Structural Causal Bottleneck Models

Abstract

We introduce structural causal bottleneck models (SCBMs), a novel class of structural causal models. At the core of SCBMs lies the assumption that causal effects between high-dimensional variables only depend on low-dimensional summary statistics, or bottlenecks, of the causes. SCBMs provide a flexible framework for task-specific dimension reduction while being estimable via standard, simple learning algorithms in practice. In addition to an analysis of identifiability in SCBMs, we provide experimental results evidencing that we can estimate bottlenecks in practice. We also demonstrate the benefit of bottlenecks for effect estimation in low-sample transfer learning settings.

@inproceedings{bingStructuralCausalBottleneckModels2025,
 author = {Simon Bing and Jonas Wahl and Jakob Runge},
 booktitle = {UAI 2025 Workshop on Causal Abstractions and Representations. Causal Abstractions and Representations (CAR), located at UAI 2025, July 25, Rio de Janeiro, Brazil},
 month = {6},
 publisher = {OpenReview},
 title = {Structural Causal Bottleneck Models},
 url = {https://openreview.net/forum?id=EFNvKMSITB},
 year = {2025}
}