I'll concede the point about the robot who's goal is to die in 5 minutes. I could try to come up with more and more absurd examples of agents without need for modeling their environment, but that gets far enough from the examples we know in real life (like people) that it's somewhat silly to base any decisions on what language to use on them. If you're not appropriately responsive to evidence, then you're not even having beliefs ... to count as a believer at all one needs to be approximately Bayesian, and to be an ideal believer, one should be as perfectly Bayesian as possible.
Depending on how you mean this, this actually sounds very radical. Although interesting. Given this view, what would you say to someone who claims to believe things on faith? Faith doesn't seem to be even a logical possibility in this way of thinking about it. And I'm somewhat sympathetic to that, although I think it might really piss of a lot of "people of faith" to hear that. Especially the more adament ones who argue that you're not a "true believer" unless you have faith. It sounds like you are saying exactly the opposite. You are not a true believe if you have any faith whatsoever!
Re: the is-ought gap
If you're not appropriately responsive to evidence, then you're not even having beliefs
...
to count as a believer at all one needs to be approximately Bayesian, and to be an ideal believer, one should be as perfectly Bayesian as possible.
Depending on how you mean this, this actually sounds very radical. Although interesting. Given this view, what would you say to someone who claims to believe things on faith? Faith doesn't seem to be even a logical possibility in this way of thinking about it. And I'm somewhat sympathetic to that, although I think it might really piss of a lot of "people of faith" to hear that. Especially the more adament ones who argue that you're not a "true believer" unless you have faith. It sounds like you are saying exactly the opposite. You are not a true believe if you have any faith whatsoever!