![[2602.16612] Causal and Compositional Abstraction](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.16612] Causal and Compositional Abstraction
Summary
The paper presents a formal framework for causal and compositional abstraction, emphasizing its significance in AI and scientific practice through category theory.
Why It Matters
Understanding causal and compositional abstraction is crucial for developing interpretable AI systems and improving causal inference methodologies. This research unifies various abstraction concepts, which may lead to advancements in both classical and quantum AI applications.
Key Takeaways
- Introduces a formal framework for causal abstraction using category theory.
- Distinguishes between downward and upward abstractions in causal models.
- Explores implications for explainable quantum AI through compositional models.
Computer Science > Logic in Computer Science arXiv:2602.16612 (cs) [Submitted on 18 Feb 2026] Title:Causal and Compositional Abstraction Authors:Robin Lorenz, Sean Tull View a PDF of the paper titled Causal and Compositional Abstraction, by Robin Lorenz and 1 other authors View PDF Abstract:Abstracting from a low level to a more explanatory high level of description, and ideally while preserving causal structure, is fundamental to scientific practice, to causal inference problems, and to robust, efficient and interpretable AI. We present a general account of abstractions between low and high level models as natural transformations, focusing on the case of causal models. This provides a new formalisation of causal abstraction, unifying several notions in the literature, including constructive causal abstraction, Q-$\tau$ consistency, abstractions based on interchange interventions, and `distributed' causal abstractions. Our approach is formalised in terms of category theory, and uses the general notion of a compositional model with a given set of queries and semantics in a monoidal, cd- or Markov category; causal models and their queries such as interventions being special cases. We identify two basic notions of abstraction: downward abstractions mapping queries from high to low level; and upward abstractions, mapping concrete queries such as Do-interventions from low to high. Although usually presented as the latter, we show how common causal abstractions may, more fundame...
