J. A. FODOR

116 J. A. FODOR
2. Though several recent papers have sought to make the putative morals for cognitive
science explicit;see [12], [13], [8], and [1].
3. Of course they are not really, since, as Geach [4] points out, there are molecules of
H2O in us but not in them. I take it that this detail will be seen not to prejudice the
spirit of Putnam's example.
4. From here on, "cognitive science" usually denotes not a problem domain but a body
of doctrine; specifically, a theory whose main tenets are RTM together with the claim
that mental processes are computational. The locution "cognitive science" should
thus be construed on the model of, say, "Christian Science" rather than of, say,
"cognitive psychology".
5. It is unclear to what extent earlier versions of RTM accepted this doctrine implicitly.
There was, for example, a traditional disagreement over whether association "by
similarity" could be a fundamental psychological mechanism, along with association
by spatio-temporal contiguity. It is possible to view this debate as in part about
whether mental operations are computational. Since the sort of similarity at issue
was similarity of content, to acknowledge association by similarity as irreducible
would have been to violate the condition that mental operations apply in virtue of
the form of representations in their domains. Contiguity, by contrast, counts as a
formal property within the meaning of the act.
6. I shall often use "de dicto propositional attitude" as short for "propositional attitude
under de dicto specification". No propositional attitude can be de dicto per se, or
de re stricto dictu. Of course.
7. In a recent paper [8], however, Putnam opts for a "verificationist" notion of concept
such that "water" and "water2", though nonsynonymous, are nevertheless used to
express the same concept so long as speakers are ignorant of the relevant microchemical
facts. The effect is thus to detach the question "what does the word W mean?" from the
question "what concept is the word W used to express?" and, more generally, to detach
semantic questions from questions about the propositional attitudes of speaker/hearers.
I think that there is something right about this since, as will appear, I think that the
content of a belief does not determine its truth conditions (or those of the forms of
words used to express it). My reasons for thinking this are, however, a good deal less
dramatic than Putnam's. Though it will turn out that beliefs of identical content may
differ in their truth conditions, the considerations that lead to this conclusion are
philosophically innocuous. They do not, in particular, lend any comfort to verificationist
s.
8. I propose to use such expressions as "Grice's principle", "Gricean theory", "Gricean
reduction", and so forth to refer to any of a wide variety of doctrines which have in
common the idea that meanings are somehow constructs out of propositional attitudes.
I do not mean to assume the particular account of that reduction which Grice sets out
in [6]; nor, of course, is Grice to be blamed for the use I have made of his leading
doctrine.
9. This is pointed out by Burge in [1], although the morals he draws are quite different
from the ones that I shall endorse.
10. Professor Michael Lipton has suggested to me the following snazzy argument, which
makes trouble for all forms of the indexical proposal. What we are trying to do is pick