Introduction
C) Finally, there is a risk that the kind of meta-linguistic primary intension we are discussing, as well as not allowing for a priori knowledge when this would seem desirable,
2.4. Saving the Ideal (-ized Rational Reflection)?
2.4.1. Partial Understanding
Although Chalmers [2002] explicitly advocates the internalistic nature of primary intensions, and this seems clearly the spirit of two-dimensionalism in general, maybe one could concede the falsity of 1 and still maintain a similar enough view to deserve the name of ‘two-dimensionalism’. What would be needed is a strategy to maintain 2, while giving up 1. Let us remind ourselves what 2 says:
2. There is a primary level of content which supervenes on judgements about possible cases under idealized rational reflection (i.e. putting aside confusion, lack of time, and so on).
Is 2 incompatible with the negation of 1? If the relevant judgements about possible cases supervene on the internal state of the individual, then, since supervenience is transitive, the content that supervenes on those judgments will also supervene on the internal state of the individual, and so 2 will entail 1, and obviously the negation of 1 will entail the negation of 2. Why should we think though that judgments about possible cases supervene on the internal states of the individual? Consider Bert and TwinBert; on Idealized Rational Reflection (IRR), it seems, they would assent to ‘I have arthritis in my thigh’ at exactly the same centered possible worlds; therefore, 2 predicts that utterances of that sentence have the same primary content for them. It seems these considerations will generalize to any two internally identical subjects and any sentence; so 2 seems to entail 1. Obviously then the negation of 1 would entail the negation of 2. One might think that the two-dimensionalist could individuate judgments under IRR in a different way, so that they fail to supervene on internal states, and so Bert and TwinBert do not accept the same judgements. But what would be the motivation for accepting 2? There is a trivial way in which one could defend 2. If we individuate the judgments Bert makes by their content, the content will supervene on the judgments; not only judgments under IRR, but also actual judgments. I cannot assent to a proposition I am not in a position to entertain. It is not this triviality that the two-dimensionalist intended to defend. The idea was that judgments under IRR determine the set of centered possible worlds at which an utterance is true, simply because such judgments are correct, as it were, by definition. By stipulating enough into the idea of IRR, the result of saving 2 may be achieved, but its interest is limited. Clearly, an omniscient subject would judge correctly on any possible application of any concept, and so the content of the concept is modally tied to the judgments of the omniscient subject; but it is equally clear that there is no explanatory value in this observation. We will come back to the problem of understanding IRR in the next section.
A further possibility, closer to the spirit of epistemic two-dimensionalism if not to its letter, would be to modify the framework by re-introducing the distinction, which we considered in the section 2.1 above, between partial and full understanding, and limiting both 1 and 2 to subjects with full understanding. Some of the problems that I discussed in section 1 will re-emerge; however, the two-dimensionalist is independently committed to saying that deviant experts, like McGee, would on IRR change their mind, or else they are using a
different concept altogether. I will come back to the plausibility of this strategy in the next section. But at least the distinction between partial and full understanding could be thought to have a point about the arthritis case, since there seems to be some pre-theoretical plausibility to the claim that Bert only has partial understanding of the term. Let us see how this strategy would work. Consider first Bert and TwinBert; TwinBert, it seems, on IRR will arrive at the conclusion that ‘arthritis’, as he uses it, could possibly apply to the ailment in his thigh; and that conclusion will be correct. How about Bert? It might be claimed he is not in a position to start, so to speak, IRR, because by hypothesis he has only partial understanding of his own concept. If he were to have full competence, he would arrive correctly at the conclusion that ‘arthritis’ could not possibly apply to the ailment in his thigh. But then having full understanding of a concept will not be itself a condition supervening on the internal states of a subject. Or else we need to deny that TwinBert has full understanding; but then, by analogy, for the considerations discussed in section 2.3.2 above, full understanding will be very rare indeed. The more general problem is that we don’t seem to have a grasp of ‘full understanding’ which is not just ‘understanding plus the avoidance of error’. All these questions would need to be addressed if one wanted to make use of the distinction between partial and complete understanding in the context of a two-dimensionalist theory of content.
Even if one were ready to introduce the distinction and able to clarify it in a useful way so that it could apply to Bert’s case, there would still be a major problem for extending this strategy to A’s case. Like Bert, it would have to be claimed, A has an imperfect understanding of the concept and therefore is not in a position to engage in IRR. However, A
does not need linguistic information to fill the gap between partial and complete understanding, he needs factual information, which is plausibly regarded as empirical. With the needed information, A would come on IRR to the correct conclusion. Now, though, if this implies that judgements about possible cases on IRR are not a priori, this consequence seems to undermine one main motivation for two-dimensionalism, which we mentioned before, i.e. that it is supposed to ground a priori knowledge of metaphysical possibilities. But at this point, given that the notion of primary content in the sense of 1 has already been abandoned, very little interest would seem to remain in epistemic two-dimensionalism and its motivations.
Could the two-dimensionalists deny that, on this reply, judgements under IRR lose their a priori status? One way they could try to do that is by invoking the distinction between the enabling and the evidential role of experience in grounding a priori knowledge. Experience, it is commonly held, can play the enabling role, i.e. permitting the subject to
acquire the relevant concepts, without playing any evidential role in supporting a claim, and hence without depriving the claim of a priori justification. In chapter 3, I will discuss the shortcomings of this distinction, expanding on a challenge already put forward in Williamson [2007]. But here the only claim I need is that the enabling/evidential distinction breaks down if one combines it in a particular way with a partial/full understanding distinction. On the imagined view, one can have partial understanding of a claim, without having the grounds to judge it to be true, which would be acquired only with full understanding. The subject needs, in other words, empirical information to move from a false belief to a true one. But then the resulting true belief would be paradigmatically a posteriori.