We have so far been supposing a Fodorian model; we shall now sketch an
extremely simple version of a mentalese account of the structure of
cognitive states. We note that states are repeatable types; we shall
speak of instances of such. Instances are concrete, nonrepeatable
episodes. We shall assume that there are three functionally specifed
components of the mental states of our agent: the belief component,
the appetitive component, and the volitional component. These three
components are realized in three distinct, but connected, concrete
structures. Conceived of abstractly, that is, functionally, each of
these is like a file, into which tokens of sentences are read, in
which they can be stored and manipulated in various ways, and from
which they can be read. We shall further assume that mentalese is a
lot like English; indeed, we shall assume that it is
English.
With reference to the first
example, we might expect to find the following in Jerry's mind:
These tokens are concrete structures, characterizable in many
different ways. We have classified them by their syntactic type,
given that they are tokens of English. This leaves open many
difficult questions about the relations between type and token; we
shall ignore these here. We can classify agents in many different
ways, too, of course. We introduce three relational symbols,
, 
, and 
for the relations that
hold of an agent A, a time t, and a type T, just in case there is
a token of type T in A's belief (appetitive, volitional) structure
at t. We shall usually ignore the temporal relatum. We shall
also treat these symbols as if they were transitive verbs. Thus
``I drink some decaffeinated coffee'' is an appetitive
state and is also a (partial) mental state. Many different agents,
at different times and locations, can be in that mental state, with or
without being in the state we supposed Jerry to be in.
Of course, no two instances of these states are identical; we leave
open the possibility that a single agent, at a single time, can have
two distinct tokens of the same type in one of its mental structures.
Following Fodor, we shall assume that there is a single central processor, which can read the tokens in the various structures and can perform various operations on them. The processor is a deterministic device. In this respect, then, the agent's psychology is lawlike. We further assume that every agent of a given kind or species has the same kind of central processor, and indeed this is partly definitive of what we mean by a kind or species of agent. Of course, different experiences will have led to quite different sentences being written in the structures of various agents of the same species.
As noted, the psychologies of our agents are law-like. There are laws relating the various states. Here is a candidate law, implausibly simple, that we will assume to hold:
(): If an agent
``I pick up the cup and bring it to my lips.''
The above statement of the law conceals reference to tokens of the
displayed types, but for there to be laws like , or like except for being much more complex, the syntactic type of a
mentalese token must be a property of tokens that the processor can
detect. We assume that the processor can only detect local, physical
properties of the tokens. It is not quite clear what this includes,
but there are a number of things that are clearly not included. For
example, the processor cannot detect anything about the cup or the
coffee. This is not to say that the agent's sensors can't detect such
things; nor is it to say that the agent can't. In any event the law,
as stated, is quite independent of the meaningfulness of the tokens
and of whatever particular contents they have.
We need to make three points about and the extremely simple
psychology it reflects. First, we are ignoring background beliefs and
a number of interesting questions they raise. The use of the phrase
``normally or ceteris paribus'' is simply an indication that we are
aware of these issues, not an attempt to treat them. Second,
we are making no allowance whatsoever for weighing the pros and cons
of various alternative courses of action or for deliberation to
resolve conflicts among appetites. Finally, we are using the
volitional structure to model central motor control functions. Belief
states and appetitive states lead to volitional states; these, in
turn, cause bodily movements. A given volitional structure could be
wired up to the wrong kind of body, one that had nothing like hands
with opposable digits, or to a body with arms that were too short,
etc. It could also be ill-wired to the right kind of body. We ignore
all such unhappy possibilities.
We assume that a cognitive psychology assigns contents to tokens of mentalese in virtue of (i) the basic meanings associated with the types of the tokens and (ii) other facts. These other facts we gather into the following basic categories:
This may
include facts that are not in any clear and intuitive sense about
either the token itself or its owner.
We take the meanings associated with the types to be functions from circumstances to contents, where contents may have as constituents individuals external to the mind of the agent. In these respects, our semantics follows the semantics for ``schemata'' in Chapter Ten of Situations and Attitudes. But we differ with that approach in two related ways. First, we do not assign a single content to a token. Each token will have three contents: a fully opaque content, an opaque content, and a transparent content. Second, we do not take the function from context to content(s) to be the basic fact about meaning, but to result from a basic assignment of token reflexive conditions of truth. Thus our semantics will assign (up to) three contents to each expression: truth conditions--which we will take to be the fully opaque content--opaque, and transparent content. All are assigned circumstantially.
The following is an attempt to indicate the form of our semantic account by treating the sentence ``The cup in front of me contains decaffeinated coffee'' in some detail. We start with the content of the terms ``me'' and ``the cup in front of me'':
T = ``me''
T = ``the cup in front of me''
T = ``The cup in front of me contains decaffeinated coffee.''
To get at the level 1 and 2 contents (fully opaque and opaque), we shall use the sentences ``The cup in front of the owner of t contains decaffeinated coffee'' and ``The cup in front of Jerry contains decaffeinated coffee.'' When we do this we will be using the descriptions attributively. Thus in the first sentence, t is referred to, but the owner of t and the cup are not referred to. In the second, Jerry is referred to, but the cup is not.
We use the notions of loading and unloading to get at
relations between the fully opaque and the opaque, and the opaque and
the transparent contents of a token of T. Loading is an operation
that takes us from a proposition that contains a complex property, like
being the owner of t, to a proposition that contains an object that
uniquely instantiates the property, like Jerry. More precisely, we
load a proposition with respect to a complex property and a set of
circumstances. In going from the level 1 to level 2, we are loading
the proposition with respect to the property of being the unique owner
of t and the circumstance that Jerry is that owner. In going from
the level 2 to level 3, we are loading the 2-level proposition with
respect to the property of being the unique cup in front of Jerry, and
the circumstance that c is that cup. Unloading is just the opposite
of loading. The proposition that the cup in front of Jerry contains
coffee is the result of unloading the proposition that c contains
coffee with the circumstance that c is the cup in front of
Jerry.
We take the truth conditions of a mentalese token to be its fully opaque content. This is the only content it has that depends only on the form of the token and the interpretive function.
We said at the beginning of this essay, ``Fodor takes propositional attitudes to be relations to tokens of an internal language...that have content. If Jerry believes that S, Jerry has a token of mentalese in his belief structure that has the content that S.'' The picture we have arrived at requires an account that is a bit more complicated, however. First, we should note that when the context of the attitude reporter and the agent differ, different sentences will be required to get at the same content. Thus to report what Jerry believes in virtue of his having a token of ``I am sitting'' in his belief structure, I'll have to say something like ``Jerry believes that he is sitting'' or ``The author of Methodological Solipsism believes that he is sitting.'' If I use a token of the same sentence, and say ``Jerry believes that I am sitting,'' I do not convey the right message.
Second, we now have three levels of content. We shall introduce subscripts and say things like
Jerry(
) that...
Jerry
(
) that...
Jerry
(
) that...
This notation is not really adequate, since there are cases in which one term in a content sentence should be taken opaquely and another transparently. For the purposes of this essay, however, this notation will suffice.
We should emphasize that is not one of our
ordinary belief concepts. In the ordinary senses (opaque and
transparent) of ``believes,'' people don't usually believe the fully
opaque content of their beliefs, since most people do not have beliefs
about tokens of mentalese in their heads. And note further that even
for people who do have such beliefs, the belief in the fully opaque
content of a belief token b will not be the belief one has in virtue
of having b. To see this, note that Jerry and Georges might both
believe, of Jerry's belief token b, that the cup in front of the
owner of it contains decaffeinated coffee. Imagine Jerry and Georges
are talking about mentalese, using one of Jerry's beliefs as an
example, while Jerry drinks decaffeinated coffee. This belief about
Jerry's token is not the same belief that Jerry has in virtue of having the
token. This may seem a bit puzzling. But suppose that Jerry uttered
``The cup in front of me contains decaffeinated coffee.'' Call the
utterance u. The belief about u, that the cup in front of the
person who made it contains decaffeinated coffee, has to be
distinguished from what Jerry said (the proposition he expressed) with
u. What Jerry said could be true, even if he never spoke, and u
never existed. This is analagous to the belief case, and perhaps will
make it seem less puzzling. Our ordinary propositional attitude
reports simply do not focus on fully opaque content.
So we have to be careful with and remember that its
meaning derives from the theory, not from common usage. Saying that
Jerry that the cup in front of the owner of b
contains decaffeinated coffee is just saying that b is a token in
Jerry's head, whose fully opaque content is that the cup in front of
its owner contains decaffeinated coffee. Jerry doesn't
believe this in the ordinary sense. That is, he doesn't
it. Georges does it, but doesn't
it. So, of Georges, but not of Jerry, it can
be said that he beleives that the cup in front of the owner
of b contains decaf.