Characterizing Network Explanations in Complex Systems: Patterns, Processes and Epistemic Values

Network modeling is pervasive in the sciences that address complex systems, especially when they require big data. In the philosophical literature some argued that network analysis in some cases gives rise to explanations, often termed topological. In this paper I use several classical distinctions used in the scientist’s talk about modeling in order to characterize such purported explanations. I show that the distinctions between phenomenological and mechanistic models, as well as the claim that topological explanations are abstractions to be connected to genuine mechanistic explanations, don’t capture what’s at stake with these purported explanations. Finally, I consider the distinctions between epistemic values inspired by Levins (1966) and show that, contrary to what may seem intuitive, the difference between topological and mechanistic explanations is not one between genericity and realism. I call for the recognition of the plurality of uses of network modeling and explanation.