David Lewis is notoriously known for believing in the existence of concrete possible worlds equally real as our own. It came to be known as (genuine) modal realism. His systematic presentation and defence of the view appeared in On the Plurality of Worlds (Blackwell, 1986). But when exactly did he adopt the view? And what exactly was the original motivation? In this letter to W.V. Quine we get the answer!? Lewis’ letter is in response to Quine sending his Foreword for Lewis’ first book Convention (Harvard UP, 1969). Lewis is reacting to Quine’s final paragraph on p. xii. There Quine writes:
The problem of distinguishing between analytic and synthetic truths was apparently one motive of the study. In the end, Lewis concludes that the notion of convention is not the crux of this distinction. He does not for this reason find the analyticity notion unacceptable, however. On the contrary, he wonders why I questioned the notion in the first place, any more than the notion of truth. This is not the place to repeat my reason for regarding truth as much the firmer notion; anyway, he must have found my reason unconvincing. He ends up rather where some began, resting the notion of analyticity on the notion of possible worlds. His contentment with this disposition of the analyticity problem makes one wonder, after all, how it could have been much of a motive for his study of convention; but we may be thankful for whatever motives he had. For in the course of the book the reader comes to appreciate convention, not analyticity, as a key concept in the philosophy of language (1969, p. xii).
If you have any reactions, please record them in the comments below. (© Estate of David K. Lewis.)
This letter is obviously very interesting in providing information about Lewis’s shift from a less-committal view of possible worlds to the view that possible worlds are “unreduced”, but I don’t think it quite shows when he came to adopt the view now known as (genuine) modal realism. From the look of this letter, at this stage it isn’t clear that he has even taken a stand on whether worlds are abstract or not. Or, given the unclarity of that notion, the related question of whether possible objects have the features they possibly have: e.g. whether possible blue swans are _blue_ or _swans_. The fact that he didn’t intend the Counterpart Theory paper to force that interpretation is some evidence that he hadn’t committed on that question one way or another.
He does, in this letter, reject the idea that possible worlds must be “reduced to state descriptions, maximal consistent sets of sentences, or the like”: but for all this letter shows, he still had not rejected options that he later came to call “magical ersatz” options: that worlds were abstract and “represented” propositions in some more sui generis way.
I suspect that when he did settle against the “linguistic ersatz” options such as construing worlds as maximal consistent sets of sentences, it was not long before he accepted genuine modal realism of the sort he defended in his 1986 book. But I have a hunch that it was after he wrote this letter. In this letter, after all, he talks of worlds being not yet “clear enough to be content with”, and mentions again the “unclarity of possible worlds”. By 1986 he seemed more comfortable in thinking that we had a clear notion of possible worlds.
Very interesting! It’s not clear to me from this letter, though, that Lewis was adopting modal realism here. He doesn’t say anything about the possible worlds being concrete or the same sort of thing as the actual world. (Maybe he did already hold that view, or was leaning toward it, and was just a bit embarrassed to tell Quine!) All he seems to be saying is that he now thinks that possible world talk is in good enough standing to use philosophically without having an account of possible worlds, and that seems quite compatible with not holding to full-on (genuine) modal realism.
In case anyone is wondering why I am apparently being so redundant: the comment from Daniel Nolan had not yet appeared when I wrote mine.
I’m wondering whether Lewis’s presentation of counterpart theory in ‘Counterpart theory and quantified modal logic’ (J.Phil. 1968) in effect commits him to (genuine) modal realism. If the counterpart relation is determined by similarity, wouldn’t any counterpart of a concrete object have to be concrete? It’s a slightly different case, but Lewis explicitly floats the idea that all of a corporeal object’s counterparts are corporeal (p.122). That would seem to entail that they are concrete. I guess you could think of the cp relation as a matter of similarity between (say) an actual elephant and what is represented by some set of sentences or whatever, but then similarity would (in this case) have to be a relation between an existent thing and a non-existent thing, which doesn’t sound like the sort of thing Lewis would think! Am I missing something? (Fraser mentioned in our reading group today that Lewis considers this kind of objection to Armstrong’s combinatorial theory of possibility in a review of DA’s book and is not much taken with it, but I haven’t chased that up …)
Another thought I had about all this is that while it might be true that worlds qua unreduced material could be abstract simples and thus what Lewis says here is compatible with the ontology of a magical ersatzer Lewis at this point is a set-theoretic nominalist. His ontology consists of sets/classes and individuals. So from his perspective if he adds worlds qua unreduced material he would naturally incorporate them into his pre-existing ontological framework or at least prefer to do so before adding extra primitive ontology.
And of course he goes on a bit later to do just this explicitly.