I was looking at a paper of Pruss in the latest issue of Faith and Philosophy, and he ends it with an interesting remark. He says that probabilities are in general a problem for non-cognitivist accounts. For instance, he says, that if emotivism is right, then it's hard to assign a sense to abortion's being wrong having such-and-such a probability.
I was thinking about this, and it could make a useful template for anti-non-cognitivist arguments. For instance, suppose you think that necessity claims are an expression but not assertion of ungiveupability. But what sense, then, would there be in saying that the evidence indicates with moderate probability that necessarily freedom requires alternate possibility?