Here’s a somewhat puzzling phenomenon. The justification of beliefs involves two different notions. The believing itself might be justified, which we might call attitudinal justification. Sometimes epistemologists, including me, have called it ‘doxastic justification’, but I now see that this term is a bit myopic, since it blocks us from seeing the similarity between the justification of beliefs, hopes, fears, etc. All of these are capable of attitudinal justification, though only beliefs are capable of doxastic justification.
Back to the puzzling (or strange or unusual) fact about doxastic justification. We can explain how the believing itself is justified by reference to propositional justification: the believing is justified, in part, because the content of the belief is justified. That is, there is propositional justification for the content of belief, so that in the normal case, if the belief itself were absent, such an absence would be irrational (unjustified) because the proposition in question is justified for that person.
Note that this phenomenon does not generalize to other kinds of attitudinal justification. Hopes are not justified because of any justification for their content, and neither are wishes, desires, etc. Perhaps assumptions and presuppositions are, though, leading one to think that the propositional/attitudinal distinction is at home in a way among purely cognitive states that it is not at home among affective states. If that point is right, then one should expect the explanatory story of justification to be different for cognitive states than for affective ones.