Fodor uses the phrase ``object of an attitude'' to refer to the
mentalese tokens that are involved in cognition according to his
theory. We try to avoid this phrase, which is often used instead for
what we shall call the contents of the propositional attitudes.
We assume that the tokens have contents, and that these contents are
referred to by the complements of attitude reports. So ``A believes
that P'' means, roughly, that A has in his head a belief token,
whose content is that P. We take these contents, the designata of
phrases of the form ``that P,'' to be propositions. Propositions
are truth-evaluable; that is, not true or false as spoken by one
person, or at one time, but simply true or false.
We use ``believes that
...,'' ``desires that...,'' and ``intends that...'' as canonical forms
for reporting beliefs, desires, and volitions.
Sentences in ordinary language may be separated into those that are
eternal and those that are context-sensitive. Different
utterances of an eternal sentence express the same proposition. The
class of eternal sentences is arguably empty, but sentences such as
``7+5=12'' and ``The first person born in the twenty-first century was,
is, or will be a philosopher'' are at least candidates. Different
utterances of context-sensitive sentences express different
propositions, depending on such factors of the utterance as the agent,
the time, the persons the agent is addressing, the objects the agent
is attending to, and other wider circumstances. The class of
context-sensitive sentences includes those with indexicals,
demonstratives, tense markers, and, arguably, proper names. Thus if
Jerry and Zenon each say, ``The cup in front of me contains coffee,''
they express different propositions that may be true and false
independently. Context-sensitive sentences in mentalese will work the
same way; if Jerry and Zenon each have ``The cup in front of me
contains coffee'' (= S) written in their belief structure, they will
believe different things. But is there any reason to suppose that
there are context-sensitive sentences in mentalese? What reasons
would a psychological theory that adopted Fodor's language of thought
hypothesis have to include such sentences in the language postulated?
Suppose Jerry and Zenon are each thirsty, and each has a cup of coffee
in front of him. They each execute movement M, and each satisfies
his desire. Both the similarities and differences are easy to
conceive, if we suppose that the beliefs Jerry and Zenon have are
context-sensitive--in particular, if what they believe, in virtue of
the occurence of a token of S in their heads, is systematically
related to the same simple fact about that token, namely in whose head
it does occur. For instance, we can imagine that Jerry and Zenon have
different instances of a certain type V of visual
impression; that because of these impressions
a token of S is written in the belief structure of each; and that
this, given the presence of ``I drink coffee'' in the desire structure
of each, causes a token of ``I pick up the cup in front of me and
drink from it,'' to be written in their volitional structure, which
causes each of them to execute M. Jerry and Zenon thus go through
the same succession of states, and these processes are instances of
the same psychological law. But Jerry and Zenon see, believe, desire,
intend, and do different things. In particular, what Jerry believes
could be true, while what Zenon believes is false, and vice versa.
What Jerry believes (that there is a cup in front of Jerry) would not
be a good reason for Zenon to execute M. It might be true that
there is a cup in front of Jerry, while there is an irritable gorilla
in front of Zenon. Executing M in these circumstances would not be
a good idea.
We want to emphasize that it is also important to have beliefs that are less sensitive to context. Consider the case with which we began. Jerry doesn't just want coffee; he wants decaffeinated coffee. A moment or two ago he saw that a certain cup contained decaffeinated coffee. Now suppose that since that time he has left the room; the cup became perceptually inacessible to him. He needs to be able to store the relevant information about the cup, in a way that retains the same content through the change in context, as he diverts his attention from the cup and leaves the room, and throughout the [short] interval of his absence. And he needs to retain it in a form that will allow him to reidentify the cup upon his return and make use of his knowledge. This requires a way of thinking of the cup that continues to pick out that particular cup through the changes of context. This way of thinking will probably be context-sensitive, but not as context-sensitive as the way of thinking of the cup that is associated with a contemporaneous visual perception of it. For example, he might think of it as ``the cup I saw at my table a moment ago.'' This way of thinking does not shift its reference as Jerry moves about or shifts his attention to other objects.
A (relatively) basic physical action, like the execution of M, has different results depending on who does it, when, and in what circumstances. A system of beliefs should lead to the actions that will be successful, relative to one's desires, in the circumstances that make the beliefs true. But humans do not just depend on their perceptions at the time of action to provide them with the beliefs needed for decision and volition, but earlier perceptions, inferences, and the perceptions and inferences of others. In all of these cases in which the pragmatic effect of information is delayed or distanced, ways of preserving content through change of context are needed. It would be very misguided to take the simple and relatively automatic case of picking up a cup of coffee when thirsty as a complete guide to the role of the attitudes. Still, the need for context-sensitive representation never goes away. If he wants to drink the coffee, Jerry still has to pick up the cup in front of him.
Consider, for a different example, Jerry's making a phone call to Zenon. Here the belief, say ``Zenon's phone number is 555-5555,'' is relatively context insensitive. There is the present tense marker, but phone numbers are relatively stable properties of persons, so we can ignore that. When Jerry calls Zenon, the perceptions that originally gave rise to the belief may be remote in time, and the effect of the action that fulfills the goal--the ringing of Zenon's telephone and his answering it--are remote in space. Neither the number nor Zenon needs to be thought of demonstratively or indexically to understand the transaction. But, like all human action--except perhaps pure ratiocination--the crux of the matter involves physical interactions with a physical object. The practical reasoning involved will terminate with some context-sensitive way of thinking about the telephone. The Fodorian version of half of Kant's maxim about concepts and intuitions should be ``Eternal tokens of mentalese without context-sensitive tokens of mentalese are blind.''