I endorse this refinement. What brain damage demonstrates is not dependency of talking on the brain, but that the complex computations of thought can be damaged in internal detail by damaging a specific brain part, whereupon its outputs to other parts of thought are damaged. This is strong evidence that the brain is doing the internal computations of thought; it is part of the inner process producing thoughts. The radio hypothesis, in which the output is produced elsewhere and received, decisively fails at that point.

We could suppose that you had a hundred soul-parts, all of which can only communicate with each other through brain-area radio transceivers which receive a call from one soul-part, and then retransmit it to another. But leaving out the epicycleness of this idea, the degree to which it contradicts the intuitive notion of a soul, and its, if you'll pardon the phrase, sheer stupidity, the end result would still be that destroying the brain leaves the soul incapable of thought. You're not likely to find a remotely reasonable hypothesis, even in the Methodsverse where magic abounds, by which the internal parts of a thinking computation can be damaged by damaging the brain, and yet removing the whole brain leaves the soul capable of internal thinking.