wetness of H2O? Or, if you do not like,

it that molecularly identical people can differ in de dicto propositional
attitudes. But how do you capture the intuition that "water2 is wet" is true
in virtue of the wetness of XYZ whereas "water is wet" is true in virtue of
wetness of H2O? Or, if you do not like "true in virtue o f talk, we can put
the question this way: If "water is wet" and "water2 is wet" express the same
de dicto belief, how do you account for the intuition that "water is wet iff
water2 is wet" is contingent?
This is, of course, where the indexical analysis did its thing; to say that
"water2" is used to pick out stuff of the same kind as certain (ostensively
specified) stuff "around here" is to guarantee that "water2 is wet" and
"water is wet" are evaluated with respect to local samples of the wet, transparent,
potable stuff that people sail on; hence utterances of the phonological
form /water is wet/ get evaluated with respect to XYZ when they occur on
Earth2 but with respect to H2O when they occur on Earth. We have, however,
given up the indexical analysis on internal grounds and we are heuristically
committed to: "water is wet" and "water2 is wet" both express the phenomenological
belief. Now what?
I propose to resolve the difficulty by distinguishing between the content
of a belief and its truth conditions. By the content of a belief I mean approximately
what we would specify if we were asked to write down its logical form,
with constants for the predicate terms. So, the content of the phenomenological
belief is something like: (x) (x is drinkable, transparent, sailable-on,. . .
etc., only if x is wet).15 I assume that the contents of beliefs and the contents
of sentences are connected by the (Gricean) principle that sentences share the
contents of the beliefs they are used to express.
The important claim is this: you cannot go directly from the content
of a belief/sentence to its truth conditions (to the conditions for its evaluation);
you need at least to specify the universe of discourse for the bound
variables. Moreover, I want to suggest, there is a sort of Principle of Reasonableness
that operates in deciding how the universe of discourse of bound
variables is to be assigned, and the effect of this principle is to determine that
the evaluation of universally quantified standing sentences is relevantly local.
Specifically, it ensures that such sentences are evaluated in much the way they
would be if they contained demonstratives. The difference between the truth
conditions of "water is wet" and of "water2 is wet" are thus the consequence
of the application of the Principle of Reasonableness to the two cases, or so
I am about to claim.
A rough formulation of the Principle of Reasonableness might go: do not
be bloody-minded in deciding what universe of discourse sentences and beliefs
will be evaluated with respect to. I shall refine this principle, slightly, a little
farther on. For the moment, consider an example of its operation in respect
to quantification over times. Marco Polo wrote ([7], p. 62): "Kesmur is a
province distant from Bascia seven days' journey". The first point to notice
is that this looks to be a universal standing sentence with bound variables
ranging over times; something along the lines of: for all pairs (t,tf), if t is the
time of the start of a journey from Kesmur to Bascia and t' is the time of the
end of that journey, then t1 = (f + 7 days). And similarly the other way around
for journeys from Bascia to Kesmur. The second point to notice is that there