Dancy's Practical Reality and Expressive beliefs

So I've finally gotten around to rereading Jonathan Dancy's Practical Reality. There's something puzzling going on with respect to Dancy's view on beliefs. Errol Lord has an interesting critical note posted at the Excluded Middle, and I can see his worries. I think Dancy believes both:
A: Things believed are what motivate us to act, as well as potential justifications for why we act (in the jargon he rejects, both normative and motivating reasons are things believed).
B: Propositions do not motivate us, nor can they justify actions.

From which it appears to follow:
C: Things believed are not propositions.

But Errol rightly notes a quote from p121 which seems to oppose C, so this leaves me somewhat confused.

It seems to me that he should indeed accept C if his position is to be consistent (I noted some support for C here). Perhaps believing something requires that you stand in some relationship to a proposition, but that relationship isn't one of believing. That might be sufficient for propositions to do the work that we want them for.

But Dancy also approvingly quotes Collins, who seems (I've yet to read the original) to endorse a view according to which beliefs are more like stances, or attitudes, than representations (c.f. Wittgenstein?). Doing so might seem like the best way to account for Moore's paradox. But once we grant this, I'm not sure why Dancy wants to say that cognitivism is less wrong than non-cognitivism (in both Practical Reality and Moral Reasons). It looks like the Collins-esq view is that beliefs are themselves expressive. But if we're hoping to distinguish cognitivism from non-cognitivism in terms of expression vs description, it looks like Collins' view might erode the very possibility of cognitivism.

Can anyone help? I don't have the book on me, so I might add some quotes or references later.