I found Brette's blog to be immensely useful and have now referred to it in a commentary, chapter and article.
http://romainbrette.fr/2013/06/
Specifically, from the beginning of our 2017 paper
"The goals of mathematical modeling in neuroscience are many and varied. For any particular study, modeling goals need to be clear as they guide our decisions during model development. Developed models can be used to generate new hypotheses and to investigate interactions across different scales. In doing this, it is helpful to consider what is meant by an explanation. Aristotle’s doctrine of the four causes (Falcon, 2015) – material, formal, efficient and final, as described and interpreted in Brette (2013) is useful. The efficient cause is what triggers the phenomenon to be explained, the material cause refers to the physical substrate of the phenomenon, the formal cause is the specific pattern (’balance’) responsible for the phenomenon, and the final cause is the function of the phenomenon. Typically, as stated by Brette, theoretical approaches tend to focus on formal and final causes, and experimental approaches on material and efficient causes. While all four causes may be needed to obtain a complete understanding, considering these causes can serve to clarify modeling goals and guide usage of developed models."