Dorothy Edgington thinks that conditionals are not used to express believed propositions, the way that ordinary factual assertions like this one are. Instead, they’re used to express high conditional probabilities: when I say that if Iro escapes, he will help Aang, I am not committing myself to the truth of any particular proposition; instead, I’m expressing a high credence in Iro’s helping Aang, conditional on his escaping. Jonathan Bennett, if I understand/remember right, shares this view with respect to indicatives.
Here is an objection to that view. I’m not at all sure that it’s either sound or original, much less both, but it might be.
With ordinary factual discourse, there’s something pragmatically inappropriate with statements like this:
(1) Katara is from the water tribe, but maybe she isn’t.
The explanation for this sort of oddness is something like this: you shouldn’t say that Katara is from the water tribe unless you know that she is from the water tribe, and you shouldn’t say that Katara might not be from the water tribe unless you don’t know that she is from the water tribe. So under no circumstances should it be appropriate to utter the conjunction, so it’s just right that the conjunction sounds so bad.
The problem for the NTV approach to conditionals is that this sort of explanation cannot generalize to the oddness of inappropriate conditional utterances like this one:
(2) If Iroh escapes, he will help Aang, but if he escapes, maybe he won’t help Aang.
On the Edgington view, when I utter the first conjunct, I’m expressing a high conditional probability of Iroh’s helping Aang, given that he escapes. This is totally consistent, and indeed sometimes co-felicitous, with expressing conditional uncertainty. “The conditional probability is high, but it isn’t one.”
If conditionals are just expressions of high conditional probability, then we’re owed an explanation for why utterances like (2) look contradictory in the same way that utterances like (1) do.