Chalmers on Jackson on A-Intensions
David Chalmers has an interesting post on the differences between his and Frank Jackson's versions of two-dimensionalism. It turns out that my reading of a certain passage in "Why we need A-intensions" was right: Jackson believes that truth at a world considered as actual is somehow reducible via de-rigidification to truth at a world considered as counterfactual.
Dave reports that Jackson considers this to be a byproduct of descriptivism: if rigid designators are generally equivalent to explicitly rigidified descriptions, we can simply drop the rigidification when evaluating them at other worlds.
However, this commits Jackson to a much stronger kind of descriptivism than I thought he would hold. Doesn't he say somewhere that the 'descriptions' associated with a name needn't be real linguistic descriptions, that they only need to be associated properties -- that is, functions from centered worlds to individuals, that is, A-intensions? This is a good thing to say, as it avoids the "passing the buck" problem for descriptivism, and does not make implausible demands on the expressive power of languages: it surely seems possible that a community uses a language with rigid designators but without enough non-rigid expressions to explicitly give all the analyses. (Appeal to a (necessarily) rich enough language of though would help, but this is not a move I would expect from Jackson.)
Moreover, I think it is actually better to say, as a descriptivist, that the descriptions associated with our terms are non-rigid. For these descriptions are supposed to express properties competent speakers know about the referent, whereby they 'identify' the referent, etc. And such properties are ordinary, variable functions from (centered) worlds to individuals, like the function expressed by "the teacher of Alexander". Rigidification is introduced to account for embeddings in modal operators, but the same effect can be achieved by simply saying that in modal contexts, the non-rigid descriptions associated with names always (or, better, often) have wide scope: "Necessarily, Aristotle was a philosopher" gets analysed not as "Necessarily, the teacher of Alexander was a philosopher" but as "The teacher of Alexander is such that necessarily, he was a philosopher".
I also wonder why Jackson believes that truth at a world as counterfactual is a better primitive than truth at a world as actual. If semantics is understood as a theory of the informational content communicated by users of a language, then A-intensions are the most fundamental semantic values: they are the truth-conditions of our sentences, the functions from contexts to truth values a compositional semantics is built to deliver. C-intensions can't do that job. (For one, the C-intension of "Tony Blair is the actual Prime minister" is the same as that of "Tony Blair is Tony Blair", but these sentences do not convey the same information. Secondly, the domain of C-intensions are non-centered worlds, but if our language contains non-eternal sentences we need an assignment of truth values to contexts, not only to worlds.) The purpose of C-intensions is only to account for modal embeddings. As such they are as secondary and artificial as the functions from standards of precision to truth values sometimes used to account for embeddings in operators like "roughly" and "exactly".
(I'm not sure why some philosophers find truth at a world as actual puzzling, but have no problems with truth at a world as counterfactual. I guess that's partly because philosophers tend to (wrongly) think of other worlds as if they were other, far-away planets. (E.g. philosophers often don't distinguish between Twin Earth as a planet in our world and Twin Earth as a planet in another world.) Also, from its origins in the model theory of modal logic, "true at a world" is traditionally used for truth at a world as counterfactual, so for many philosophers this is a very familiar notion, whereas the other one needs explaining.)