I think part of the reason I favor integrative models in biology is that the way I've learned to understand complex things is through syntactically complex linked structures like program code that, in a (somewhat loose) sense, recapitulate the structure of what they describe. The way I've learned biology up to this point is by linking together, in my head, a variety of narrative descriptions and figures that I have to integrate in a loosely structured manner. An integrated, multi-level model (a schema, perhaps) seems like an effective alternative, though it may ultimately only be legible to the person who creates it.