Evolution and Analysis of Model CPGs for Walking: I. Dynamical Modules Chiel Beer Gallagher 1999

This paper introduces the idea of dynamical modules, which is pretty much what it sounds like: what is a module when you are looking at a system from a dynamical perspective. This particular approach to them is very thorough, and is only tractable for networks of <=4 nodes. Which is the same as saying that everything interesting starts to happen at 5 nodes. Part of my future work will be determining for myself the relationship between this approach to structure on dynamic networks and more scaled out statistical measures. The thoroughness of this approach will account for Randy's apprehensiveness about tackling large networks.

"Are they more similar to the carburetor of a car ( a replaceable functional unit) or to flour in a cake (whose functional identity is dissolved into the final product)?"

"Specifically, if one thinks of the pattern as a limit cycle, a modular decomposition is equibalent to a shape consraint on the limit cycle-that is, the transiitiosn of the BS module are confied to edges and the transitions of the FS/FT module are confined to faces. However, unless there is a compelling reason to select for limit cycles with these particular constraints, the probability of observing such limit cycles is likely to be small."

i didn't skim this article, but I didn't scrutinize it either. The companion article I both skimmed and scrutinized.