It is among the most observation-based arguments we have and observation-based arguments are usually much stronger than a priori reasoning.
Well, yes, except that the observations being called on here don't actually differentiate between soul theories and non-soul theories. That was the whole point of my remark: both kinds of theory save the phenomena.
The problem with this (and related theories) is that the soul believers believe that the soul itself can live and think without the body. Much of thinking is mediated by language. I don't think a believer in soul would accept that their soul after death will be incapable of thought until God provides it a substitute pineal gland.
This just looks like a non sequitur to me. If the soul thinks on its own but only gets to communicate to other souls via a body, you get the same observational consequences. If souls have their own language of thought, then thinking might still be mediated by language and yet that mediation not require a body. (On a side note, medieval theologian-philosophers had a very serious argument about whether dis-embodied souls could sense anything. I don't think it's a huge stretch from "a soul requires a body in order to have sensations" to "a soul requires a body in order to think anything." One question you might want to ask here is whether you are interested in going after the best possible soul theory or going after the probably-confused beliefs of typical soul-believers.)
I agree with your last remark -- that soul theories are less probable a priori. But that wasn't the issue that Richard was raising. The question is how the evidence of specific functional losses due to specific, localized brain damage bears on soul theories (or dualism more generally). Evidence only comes in through the likelihood. Simplicity arguments are, I think, good arguments against souls: souls get shaved right off by Ockham's razor. But simplicity arguments are not arguments from evidence, they are arguments from method. That is, the evidence could be equally likely on two different hypotheses but one of them could still be preferred -- not because it fails to save the phenomena but because it requires extra machinery in order to do so.
If the soul thinks on its own but only gets to communicate to other souls via a body, you get the same observational consequences. If souls have their own language of thought, then thinking might still be mediated by language and yet that mediation not require a body.
I haven't said "thinking is mediated by language, therefore souls need body to think". I have said "if souls need body to understand and actively use language, then souls need body to think".
...One question you might want to ask here is whether you are interested in going
No plot spoilers here, just wanted to flag a bit of poor reasoning that shows up in Chapter 39:
This is a surprisingly common fallacy. Just because X depends on Y, it doesn't follow that X depends on nothing but Y. A phenomenon may involve more than just its most obvious failure point.
To illustrate: Suppose I'm trapped in a box, and my only way to communicate with the outside world is via radio communication. Someone on the other end argues that I don't really exist -- "There's no person beyond the radio receiver, for if there was then there wouldn't be any such thing as damaged radios!" Pretty silly, huh? But people say this kind of thing in defense of physicalism all the time.
(N.B. This is not to defend the existence of souls. It's just to point out that this particular argument against them is invalid.)