The circuits that light up do so stochastically as a function of inputs. If you clamp these, you get neural activation that is not mysterious, even if unknowable by you (or anyone). That's the heart of the problem with the "unknowable" critique.