Here’s another use for semantic relativism. In Indicatives and Subjunctives I defended an epistemic account of indicative conditionals. Epistemic accounts have a lot of attractions, but it’s always been a little hard to say just whose knowledge is relevant to the truth of a particular indicative. I hadn’t previously considered that we should make this evaluator-relative, but now I think about it this seems to avoid several problems.
Here are the kind of problems that I never knew how to solve. Andy knows that I’m planning to do a paper on relativism at the conference tomorrow, and knows that most of the audience are not disposed to accept relativist theories. So he utters (1)
(1) If Brian does the paper on relativism he’ll get a pretty hostile reception.
But what Andy doesn’t know is that I’ve arranged for everyone’s drinks tonight to be drugged, so tomorrow they’ll be so happy they’ll accept any crazy theory I put forward. I know that the conjunction (Brian does the paper on relativism AND Brian gets a pretty hostile reception) is false, although I don’t know that I’ll do the paper. (To avoid some complications, say that I end up consuming too many of the drugs in question and spending the day in bed recovering. This is not assumed to be known at the time.) It seems I can say that what Andy said was false.
More generally, whenever an evaluator knows ~(A & C) without knowing ~A, she can evaluate If A then C as false. And she can do this whether or not she was part of the audience of the original comment. (E.g. she can do it when she’s an eavesdropper.)
On the other hand, if the evaluator knows ~(A & ~C) without knowing ~A, i.e. she knows the material conditional A hook C, she will evaluate the conditional If A then C as true. For instance, someone who doesn’t know whether I’ll do the paper but has been told by a reliable oracle that everyone who presents a paper tomorrow will get a hostile reception, will evaluate (1) as true.
One puzzle is that it seems neither of these evaluators is making a mistake, given their evidence. Moreover, it isn’t clear that adding more evidence will move them towards the objective truth. (So far this should look like Gibbard’s Sly Pete case – it seems some days that every puzzle about indicatives ends up looking a bit like the Sly Pete case.) Another puzzle is that Andy’s knowledge, or even Andy’s audience’s knowledge, seems irrelevant to these evaluators.
Semantic relativism to the rescue! In the old paper I said that A → C is true in a context if there is some S such that some person in that context knows S, and A and S together entail C. At the time I was thinking of contexts as being like the things that fix references of indexicals. So I thought that the people ‘in the context’ were the speaker, the (intended) hearer(s) and whatever other individuals are salient. But the eavesdropper cases suggest this won’t work. What we should say is that the contexts are contexts of evaluation, and who is ‘in’ the context can vary from one context of evaluation to another.
Obviously there would need to be a lot of work on the pros and cons of this view before we accepted or rejected it, but I think that if we take epistemic theories of indicative conditionals seriously, and I for one very much do so, then we should take relative truth accounts of these conditionals seriously.
I imagine this is already covered in something or other John MacFarlane has written, but if not this can go down as another contribution of mine to the ‘post-modernization’ of analytic philosophy.