I’ve been reading a lot about network community structure, in the context of large-scale brain networks. One of my favorite papers on the subject was Betzel and Bassett (2016).

However, most papers I’ve read, including Betzel and Bassett (2016), partition a network into disjoint communities. I’ve also encountered some publications that use overlapping communities instead, such as Najafi et al. (2016).

This seems like an important choice. I think I agree with Najafi et al. (2016) that overlapping communities can reveal more about brain structure.

For example, consider a case of two overlapping communities. They will share a common core – their intersection. Due to how communities are generated, the intersection would necessarily be highly coherent and form a community in its own right at a different resolution.

In other words, we’d have two systems that share the same structure, reusing it for different purposes. This kind of reuse is a core principle in system architecture. It makes sense to see it in the brain.

However, the two papers I’ve cited are pretty old at this point. Has there developed a consensus about what kind of community model is best suited for studying the brain?