[2508.11214] How Causal Abstraction Underpins Computational Explanation

Abstract:Explanations of cognitive behavior often appeal to computations over representations. What does it take for a system to implement a given computation over suitable representational vehicles within that system? We argue that the language of causality -- and specifically the theory of causal abstraction -- provides a fruitful lens on this topic. Drawing on current discussions in deep learning with artificial neural networks, we illustrate how classical themes in the philosophy of computation and cognition resurface in contemporary machine learning. We offer an account of computational implementation grounded in causal abstraction, and examine the role for representation in the resulting picture. We argue that these issues are most profitably explored in connection with generalization and prediction.

View PDF HTML (experimental) Abstract:Explanations of cognitive behavior often appeal to computations over representations. What does it take for a system to implement a given computation over suitable representational vehicles within that system? We argue that the language of causality -- and specifically the theory of causal abstraction -- provides a fruitful lens on this topic. Drawing on current discussions in deep learning with artificial neural networks, we illustrate how classical themes in the philosophy of computation and cognition resurface in contemporary machine learning. We…

saved by

• Mackenzie Dion

related reading

• [2508.11214] How Causal Abstraction Underpins Computational Explanationarxiv.org
• Faithful, Interpretable Model Explanations via Causal Abstraction | SAIL Blogai.stanford.edu
• Verbalizable Representations Form a Global Workspace in Language Modelstransformer-circuits.pub
• Circuit Tracing: Revealing Computational Graphs in Language Modelstransformer-circuits.pub
• [2604.21691] There Will Be a Scientific Theory of Deep Learningarxiv.org
• [2511.08579] Training Language Models to Explain Their Own Computationsarxiv.org
• On the Learnability of Physical Concepts: Can a Neural Network Understand What’s Real?arxiv.org
• [2604.21691] There Will Be a Scientific Theory of Deep Learningarxiv.org
• Training Language Models to Explain Their Own Computationsarxiv.org
• Causal Scrubbing: a method for rigorously testing interpretability hypotheses [Redwood Research] — LessWronglesswrong.com
• The Building Blocks of Interpretabilitydistill.pub
• Weakly supervised causal representation learningarxiv.org