There's nothing wrong with presenting (what I consider) the same, relevant dynamic in a hypothetical context in order to make a point about the conclusions you should draw in a different one.
If the relevant dynamic in the actual situation is really the same, then why not just refer to the actual situation? If you have to "amplify" the problematic elements, then you are giving yourself the burden of proving that you haven't amplified them to the point that they yield a different conclusion than the original setting would.
In religions, it's hard to see the disconnect between the professed beliefs and the adherent's actual internal predictive model of reality. My example made the disconnect obvious, and also showed the surrounding motives that give evidence as to what they really believe.
If the disconnect is "hard to see" in the case of religion, then you ipso facto need strong evidence to establish that the disconnect exists. By moving to a situation where the disconnect is easier to see, so that less evidence is necessary, you are moving to a situation where your burden of proof is less. Therefore, establishing your claim in your hypothetical does not suffice to establish your claim in the original situation.
In such a scenario, I would conclude that the person is using the term "believe" differently than the term is normally used. You would conclude that the person knowingly puts himself in a situation where he expects to die, despite not wanting to die.
Your conclusion would be one real possibility. Another possibility is that, although he doesn't want to die, he prefers it to sleeping somewhere other than his bed. Perhaps sleeping elsewhere seems, to him, a fate worse than death. Since I'm manifestly dealing with a crazy person, that remains a real possibility, at least until I learn more about how he thinks.
The more someone professes different beliefs from yours, the more evidence there is that their mind works differently from yours in some crucial respect, and so the less credit you should give to your mental model of them.
In approximate order of your objections to the hypothetical:
Tyrrell, it's not my fault if you can't handle reasoning from hypotheticals, but it certainly doesn't make that form of reasoning -- used in the rest of the civilized world -- off-limits. If you think the analogy I'm making doesn't hold, you can politely show where it breaks down. I had already attempted to speak to the specific situation in dispute -- about religion -- but in that case it's less obvious how one's actions aren't following from one's beliefs if the professed beliefs are the real ...
Anticipation and faith are both aspects of the human decision process, in a sense just subroutines of a larger program, but they also generate subjective experiences (qualia) that we value for their own sake. Suppose you ask a religious friend why he doesn’t give up religion, he might say something like “Having faith in God comforts me and I think it is a central part of the human experience. Intellectually I know it’s irrational, but I want to keep my faith anyway. My friends and the government will protect me from making any truly serious mistakes as a result of having too much faith (like falling into dangerous cults or refusing to give medical treatment to my children)."
Personally I've never been religious, so this is just a guess of what someone might say. But these are the kinds of thoughts I have when faced with the prospect of giving up the anticipation of future experiences (after being prompted by Dan Armak). We don't know for sure yet that anticipation is irrational, but it's hard to see how it can be patched up to work in an environment where mind copying and merging are possible, and in the mean time, we have a decision theory (UDT) that seems to work fine, but does not involve any notion of anticipation.
What would you do if true rationality requires giving up something even more fundamental to the human experience than faith? I wonder if anyone is actually willing to take this step, or is this the limit of human rationality, the end of a short journey across the space of possible minds?