That's a false dilemma. It is the same with any process. On a functional level, input output correlation is defining. On a mechanical level, we may recognize a

That's a false dilemma. It is the same with any process. On a functional level, input output correlation is defining. On a mechanical level, we may recognize a structure equal to another black box. Indeed, dividing a structure into separate components is the key technique of learning. The problem is just how highly complex the brain is and that the "output" here is rather different from that of a blood cell or a dna strand. In fact, being based on and perhaps even interacting with those and other building blocks, is what makes the brain so abstract. So speech modeled as the output of dna is necessairly higher up the chain of abstraction than a microscopic ribose enzyme or what. But it appears to as us macro precisely because we don't understand all of it. You have effect and cause reversed. If we knew where to look better, it wouldn't be a problem. Chances are experts know way better, but there are really few who know enough to see a bigger picture and can guide the search. I'm trying to say they are breaking it open, inserting probes and imaging it working. But as far as the "output" is concerned, everyone is naturally aspiring to be an expert. A lot of unguided superficial theorizing may be informative as well, to great length as it concerns psychology and related fields.