Can we characterize more formally the idea Scanlon's distinction is meant to capture?
Certainly, Scanlon's idea presupposes two ideas:
1. the moral worth of a life (or of the choices that the agent makes when he lives) is one of the aspect of a life's "total" worth, or its "choiceworthiness all-things-considered"
2. the moral worth of a life can be augmented by a sacrifice of well-being.
Claim 1 could be regarded as true ex-costructo : it is part of the way the concept of choiceworthiness is introduced that it includes what we may call "the moral worth of a life". But the claim also follows from the meaning of "choiceworthiness" and our intuitions about what has"worth". Many philosophers (Nietzsche being a relevant example of the opposite) think that acting morally gives our life some sort of objective worth.
Claim 2, is, I believe, really fundamental. Saying that self-sacrifice can have moral worth, I believe, entails the truth of the following claim:
(that I shall call M1, because there is an even more basic principle that I shall call M)
M1: there can be cases in which morality requires from person p that she Fs, even when F-ing decreases p’s well-being or fail to maximize it, p desires to F (despite her awareness of well-being cost) and she is committed to respect values which justify the requirement in question