The principle of charity

114 J. A. FODOR
c. the principle of charity
d. the de re/de die to distinction.
Roughly, to preserve the first two we must hold that /water is wet/ expresses
different beliefs on Earth and Earth2. But the different beliefs that commend
themselves are XYZ is wet and H2O is wet, and these beliefs can be ascribed
only de re if the principle of charity is to be respected. To put it slightly
differently, the de re/de dicto distinction seems to be the distinction between
what is in the head and what is not; so we cannot both say (with Burge and
Putnam) that meanings are social and say (with Grice) that meanings are logical
constructs out of de dicto propositional attitudes.
Giving up the Gricean reduction would get us out of this, of course,
but only by blunting Putnam's polemical darts. For, if meanings are not
constructs out of de dicto propositional attitudes, it is perfectly possible
that nonsynonymous formulas (such as "water is wet" and "water2 is wet",
assuming Putnam's lexicography) may function to express the same de dicto
propositional attitudes. But if this is possible, then even if Putnam is right
about what "water" and uwater2" mean, he has provided no argument that
molecularly identical people can differ in what they de dicto believe. If, in
short, he gives up Grice's principle, there would seem to be nothing in Putnam's
lexicography that is relevant to the cognitive science project.
Alternatively, we could imagine giving up the distinction between de re
and de dicto propositional attitudes. But that really would make the sky fall
down. For one thing, believing that water is wet looks to be a different state
of mind from believing that H2O is, and we need the de re/de dicto distinction
to say what the difference amounts to. Moreover, as I remarked above, it
looks as though we are going to need the notion content of a de dicto belief
if we are to have any psychology of the contingency of behavior upon mental
states at all. What would it be like to give that up?
Anyhow, it is just as well that we do not have to. What we can do instead
is account for the fact that /water is wet/ has different truth conditions here
and on Earth2 by appealing to the operation of what are essentially pragmatic
rather than semantic considerations: viz., by appealing to the Principle of
Reasonableness rather than the putative nonsynonymy of "water" and
"water2". This leaves the lexicographic issues wide open; and if it turns out
that such issues have no principled resolutions, maybe that is all right too.
Here are some questions that people have asked me about this paper:
1. "Aren't you saying that 'water' isn't a rigid designator for H2O?"
Answer: no, I am not saying that. What I do claim is that, if the evidence
that "water" rigidly designates H2O is the difference in truth conditions
between "water is wet" and "water2 is wet", then we have no evidence
for the rigid designation claim. For, as we have seen, that difference in truth
conditions is compatible with "water" and "water2" both expressing the
phenomenological concept.
On the other hand, the fact that "water is not H2O" is not of the de dicto
form P and ~P is consistent with "water" being a rigid designator for H2O.
What would make them inconsistent is the principle that if two terms rigidly
designate the same thing then they have the same meaning. But we have