Chapter 8

ARE MEANINGS IN THE HEAD?

The fundamental question of the philosophy of language has

always been: How does language relate to reality? The answer I
proposed to that question in Speech Acts was that language relates
to reality in virtue of the fact that speakers so relate it in the
performance of linguistic acts. The original question then reduces
to one of analyzing the nature and conditions of the possibility of
these acts. In this book I have tried to ground that analysis further
in the Intentionality of the mind: the question, "How does
language relate to reality?" is only a special case of the question,
"How does the mind relate to reality?", and just as the question
about language reduced to one about various sorts of speech acts,
so the question about the mind reduces to one about the various
forms of Intentionality, the representational capacities of speech
acts being simply a special case of derived Intentionality.
On one interpretation of Frege, my general approach to
Intentionality is a matter of revising and extending Frege's
conception of "Sinn" to Intentionality in general, including
perception and other forms of self-reference; and my approach to
the special problem of reference is in some respects Fregean in
spirit, though, of course, not in detail. Specifically, it is possible to
distinguish at least two independent strands in Frege's account of
the relations between expressions and objects. First, in his account
of the Sinn and Bedeutung of Eigennamen, an expression refers to an
object because the object fits or satisfies the Sinn associated with the
expression. Second, in his fight against psychologism Frege felt it
necessary to postulate the existence of a "third realm" of abstract
entities: senses, propositions, etc. Communication in the utterance
of an expression is possible only because both the speaker and the
hearer can grasp a common abstract sense associated with the
expression. My own account is Fregean in accepting the first of
these strands, but I reject the second. Linguistic reference is a
 Are meanings in the head?

special case of Intentional reference, and Intentional reference is

always by way of the relation of fitting or satisfaction. But it is not
necessary to postulate any special metaphysical realms in order to
account for communication and shared Intentionality. If you think
about the Evening Star under the mode of presentation "Evening
Star", and I think about the same planet under the same mode of
presentation, the sense in which we have an abstract entity in
common is the utterly trivial sense in which, if I go for a walk in the
Berkeley hills and you go for exactly the same walk, we share an
abstract entity, the same walk, in common. The possibility of
shared Intentional contents does not require a heavy metaphysical
apparatus any more than the possibility of shared walks.
Both the Fregean and the present account of meaning are
internalist in the sense that it is in virtue of some mental state in the
head of a speaker and hearer — the mental state of grasping an
abstract entity or simply having a certain Intentional content — that
speaker and hearer can understand linguistic references. At the
time of this writing, the most influential theories of reference and
meaning reject a Fregean or internalist analysis. There is a variety
of reasons for which the anti-internalist position has become
fashionable, and there is considerable disagreement among the
anti-internalists as to what the correct analysis of reference and
meaning is. In this chapter and the next I will consider and answer
at least some of the more influential attacks on the internalist,
Fregean, or Intentionalistic tradition. These chapters, therefore,
are more argumentative than those which preceded them: my aim
is not only to present an Intentionalistic account of reference but to
do so by way of answering what I believe is a family of mistaken
doctrines in contemporary philosophy. Here, in no special order,
are some of the most influential theses urged against the internalist
picture.
1. There is supposed to be a fundamental distinction between de
re and de dicto beliefs and other sorts of propositional attitudes. De
re beliefs are relations between agents and objects, they cannot be
individuated solely in terms of their mental contents (de dicto),
because the object itself (res) has to be part of the principle of
individuation of the belief.
2. There is supposed to be a fundamental distinction between
the "referential" and the "attributive" use of definite descrip-
198
 Are meanings in the head?

tions. Only in the case of attributive uses of definite descriptions

does a speaker "refer" to an object in virtue of the fact that his
Intentional content sets conditions which the object satisfies, but
these are not genuine cases of referring at all; in the referential use
of definite descriptions the speaker need not use an expression that
the object referred to satisfies.1
3. Indexical expressions, e.g., " I " , "you", "this", "that",
"here", "now", are supposed to be impossible for an internalist
theory to account for, since their utterance lacks a "completing
Fregean sense".
4. Exponents of the so-called causal theory of names and the
causal theory of reference are supposed to have refuted something
called the "descriptivist theory" of names and of reference, and
thereby to have refuted any internalist or Fregean account, and to
have shown that reference is achieved in virtue of some external
causal relations.
5. The causal theory of reference is supposed to be applicable to
a large class of general terms, the natural kind terms and perhaps
others; and for these terms there are supposed to be decisive
arguments showing that knowing their meanings cannot consist in
being in psychological states of any sorts, but must involve some
more direct causal relations with the world. It is supposed to have
been shown that "meanings are not in the head".
I believe that all these views are false. Furthermore, they share a
family resemblance; they suggest a picture of reference and
meaning in which the speaker's internal Intentional content is
insufficient to determine what he is referring to, either in his
thoughts or in his utterances. They share the view that in order to
account for the relations between words and the world we need to
introduce (for some? for all? cases) external contextual, non-
conceptual, causal relations between the utterance of expressions
and the features of the world that the utterance is about. If these
views are correct then the account I have given of Intentionality
must be mistaken. At this point then I see no alternative to
mounting a series of set piece philosophical arguments. The

1 will not discuss this view further in this book since I have attempted to refute it
elsewhere; see, 'Referential and attributive', in J. R. Searle, Expression and Meaning
(Cambridge: Cambridge University Press, 1979), pp. 137—61.

199
 Are meanings in the head?

justification for making such a fuss over views I believe are false
anyway has to do with the size of the issues involved. If we are
unable to account for the relation of reference in terms of internal
Intentional contents, either the contents of the individual speaker
or the linguistic community of which he is a part, then the entire
philosophical tradition since Frege, both the analytic and the
phenomenological strands, is mistaken and we need to start over
with some external causal account of reference in particular, and
the relation of words to the world in general.

I. MEANINGS IN THE HEAD

I shall begin by considering Hilary Putnam's argument that

"meanings are not in the head".2 I think in the relevant sense that
meanings are precisely in the head — there is nowhere else for them
to be — and that Putnam's arguments fail to show anything to the
contrary.
Putnam considers two views:
(1) Knowing the meaning of a word or expression consists in
being in a certain psychological state.
(2) Meaning (intension) determines extension.
Appropriately construed these two entail a third:
(3) Psychological states determine extension.
Putnam tries to show that we cannot hold both (1) and (2) together
and that (3) is false. He proposes to reject (1) and (3) while
accepting a revised version of (2). In the discussion which follows
it is important to point out that nothing hangs on accepting the
traditional analytic—synthetic distinction; for the purposes of this
discussion both Putnam and I accept holism, and nothing in our
dispute turns on that issue.
Putnam's strategy is to try to construct intuitively plausible
cases where the same psychological state will determine different
extensions. If type-identical psychological states can determine
different extensions, then there must be more to the determination
of extension than psychological states, and the traditional view is,
therefore, false. Putnam offers two independent arguments to

2 H. Putnam, 'The meaning of meaning', in Philosophical Papers, vol. 2, Mind, Language and
Reality (Cambridge: Cambridge University Press, 1975), pp. 215-71.

2OO
 Are meanings in the head?

show how the same psychological state can determine different

extensions. He sometimes talks as if they were part of the same
argument but, in fact, they are quite independent and, I believe,
only the second is really serious. I will, therefore, deal rather briefly
with the first.
The first argument concerns what he calls the principle of "the
linguistic division of labor", i.e., the principle that in any
linguistic community some people have more expertise in applying
certain terms than others do. For example, in our community some
people know more about trees than others and so can tell which
trees are, for example, beeches and which are elms. Others, such as
myself, don't know much about the difference between beech trees
and elm trees, so insofar as there is any concept attaching to the
words "beech" and "elm" for me, they are pretty much the same
concept. In both cases I have the concept of a big, deciduous tree
growing in the Eastern part of the United States. Therefore,
according to Putnam, in my idiolect the concept or "intension" is
the same, but the extension is clearly different. "Beech" denotes
beech trees and "elm" denotes elm trees: same psychological state,
different extensions.
I really don't believe any defender of the traditional view would
be worried by this argument. The thesis that meaning determines
reference can hardly be refuted by considering cases of speakers
who don't even know the meaning or know it only imperfectly. Or
to put the same point another way, the notions of intension and
extension are not defined relative to idiolects. As traditionally
conceived, an intension or Fregean Sinn is an abstract entity which
may be more or less imperfectly grasped by individual speakers.
But it does not show that intension does not determine extension
to show that some speaker might not have grasped the intension,
or grasped it only imperfectly; for such a speaker hasn't got a
relevant extension either. The notion of the "extension in my
idiolect" has no application for cases where one does not know the
meaning of the word.
To make out the case, Putnam would have to argue that the
collectivity of speakers' Intentional states, including those of all
the ideal experts, does not determine the correct extensions. But if
the argument is to be based on linguistic and factual ignorance, the
very doctrine of the linguistic division of labor would seem to
201
 Are meanings in the head?

refute the argument from the start, because the doctrine is that
where one speaker is ignorant he can appeal to the experts: what is
and what is not an elm is for the experts to decide. That is, where his
intension is inadequate he lets their intension determine extension.
Furthermore, if we assume that Putnam knows this argument to be
valid we get something very much like an inconsistency as follows:
1. My (Putnam's) concept of " elm " = my concept of " beech "
but
2. The extension of "elm" in my idiolect =jt the extension of
"beech" in my idiolect.
How do I know 2 to be true? Obviously because
3. I know that beeches are not elms and elms are not beeches.
And how do I know that? I know that because I know that elms
and beeches are two different species of trees. Imperfect as my grasp
of the relevant concepts is, at least I have enough conceptual
knowledge to know that the two are distinct species. But for this
very reason,
4. Number 3 states conceptual knowledge.
If such knowledge is not conceptual knowledge, nothing is.
Therefore,
5. Contrary to 1, my concept of "elm"=^=my concept of
"beech".
In his more important and influential second argument Putnam
tries to show that even the collectivity of speakers' Intentional
states might be insufficient to determine extension, for there might
be two communities with the same set of collective intensions but
with different extensions. Imagine that in a distant galaxy there
was a planet very similar to ours with people like ourselves
speaking a language indistinguishable from English. Imagine,
however, that on this twin earth the stuff they call "water" is
perceptually indistinguishable from what we call "water", but in
fact it has a different chemical composition. What is called "water"
on twin earth is a very complicated chemical compound, the
formula for which we will abbreviate as "XYZ". According to
202
 Are meanings in the head?

Putnam's intuitions, the expression "water" on earth in 1750,

before anything was known about the chemical composition of
water referred to H2O; and "water" on twin earth in 1750 referred
to XYZ. Thus, even though the people on both earth and twin
earth were all in the same psychological state relative to the word
"water", they had different extensions and therefore Putnam
concludes that psychological states do not determine extension.
Most people who have criticized Putnam's argument have
challenged his intuitions about what we would say concerning the
twin earth example. My own strategy will be to accept his
intuitions entirely for the purpose of this discussion, and then
argue that they fail to show that meanings are not in the head. But I
want to digress for a moment and consider what the traditional
theorists would say about the example as presented so far. I think it
would go something like this: Up to 1750 "water" meant the same
on both earth and twin earth and had the same extension. After it
had been discovered that there were two different chemical
compositions, one for earth and one for twin earth, we would have
a choice. We could define "water" as H 2O, which is what we have,
in fact, done; or we could just say that there are two kinds of water,
and that water on twin earth is constructed differently from water
on earth. There is, indeed, some support for these intuitions.
Suppose, for example, there had been a great deal of going and
coming between earth and twin earth, so that speakers were likely
to have encountered both. Then it seems likely that we would
construe water as we now construe jade. Just as there are two kinds
of jade, nephrite and jadeite (Putnam's example), so there would be
two kinds of water, H2O and XYZ. Furthermore, it looks like we
pay a high price for accepting his intuitions. A very large number
of things have water as one of their essential components, so if the
stuff on twin earth is not water then presumably their mud is not
mud, their beer is not beer, their snow is not snow, their ice cream
is not ice cream, etc. If we take it really seriously, indeed, it looks as
if their chemistry is going to be radically different. On our earth if
we drive cars we get H2O, CO and CO2 as products of the
combustion of hydrocarbons. What is supposed to come out of the
cars on twin earth? I think that a defender of the traditional view
might also point out that it is odd that Putnam assumes that
" H 2 O " isfixedand that "water" is problematic. We could equally

203
 Are meanings in the head?

well imagine cases where H2O is slightly different on twin earth

from what it is on earth. However, I don't want to pursue these
alternative intuitions to Putnam's, rather I want to accept his
intuitions for the purpose of the argument and continue with his
positive account of how extension is determined.
On Putnam's theory the extension of a general term like
"water", and indeed on his theory just about any general term, is
determined indexically as follows. We identify a kind of substance
such as water by certain surface features. These are such things as
that water is a clear, tasteless, colorless liquid, etc. The crucial point
is that the extension of the word "water" is then determined as
whatever is identical in structure with this stuff, whatever that
structure is. Thus, on his account the reason that "water" on twin
earth has a different extension from "water" on earth is that the
stuff identified indexically has a different structure on twin earth
from the structure that it has on earth, and "water" is simply
defined as whatever bears the relation "same L " to this stuff.
Now from the point of view of a traditional theorist what
exactly does this argument achieve? Even supposing Putnam is
right about his intuitions, all he has done is substitute one
Intentional content for another. For the traditional cluster-of-
concepts Intentional content, Putnam has substituted an indexical
Intentional content. In each case it is a meaning in the head that
determines extension. In fact, Putnam's suggestion is a rather
traditional approach to natural kind terms: a word is defined
ostensively as whatever bears the right relation to the denotation
of the original ostension. "Water" has simply been defined as
whatever is identical in structure to this stuff whatever that
structure is. And this is simply one case among others in which
intensions, which are in the head, determine extensions.
On the traditional Lockean view, water is defined (nominal
essence) by a check list of concepts: liquid, colorless, tasteless, etc.
On the Putnam proposal, water is defined (real essence) indexically
by identifying something that satisfies the nominal essence and
then declaring that water is to be defined as whatever has the same
real essence as the stuff so identified. This may be an improvement
on Locke but it certainly does not show that meanings are not in
the head.
I believe Putnam would not regard this as an adequate response,
204
 Are meanings in the head?

since the whole tone of his writings on this topic is to suggest that
he takes himself not to be proposing a variation of the traditional
view that meanings are in the head but to be rejecting the tradition
altogether. The interest of this discussion for the present work
only becomes clear when we examine the underlying assumptions
about Intentionality that lead him to suppose that the alternative
account of meaning that he proposes is somehow fundamentally
inconsistent with the view that meanings are in the head. Let us try
to state his position a little more precisely. We can distinguish three
theses:
(1) The associated cluster of concepts does not determine
extension,
(2) The indexical definition does determine extension,
(3) What is in the head does not determine extension.
Now (3) does not follow from (1) and (2). To suppose that it
does one must assume that the indexical definition is not in the
head. Putnam uses (1) and (2) to argue for (3) and thereby assumes
that the indexical definition is not in the head. Now, why does he
think that? Why does he think that in the case of these indexical
definitions what is in the head does not determine extension? I
believe that there are two reasons why he makes this fallacious
move. First, he supposes that since we don't know the micro-
structure, and since it is the micro-structure that determines
extension, then what is in the head is insufficient to determine
extension.
But that, I believe, is simply a mistake; and we can illustrate the
way it is a mistake by considering the following example. The
expression, "The murderer of Brown", has an intension which
determines as its extension the murderer of Brown.3 The intension,
"The murderer of Brown", fixes the extension even though it is a
fact about the world who murdered Brown. For someone who
does not know who murdered Brown the extension of the
expression, "The murderer of Brown", is still the murderer of
Brown even though he does not know who it is. Now analogously,
the Intentional content "identical in structure with this (indexically
identified) stuff" is an Intentional content that would determine an

3 Strictly speaking it determines the unit class whose sole member is the murderer of
Brown, but for the purposes of this argument we can ignore this distinction.

205
 Are meanings in the head?

extension, even if we don't know what that structure is. The theory
that intension determines extension is the theory that intensions set
certain conditions which anything has to meet in order to be part of
the extension of the relevant intension. But that condition is
satisfied by Putnam's example: the indexical definition of water has
an Intentional content, that is, it sets certain conditions which any
potential sample has to meet if it is to be part of the extension of
"water", in exactly the same sense that the expression "The
murderer of Brown" sets certain conditions which any potential
candidate has to meet if he or she is to be the extension of "The
murderer of Brown". But in both cases it is a matter of fact about
the world, whether or not some existing entities satisfy the
Intentional content. It is, therefore, just a mistake to suppose that
because we define "water" in terms of an unknown micro-
structure, that intension does not determine extension.
But there is a second and deeper reason why Putnam supposes
that his analysis shows that meanings are not in the head. He makes
certain assumptions about the nature of Intentional contents and
the nature of indexical expressions and especially about the way
Intentional contents relate to indexical expressions, which we must
now explore. The assumptions emerge when he says:
For these (indexical) words no one has ever suggested the
traditional theory that 'intension determines extension'. To
take our Twin Earth example: if I have a Doppelganger on Twin
Earth, then when I think,' I have a headache', he thinks ' I have a
headache'. But the extension of the particular token of'I' in his
verbalized thought is himself (or his unit class, to be precise),
while the extension of the token of ' I ' in my verbalized thought
is me (or my unit class, to be precise). So the same word, ' I ' , has
two different extensions in two different idiolects; but it does
not follow that the concept I have of myself is in any way
different from the concept my Doppelganger has of himself.4
This passage makes it clear that Putnam supposes both that the
traditional view that what is in the head determines extension
cannot be applied to indexicals and that if two speakers, I and my
"'Doppelganger", have type-identical mental states our states must

4 Op. cit. p. 234.

206
 Are meanings in the head?

have the same conditions of satisfaction. I believe both these

assumptions are false. I want to argue, first, that if by "intension"
we mean Intentional content then the intension of an utterance of
an indexical expression precisely does determine extension; and,
second, that in perceptual cases two people can be in type-identical
mental states, indeed we can even suppose that a man and his
Doppelgdnger can be type-identical down to the last microparticle,
and their Intentional contents can still be different; they can have
different conditions of satisfaction. Both perceptual Intentionality
and indexicality are cases of self-referentiality of Intentional or
semantic content. We will explore the self-referentiality of in-
dexical propositions later in this chapter. For present purposes it is
sufificient to remind ourselves of the causal self-referentiality of
perceptual experience that we explored in Chapters 2 and 4 and to
show how it is relevant to the twin earth argument.
Let us suppose that Jones on the earth in 1750 indexically
identifies and baptizes something as "water" and twin Jones on
twin earth also indexically identifies and baptizes something as
"water". Let us also suppose that they have type-identical mental
contents and type-identical visual and other sorts of experiences
when they make the indexical identification. Now, since they give
the same type-identical definitions, namely, "water" is defined as
whatever is identical in structure with this stuff, and since they are
having type-identical experiences, Putnam supposes that we
cannot account for how "water" has a different extension on earth
from the extension on twin earth in terms of their mental contents.
If their experiences are the same, how can their mental contents be
different? On the account of Intentionality presented in this book
the answer to that problem is simple. Though they have type-
identical visual experiences in the situation where "water" is for
each indexically identified, they do not have type-identical Inten-
tional contents. On the contrary, their Intentional contents can be
different because each Intentional content is causally self-
referential in the sense that I explained earlier. The indexical
definitions given by Jones on earth of "water" can be analyzed as
follows: "water" is defined indexically as whatever is identical in
structure with the stuff causing this visual experience, whatever
that structure is. And the analysis for twin Jones on twin earth is:
"water" is defined indexically as whatever is identical in structure
207
 Are meanings in the head?

with the stuff causing this visual experience, whatever that

structure is. Thus, in each case we have type-identical experiences,
type-identical utterances, but in fact in each case something
different is meant. That is, in each case the conditions of
satisfaction established by the mental content (in the head) is
different because of the causal self-referentiality of perceptual
experiences.
This account does not have the consequence that different
speakers on earth must mean something different by "water".
Most people do not go around baptizing natural kinds; they just
intend to use words to mean and refer to whatever the community
at large, including the experts, use the words to mean and refer to.
And even when there are such public baptisms they would
normally involve on the part of the participants shared visual and
other experiences of the sort that we discussed in Chapter 2. But
the account does have the consequence that, in making indexical
definitions, different speakers can mean something different be-
cause their Intentional contents are self-referential to the token
Intentional experiences. I conclude, then, that even if we accept all
of his intuitions — which many of us will not — Putnam's arguments
do not show that meanings are not in the head. Quite the contrary,
what he has done is to offer us an alternative Intentionalistic
account, based on indexical presentations, of the meanings of a
certain class of general terms.

II. ARE THERE IRREDUCIBLY DE RE BELIEFS?5

I have never seen a clear and precise statement of what exactly the
de dictojde re distinction as applied to propositional attitudes is
supposed to be. Perhaps there are as many versions of it as there are
authors on the subject, and certainly the notions have gone far
beyond the literal Latin meanings, "of words" and "of things".
Suppose one believes, as I do, that all Intentional states are entirely
constituted by their Intentional content and their psychological
mode, both of which are in the head. On such an account all beliefs
are de dicto. They are entirely individuated by their Intentional

5 Like other authors who write on this topic I will use belief as an example for the whole
class of propositional attitudes.

2O8
 Are meanings in the head?

content and psychological mode. Some beliefs, however, are also

actually about real objects in the real world. One might say that
such beliefs are de re beliefs, in the sense that they refer to actual
objects. De re beliefs would then be a subclass of de dicto beliefs, in
the same way that true beliefs are a subclass of de dicto beliefs, and
the term "de dicto belief" would be redundant since it just means
belief.
On such a view, the belief that Santa Claus comes on Christmas
Eve and the belief that de Gaulle was President of France are both
de dicto, and the second is also de re since it is about real objects, de
Gaulle and France.
With such an account of the de rejde dicto distinction I would have
no quarrel. But several accounts in the philosophical literature
since Quine's original article6 advance a much stronger thesis: The
intuitive idea is that in addition to the class oide dicto beliefs which
are entirely individuated by their content and mode, by what is in
the head, there is a class of beliefs for which what is in the head is
insufficient to individuate the beliefs because such beliefs involve
relations between believers and objects as part of the identity of the
belief. Such beliefs are not a subclass of de dicto beliefs, but are
irreducibly de re. Purely de dicto beliefs could be held by a brain in a
vat; they are independent of how the world is in fact. But de re
beliefs, on this view, are relations between believers and objects;
for them, if the world were different in certain ways, the beliefs
themselves would be different even though what is in the head
remained unchanged.
There are as near as I can tell three sets of considerations that
incline people to the view that there are irreducibly de re beliefs.
First, there just does seem to be a class of beliefs which are
irreducibly about objects, that is, beliefs which relate the believer
to an object and not just to a proposition and in that sense are de re
rather than de dicto. For example, suppose that George Bush
believes that Ronald Reagan is President of the United States. Now
that is clearly a fact about Bush, but under the circumstances isn't it
equally clearly a fact about Reagan? Isn't it just a plain fact about
Reagan that Bush believes him to be President? Furthermore, there

6 W. V. Quine, 'Quantifiers and propositional attitudes', in Ways of Paradox (New York:

Random House, 1966), pp. 183-94.

209
 Are meanings in the head?

is no way to account for the fact simply in terms of facts about

Bush, including facts which relate him to propositions. The fact in
question is stated by a proposition of the form
About Reagan, Bush believes him to be President of the United
States
or, more pretentiously,
Reagan is such that Bush believes of him that he is President of
the United States.
Such propositions, describing de re beliefs, permit quantification
into "belief contexts"; that is, each permits an inference to
(3x) (Bush believes (j is President of the United States) of x)
According to received opinion, both our logical theory and OUJ
theory of mind compel us to such an analysis.
Second, there is clearly a distinction between propositional
attitudes which are directed at particular objects and those which
are not. In Quine's example, we need to make a distinction
between the desire a man might have for a sloop where any old
sloop will do, and the desire a man might have which is directed at
a particular sloop, the sloop Nellie parked at the Sausalito Yacht
Harbor. In the first or de dicto desire, the man seeks - as Quine says
-mere "relief from slooplessness", in the second or de re desire the
man's desire relates him to a particular object. The difference
according to Quine is expressed in the following two sentences?
de dicto: I wish that (3x) (x is a sloop & I have x)
de re: (3x (x is a sloop & I wish that I have x)
Third, and I believe most important, there is supposed to be a
class of beliefs which contain a "contextual", "nonconceptual"
element, and for that reason are not subject to an internalist or de
dicto account. As Tyler Burge writes,8 "A de re belief is a belief
whose correct ascription places a believer in an appropriate
nonconceptual contextual relation to objects the belief is about . . .
The crucial point is that the relation not be merely that of concepts

7 See Quine, op. cit., p. 184.

8 T. Burge, 'Belief de re', Journal of Philosophy, vol. 74, no. 6 (June 1977), pp. 338-62.

2IO
 Are meanings in the head?

being concepts 0/the object — concepts that denote or apply to it"

(first italics mine). According to Burge such beliefs cannot be
completely or exhaustively characterized in terms of their Inten-
tional contents, because, as he puts it, there are contextual,
nonconceptual elements which are crucial to the identity of the
belief.
I believe that all three of these reasons can be answered rather
swiftly, and that all three embody various confused notions of
Intentionality. I begin with the third set of reasons, as a discussion
of them prepares the way for a discussion of the earlier two; and I
will confine my remarks to Burge because he gives the strongest
statement of the de re thesis known to me.
Implicit in Burge's account is a contrast between the conceptual
and the contextual. A fully conceptual belief is de die to and
completely analyzable in general terms. A contextual belief is
individuated in part by relations between the believer and objects
in the world and is therefore de re. His strategy is to argue by way of
examples that there are beliefs that are not fully conceptual but are
contextual. I agree that there are beliefs that are not fully
conceptual in the sense that they do not consist in verbal
descriptions in general terms but that does not show that they are
contextual or de re in his sense. In addition to the two options of
"conceptual" or "contextual" there is a third possibility, there are
forms of Intentionality which are not general but particular and yet
are entirely in the head, entirely internal. Intentionality may
contain self-referential elements both of the causal kind that we
considered in our discussion of perception, memory, intention,
and action, and of the indexical kind that I alluded to briefly in the
discussion of Putnam and will say more about later in this chapter.
A proper understanding of the self-referentiality of certain forms
of Intentionality is, I believe, sufficient to account for all of Burge's
examples of allegedly de re beliefs, since in each case the Intentional
content can be shown to account completely for the content of the
belief. And that is just another way of saying that, in the relevant
sense, the belief is de die to.
His first example is of a man seen coming from the distance in a
swirling fog. Of this example he says, "We may plausibly be said to
believe of him that he is wearing a red cap, but we do not see the
man well enough to describe or image him in such a way as to
211
 Are meanings in the head?

individuate him fully. Of course, we could individuate him

ostensively with the help of the descriptions that we can apply but
there is no reason to believe that we can always describe or
conceptualize the entities or spatio-temporal positions that we rely
on in our demonstration."
I find this passage very revealing, since it says nothing at all
about the Intentional content of the visual experience itself which
in this case is part of the content of the belief. Once you understand
that the visual experience has a causally self-referential propo-
sitional content you don't need to worry about "describing" or
"conceptualizing" anything in words in order to individuate the
man: the Intentional content of the visual experience has already
done it. On my account the (de die to) Intentional content of the
visual experience individuates the man, and that content is part of
the (de dicto) content of the belief. The relevant de dicto Intentional
content of the belief can be expressed as follows:
(There is a man there causing this visual experience and that man
is wearing a red cap.)
In such a case the "contextual" elements are indeed present, but
they are fully internalized in the sense that they are part of the
Intentional content. Notice that this de dicto belief is quite sufficient
to individuate any alleged de re analogue but at the same time it is
consistent with the hypothesis that there is no man there at all.
Such a belief as this could be held by a brain in a vat. It might be
objected that this analysis has the consequence that it is in principle
impossible for two different people to have the same perceptual
belief. But that consequence does not follow, for the same man
may be part of the conditions of satisfaction of two different
perceptual beliefs; and it may be even part of the content of two
perceptual beliefs that they should have exactly the same man as
part of their conditions of satisfaction. Thus, in the case of shared
visual experiences, I may believe not only that I am seeing a man
and that you are seeing a man but that we are both seeing the same
man. In such a case, the conditions of satisfaction will require not
only that there is a man causing my visual experience, but that the
same man is also causing your visual experience. Of course our
beliefs will be different in the trivial sense that any self-referential
perceptual content makes reference to a particular token and not to
212
 Are meanings in the head?

qualitatively similar tokens, but that is a result we want anyway,

since, when you and I share a visual experience, what we share is a
common set of conditions of satisfaction and not the same token
visual experiences. Your experience will be numerically different
from mine even though they may be qualitatively similar.
The next class of cases considered by Burge are indexicals. His
example is that of a man who believes of the present moment that it
is in the twentieth century. But this is subject to Intentionalistic
analysis formally similar to that we gave in the perceptual case. The
method here as before is always to ask what must be the case in
order that the Intentional content is satisfied. In the case of visual
perception, the visual experience itself must figure causally in the
conditions of satisfaction. In the case of the indexicals, there is an
analogous self-referentiality though this time it is not causal. The
truth conditions of" This moment is in the twentieth century", are
that the moment of this utterance is in the twentieth century. Just
as the perceptual case is self-referential to the experience, so the
indexical case is self-referential to the utterance. I hasten to add that
this statement of the conditions of satisfaction is not meant as a
translation of the original sentence: I am not saying that "this
moment" just means "the moment of this utterance". Rather, what
I am arguing is that the indexical operator in the sentence indicates\
though does not represent or describe, the form of the self-
referentiality. The self-referentiality of indexical expressions is in
that sense shown but not said, just as the self-referentiality of visual
experience is 'shown' but not 'seen'. In the case of the statement
of the conditions of satisfaction, I describe or represent or say what
was indicated or shown in the original.
I conclude, then, that there is nothing irreducibly de re about
either perceptual or indexical beliefs. They are subject to an
Intentionalistic or de dicto analysis and the mistake of supposing
there must be irreducibly de re sets of perceptual or indexical beliefs
seems to rest on the assumption that all de dicto Intentionalistic
analyses must be given using purely general words. Once the self-
referential forms of indexicality and perceptual experience are
explicated it is easy to see that there are forms of Intentionality
where the Intentional contents are sufficient to determine the entire
sets of conditions of satisfaction but they do not do so by setting
purely general conditions, but rather by indicating relations in
213
 Are meanings in the head?

which the rest of the conditions of satisfaction must stand to the

Intentional state or event itself.
The diagnosis, then, of the mistake made by the de re theorists
who rely on perceptual and indexical beliefs is the following: they
correctly see that there is a class of beliefs that cannot be accounted
for in purely general terms. They also see that these beliefs depend
on contextual features, and they then mistakenly suppose that
these contextual features cannot themselves be entirely represented
as part of the Intentional content. Having contrasted the con-
ceptual (in general terms) with the contextual (involving the real
world) they then ignore the possibility of a completely internalist
account of nonconceptual beliefs. I am arguing for forms of
Intentionality that are not conceptual but not de re either.
Part of the difficulty here, I am convinced, comes from this
archaic terminology which seemingly forces us to choose between
the views that all beliefs are in words {dicta) and that some involve
things {res). We can sort this out if we distinguish between several
different questions. The question, "Are all beliefs de dicto?", tends
to oscillate between at least four different interpretations.
1. Are all beliefs expressible using purely general terms?
2. Do all our beliefs occur to us in words which are sufficient to
exhaust their content?
3. Do all of our beliefs consist entirely in an Intentional
content?
4. Do some beliefs relate the believer directly to an object
without the mediation of an Intentional content which is
sufficient to individuate the object? Are they such that a
change in the world would necessarily mean a change in the
belief even if what is in the head remained constant?
The answer to the first two questions is no: the first, because many
beliefs contain singular terms essentially, as we will see in our
discussion of indexicals; and the second, because many beliefs
contain, for example, a perceptual content, as we saw in the case we
considered where a belief contains a visual experience as part of its
content. But a negative answer to the first two questions does not
entail a negative answer to the third: a belief can be exhaustively
characterized by its Intentional content, and in that sense be a de
dicto belief, even though it is not characterizable in general terms
and contains nonverbal forms of Intentionality. If by de dicto we
214
 Are meanings in the head?

mean verbal, in words, then not all beliefs are de dicto, but it does not
follow from that that there are irreducibly de re beliefs, because a
negative answer to the first two questions does not entail an
affirmative answer to the fourth. If the answer to 3 is yes, that is, if,
as I believe, all beliefs consist entirely in their Intentional content,
then it is consistent to claim that the answer to 1, 2, and 4 is no. In
once sense of de dicto, there are some beliefs that are not de dicto (in
words), but that does not show that there are any irreducibly de re
beliefs, because in another sense of de dicto (Intentional content) all
beliefs are de dicto (which illustrates, among other things, that this
terminology is muddled).
Using these results we can now turn to the other two arguments
for the belief in irreducibly de re attitudes. The first argument says
correctly that it is a fact about Ronald Reagan that Bush believes
him to be President. But in what does this fact consist? On my
account it consists simply in the fact that Bush believes the de dicto
proposition that Ronald Reagan is President of the United States,
and that Ronald Reagan satisfies the Intentional content associated
with Bush's use of the name "Ronald Reagan". Some of this
content is perceptual, some indexical, much of it is causal; but all of
it is de dicto in the sense that it consists entirely in an Intentional
content. Bush could have had exactly the same belief if Ronald
Reagan had never existed and the whole thing, perceptions and all,
had been a massive hallucination. In such a case Bush would have
had a lot of perceptual, indexical, and causal Intentional contents
which nothing satisfied.
Quine's argument I believe rests on confusing the distinction
between particular and general propositional attitudes with a
distinction between de re and de dicto propositional attitudes. There
really is a distinction between those Intentional states that make
reference to a particular object and those that do not. But in each
case the state is de dicto. On this view, the sentence that Quine gives
to express the de re attitude cannot be correct, because the sentence
expressing the desire for a particular sloop is incomplete: there is
no way an agent can have a desire for a particular object without
representing that object to himself in some way, and Quine's
formahzation does not tell us how the object is represented. In the
example as stated the agent would have to have a belief in the
existence of a particular sloop and a desire to have that very sloop.
 Are meanings in the head?

The only way to express the relation between the belief in the
existence of a particular sloop and the desire to have it in the
quantifier notation is to allow the scope of the quantifier to cross
over the scope of the Intentional operators. That this is the correct
way to represent the facts is at least suggested by the fact that we
would so express the man's mental state in ordinary language.
Suppose the man who wants a particular sloop gave expression to
his whole mental state including his representation of the sloop.
He might say,
There is this very nice sloop in the yacht harbor and I sure
wish I had it.
The mental states he expressed here are, first, a belief in the
existence of a particular sloop and, then, a desire to have that sloop.
In English,
I believe that there is this very nice sloop in the yacht harbor and I
wish I had it.
Notice that in this formulation the scope of the quantifier in the
content of the belief extends to the content of the desire even
though the desire is not within the scope of the belief. Thus, using
square brackets for the scope of the Intentional verbs and round
brackets for the quantifier and F for the Intentional content which
identifies the sloop in question, we have:
Bel [(3x) ((sloop x & Fx) & (Vj) (sloopj & Fj -+y = x)\ &
Des [I have x])
This de dicto form represents the entire content of the desire
directed at a particular object.
We have so far considered and rejected some arguments in favor
of the belief in de re propositional attitudes. I want to conclude with
a Wittgensteinian diagnosis of what I believe to be the deepest but
unstated motives for the belief in irreducibly de re attitudes. The
belief in two fundamentally different kinds of propositional
attitudes, de re and de dicto, derives from the possibility that our
language provides of giving two different kinds of reports of
propositional attitudes, de re reports and de dicto reports. Suppose,
for example, that Ralph believes that the man in the brown hat is a

216
 Are meanings in the head?

spy.9 Now about Ralph's belief wTe can either say, "About the man
in the brown hat, Ralph believes he is a spy", or, "Ralph believes
that the man in the brown hat is a spy". The first report commits
us, the reporters, to the existence of the man in the brown hat. The
second report commits us only to reporting the content of Ralph's
belief. Now since sentences about beliefs can differ in this way, and
indeed can have different truth conditions, we are inclined to think
that there must be a difference in the phenomena reported. But
notice that the distinction we can make between the de re report of
Ralph's belief and the de dicto report is not a distinction that Ralph
can make. Suppose Ralph says, "About the man in the brown hat, I
believe he is a spy", or he says, "I believe that the man in the
brown hat is a spy". From Ralph's point of view these amount to
exactly the same belief. Imagine the craziness of the following
conversation:
Quine: About the man in the brown hat, Ralph, do you believe
he is a spy?
Ralph: No Quine, you've asked me if I hold a de re belief, but it
is not the case that about the man in the brown hat I
believe that he is a spy. Rather, I believe the de dicto
belief, I believe that the man in the brown hat is a spy.
Just as the belief that Intentional-with-a-t states are somehow
intrinsically intensional-with-an-s entities is founded on the
confusion between logical properties of reports of Intentional
states with logical properties of the Intentional states themselves,
so the belief that there are two different kinds of Intentional states,
de re and de dicto, is founded on confusing two different kinds of
reports of Intentional states, de re and de dicto reports, with logical
features of the Intentional states themselves. I conclude, then, that
there is a genuine de re\de dicto distinction, but it is only a distinction
in kinds of reports. If de re propositional attitudes are supposed to
be those in which Intentional content is insufficient to individuate
the mental state, then there are no such things as de re propositional
attitudes; though there are de re reports of propositional attitudes in
the sense that there are reports that commit the reporter to the
existence of objects that the propositional attitudes are about.
9 The example is, of course, Quine's, 'Quantifiers and propositional attitudes'.

217
 Are meanings in the head?

III. INDEXICAL EXPRESSIONS

In both our discussion of Putnam's attack on internalism in

semantics and our discussion of the alleged existence of irreducibly
de re beliefs, we have suggested an account of indexical expressions
and it is now time to make that account fully explicit.
There is at least one big difference between the problem of de re
attitudes and the problem of indexicals: there are no such things as
irreducibly de re propositional attitudes, but there really are
indexical expressions and indexical propositions. The strategy
therefore in this section will differ from that of the previous
sections. First, we need to develop a theory of indexicals; second,
to do it in such a way as to show how it fits in with the general
account of Intentionality developed in this book; and, third, in so
doing to answer those accounts of indexicals which claim that it is
impossible to assimilate indexicals to an internalist or Fregean
account of language. I begin with some of the arguments of the
opposition.
Various authors, notably Perry10 and Kaplan11 maintain that
there are thought contents which are essentially indexical. Con-
sider, for example, the belief I might have if I come to believe that I
am inadvertently making a mess in a supermarket by spilling sugar
out of my cart. If I come to believe that I am making a mess, the
content of my Intentional state seems to contain an essential
indexical element; and this is shown by the fact that no paraphrase
of my belief into any nonindexical terms will capture exactly the
belief I have when I believe that I am making a mess. If I try to
specify the belief using space and time coordinates, I will not be
able to specify the content of my belief. For example, my
possession of the belief that person^ is making a mess at location /
and at time / would not explain how my behavior changes when I
discover that it is me that is making the mess, since I might have the
belief that some person satisfying certain space—time coordinates is
making a mess without realizing that it is me. Analogous remarks
apply to definite descriptions and proper names: the belief that I am

10 J. Perry, 'The problem of the essential indexical', NOUS, vol. 13, no. 1 (March 1979),
pp. 3-21.
11 D. Kaplan, 'Demonstratives', mimeo, UCLA, 1977.

218
 Are meanings in the head?

making a mess is not the same as the belief that the only non-
bearded philosopher in the Berkeley Co Op is making a mess or the
belief that JS is making a mess, for I might have these latter beliefs
without knowing that I am the only non-bearded philosopher in
the Berkeley Co Op or that I am JS. The content of my belief
seems, then, to be essentially indexical.
As I am sure both Perry and Kaplan are aware, there is nothing
so far that is anti-Fregean or anti-internalist about this point. In
fact, it looks like a paradigm example of Frege's distinction
between sense and reference. Just as the proposition that the
Evening Star shines near the horizon is different from the
proposition that the Morning Star shines near the horizon, so the
proposition that I am making a mess is different from the
proposition that JS is making a mess. So far, so Fregean.
The anti-Fregean point comes next. According to Perry12 and
Kaplan13 there is no way a Fregean can account for such essentially
indexical Intentional contents, because in such cases there is no
"completing Fregean sense" which is sufficient by itself to
determine the conditions of satisfaction. To illustrate and support
this claim Perry introduces the following sort of example. Suppose
David Hume believes "I am David Hume". Suppose also that
Heimson believes "I am David Hume", and just to make the
strongest case let us suppose that Heimson is David Hume's
Doppelgdnger on twin earth and that he has type-identical mental
states with David Hume, and indeed we can suppose that he is type
identical with Hume down to the last microparticle. Now the
sentence that Hume and Heimson both utter (or think), "I am
David Hume", has the same Fregean sense on both occasions and
Heimson and Hume are in type-identical mental states. But the
propositions expressed must be different because they have
different truth values. Hume's is true, Heimson's is false. There is a
Fregean sense to the sentence, "I am David Hume", but it is not
sufficient to determine which proposition is expressed. Kaplan and
Perry conclude from such examples that the Fregean account of
sense and reference and the Fregean account of propositions must

12 J. Perry, 'Frege on demonstratives', The Philosophical Review, vol. 86, no. 4 (October
X
977), PP- 474-97-
13 Op. cit.

219
 Are meanings in the head?

be inadequate to explain indqxicals. Since what is expressed in such

utterances is essentially indexical and since there is no completing
Fregean sense, we need another theory of propositions at least for
such cases.
At this point they adopt what I believe is a desperate expedient,
the theory of "direct reference" and "singular propositions".
According to them, in such cases the proposition is not the
Intentional content in the mind of the speaker but rather the
proposition must contain the actual objects referred to. Hume's
proposition contains Hume, the actual man and not some
representation of him, and Heimson's proposition contains Heim-
son, the actual man and not some representation of him.
Expressions which (like Russell's logically proper names) intro-
duce objects themselves into propositions are said to be "directly
referential" and the propositions in question are (misleadingly)
said to be "singular propositions".
I am, frankly, unable to make any sense of the theory of direct
reference and singular propositions, but for the purposes of this
argument I am not attacking its intelligibility but its necessity to
account for the data: I think that the arguments for it are
inadequate and rest on a misconception of the nature of Inten-
tionality and of the nature of the functioning of indexicals.

(i) How do indexical expressions work?

We need to develop an account of indexicals which will show how
the utterance of an indexical expression can have a "completing
Fregean sense":14 that is, we need to show how in the utterance of
an indexical expression a speaker can express an Intentional
content which is sufficient to identify the object he is referring to in
virtue of the fact that the object satisfies or fits that Intentional
content.
In what follows I will confine the discussion to indexical referring
expressions such as " I " , "you", "this", "that", "here", "now",

14 Though remember, the account is not Fregean in postulating a third realm of abstract
entities. Ordinary Intentional contents will do the job. When I say "completing Fregean
sense" I do not mean to imply that such senses are abstract entities, but rather that they
are sufficient to provide adequate "modes of presentation".

22O
 Are meanings in the head?

"he", "she", etc. But it is worth pointing out that the phenomenon
of indexicality - the phenomenon of the conditions of satisfaction
being determined in virtue of relations things have to the
realization of the Intentional content itself — is quite general and
extends beyond just referring expressions and indeed even beyond
cases of indexical expressions. Various forms of indexicality are part
of the nonrepresentational Background. For example, I now
believe that Benjamin Franklin was the inventor of bifocals.
Suppose that it was discovered that 80 billion years before
Benjamin Franklin's discovery, in a distant galaxy, populated by
organisms somewhat like humans, some humanoid invented the
functional equivalent of bifocals. Would I regard my view that
Benjamin Franklin had invented bifocals as false? I think not.
When I say Benjamin Franklin invented bifocals there is a
concealed indexical in the background: the functioning of the
Background in such cases assigns an indexical interpretation to the
sentence. Relative to our earth and our history, Benjamin Franklin
invented bifocals; the statement that Benjamin Franklin invented
bifocals is, therefore, like most statements, indexical; even though
there are no indexical expressions (other than the tense of the verb)
contained in the sentence used to make the statement.
Let us begin by asking what indexical referring expressions have
in common that makes them indexical? What is the essence of
indexicality? The defining trait of indexical referring expressions is
simply this: In uttering indexical referring expressions, speakers
refer by means of indicating relations in which the object referred
to stands to the utterance of the expression itself. " I " refers to the
person uttering the expression, "you" refers to the person
addressed in the utterance of the expression, "here" refers to the
place of the utterance of the expression, "now" refers to the time
of the utterance of the expression, and so on. Notice that in every
case the speaker will refer to a particular entity because his
utterance expresses an Intentional content that indicates relations
that the object he is referring to has to the utterance itself. The
utterance of indexical expressions, therefore, has a form of self-
referentiality which is similar to the self-referentiality of certain
Intentional states and events, and we will need to explore it in more
detail. But at this point we need only note that this self-referential
feature is sufficient to account for how the utterance of an indexical
221
 Are meanings in the head?

expression can have a completing Fregean sense. The problem for

a Fregean (internalist or Intentionalist) account of reference is to
show in every case how reference succeeds in virtue of the fact that
the utterance sets conditions of satisfaction and an object is
referred to in virtue of the fact that it meets those conditions. An
object is referred to in virtue of satisfying an Intentional content,
normally expressed by a speaker in the utterance of an expression.
This is the basic idea of Frege's notion of the "Sinn" of
"Eigennamen". His favorite examples are cases such as "the Morning
Star", where the lexical meaning of the expression is supposedly
sufficient to determine which object is referred to. What is special
about indexical expressions is that the lexical meaning of the
expression by itself does not determine which object it can be used to
refer to, rather the lexical meaning gives a rule for determining
reference relative to each utterance of the expression. Thus the same
unambiguous expression used with the same lexical meaning can
be used to refer to different objects because the lexical meaning
determines that the conditions laid down by the utterance of the
expression, viz., the completing sense expressed by the speaker in
its utterance, is always self-referential to the utterance itself. Thus,
for example," I " has the same lexical meaning when uttered by you
or me, but the reference in each case is different because the sense
expressed by my utterance is self-referential to that very utterance
and the sense expressed by your utterance is self-referential to your
utterance: in any utterance " I " refers to the person who utters it.
There are, then, three components to the Fregean sense
expressed by a speaker in the utterance of indexical expressions: the
self-referential feature which is the defining trait or essence of
indexicality; the rest of the lexical meaning, which can be expressed
in general terms; and for many indexical utterances, the awareness
by the speaker and the hearer of the relevant features of the actual
context of the utterance, as in, for example, perceptual demonstra-
tives, e.g., "that man over there". We need to explore each of these
features in turn.

Self-referentiality. How does it work? Recall that for visual

experiences the specification of the conditions of satisfaction
makes reference to the visual experience itself. If I see my hand in
front of my face then the conditions of satisfaction are
222
 Are meanings in the head?

Vis Exp (there is a hand there and the fact that there is a hand
there is causing this Vis Exp).
The form of the conditions of satisfaction of indexical pro-
positions is analogously self-referential; though there is a dif-
ference in that the self-referentiality of the indexical cases is not
causal. The sense in which the indexical cases are self-referential,
like the case of Intentional self-reference, does not imply that the
speaker in making the utterance performs a speech act of referring to
the utterance, nor is the utterance explicitly represented in itself.
Rather, the specification of the conditions of satisfaction, e.g., the
truth conditions, requires reference to the utterance itself. Con-
sider any utterance of the sentence, "I am now hungry". That
utterance will be the making of a true statement iff the person
uttering the sentence is hungry at the time of the utterance of
the sentence. The conditions of satisfaction can therefore be
represented as follows:
(the person making this utterance, " I " , is hungry at the time of
this utterance, "now").
This analysis does not imply that " I " is synonymous with "the
person making this utterance", nor is "now" synonymous with
"the time of this utterance". They could not be synonymous
because the self-referentiality of the original is shown but not
stated, and in the statement of the truth conditions we have stated
it and not shown it. Just as we do not see the visual experience even
though the visual experience is part of its own conditions of
satisfaction, and is in that sense self-referential, so we do not refer
to (in the speech act sense) the utterance of the indexical
expression, even though the utterance is part of its own truth
conditions and is in that sense self-referential. The self-
referentiality of the visual experience is shown but not seen; the self-
referentiality of the indexical utterance is shown but not stated. If we
wanted to introduce a synonym which showed the indexicality we
could introduce an arbitrary device, such as the asterisk symbol (*)
to indicate the indexicality, i.e., to express the fact without stating
it that the expression was being used to refer by way of indicating
relations in which the referred to object stood to the utterance of
the expression itself. Such a form of expression would give a
223
 Are meanings in the head?

canonical notation for isolating the self-referentiality of indexical

expressions:
I = *person uttering
you = *person addressed
here = *cospatial
now = *cotemporal
and so on. All of these equivalences give us a display of the meaning
of expressions, and consequently a display of the meanings of
sentences containing these expressions. Thus, for example, the
meaning of the sentence "I am hungry" is given by
*person uttering is hungry at *cotemporal.
Nonindexical descriptive content. We will deepen our understanding
of the self-referential feature of indexical expressions if we see how
it latches on to the rest of the lexical meaning, the nonindexical
descriptive content, of the expression. I said that all indexical
referring expressions refer by indicating relations in which the
object referred to stands to the utterance of the expression. This
naturally raises the question, how many kinds of relations are
indicated in this manner? In English and in other languages known
to me there are certainly four, and possibly five, relations indicated
by the literal meaning of indexical expressions. These four are:
(1) time: examples of such expressions are "now", "yester-
day", "tomorrow", and "later on";
(2) place: e.g., "here" and "there";
(3) utterance directionality: "you" refers to the person being
addressed in the utterance, " I " refers to the person uttering;
(4) discoursal relations: anaphoric pronouns and expressions
such as "the former" and "the latter" refer to something in
virtue of its relation to the rest of the discourse in which the
indexical utterance is embedded.
Notice that in each of these examples the nonindexical descriptive
lexical meaning contains two elements: a sense which expresses the
particular determinate form of the determinable relation indicated,
and a sense which expresses the sort of entity being referred to.
Thus, "yesterday" expresses the determinate time indication "one
day before", and the type of entity referred to is a day. Thus the
224
 Are meanings in the head?

entire set of conditions of satisfaction expressed by "yesterday"

are: the day which is one day before the day of this utterance. Not
all indexicals have a lexical meaning which is in this way complete,
for example, the demonstratives "this" and "that" usually require
an extra expression ("this man" or "that tree"), as well as an
awareness of the context in order to express a completing Fregean
sense in a given utterance. More about this later.
These four are certainly forms of indexical relations expressed in
the literal meaning of English indexical expressions. It has been
argued that another relation is indicated by such words as " actual"
and "real", the idea being that the word "actual" expresses its
sense indexically by referring to the world in which it is uttered; and
thus among possible worlds the actual world is picked out
indexically. I think this claim is completely false; however, since it
involves modal issues that go beyond the scope of this book I will
not discuss it further here.15
Though there are only four (or arguably five) forms of indexical
relations indicated in the lexical meaning of expressions in actual
languages such as English, there is no limit in^ principle to
introducing new forms of indexicality. We might, for example,
have an expression which when uttered at a certain pitch would
indicate sounds of a higher or lower pitch or of the same pitch.
That is, we could imagine a class of indexical expressions that are
used to refer to tonal qualities by indicating relations in which the
tonal qualities stand to the tonal quality of the utterance analog-
ously to the way that "today", "yesterday", and "tomorrow"
refer to days by indicating relations in which they stand to the day
of the utterance of the expression itself.

Awareness of the context of utterance. Often the literal utterance of an

indexical expression will not by itself carry a completing Fregean
sense, but the completing Fregean sense is provided by the
Intentional content of the indexical utterance together with the
Intentional content of the awareness by the speaker and the hearer
of the context of the utterance. One sees this most clearly in the
case of the utterance of the demonstratives "this" and "that".

15 For a criticism of the view see P. van Inwagen, 'Indexicality and actuality', The
"Philosophical Review, vol. 89, no. 3 (July 1980), pp. 403-26.

225
 Are meanings in the head?

Suppose upon seeing a man behave strangely at a party I say " That
man is drunk". Now, in this case the descriptive content of "man"
together with the indexical does not provide the completing
Fregean sense because the utterance is only meant and understood
in the context of an accompanying visual perception of which man
is meant, and the proposition expressed has to contain the
Intentional content of the perceptual experience that accompanied
the utterance. The argument for this is simply that somebody who
doesn't have the relevant perceptual experiences, e.g., because he is
listening to me on the telephone or is blind or overhears me from
the next room, cannot fully grasp the proposition I express;
without the perceptual experience, he literally doesn't understand
the entire proposition even though he understands all the words
uttered.
In such cases a complete analysis of the proposition which
makes the completing Fregean sense fully explicit would have to
include both the Intentional content of the utterance and the
Intentional content of the visual experience, and it would have to
show how the latter is nested in the former. Here is how it works.
The indexical expression refers by indicating relations in which the
object stands to the utterance of the expression itself. In this case,
then, there is some relation R such that the truth conditions of the
utterance are expressible as
The man who stands in relation R to this utterance is drunk.
And, in the case as described, R is perceptual and temporal; the
man who is referred to is the man we are seeing at the time of this
utterance. But if we are seeing someone at the time of this utterance
each of us will also have a visual experience with its own present-
tense propositional content:
Vis exp (there is a man there and the fact that there is a man there
is causing this visual experience).
Now that Intentional content simply plugs into the Intentional
content of the rest of the utterance to give us the completing
Fregean sense which identifies the man uniquely in virtue of both
the self-referentiality of the utterance and the self-referentiality of
the visual experience. The entire conditions of satisfaction of the

226
 Are meanings in the head?

whole proposition (with the self-referential parts italicized) are

expressible as follows:
((there is a man, x, there, and the fact that x is there is causing this
pis exp) and x is the man visually experienced at the time of this
utterance and x is drunk).
This may look strange, but I think that the reader who is prepared
to recognize the Intentionality of the visual experience, its role in
the Intentionality of the proposition expressed by the utterance,
the self-referentiality of the visual experience, and the self-
referentiality of the indexical utterance, will see that something like
this formulation has to be right. It is intended to capture both the
indexical and the perceptual content of the proposition and the
relations between them. In the case of the perceptual use of the
demonstratives, both the sense of the indexical expression and the
Intentional content contained in the perceptual experience that
accompanies the utterance contribute to the propositional content
expressed in the utterance. Notice that in these cases we have a
completing Fregean sense sufficient to identify the object. Notice,
further, that there is no twin earth problem for these cases. I, on
this earth, and my Doppelganger, on twin earth, will express
different Fregean senses in our use of the demonstrative "That
man", even though our utterances and our experiences are
qualitatively type identical. His perception and his utterance are
both self-referential, as are mine.
Let us now summarize the account. We need to distinguish
between an indexical expression with its literal meaning, the literal
utterance of an indexical expression, and the sense expressed by a
speaker in the literal utterance of the expression. Analogously, we
need to distinguish the indexical sentence (i.e., any sentence
containing an indexical expression or morpheme, such as the tense
of a verb) with its literal meaning, the literal utterance of an
indexical sentence, and the proposition expressed by the speaker in
a literal utterance of an indexical sentence. The meaning of the
indexical expression by itself is not sufficient to provide the
completing Fregean sense, since the same expression with the same
meaning can be used to refer to different objects, e.g., different
people refer to themselves by uttering " I " . But the literal indexical

227
 Are meanings in the head?

meaning is such as to determine that when a speaker makes an

utterance of that expression the sense he expresses will be relative
to that utterance. So the sense of the expression can become a
completing Fregean sense relative to an utterance because the
lexical sense determines that any utterance is self-referential to that
very utterance. And this explains how two different speakers can
utter the same sentence with the same meaning, e.g., "I am
hungry", and still express different Fregean propositions: each
proposition expressed is self-referential to the utterance in which it
is expressed. It is the completing Fregean sense expressed which
determines the reference and it is the Fregean sense and not the
reference which is a constituent of the proposition. It cannot be
emphasized too strongly that there is nothing reductionist or
eliminative about this account of indexicality. I am not trying to
show that indexicality is really something else, but rather I am
trying to show what it is and how it works in utterances to express
Intentional contents.

(ii) How this account answers the objection to an internalist account of

indexicals
In the course of developing an independently motivated account
of indexicals we have, in passing, answered the objection of Perry
and Kaplan that no Frege-like account of indexicals can provide a
completing Fregean sense. Hume and Heimson utter the same
sentence with the same literal meaning but each utterance
expresses a different Intentional content; and each, therefore, has a
different completing Fregean sense, because each proposition
expressed is self-referential to the utterance which expresses the
proposition. In every case we have shown how the. self-
referentiality of the indexical utterance, as determined by the rule
for using the indexical expression, sets the conditions which an
object has to meet in order to be the referent of that utterance.
Perry argues correctly that there are essentially indexical thought
contents (propositions, in my sense), but he argues, in my view
incorrectly, that there is no completing Fregean sense for
essentially indexical thought contents. And from these two
premises he concludes that the propositions expressed in such
cases can only be accounted for on a direct reference theory. I
228
 Are meanings in the head?

accept the first of his premises but reject the second and his
conclusion. Indexical expressions are not counterexamples to the
claim of the theory of Intentionality that objects are referred to by
utterances only in virtue of the sense of the utterance, only by
virtue of the fact that the utterance sets conditions of satisfaction
which the objects referred to must meet.
Two concluding remarks: First, I have called my account of
indexicals "Fregean" in spirit, but it is quite different from Frege's
few actual remarks about indexicals. What little Frege did say
seems both mistaken and inconsistent with his general account of
sense and reference. About " I " he says that since each of us is
aware of himself in a special, private way, " I " has both a public
and a private sense. About "yesterday" and "today" he says
that if we want to express today the same proposition that was
expressed yesterday by an utterance containing "today" we must
use the word "yesterday",16 thus he seems to adopt a de re account
of such indexical propositions. What is one to make of these
remarks? The idea of incommunicable senses of expressions is
profoundly anti-Fregean, since the notion of sense was introduced,
in part, to provide a publicly graspable content to be shared by
speaker and hearer. And the example of "yesterday" and "today"
looks like a stock example of the sort of case where different senses
can determine the same reference. Just as "the Evening Star" and
"the Morning Star" can have the same reference with different
senses because the referent is presented in each case with a different
"mode of presentation", so "today" said yesterday and "yester-
day" said today have different senses and hence are parts of the
expression of different Fregean propositions, even though they
both are used to refer to the same day. I believe Frege failed to see
that it was possible to give a Fregean account of indexicals because
he failed to see their self-referential character, and this failure is
part of a larger failure to see the nature of Intentionality.
Second, discussions like this can tend to degenerate into a kind
of fussy scholasticism which conceals the basic 'metaphysical'
assumptions at issue, and, as far as possible, I believe, we should
allow those assumptions to surface. My basic assumption is simply

16 G. Frege, 'The thought: a logical inquiry', reprinted in P. F. Strawson (ed.) Philosophical

"Logic (Oxford: Oxford University Press, 1967), pp. 17-38.

229
 Are meanings in the head?

this: causal and other sorts of natural relations to the real world are
only relevant to language and other sorts of Intentionality insofar
as they impact on the brain (and the rest of the central nervous
system), and the only impacts that matter are those that produce
Intentionality, including the Network and the Background. Some
form of internalism must be right because there isn't anything else
to do the job. The brain is all we have for the purpose of
representing the world to ourselves and everything we can use
must be inside the brain. Each of our beliefs must be possible for a
being who is a brain in a vat because each of us is precisely a brain
in a vat; the vat is a skull and the ' messages' coming in are coming
in by way of impacts on the nervous system. The necessity of this
internalism is masked from us in many of these discussions by the
adoption of a third-person point of view. By adopting a God's eye
view we think we can see what Ralph's real beliefs are even if he
can't. But what we forget when we try to construct a belief that is
not entirely in Ralph's head is that we have only constructed it in
our head. Or, to put the point another way, even if there were a set
of external semantic concepts they would have to be parasitic on
and entirely reducible to a set of internal concepts.
Paradoxically, then, the point of view from which I defend a
'Fregean' account of reference is one Frege would have found
utterly foreign, a kind of biological naturalism. Intentionality is a
biological phenomenon and it is part of the natural world like any
other biological phenomenon.

230