COGNITIVE SCIENCE

out the de dicto belief that "water2 is wet" is used to express. But you cannot do this
with a formula of the form "the belief that this stuff i s . . . " because, since indexicals
always occur transparently in descriptions of propositional attitudes, no such formula
can, even in principle, specify a belief de dicto. What I suspect that this argument
shows is that there can be no word that is defined in terms of an indexical (which is
not, of course, to say that there can be no indexical words).
11. Notice that 2 is not a version of Grice's principle, for although it connects semantic
properties with properties of concepts, the latter are not being assumed by Burge
to be mental in anything like the sense that Gricean reductions require.
12. As the reader will have divined, I am pretending that verbal contracts bind exhausts
the concept of contract instead of just constituting part of it as per assumption 1.
It simplifies the discussion and changes nothing to do so.
13. More precisely, I take the following principle to be valid: Let a and b be distinct
expressions, and let believes that. .. a . . . and believes that... b . . . be formulas
that specify beliefs de dicto. Then, if a and b are synonyms (express the same concept)
either D [(x) (x believes t h a t . . . a ...) = (x believes t h a t . . . b . . .)] or believes that...
a . . . and believes that. .. b . . . designate distinct beliefs.
14. I have no story at all to tell about Burge's examples insofar as they do not involve kind
terms, since I find that I do not share many of the intuitions that motivate Burge's
solutions. A natural thing to do in the case of "contract" would be to take it to express
some such concept as legally binding agreement, so that the belief that contracts must
be written would be consistent but false. For what it is worth, The American College
Dictionary says that a contract is "an agreement enforceable by law", thereby leaving
it as a legal (rather than a conceptual) issue whether verbal contracts bind. This seems to
be entirely plausible, for it seems to me not incoherent to wonder whether, for
example, verbal contracts are binding in France.
15. For convenience, I am assuming that "water is wet" expresses an (implicit) universal
generalization. But the analysis I shall propose has an obvious extension to the
assumption that "water" is a singular term. Indeed, the principles involved in the
analysis appear to be quite general in their application. See note 17 below.
16. The "the" is there to indicate that the belief that water is wet purports to be about a
single kind of stuff, and the "kind" is there to indicate that the belief that water
is wet purports to be about a kind. In these respects the present analysis shares
Putnam's assumption that "water" is (or, anyhow, purports to be) a kind term.
17. Analogous considerations apply to ensure that the evaluation of existentially quantified
variables should also be local, a point that has been widely noticed. For example,
nobody would evaluate "there are cookies to eat" with respect to cookies in China.
Similarly with respect to singular terms, whose referential ambiguity would otherwise
make life miserable. Contemporary tokens of "John is in the shower" are not, of
course, evaluated with reference to John the Baptist.
18. Isotropy does not, of course, supply an independent characterization of the relevant
universe of discourse since isotropic regions of space-time just are those that are homogeneous
with respect to the sorts of descriptions that laws deploy.
19. This is not a fudge since, on anybody's story, explicit indexicals can presumably be
distinguished from words like "water" on grounds independent of their indexicality.
For example, the explicit indexicals belong to "closed class" vocabulary.