The generalization that the practical syllogism,

COGNITIVE SCIENCE 101
prefer his behavior to exhibit) given that he has certain specified beliefs and
desires. The generalization that the practical syllogism (and, mutatis mutandis,
other decision theories) articulates thus applies to behavior under intentional
description; and so, in similar ways, do the rest of our folk psychology and
practically all of our cognitive science.
Now, RTM proposes a two-step reduction of the intentionality of
behavior to the content of mental representations. Obviously, the details
are much in dispute. But, very schematically, the idea is that (step 1) for
behavior to have such and such an intentional property involves its being
caused by a mental state having the corresponding propositional content;
and (step 2) to have a mental state with the propositional content that P is
to be related, in a certain way, to a mental representation which expresses
the proposition that P. Attempts to bring it about that P are thus explained
by reference to intentions to bring it about that P, and intentions to bring it
about that P are in turn explained by reference to mental representations
which, in effect, mean that P. What is essential to this pattern of explanation
is that the first step accounts for the intentionality of behavior by reference
to the content of a (causally efficacious) propositional attitude, and the
second step accounts for the content of the propositional attitude by reference
to the content of a mental representation.
It is worth emphasizing that, in canonical psychological explanations
of the sort that RTM contemplates, the required specifications of propositional
attitudes are characteristically de dicto rather than de re. (By a de dicto
specification of a propositional attitude, I mean approximately one in which
substitution of coreferring expressions does not, in general, preserve truth
unless the expressions are synonymous.) The point here is that de re specifications
of propositional attitudes are generally too weak to support explanations
of behavior when the latter is intentionally characterized. So, de re,
Oedipus' desire to marry Jocasta = his desire to marry his mother = his desire
to marry the tallest woman in Greece (assuming that Jocasta was the tallest
woman in Greece at the time when Oedipus desired to marry her). But it
is only the first of these specifications of what Oedipus desired (or maybe
the first two, depending on how you feel about depth psychology) that
figures in canonical explanations of the behavior that Oedipus tried/intended
to produce.
What all this comes down to, then, is that we need the notion of the
content of a mental representation to reconstruct the notion of the content
of a de dicto propositional attitude6; and we need the notion of a de dicto
propositional attitude in order to reconstruct the notion of the intentionality
of behavior; and we need the notion of the intentionality of behavior in order
to state a variety of psychological generalizations which appear to be (more
or less) counterfactual supporting and true, and which subsume behavior in
virtue of its satisfaction of intentional descriptions.
This does not, however, quite settle the issue; it still is not out of the
question that some purely syntactic (formal) specification of mental representations
might do the job. For example, it is conceivable that there should
be some formal property (call it "£/") that mental representations have iff
they express the property of being a unicorn, and some (different) formal