The question you raise is an important one indeed. The field is still coming to grips with it. And there may not be just one answer. Although people tend to focus on the cerebral cortex as having the exemplar architecture/connectivity for the whole brain, it is the case that some of the most critical brain areas have very different architectures. For example, the cortex is a six-layered structure, while the reticular formation in the midbrain seems to be a diffuse network of neurons with a structure that cross-cuts many other midbrain nuclei. It may be the case that we discover that structures like cortex can be described with statistical connectivity, but that the reticular formation requires much more precise wiring in order to explain. At the end of the day, it will come down to how well our models of these systems explain and predict the data we collect.