I recently read Delia Graff Fara's paper: "Desires, Scope and Tense", . Delia here questions what Kent Bach calls 'the specification assumption':
(SA) A sentence or clause embedded in a propositional attitude context, when disambiguated, specifies precisely the content to which the subject in question is related.
(SA) seems prima facie plausible. However, Delia rejects it. She offers examples like the following:
a. Joe wants to smoke a cigarette
b. Monte hopes he will find a cheap place to live
c. Jim wishes he were somewhere warmer
Intuitively, (a) might be true even if Joe has no desire that could be satisfied by smoking a cigarette that has been floating in a glass of beer most of the night, (b) might be true even if Monte has no hope that is satisfied if he finds an empty dorm room, and (c) might be true even if Jim has no wish that would be satisfied if he were in Sahara during the warmest month of the year. (SA) thus seems be at odds with our normal ways of reporting propositional attitudes. More carefully, Graff Fara adheres to what she calls 'the Content-Satisfaction Principle':
The Content-Satisfaction Principle
If one has a desire/hope/wish with the exact content that p, then one has a desire/hope/wish which is satisfied in any possible world in which it is true that p.
Since Joe's desire to smoke a cigarette isn't satisfied in worlds where he smokes an old, beer-soaked cigarette, Joe's desire does not have the exact content Joe smokes a cigarette. Since Monte's hope that he will find a cheap place to live isn't satisfied in worlds where he finds a cheap dorm room, Monte's hope doesn't have the exact content Monte finds a cheap place to live. And since Jim's wish to be somewhere warmer isn't satisfied in worlds where he is in Sahara during the warmest month of the year, Jim's wish does not have the exact content Jim is somewhere warmer. So, she says, (SA) is false.
It is tempting to think that sentences like (a)-(c) are cases of loose talk (false yet close enough to the truth). But, as Delia points out, normally when we speak 'loosely', we are able to recognize that we are speaking that way. However, this is not obviously so for (a)-(c).
I think that there are two further reasons for denying that (a)-(c) are cases of loose talk. First, if Joe wants a dry cigarette but has no desire that is satisfied if he is offered a soaked one, it sounds much worse to say that Joe doesn't want a cigarette than saying that he wants one. Likewise, if Monte hopes that he will find a cheap place to live but has no hope that is satisfied if he finds an empty dorm room, it sounds much worse to say that Monte has no wish to find a cheap place to live than saying he hopes he will find one. Second, the following cases of loose talk,
France is roughly hexagonal, but it is not hexagonal
It is almost 3 o'clock, but it is not 3 o'clock
do not sound too awful. But the analogous attitude reports sound terrible:
Joe wants to smoke a dry cigarette, but he doesn't want to smoke a cigarette.
Monte hopes that he will find a cheap and decent place to live, but he has no desire to find a cheap place to live.
Jim wishes he were somewhere a bit warmer, but he has no wish that would be satisfied if he were somewhere warmer.
It appears that (SA) is false. Quite surprising!
Thursday, July 13, 2006
The Specification Assumption
Subscribe to:
Post Comments (Atom)
9 comments:
Without having read the paper, my immediate reaction is that surely the Content-Satisfaction Principle is wrong. Here's a counterexample: I have a desire that Peter smoke a cigarette. But clearly there are some possible worlds where Peter smokes a cigarette but where that desire is not satisfied because I - and my desire - do not exist. (An even more obvious case: consider the desire that one not exist.) But that doesn't mean the exact content of my desire really must be something else (e.g. that Peter smoke a cigarette and I exist).
Maybe something like the following has a better shot at being true:
If one has a desire/hope/wish with the exact content that p, then one has a desire/hope/wish which is satisfied in some/many/most close worlds in which it is true that p.
But this would have a shot at dealing with the wet cigarette cases, since it could be argued that context selects a closeness metric which sends wet-cigarette worlds sufficiently far out for them not to matter.
Hi Carrie. Thanks for the comment (the first one on my blog!) The following fix seems obvious at first:
If S has a desire/hope/wish with the exact content that p, then S has a desire/hope/wish which is satisfied in any possible world in which it is true that p and S and her desire/hope/wish exist.
But you are right, that won't quite work b/c of cases like 'I wish I didn't exist'.
But I think maybe Delia intends it to be read as follows (though I can't be sure):
If one has a desire/hope/wish with the exact content that p, then one actually has a desire/hope/wish x which is such that for any possible world w, x is satisfied in w (where 'x is satisfied in w' means that the content of x is true in w).
Is there a sense in which we can view the speaker of a sentence like "Joe wants to smoke a cigarette" as expecting the content of the utterance to be understood to set fairly obvious parameters on what the possibilities for the satisfaction of his desire would be.
It is not the case that it is impossible for the speaker to identify and make explicit enough of her parameters of satisfaction so that a rough parameter may be understood, it is just laborious. I am sympathetic to Brit's claim that the following sentence sounds horrible, "Jim wishes he were somewhere a bit warmer, but he has no wish that would be satisfied if he were somewhere warmer." And it seems that sentences like these would occur within the SA framework.
"If one has a desire/hope/wish with the exact content that p, then one actually has a desire/hope/wish x which is such that for any possible world w, x is satisfied in w (where 'x is satisfied in w' means that the content of x is true in w)."
I'm not sure if this is quite what you mean - doesn't it imply that the content of each desire is necessarily true? Did you mean 'for any possible world w such that p is true at w'? (But if so, isn't the condition trivial?)
Hi Jeff. Thanks for your comments. Kent Bach has a very interesting article "Do Belief Reports Report Beliefs?", where he gives some general rules for determining whether a belief report adequately describes the belief in question. The general idea in the article is that belief content can be more specific than the proposition expressed by the that-clause. There are obvious cases like "Lois Lane believes that Clark Kent can fly", which may be false even though Lois Lane believes that Superman can fly, and Superman = Clark Kent. But Kent gives a number of other interesting cases as well. Maybe I will post something on this later today.
"Did you mean 'for any possible world w such that p is true at w'? (But if so, isn't the condition trivial?)"
Yes! Thank you, Carrie! And yes, it's trivial.
Perhaps this principle is better:
'S wishes/desires/believes that p' is true iff for any world w compatible with the exact content of S's wich/desire/belief, p is true in w.
'Joe wants to smoke a cigarette' (= 'Joe wants that Joe smokes a cigarette') is true iff for any world w compatible with the exact content of Joe's desire (= Joe-smokes-non-beer-soaked cigarette worlds), 'Joe smokes a cigarette' is true in w.
Likewise, 'Mary believes that Bush is president' is true iff for any world w compatible with the exact content of Mary's belief (= say, Bush is a poor Republican president of the United States, etc.), 'Bush is president' is true.
This seems obviously true but it's not trivial (right-to-left). One might, for example, agree that at all worlds compatible with the exact content of Joe's desire, 'Joe smokes a cigarette' is true, and yet deny that 'Joe wants to smoke a cigarette' is true (if the specification assumption is true, then it is false that Joe wants to smoke a cigarette, as he doesn't want to smoke a beer-soaked cigarette).
I do think SA is false and that even if we could exhaust all the different permutatons of un-smokeable cigarettes, it still would not follow that this set "specifies precisely the content" of a belief. Such canonical sentences like 'It is raining' are easy; the sentences offered by Delia seem to force an invocation of Quines "web of belief": we just cannot specify the content of a belief without including indefinitely large amount of other beliefs (this is a rough sketch of how I take the "web of belief. Would Joe smoke a cigarette that just belonged to Lance Armstrong (maybe he admires Lance so much, he would treasure the cigarette as a keepsake)? How many other beliefs would have to come into play?
And these considerations lead to the question: is the content of the proposition 'Joe wants to smoke a cigarette' constituted by some indefinitely discursive list of beliefs, some which do no contain the referent of 'cigarette' ('I really admire Lance Armstrong').
The very notion of a "principle" around propositional attitudes seems like a messy business, as Graff's examples illustrate.
Sorry, Brit, for adding my comments so late in the game, but I just discovered the blog. (It looks fabulous, by the way, and does your research!)
It seems to me that the Content-Satisfaction Principle stated by Delia was intended to be trivial, and that it could be rendered so, even in light of Carrie’s initial objection to it, by changing one word in its original formulation. Let me explain.
The principle, as formulated in Delia’s paper is this:
Content-Satisfaction Principle (CSP): If one has a desire/hope/wish with the exact content that
P, then one has a desire/hope/wish which is satisfied in any possible world in which it is true
that P.
Carrie sensibly objected that this principle runs into trouble with desires such as my desire not to exist. The reasoning here seems to be that though such a desire has an (exact) content, it fails to be satisfied in any (let alone every) world in which that content (viz. that Scott doesn’t exist) is true. And again, the reasoning here seems to be that this is because the desire isn’t around in that world to be satisfied in it.
Notice that the same general “problem” afflicts the following amendment of the C-SP. (Here I’m following a formulation you offer, as amended by Carrie):
CSP*: If one has a desire/hope/wish with the exact content that P, then one has a desire/hope/wish x such that for any possible world w [in which it is true that P], x is satisfied in w (where 'x is satisfied in w' means that the content of x is true in w).
For consider my desire that contents not exist. That desire has a definite content (viz. that contents not exist), yet the content of that desire isn’t around to be true in worlds in which it is true that contents don’t exist. (This assumes, perhaps controversially, that contents don’t necessarily exist.)
The way around these objections is to substitute ‘at’ for ‘in’ in one’s formulation of the CSP:
CSP**: If one has a desire/hope/wish with the exact content that P, then one has a desire/hope/wish which is satisfied AT any possible world in (I’d say ‘at’ here too) which it is true that P.
The idea is that a desire may be satisfied AT a possible world without existing IN that possible world. (The same goes for contents and being true.) A desire’s being satisfied AT a world involves things being as the desirer wishes them to be; it does not involve the desire (or even its content) existing in the world in question, as perhaps the desire’s being satisfied IN the world does. Compare: A painting may depict truly a world barren of depictions, whether they be concrete (paintings) or abstract (contents). Yet its depicting truly does not require its (or its content’s) existence in the world being depicted; in fact it requires its non-existence there.
The distinction between satisfaction at a world and satisfaction in it is a part of a generalization of the distinction between truth at a world and truth in it. (Cf. Iacona’s book _Propositions_ and McGrath’s reference to it in his entry “propositions” in SEP. I think the distinction originally(?) is made by Fine (1985?) in response to an argument of Plantinga’s.)
Sorry this last bit is so compressed. Thanks again for the great blog!
Thanks, Scott! Very interesting. The 'at'/'in' distinction really does seem to do the trick. That's neat! I am familiar with Fine's and Plantinga's work and the McGrath article, but I should definitely read lacona's book.
Post a Comment