Lots of predicates are claimed to be context-sensitive. Familiar examples include 'know', 'come', 'go', 'nearby', 'local', 'left', 'right', 'tall', and 'rich'. Here I shall mostly be concerned with a different group of predicates which may or may not turn out to be context-sensitive, namely predicates such as 'is the brightest object in the evening sky'. At first glance predicates like this one seem to be context-sensitive. After all, nothing can be the brightest without being the brightest to someone. Besides your evening sky isn't my evening sky. So, a good guess is that this sort of predicate is context-sensitive. This would mean that its 2-intension would vary across contexts. So, for some speakers, its extension may include Venus. For others not. But how then do we account for its 1-intension (its Fregean sense). One possibility is to take the predicate's 1-extension at a scenario to be relative to speakers. The idea is simple enough: scenarios are worlds with or without a centered individual. What counts as the brightest in the evening sky will be relative to the individual in the center (I am pretty sure I have seen this account of the predicate in David Chalmers' work, but when I went back to look I couldn't find the place).
But this sort of approach comes at a cost. For the individual in the center is also the 1-extension of 'I'. So, it would seem that I can know the following things on a priori grounds: 'Hesperus and I both exist iff Hesperus is the brightest object in the evening sky', 'if Hesperus and I both exist, then I am not blind', and 'If Hesperus and I both exist, then I am not a zombie'. Let's run the argument for the first case. 'Hesperus' (suppose) was acquired via the description 'the brightest object in the evening sky'. Now, suppose the scenario has no center, then the left-hand side is false. But so is the right-hand side. For there is no brightest object if the centered individual is not there to observe it. Suppose scenarios have centers. Then the left-hand side is true just when the right-hand side is. But now if these conditionals are true at all scenarios, then by definition one is in a position to know them on a priori grounds. But these conditionals do not seem a priori.
What to do? Well, we can simply reject the supposition that 'is the brightest object in the evening sky' is context-sensitive. Ignoring 'the evening sky' part of the predicate, there is a simple way of accounting for such predicates as expressing dispositional properties. x is the brightest object in the evening sky iff x would appear as the brightest in the evening sky to a standard observer under standard circumstances. This will get around the problem. But there is something odd about this. 'Bright', 'red', 'blue', and so on, no doubt express dispositional properties. It is easy to imagine what 'standard conditions' means here. It means something like 'normal lighting conditions, no obstructions, no evil deceivers, etc.'. But standard conditions do not usually include information about location. What would a normal or standard location be anyway? To account for the 'the evening sky' part it thus seems that we need to contextualize after all. Perhaps a mixed approach will do. On a mixed approach, 'x is the brightest object in the evening sky' is true at a scenario iff x is the brightest object to a standard observer located at the center under standard conditions. Even though we appeal to the center here, we do not run into trouble, as the analysans is a modal claim.
Tuesday, December 12, 2006
2D and Context-Sensitive Predicates
Subscribe to:
Post Comments (Atom)
5 comments:
Hi Berit. This is interesting. As a sort of background issue, I think that on the 2D framework there are actually two sorts of "context-sensitivity". There are cases where 1-intensions (and 2-intensions) of expressions vary between contexts, due to differences in speaker's intentions: e.g. 'tall' and 'ready' might be like this. And there are cases where 1-intensions are fixed, but extensions and 2-intensions vary because of relations to a center: e.g. 'I' and 'now' might be like this. The latter kind, which might also be called "scenario-sensitivity", correlates with automatic indexicality (here varying intentions aren't relevant; a speaker's location does the job automatically). Of course there are mixed cases too. I take it that the sort of context-sensitivity that you have in mind in your example is scenario-sensitivity.
Re your a priori sentences: actually, I think it's not out of the question that some utterances of this sort express a priori knowledge. Some related cases in the vicinity include 'if pink things exist, I am not blind' (said by a nondeferential user of 'pink'), etc. But I agree that it's not obvious that these sentences are a priori.
One can avoid the conclusion that they are a priori as long as we understand the 1-intensions in the right way. E.g. for the first sentence, one will need a 1-intension that has extensions at some uncentered worlds as well as centered worlds. For the other two, one will need a 1-intension that has extensions either at some uncentered worlds or at some centered worlds where the subject at the center lacks visual experience. If that's right, then the 1-intension can't be the one captured by 'the object that appears brightest to the subject at the center'. But that's at most a problem for the hypothesis that the 1-intension is that one, not for the framework as a whole. There are plenty of other 1-intensions in the vicinity that will do the job, while at the same time capturing the requisite scenario-sensitivity.
(What's the right 1-intension? This depends on the speaker. E.g. for a deferential user of 'Hesperus' it might involve a causal chain via uses of the world. For some it might involve astronomical theory, rather than vision per se. Or it might involve broadly disjunctive combination of theory, appearance, deference. As usual, it may be hard to encapsulate a 1-intension in a description, and the 1-intensions may vary greatly between speakers. It's worth noting that the latter phenomenon alone suggests that there's not much chance that the sentences will be a priori. But it's possible that some utterances of those sentences will be a priori (i.e. will express a priori knowledge), for speakers who use the terms in the relevant 1-intensions.)
I just noticed that you start by talking about the predicate 'is the brightest object in the evening sky' and then switch to 'Hesperus'. Of course these raise slightly different issues. My discussion above is meant to apply to 'Hesperus'. Most of it also applies to the predicate, except bits of the last paragraph. Re the predicate specifically: some uses of it will in effect be equivalent to 'is the brightest object in the evening sky *to me*'. For those uses, I think it's very plausible that utterances of versions of the three sentences you mention (with 'the brightest...' in place of 'Hesperus' and minor other changes) will be a priori. Other uses of the predicate may in effect be equivalent to 'is the brightest ...*to my community*', in which case the first sort of utterance will be a priori but not the other two. There may be other uses that behave differently, too.
Thanks David! Yes, I agree that one can find an appropriate 1-intension. I think a dispositional analysis of the sort outlined in the post might work. You mention a number of other possibilities. But I think I would actually prefer if 'If pink things exist, I am not blind' does not come out a priori, and it does not if one offers a dispositional analysis of color concepts.
Right, a dispositional analysis is a definite possibility. I don't think it entirely gets around the issue, though, as one still needs to specify which individual or group "normal" is indexed to. If it's in effect "normal for me" or "normal for my community" (the two most obvious possibilities), one still needs a center. So the first of your three sentences will still come out a priori, though on the "my community" proposal, the latter two sentences won't. Alternatively, normal conditions could be specified qualitatively, as conditions where e.g. light has such and such properties, and so on. This wouldn't require a center, but it's not obvious that ordinary speakers are in a position to invoke these qualitative conditions, at least in a manner that's specific enough to plausibly fix reference.
Right. I like the *to my community* part. What counts as 'normal' may be determined by use. As you make clear in your work, the reason that there is not always a description whose 1-intension captures that of a given term is that it is actual use of the term that determines its 1-intension. As you say in reply to Kripke's epistemic argument, a speaker may be able to use 'Godel' to refer to Godel even though she mistakenly believes Godel is the discoverer of the imcompleteness of arithmetic. And that then just shows that, as she is using the term, the 1-intension of 'Godel' is not that of 'the discoverer of the incompleteness of arithmetic'. So, the 1-intension of 'Godel' is determined by her actual use of the term, not by what she believes concerning her use of the term. Likewise, what counts as normal may be determined by the speakers in the relevant community (it might depend on linguistic practice and the cognitive apparatus of the relevant speakers).
Post a Comment