I think you can factor it into two parts, one of which is normative and the other of which is positive. The positive question is whether the observation makes it more likely that the treatment is helpful. And the normative part is to what extent we should react by feeling confident.
I think you should be careful about this factorization. It seems like a good one at first, but I think we should question what "more likely" means. As a matter of fact, the treatment either is helpful or it isn't - so if "likely" is talking about physical chance, the chance is already either 1 or 0 and it stays that way, so nothing gets more or less likely as a result of the observation. If "likely" is about frequencies, then it doesn't even enter into the question - again, either the treatment is helpful or it isn't, and the observation doesn't affect anything.
So you have to come up with some other meaning of what "likely" means here. The most natural one is some notion of "evidential probability", which is supposed to be a notion of probability given some evidence. But if that's what you're going to say, then I don't know what that might mean other than what someone's beliefs should be if they have the relevant evidence. If it's not that, then it seems that you're positing some primitive notion of evidential probability that isn't directly connected to anything else.
Of course, you could just go my way and say that "likely" here is a Bayesian probability, and that to say that an observation makes something more likely for someone is to say that that person should become more confident in that thing.
But maybe there's some other understanding of "more likely" that I'm missing, which is not normative, and also refers to something that can change as a result of evidence, in a way that chances and frequencies don't.
Re: the is-ought gap
I think you should be careful about this factorization. It seems like a good one at first, but I think we should question what "more likely" means. As a matter of fact, the treatment either is helpful or it isn't - so if "likely" is talking about physical chance, the chance is already either 1 or 0 and it stays that way, so nothing gets more or less likely as a result of the observation. If "likely" is about frequencies, then it doesn't even enter into the question - again, either the treatment is helpful or it isn't, and the observation doesn't affect anything.
So you have to come up with some other meaning of what "likely" means here. The most natural one is some notion of "evidential probability", which is supposed to be a notion of probability given some evidence. But if that's what you're going to say, then I don't know what that might mean other than what someone's beliefs should be if they have the relevant evidence. If it's not that, then it seems that you're positing some primitive notion of evidential probability that isn't directly connected to anything else.
Of course, you could just go my way and say that "likely" here is a Bayesian probability, and that to say that an observation makes something more likely for someone is to say that that person should become more confident in that thing.
But maybe there's some other understanding of "more likely" that I'm missing, which is not normative, and also refers to something that can change as a result of evidence, in a way that chances and frequencies don't.