There have been some discussions, starting here and continuing here, about the relationship between certain epistemological worries for modal realism and some other worries for consequentialism. I find the discussion of modal realism puzzling. I certainly don't want to saddle Lewis with outright pragmatism and verificationism, but he did have a certain sympathy with both of those views, and as a result I'm not sure the kind of case that's described makes sense on the Lewis view. Certainly his argument for modal realism is that it's a very useful theory. Lewis doesn't consider any kind of evidence for or against modal realism except evidence of this kind, and it is hard to see how there could be any other evidence for or against it (for the usual causal reasons). It's not clear that it even makes sense to speak of the oracle as knowing there are no other concrete worlds if it isn't based on something like this; the oracle cannot, of course, have causal contact with the other worlds or anything of that sort. Perhaps the oracle doesn't know anything, but only says things that are true. But how do we know what the oracle's statements mean? How do we know the oracle is talking about modal reality? Again, for Lewis, the only evidence concerning modal reality outside the actual world is usefulness; we could understand the oracle as speaking of modal reality if the oracle could be interpreted as talking about useful theories of modality, but we can't, for example, understand the oracle as speaking about modal reality because of any causal contact between the oracle and any modal reality (again, that's metaphysically impossible for Lewis, so we can't even give that power to hypothetical oracles in thought experiments).
Perhaps it would be more useful to put the point by analogy. What should anyone's reaction be if the oracle tells us there are no sets, or that there are some sets, but not others (perhaps there are all other kinds of sets, but no sets contain donkeys as elements, by analogy with Richard's case). How would that affect our beliefs about sets? Would the latter even make sense?
Of course, one obvious response is to reject the analogy on the basis that sets are abstract and the Lewis worlds are supposed to be concrete, but another part of the Lewis argument for modal realism is that the concrete/abstract distinction is too messy to do any serious philosophical work, so anybody who took this line in response to Lewis would need to answer those arguments and provide a convincing account of this concrete/abstract division and why it matters to comparing sets to the Lewis worlds.
One shouldn't take "oracle" talk too seriously. It's really just a shorthand to get us thinking about conditional probabilities/credences. E.g. Cr(Modal Realism | only one concrete world) is near-zero, whatever one's prior credence in Modal Realism. By contrast, Cr(Utilitarianism | murdering Bob increases happiness) = Cr(Utilitarianism).
Posted by: www.philosophyetc.net | December 12, 2009 at 01:45 PM
Hmmm. Not sure exactly why Cr(modal realism|only one concrete world) shouldn't be equal to Cr(modal realism). Modal realism with only one concrete world sounds like Quine's view, and I'm not sure that lowering one's credence in modal realism, rather than appropriately increasing one's credence in skepticism/nihilism about modality generally, is the rationally appropriate response for someone who puts significant credence in modal realism and who obtains evidence that there's only one concrete world.
I'm also a bit skeptical about trying to apply probabilities to a priori matters; I suppose there has to be some way to talk about appropriate credence in such matters, but it seems dangerous to assume that we can just talk about them in exactly the same way we talk about contingent matters.
Posted by: Aaron Boyden | December 12, 2009 at 02:02 PM
Lewis, at least, argued for modal realism on the basis of various theoretical payoffs (accounts of propositions, properties, etc.) that would be completely undercut by the loss of a plenitude of worlds.
Things might be different for someone who accepted modal realism on different grounds, e.g. skepticism about abstracta. They would then be subject to the epistemic objections Ross raises. If my account (of when it's legitimate for a theorist to reverse the metaphysical and epistemic orders of explanation) is correct, then it would *not* be legitimate for your kind of modal realist to offer the Lewisian inference, "talking donkeys are possible, therefore there's a concrete world containing a talking donkey".
Posted by: www.philosophyetc.net | December 12, 2009 at 02:55 PM