In what situation WOULD you be able to expect to extract "something useful" about "our cognitive fucntions or our brain disorders"? It seems silly to think we could learn anything about such complex things without understanding something simpler first, hence the approach of validating models of simpler structures.

What would you recommennd?