preset in advance.

Limitations

Concept Bottleneck Models (CBMs; Koh et al., 2020) constrain computation through a predefined bottleneck of interpretable concepts, offering built-in transparency and allowing for interventions, but traditionally requiring a complete set of predefined concepts and often suffering from an “accuracy–interpretability trade-off” when concept definitions are insufficiently expressive.

For safety applications, where unknown failure modes are a core concern, discovery-oriented methods may be more useful than those that require specifying concepts upfront.

Related methods

• Probing Classifiers
• Representation Engineering and Steering (Steering Vectors)

Compare side by side →

Key papers

Cited in §3.2.

• ■ Atticus Geiger, Hanson Lu, Thomas Icard, and Christopher Potts. Causal abstractions of neural networks. In M. Ranzato, A. Beygelzimer, Y. Dauphin, P.S. Liang, and J. Wortman Vaughan (eds.), Advances in Neural Information Processing Systems, volume 34, pp. 9574–9586. Curran Associates, Inc., 2021. URL https://proceedings.neurips.cc/paper_files/paper/2021/file/4f5c422f4d49a5a807eda27434231040-Paper.pdf.
• ■ Atticus Geiger, Zhengxuan Wu, Hanson Lu, Josh Rozner, Elisa Kreiss, Thomas Icard, Noah Goodman, and Christopher Potts. Inducing causal structure for interpretable neural networks. In Kamalika Chaudhuri, Stefanie Jegelka, Le Song, Csaba Szepesvari, Gang Niu, and Sivan Sabato (eds.), Proceedings of the 39th International Conference on Machine Learning, volume 162 of Proceedings of Machine Learning Research, pp. 7324–7338. PMLR, 17–23 Jul 2022. URL https://proceedings.mlr.press/v162/geiger22a.html.
• ■ Pang Wei Koh, Thao Nguyen, Yew Siang Tang, Stephen Mussmann, Emma Pierson, Been Kim, and Percy Liang. Concept bottleneck models. In Hal Daumé III and Aarti Singh (eds.), Proceedings of the 37th International Conference on Machine Learning, volume 119 of Proceedings of Machine Learning Research, pp. 5338–5348. PMLR, 13–18 Jul 2020. URL https://proceedings.mlr.press/v119/koh20a.html.
• ■ Mateo Espinosa Zarlenga, Pietro Barbiero, Gabriele Ciravegna, Giuseppe Marra, Francesco Giannini, Michelangelo Diligenti, Zohreh Shams, Frederic Precioso, Stefano Melacci, Adrian Weller, Pietro Lio, and Mateja Jamnik. Concept embedding models: beyond the accuracy-explainability trade-off. In Proceedings of the 36th International Conference on Neural Information Processing Systems, NIPS ’22, Red Hook, NY, USA, 2022. Curran Associates Inc. ISBN 9781713871088.
• ■ Gabriele Dominici, Pietro Barbiero, Mateo Espinosa Zarlenga, Alberto Termine, Martin Gjoreski, Giuseppe Marra, and Marc Langheinrich. Causal concept graph models: Beyond causal opacity in deep learning. In The Thirteenth International Conference on Learning Representations, 2025. URL https://openreview.net/forum?id=lmKJ1b6PaL.
• ■ Moritz Vandenhirtz, Sonia Laguna, Ričards Marcinkevičs, and Julia E Vogt. Stochastic concept bottleneck models. In The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024. URL https://openreview.net/forum?id=iSjqTQ5S1f.
• ■ Ruochen Zhao, Tan Wang, Yongjie Wang, and Shafiq Joty. Explaining language model predictions with high-impact concepts. In Yvette Graham and Matthew Purver (eds.), Findings of the Association for Computational Linguistics: EACL 2024, pp. 995–1012, St. Julian’s, Malta, March 2024c. Association for Computational Linguistics. . URL https://aclanthology.org/2024.findings-eacl.67/.
• ■ Fahim Dalvi, Abdul Rafae Khan, Firoj Alam, Nadir Durrani, Jia Xu, and Hassan Sajjad. Discovering latent concepts learned in BERT. In International Conference on Learning Representations, 2022. URL https://openreview.net/forum?id=POTMtpYI1xH.
• ■ Yue Yang, Artemis Panagopoulou, Shenghao Zhou, Daniel Jin, Chris Callison-Burch, and Mark Yatskar. Language in a bottle: Language model guided concept bottlenecks for interpretable image classification. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 19187–19197, 2023.
• ■ Tuomas Oikarinen, Subhro Das, Lam M. Nguyen, and Tsui-Wei Weng. Label-free concept bottleneck models. In The Eleventh International Conference on Learning Representations, 2023.
• ■ Xuemin Yu, Fahim Dalvi, Nadir Durrani, Marzia Nouri, and Hassan Sajjad. Latent concept-based explanation of NLP models. In Yaser Al-Onaizan, Mohit Bansal, and Yun-Nung Chen (eds.), Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pp. 12435–12459, Miami, Florida, USA, November 2024. Association for Computational Linguistics. . URL https://aclanthology.org/2024.emnlp-main.692/.
• ■ PreprintXuemin Yu, Ankur Garg, Samira Ebrahimi Kahou, and Hassan Sajjad. Vector quantized latent concepts: A scalable alternative to clustering-based concept discovery, 2026. URL https://arxiv.org/abs/2602.02726.

Other methods in The Interpretability Landscape

Behavioural Methods · Input Attribution and Saliency Methods