The problem is that we are seriously limited in what we can actually do/use if we base psychology entirely on this premise. I mean this both conceptually and technologically. We haven't fully mapped out the brain, we don't know about all the neurotransmitters out there, and we don't even know what fMRI shows us exactly, so we gotta be realistic with our expectations given our current limitations. Likewise, we don't actually fully know how most psychoactive compounds work to produce the effects they do, much less 'designer medications'. This isn't a reason to dismiss their usefulness though. There's a difference between being 'limited', and being 'wrong'. You could've made a similar argument about physics back before we really understood the atom (again, due to limitations).
Additionally, the notion of 'wrong' here is only relative to the philosophical theory of falsification, which is good, but has it's own set of gotchas and nuances that others have tried to address:
http://www.loyno.edu/~folse/Lakatos.html
http://philosophy.wisc.edu/forster/papers/Lakatos.htm
http://coraifeartaigh.wordpress.com/2011/02/13/the-philosoph...
http://andrewgelman.com/2004/10/14/bayes_and_poppe/
So while good, this definition of science is still far from a stable building-block to deem all nonconforming studies unequivocally 'wrong'.