Wednesday, September 17, 2008

argument against relational analysis of propositions

Bach's objection to RABR:
(RABR) verbs of propositional attitudes expresses a relation between persons and propositions; the claim that 'the semantic value of a “That” clause is a proposition': and the claim that in a true belief report, a proposition that the subject of the report believes must be specified.
(1) RABR
(2) (1) → (3)
(3) Propositional attitude verbs take as complements agents and propositions,
(4) The only thing relevant to the truth value associated with a well-formed sentence involving only an agent, a propositional attitude verb and a proposition, is which complements the relation denoted by the propositional attitude verb receives
(5) (3)&(4)
(6) (5) → ~(7)
(7) There is sometimes a change in truth value when substituting a that clause for a proposition description and vice versa, where the TC and the PD denote the same proposition.
(8) ~(7)
(9) (7)&~(7)
(10) ~(1)
This is set up as a reductio to RABR. Not much is said about how this argument is supposed to run, so I formulated an argument that fits neatly to the objections King considers. (1) is assumed for reductio. Part of the thesis of RABR is that propositional attitude verbs are relations between propositions and agents. Thus (2) is true. (4) is intuitively true. If, in fact, propositional attitude verbs are relations between agents and propositions, then the only thing that should matter to the truth value of a simple sentence involving only that relation and two relata, is whether or not the two relata are in fact related via the relations. Premise (6) is supported by the intuitive idea that proposition descriptions and that-clauses sometimes contribute only a single proposition to be evaluated as one of the relata of a propositional attitude verb. That is to say, they play the same role (contributing a proposition as a relata) and that a propositional description is capable to contributing the same proposition as some that-clause. If that's true, then substituting a TC for the appropriate PD should never change the truth value of a sentence. However, in support of (7), there are apparent cases in which substituting a TC for a PD or vice versa changes the truth value of a sentence, where intuitively the TC and PD contribute the same proposition.
King examines an objection to (4) that claims that syntax is also relevant to the truth value of propositions involving attitude verbs (142-143) but claims that the response is uninteresting. It's uninteresting in the sense that it can't explain ALL instances of substitution failure. He then goes on to examine denials of (7). There are various ways he considers of claiming that in instances of apparent substitution failure, the TC and the PD actually don't denote the same proposition (or denote other things as well) (145-146). While formally, it's possible that the TC and the PD don't denote the same proposition, he finds no plausible way of fleshing out the idea. He considers the proposition forms:
7a':[o[R[p]]]
7b':[o[R[q]]]
Where p is a that clause, and q is the corresponding PD, or vice versa. His dilemma is as follows:
(Lettered sentences are not themselves premises in the argument)
(1') 7a' and 7b' diverge in truth value
(2') (1') -> (3')
(3') (A) OR (B)
(4') ~(A)
(5') ~(B)
(6') ~(A)&~(B)
(7') ~(3')
(A) one of p or q determine some entity (or entities) o* in addition to the proposition, and this entity is relevant to the truth value of 7a' or 7b'. (the way the conglomeration is arranged will also be relevant)
(B) 7a' requires for its truth that o, R and p be arranged in one way, and 7b' requires those same things to be arranged in a different way for its truth.
In support of (2), the forms 7a' and 7b' are meant to stand for paradigmatic cases in which two sentences differ in truth value, but differ only in that a TC is substituted for a PD (or vice versa) where it appears the TC and PD denote the same proposition. King claims there are only two ways this could happen. One way is if there is more to the semantic content of the PD or the TC than merely denoting a proposition. This extra addition is denoted by o*. The only alternative King sees is that the TC and the PD make the proposition true or false in different ways, that is, they each require the proposition to be structured differently.
He claims (A) is false, simply because this o* is mysterious and elusive. Furthermore, there's not principled way to decide whether p contributes the o*, or q does. It's a merely ad-hoc construction that avoids the problem instead of addressing it.
He claims that (B) is false. I actually had a tough time phrasing (B) in such a way that it wasn't obviously false. In general, we've been speaking as if the logical structure of the propositions expressed by 7a' and 7b' were simple. In fact they are simple. And there's simply no way of restructuring them in the ways that (B) commands.
You'll notice that the last argument was a reductio of the consideration that 7a' and 7b' diverge in truth value. In fact King things they don't diverge in truth value, if the R remains constant in each case. This bring us to King's solution.
King denies (4), he does this by claiming that some propositional attitude verbs are ambiguous between two relations. So, in a case with apparent substitution failure, the TC is merely forcing one disambiguation of the propositional attitude verb, while the PD is forcing another. So, it's not the case that the only thing relevant to the truth value of the pertinent sentences is which relata the propositional attitude verb receives, it's also relevant which way the propositional attitude verb is disambiguated. He gives lots of independent evidence that these propositional attitude verbs are indeed ambiguous between relations.

1 comment:

Dan said...

I forgot to actually evaluate his argument... oops.
So let's examine a little further (although this is probably too long as it is, sorry chris).
Consider two sentences:
(T)Curtis fears that the bacon is salty and deliscious.
(P)Curtis fears the proposition that bacon is salty and deliscious.
Curtis has no reason to fear that bacon is salty and deliscious, of course it's salty and deliscious! However, Curtis fears things that the physical sciences don't address, so he fears propositions. One of the propositions he fears is the proposition that bacon is salty and deliscious.
King explains the difference in truth value between (T) and (P) by saying 'fears' is ambiguous between two relations. Let's call the relations R and R'. So the formal structure of the two sentences is thus:
(TF)[c[R[p]]]
(PF)[c[R'[q]]]
Since p is a that clause, it forces one way of dissambiguating 'fears'. Since q is a proposition description it forces another dissambiguation of 'fears'. The thing to note here is that we just have to find one example of substitution failure in which it's implausible that the propositional attitude verb is ambiguous between two meanings. If curtis actually does fear propositions in the "they're out to get me" sense, and is not scared of bacon being salty and deliscious, I think we could have such an example. If we were to ask Curtis about his current situation, he would say that he's afraid of propositions, but has no fear that bacon is deliscious. He would NOT say that he is afraid(sub1) of the proposition that bacon is deliscious and not afraid(sub2) of the proposition that bacon is deliscious.
The way out of this would be to Frege-puzzle the proposition and say that Curtis doesn't know what he's talking about.