Approximate Causal Abstraction
This work addresses the need for more realistic causal modeling in scientific domains where exact abstractions are insufficient, though it is incremental as it builds on prior exact frameworks.
The paper tackles the problem of extending exact causal abstraction to approximate causal models, which better reflect real-world discrepancies between low- and high-level descriptions, and provides an account for how one causal model approximates another, including extensions to probabilistic models.
Scientific models describe natural phenomena at different levels of abstraction. Abstract descriptions can provide the basis for interventions on the system and explanation of observed phenomena at a level of granularity that is coarser than the most fundamental account of the system. Beckers and Halpern (2019), building on work of Rubenstein et al. (2017), developed an account of abstraction for causal models that is exact. Here we extend this account to the more realistic case where an abstract causal model offers only an approximation of the underlying system. We show how the resulting account handles the discrepancy that can arise between low- and high-level causal models of the same system, and in the process provide an account of how one causal model approximates another, a topic of independent interest. Finally, we extend the account of approximate abstractions to probabilistic causal models, indicating how and where uncertainty can enter into an approximate abstraction.