In this paper, Quine is concerned with an issue that is related to those about modality we have just been discussing: the contrast beween what he calls notional and relational readings of sentences like "I want a sloop". This leads to examples involving belief, such as "Ralph believes that someone is a spy", in which, on the relational reading, we are "quantifying into" a propositional attitude context. But this, Quine says, "is a dubious business".
Quine suggests that only on the relational reading is Ralph really related to the person he believes is a spy. Why does he say that? How might that thought be related to Russell's notion of acquaintance?
The basic problem is that, if we regard it as true that "There is someone Ralph believes to be a spy", then it seems as if we ought to accept as true "Ralph believes that Ortcutt is a spy", which seems wrong. The question is what to do.
See if you can construct similar examples, and explain the problem they pose, using pairs like Hesperus–Phosphorous or Twain–Clemens.
Quine remarks that "we are scarcely prepared to sacrifice the relational construction". By contrast, he seems utterly prepared to sacrifice the relational reading of modals (i.e., to ban quantification into modalities). Why the difference?
Quine's first proposal is to distinguish belief1, which does not permit quantification into it and prohibits substitution of identicals, from belief2 which allows both. But he turns in section II to a different proposal, or perhaps better a way of developing that one: In the simplest case, we can regard belief2 as a three-place relational between a believer (Ralph), the subject of his belief (Ortcutt), and the property (attribute) Ralph believes Ortcutt to have. As Quine notes at (17), this allows a natural representation of the relational reading. (It is sometimes read: Ralph believes of Ortcutt that he is a spy; or Ralph believes Ortcutt to be a spy.)
One thing Quine mentions but does not discuss in detail is the problem of "exportation", which is the inference from "Ralph believes that Ortcutt is a spy" to "Ralph believes of Ortcutt that he is a spy" or directly to "There is someone Ralph believes to be a spy". I.e., the problem concerns the logical relations between belief1 and belief2, or however we explain the contrast between notional and relational. Can you see why an appeal to something like acquaintance might be natural here?
In section III, Quine shows how this apparatus can be used to deal with the original examples involving hunting and wanting. An additional complication arises, as Quine notes, with such examples as "Someone wants a sloop", read as: ∃x(x desires that x has a sloop"), since the quantifier binds into "that...". So it has to be read as: ∃x(x desires y(y has a sloop) of x).
In section IV, Quine reworks this analysis in terms that do not require the use of intensional entities. The idea is simply to replace mention of intensions by mention of expressions. The dyadic notion of belief becomes "N believes-true S in L", for some language L; the triadic notion becomes "N believes-true S of x in L"; etc.
Quine considers several objections to this. He impatiently dismisses the objection that it is "unnatural", but also notes that a friend of intensions cannot object to it as unintelligible, since it is definable in terms they would accept. He also dismisses an objection due to Church about translation. The worry concerns:
- Hans believes that snow is white
- Hans glaubt, dass schnee ist weiss
- Hans believes "Snow is white" in English
- Hans glaubt "Snow is white" in English
The objection is that (2) translates (1), and (3) translates (4)—note that we do not translate the quoted sentence—but (3) was supposed to give the meaning of (1), and yet (4) surely does not give the meaning of (2). Quine responds that he was seeking only "a systematic agreement in truth-value".
How worrying does Church's objection seem to you? Is there some other way to try to bring out what is weird about Quine's appeal to language (which Quine himself finds somewhat problematic)?
All the views Quine considers seem to imply that "believes" is ambiguous: between belief1 and belief2, or dyadic and triadic, or whatever. Is that plausible?