Non-causal explanation in science

Non-causal and causal explanation in science are unified under an extension of James Woodward's manipulationist account of causal explanation. Scientific explanation is about capturing and representing the modal structure of the world. Both causal and non-causal explanations
often involve implicit and explicit modelling steps. Manipulationism can be extended to models; models have an endogenous set of rules that allow the specification of model analogues of manipulations and explanatory
invariances. A pluralist view of explanation is defended. Models can explain despite, and sometimes because of, ineliminable fictions they contain.
These fictions do not undermine an ontic account of explanation if the intuitions informing ontic sensibilities are suitably disaggregated. Ontic explanation is a two-levelled process. On the one hand, if we can connect
variables with objective modal connections and those variables correspond to entities or properties of entities, or real structures in the world, then we have a correspondence explanation. If, on the other hand, we can still objectively produce modal connections but the ontology of the model is strictly false, then the variable terms do not correspond to real entities. It only appears as if they do, and we have a quasi-explanation. A quasiexplanation is only applicable in a certain empirical domains. This disaggregation has implications for realism. Often explanations will only license an attenuated realistic-or surrealistic- attitude to the ontology of
models. This extension of manipulationism to models is far reaching, and as well as unifying many types of scientific explanation it also has applications in pure mathematics.