INDEXICALITY
Adriano Palma
Submitted to the faculty of the Graduate School
in partial fulfillment of the requirements
for the degree
Doctor of Philosophy
in the Department of Philosophy
Indiana University
December 1989
i
ii
INDEXICALITY
ADRIANO PALMA
c
Adriano
Palma 1989
All rights reserved. No part of this publication may be reproduced, stored in a
retrieval system, or transmitted, in any form or by any means, electronic,
mechanical, photocopying, recording, or otherwise, without the prior written
permission of the author.
Acknowledgments
The author should like to thank all members of the thesis committee for the extent of their help.
i
Abstract
The problem of indexicality is presented and then the problems brought about by it with
several semantical approaches are examined. In particular theories which claim to have a quick
eliminative solution for indexicality are discussed and criticized. Frege’s remarks on the logic of
sense and reference are discussed to indicate the limits of his original approach when indexical
terms are seen as not eliminable in any simple and straightforward way, in particular when the
importance of indexicality is understood in its essential role in the formation of attitudes toward
action and belief.
In the last chapter a tentative reformed and amended approach is proposed, trying to
keep insofar as possible intact the insights of the Fregean semantics, but with relevant changes
to accommodate the phenomenon of indexicality.
ii
Doctoral Committee:
Hector-Neri Castañeda, Ph.D., Chairman
Romane Clark, Ph.D.
Jon Michael Dunn, Ph.D.
Linda Wessels, Ph.D.
August 29th, 1989
iii
TABLE OF CONTENTS
CHAPTER 1: Introduction
(p.2)
CHAPTER 2: Reductions
(p.9)
CHAPTER 3: Frege’s notion of sense
(p. 41)
CHAPTER 4: Direct Reference
(p. 54)
CHAPTER 5: I (and some speculations)
(p. 100)
BIBLIOGRAPHY
1
CHAPTER 1
Introduction
I have been intrigued by indexical expressions for a long time. I still am, for I think I have
not found a finally satisfactory view that takes into account all the data (linguistic, psychological,
logical, etc.) and collates them in a simple and elegant pattern.
That being my final goal, here I offer prolegomena to it: what I take to be the relevant
data, and what I think one should not have as a theory.
Some of the critical parts of this paper are inherently unfair. To criticize Frege or Russell
from my present standpoint is using against them many of the data, theories, approaches, others
have built, after them, sitting on their shoulders. Perhaps in philosophy to be criticized is to be
alive, so this is my way of paying respect to them.
What I will eventually present (in chapter 5) as my positive view is somewhat speculative.
It has its foothold in works by others from which I have largely borrowed data, terminology,
and concepts. In that respect I have to mention Hector-Neri Castañeda’s work, which had me
interested in the topic of indexical reference to begin with.
First a delimitation of the world ’indexical’. I take indexical words to be expressions such
as ‘I’, ‘you’, ‘she’, ‘he’, ‘it’, ‘this’, ‘here’, ‘that’, ‘there’, ‘now’, ‘yesterda y’, ‘today’, and so forth
(caveat: the expressions to function indexically have to b e used in their “standard” pronominal,
adverbial, adjectival role, and the list itself is not exhaustive, in part because of disagreement
on the extent of indexicality in language). The common denominator of indexical expressions is
that they have no stable referent, even though always referential1 .
The referent is determined contextually in each occasion of use. An ‘occasion of use’ is
in general an utterance. An utterance is a physical event in which a speaker uses a sentence.
3
Two notes.
1. the context in which an utterance takes place includes a speaker as an essential feature (thus
one token sentence containing indexicals uttered by two speakers will bring about two occasions
of use, even ceteris paribus).
2. an utterance will do for me double duty. Either assuming sincerity or not, I will take it to be
the expression of a thought, which may occur independently of an utterance. Given candid talk,
I will take the text of an utterance to be the text of a thought and when the speech act is not
candid, the intention to deceive will at least determine one text of –deceitful– thought.
Even though in the final analysis I will reject one alleged reason for the distinction, I will
be using two terms to indicate a difference in use: pure and demonstrative indexical (in short,
demonstrative). The distinction, I owe to David Kaplan, is the following. While both pure and
demonstrative indexical expression have their referent fixed only in a context of use, the kind
of factors to be taken into account is different. For pure indexicals (paradigmatically ‘I’, ‘now’,
‘here’) the use of them is enough to establish that there is a reference made to a real entity.
This might be taken to be the semantical principle underlying Descartes’ argument that
just by thinking ‘I’, he would be able to conclude that there was a referent for his tokening the
expression, even without relying on logical principles of inference2 .
On the other hand, according to David Kaplan at any rate, demonstrative indexicals or
demonstratives (paradigmatically ‘that’, ‘this’, ‘he’) achieve their reference by way of a context
which includes a demonstration. This might or might not be “physically executed” by a pointing,
or an ostension. The basic case is an utterance such as ‘That is a great work of art’, said by
4
someone who is pointing to, say, a sculpture in front of an audience. The fixation of the referent
of the token of ‘that’ is possible for the hearers through the speaker’s pointing. This pointing is
very loose a notion (in a sense anaphoric reference can be seen as a way of pointing back, and
it is not very obvious what constitutes a pointing when an indexical reference to abstracta takes
place), but in first approximation it is a notion helpful in indicating a certain difference among
indexicals. No matter what one thinks of demonstratives a difference is perceivable. Though I
disagree with it, I have to notice that one position I found shared by many authors (Hector-Neri
Castañeda, Roderick Chisholm, and David Kaplan, for instance) is the one which singles out as
one feature of difference between the two kinds of indexicals referential failures. As noted pure
indexicals can not have non-referential uses, but demonstratives can according to the position
I am describing. I will spell out more clearly the reasons of my disagreement with such a view
in Chapter 5. The notion of pointing is to be taken loosely here, because I am not sure we can
make sense of a physical pointing to abstract objects directly, without a mediatory role of figures
and the like.
I will argue that we can refer to entities in thought that is independently from any
utterance, although utterances as remarked before are thoughts’ witnesses. The thought, insofar
as it is expressed by a sentence, has some level of syntactic structure, at least as much as we can
describe linguistically. I assume, then, a certain correlation between thought and speech.
As Castañeda never tires to argue, a language has to be used by a speaker first, in order
to be understood by anybody else. Accordingly I will make use of his distinction. There is a
(speaker’s) thinking reference which constitutes “the zeroing in on an entity E of one’s thinking
5
of E”, and communicational speaker’s reference: “The issue here is the techniques for succeeding
and the limits of success in making others think, i.e., refer thinkingly in the first-person to, what
one thinkingly refers to in the first person”.3
I believe certain non-indexical expressions have a reference which is just fixed relative
to the language in which they occur. Whether they have by themselves denotations might be
doubted. I think it clear though, that indexical expressions get a denotation in use only, either
in thinking use or in communicational use. I shall indicate some of the reasons why I think that
in the passage from thinking reference to communicational speaker’s reference mechanisms akin
to Grice’s rule for meaning are at work4 . The aim of the distinction lies in that we do convey
information in communication, but I think there is also a dimension of strict privacy of firstperson thoughts, those thoughts can be certainly alluded to, attributed to a thinker by others in
some vicarious ways, but I do not think they are shared in Frege’s sense.
One term I will make use of is ‘field of awareness’. It stands in need of at least preliminary
explanation.
I take it that I am aware of my surroundings, and, sometimes of myself, and I tend to
attribute the same conditions to others. Now by field of awareness I’ll mean all of which I can
be aware of at any given time. The construal here has to be taken internally, ‘being aware of x’
does not entail under an “internal” reading the existence of x. Different moments in time then
will have different fields of awareness. At one time though I might have present in my field of
awareness items which were presented in prior ones: memorized entities, so to speak, can be in
a field of awareness (bringing about the problem of re-identification; i.e. of how one can make
6
a judgment of sameness between the item in field of awareness at time 0+1 and the item in the
field of awareness at time 0).
I assume that there are no boundaries to what our awareness can encompass: in our
fields there are at different times (and probably for different people) gods and humans, triangles,
concreta and abstracta, and so forth. At the basic level the items of our field of awareness
are perceptual, but awareness does not have to stop there. On what the precise structure and
contents of a field of awareness is, I will remain neutral5 . I shall eventually note how our field of
awareness is shaped by unconscious processes of categorization and reification, which will play a
role in my view of demonstrative reference. I think the idea of field of awareness is a time-sliced
equivalent to Ray Jackendoff’s projected world6 .
One final word of introduction: I do not have a mathematical theory, or a perfectly
formalized semantics. I would like to have it, if I only were able to construct one. In particular in
Chapter 5, I will indicate what I take to be the shape of the terrain a formal semantics ought to
formalize. In particular I think it an open question if we have good formal ways to characterize
the perspectival aspect of language mathematically: I certainly do not have any such. I do not
believe, though, that’s enough to refrain from trying to formalize it.
Enough of introduction: it is better to see the machinery at work than talking about it.
7
NOTES
\1/ Or, at least, always meant to be referential: as I note below some theories countenance
referential failures for some of these expressions.
\2 “When someone says ‘I am thinking, therefore I am or I exist’, he does not deduce existence
from thought by means of a syllogism, he would have to have had the previous knowledge of the
major premiss ‘Everything which thinks is, or exists’; yet in fact he learns it from experiencing in
his own case that it is impossible that he should think without existing”, wrote Rene’ Descartes in
his Second set of Replies to Objections. The objection in question was formulated by Mersenne.
See The Philosophical Writings of Descartes, Cambridge, 1984, vol. I I: p. 100.
\3/ See CASTAÑEDA {7}: p. 277.
\4/ See in particular “Utterer’s Meaning and Intentions”, now in Studies in the Way of Words,
Cambridge, Mass., 1989, pp. 86-109.
\5/ Castañeda has a specific theory of hierarchy of consciousness (see, in particular, his {5} and
{9}). This has to do, I believe, with certain views of his on the need for a “rich biography” which
requires some higher-level consciousness of the self. At the present time I do not have any such
developed connections, so I thought it correct to leave the question open. But see Chapter 5, for
some notes on self-consciousness.
\6/ See JACKENDOFF {1} and {2}.
I wrote ‘time-sliced’ because I wish to remain as noncommittal as possible w.r.t. issues of reidentification of recurring particu lars even though items of memory, I believe, can be in a field
of awareness.
8
CHAPTER 2
Reductions
Das Wort ≫ ich ≪ bedeutet nicht dasselbe wie ≫ L.W. ≪ selbst wenn
ich L.W. bin, noch bedeutet es dasselbe wie der Ausdruck ≫ die Person, die
jetzt spricht ≪. Das bedeutet jedoch nicht, daß≫ L.W. ≪ und ≫ ich ≪ zwei
verschiedene Dinge bedeuten. Es bedeutet nichts weiter, als daß diese Woerter
verschiedene Instrumente in unserer Sprache sind.
Ludwig Wittgenstein∗
Language is conceived in sin and science is its redemption.
Willard V.O. Quine∗∗
A setting of the problem
There is no visible practical problem of indexicality: every mature speaker is able to use
these expressions, these expressions are understood by audiences, the communicational traffic
seems to flow without any greater problems than in other areas of language.
The issues do not seem to be all that straightforward in philosophy of language. Indexicals bring about a problem when certain assumptions, otherwise rather reasonable, are made.
The central assumption is that linguistic entities, such as sentences, are expressive of propositions. Propositions are hold by many to be abstract objects, to be bearers of truth-values, and
to be squarely out of the space-time network.
Taking a couple of extreme examples: the propositions expressed by:
2+2 = 4
and
II + II = V
are (given the medium of numerals) referring to numbers, and respectively true and false of
numbers. I have explicitly used two different kinds of numerals to indicate that in the classical
10
notion of propositions they can be expressible by a variety of means. Obviously, modulo English,
one can express those two propositions by producing the noises (which are recognizable as) ‘twoplus-two-equal-four’ and ‘two-plus-two-equal-five’.
Note: classically propositions are objects of cognitive capacities, they can be asserted,
denied, entertained in thought, believed, disbelieved, some will add also hoped for, queried about
and so forth. Frege tried to capture this facet by distinguishing and assertoric content from the
force of a sentence1 .
Indexicality comes into the picture and a few features of the classical view of propositions do not look too convincing. For one thing there seems to be no prior ground to exclude
that sentences like ‘I am thinking now’ or ‘That chapter I wrote three days ago was not too
coherent’ are expressive of propositions. They do carry a definite cognitive significance for the
thinker/utterer, they are understandable by those who know English, and there seems to be a
content which gets exchanged communicationally. But they do not seem to be in the realm of
abstract tenseless entities (the propositions of the classical view, so to speak). Almost palpably
two distinct utterances or occurrences in thought of
‘I am hungry now’
might have diverging truth values, and so it seems that tenselessly the two occurrences of the
same sentence are expressing two different propositions after all.
On the other hand, as Frege noted in his Der Gedanke, there seems to be an irreducible
cognitive role of the first person pronoun: each user understands by it something private and not
quite shareable in the way in which the meaning and the denotatum of “II” is shareable.
11
These intuitive problems have prompted two main responses: one is to show that indexical expressions, although pragmatically convenient, are not really needed, with the sense of
“need” to be further explained; the other one is to build a formal apparatus that provides ways
to “tag” each indexical occurrence in such a way that it “hooks up” directly with its denotatum,
and then provide a separate theory w.r.t. to the cognitive significance of the sentence and of the
proposition expressed by it. To these two approaches I turn next.
My general view consists of three main tenets:
1. we do use indexical reference
2. indexical reference is irreducible to different forms of reference
3. the irreducibility of indexical reference points to a mode in which experience impinges upon
language, and insofar as language mirrors thoughts, upon thought as well.
Against this view a host of positions claims instead that indexicals are somehow reducible
to other expressions. With diverse aims and motivations theorists have been trying to show that
indexicality, while contingently present in natural language, need not be a permanent fixture in
our conceptual schemes: given enough patience, we could do without it. There is a plain sense
in which certain indexical expressions are replaceable by a non-indexical one.
My uttering a token of
I am reading
is easily understood by a hearer who knows my name as an act of stating to the effect that AP
(or ‘the speaker of the speech act in question’ if he takes me to be such an agent) is reading. Very
similar stories can be told for other indexicals as well. Indeed, generally a speaker’s indicators
12
have to be replaced by other denoting expression in hearers’ reports: by names and description
in the de re mode or by quasi-indicators if the report wants to preserve the original indexical
character of the reported utterance.
The utterance at 6 a.m. of
I am awake
is easily understood by someone who hears it, knows English and assumes that I am producing it,
and knows or believes that it is 6 a.m. and that I am (the s ame as) AP to convey the information
that AP is awake at 6 a.m.
There are still some difficulties in providing this sort of quick translation with tokenings
of pure demonstratives. The difficulty I perceive is that the referents targeted by tokens of
demonstratives like ‘this’ and ‘that’ sometimes so speedily disappear that seldom if ever, anybody
conceived of having them baptized into an everlasting (or longlasting) lexicon. On the surface of
utterances often the tokenings of ‘this’ and ‘that’ have no sortal predicative specifying appendix,
as in ‘this dolphin’, and it is left to context and intentions of the speaker to make the reference
definite.
No doubt, indexicals express ephemeral references and to capture the content of a statement containing them they have to be replaced by referentially stabler expressions. The next
question is: Is this replaceability a symptom of or a “proof” of their reducibility? To this question
I turn next.
One finds in the literature many a different “way” of reduction. One categorization I
propose is to divide the field between absolute and relative reductions.
13
Absolute reductions attempt to eliminate indexicals tout court, to show them to be
dispensable, once one realizes they are in use only as abbreviations of longer, more cumbersome,
expressions which can be redrafted into service, were need to arise, as in the case of the fictional
physicist who has to speak pretending to be a god. Relative reductions, by contrast, try to show
that indexicals can be analyzed away only via a translation into other expressions, indexical
themselves, but taken to be more basic or fundamental in some sense.
The moral of the tale lies in what one’s primitive terms are: an indication of which kind
of philosophical semantics one wishes to construct.
I am inclined to categorize Millikan, Boër and Lycan, at least under one reading, and
Frege (again, in one of his phases, as I explain later) as absolute reductionists; G.E.M. Anscombe,
Chisholm, Russell and Reichenbach as relative reductionists.
The thrust of all “duplication arguments” (introduced qua arguments about indexicality
by Castañeda, and revived later by Perry, and afterwards used by numerous authors) relies on a
general characteristic of thought: its being never closed under any form of closure. A duplication
argument, in its most general form, tries to establish the irreducibility of the meaning of one
expression to another’s by calling attention to a possible disparity in the truth-evaluation of the
utterance/thought where the allegedly dispensable expression occurs. Here the assumption is that
the minimal unit of significance from the standpoint of assigning truth values is a declarative
sentence (this is not in general the case: anaphora and all sorts of contextual factors often make
much larger segments of discourse minimal units of significance). What those arguments show
is the following: in absolutely general terms the thought X is not equival ent to the thought
14
Xt where Xt is a “translation” of X with replacements of referring terms with other coreferring
terms.
That I take to be the lesson of all the controversies on the morning star, Venus, the
evening star, Cicero and Tully, Londres and London, and so forth. Any given attempt at reduction
ought to be tested again this datum of thought: I see this character of thought as a consequence
of the finitude of minds’ activities.
To use my simple little example again:
I am awake now
is perfectly thinkable by me, whether I otherwise know or not what time it is now, or what my
(given) name is.
For I might be unaware of all information regarding my name, the time at which I am
thinking and so on and so forth, be that because I forgot it, because I just do not know it,
because I am radically mistaken, because I am in the wrong time zone, and so forth by way
of multiplication of possible situations in which there are indeed coreferring terms available to
express at some level of public communication my thought and I have not used them.
I take this character of thought to be a fact about human psychology. This openness is
what makes it impossible to have general replacements of indexical expressions. The de re character of many of the alleged replacements lacks the proper finegrained multiplicity of thoughts.
The de re modifier is to be taken in its literal meaning: it refers to that which is “about things”.
We have the tendency to think, in philosophical circles no less than in common sense, that the
component of a thought which is referential which the labeling functions, is part and parcel of
15
what is meant.
My view, on the other hand, can be put in simple words thus: the way of reference is
part of what is meant in thought and speech, possibly and not necessarily together with the
expression’s referent.
So much for introductory remarks. Let’s test my view with some alleged translations, to
see how it fares by comparison.
16
Russell’s “egocentric particulars” and Reichenbach’s “token-reflexive words”
Hans Reichenbach and Bertrand Russell had very similar approaches to the elimination
of indexicals in their purified language systems2 .
Reichenbach rather cavalierly sets up his translation-cum-elimination of indexicals by
using quotation devices for the sentences in which they occur.
In his own words, Luther in 1521 at the Reichstag at Worms utters ‘Here I stand’.
Labeling the utterance in single quotes as ’U’, the ‘I’ tokened by Luther can be given the form
‘the x that spoke U’ and the ‘here’ tokened by Luther can be translated as ‘the place z where U
was spoken’. U can then be expressed formally (allowing ourselves to use the function ‘x̂ speaks
ŷ at ẑ’, with SP for ‘speak’ and ST for ‘stand’ and a definite description operator ’dd’ for the
“iota operator” of Russell):
ST[(ddx)there is a z)SP(x,U,z),(ddz)(there is an x)SP(x, U,z)
Then, Reichenbach remarks, ‘here’ and ‘I’ used in a quotation of Luther’s utterance do not have
the original token-reflexive meaning3 .
Aware of the limitations of his proposal, Reichenbach introduces a single symbolic equivalent of the natural language type-word ‘this’, token-reflexive exactly in the same way ‘this’ is
in English, despite his insistence on the importance of the switch from types to tokens. And
his ‘this’ type has exactly the same function of ‘this’ in English, namely to make possible direct
reference to an item in the field of awareness of a speaker/thinker, the type ‘this’ has then a
constant character and a different meaning (referent) for each one of its tokenings4 . Reichenbach
in this respect is very close to Russell’s view of keeping a single primitive indexical and ‘define’
17
the meaning of all the others in its terms.
Russell, less mesmerized by the need of constructing symbolic calculi, shows more awareness of the issues at stake in the “elimination” of indexical reference.
Very boldly he states:
All egocentric words can be defined in terms of ‘this’. Thus: ‘I’ means
‘The biography to which this belongs’; ‘here’ means ‘The place of this’; ‘now’
means ‘The time of this’; and so on. We may therefore confine our inquiry to
‘this’. It does not seem equally feasible to take some other egocentric word as
fundamental, and define ‘this’ in terms of it.
. . . Before embarking upon more difficult questions, let us observe that
no egocentric particulars occur in the language of physics. Physics views spacetime impartially, as God might be supposed to view it; there is not, as in perception, a region which is specially warm and and intimate and bright, surrounded
in all directions by gradually growing darkness. A physicist will not say ‘I saw
a table’, but like Neurath or Julius Caesar, ‘Otto saw a table’. . . 5
Russell aims at an elimination of indexicality through showing that references made to
items in the speakers’ experience, such as the ones expressed by indexical terms, are dispensable,
not really needed for a sort of view of the universe from the standpoint of an impersonal observer:
“If our theory of ‘this’ is correct, it is a word which is not needed for a complete description of
the world”.6
The reason for such a claim is quite simply that Russell holds a twofold view of egocentric
particulars-words’ meaning. The use of a sentence beginning with a phrase like ‘this is’ is the end
product, so to speak, of a causal chain that begins with an external stimulus impinging on the
perceptual apparatus of the speaker/thinker: “A verbal reaction to a stimulus may be immediate
or delayed. When it is immediate, the afferent current runs into the brain and continues along
18
an efferent nerve until it affects the appropriate muscles and produces a sentence beginning ‘this
is’. When it is delayed, the afferent impulse goes into some kind of reservoir, and only produces
an efferent impulse in response to some other stimulus”.7 The second tenet of the view is that all
other forms of indexical reference are to be shown reducible to the “causal” one which constitutes
the path to the reference of ‘this’.
Now, claims that ‘I’ is replaceable by ‘The biography to which this belongs’ are refuted
by the experiencing in thinking of no such equivalences. Quite visibly, at a purely empirical level,
children of very young age do use the first person pronoun without ever being aware of any as
complex as ‘biography’. If in their thinking the meaning of ‘I’ were to be the one alleged by
Russell, we would be in the rather counterintuitive situation of having to attribute them such a
highly sophisticated conceptual network. Again I have to remind the reader that my focus is on
the level of thinking, the first person pronoun in the utterance can be understood by Russell or by
anybody else in a number of ways, among them as ‘the biography to which the event/utterance
belongs’. On a more philosophical level, we can apply duplication arguments rather smoothly to
the Russell-Reichenbach views. It is conceivable that an amnesiac has no idea whatsoever of his
own biography8 . And yet the amnesiac can say and truly believe:
I am tired now
then of the amnesiac we can say, calling him ‘A’, ‘A believes ‘I am tired now”, if we allow on our
side the capacity to attribute to others indexical thoughts we can state that ‘A believes he-himself
to be tired’.
Since Castañeda’s seminal “Indicators and Quasi-indicators” we have the means to dis19
ambiguate as done above the more idiomatic report ‘A believes he is tired’. The ‘he’ token here
can be read in radically different ways: it can be a demonstrative pronoun, with its reference
depending on the context of the utterance and, I urge, on the speaker’s intention, or it can be
used to pinpoint in the report that A made a reference to himself-as-himself. It ought to be clear
then that when a reported ‘he’ is used as a quasi-indicator (made explicit in a semi-regimented
language by the he-himself phrase) we do not have the means to replace it by any other way of
referring to A, assuming we want to preserve the content of the original first person utterance9 .
For the amnesiac’s beliefs are not correctly depicted by translating them to the effect that
‘The biography to which this belongs is tired’: A, the amnesiac, by definition has no biography
of which he is aware. As I noted in the introductory remarks we have again a disparity between
de re attribution (the way in which we can understand others’ utterances) and semantics of
thinking and uttering from the standpoint of the speaker/utterer. In most cases, probably in
a very large majority of cases of communication, default values will play the roles of referents
for singular expressions. Russell’s proposal works well for all cases of default: in most cases the
speaker is able, if necessary with some prodding, to cognize and recognize the ‘I’ in the utterance
as somehow equivalent to the biography within which the utterance itself appears as an event10 .
Thus I conclude that, even if we grant Russell his understanding of the referential path for
tokens of ‘this’, his reduction of other forms of indexicality to it does not satisfy the requirements
of a semantics for thinking, even though it may be taken as one view of what default assignments
of referents are in many cases.
20
G.E.M. Anscombe’s vanishing ‘I’.
Anscombe in her essay comes to the conclusion that first person referential acts are
dispensable:
‘I’ is neither a name nor another expression whose logical role is to make a
reference, at all −−11
Her view develops an intuition to the effect that a token of ‘I’ is not a name for an
object. It could not be a name for a body because the “loss” of a body, in putative cases of
sensory deprivation, or the loss of memory does not bring about the loss of an Ego, as long as
there are I-thoughts still thinkable under those conditions:
. . . ‘I’ is not a name: these I-thoughts are examples of reflective consciousness of
states, actions, motions, etc., not of an object I mean by ‘I’, but of this body.
These I-thoughts . . . are unmediated conceptions (knowledge or belief, true or
false) of states, motions, etc., of this object here, about which I can find out (if
I do not know it) that it is E.A. About which I did learn that it is a human
being.12
Now, first of all I suspect that behind positions like hers there is overly simplified notion
of naming: it is not clear why a name has to be the name of a physical object, and why not
the name of a state. It might well be the case that within a larger philosophical outlook one
wishes to be materialist and claim that in final analysis what ‘there is’ is just a composition of
molecules, quarks, energy or what have you. It is not at all clear to me at any rate, why names
(here I mean names of natural languages) can not be naming entities other than objects, be those
states, motions, etc., and their indefinite combinations. But even setting aside my suspicions for
the nonce, Anscombe is forced to recognize that her zero-referential value of tokened ‘I’ will not
21
quite do:
There is a real question: with what object is my consciousness of action, posture,
and movement, and are my intentions connected in such a fashion that that
object must be standing up if I have the thought that I am standing up and my
thought is true? And there is an answer to that: it is this object here.13
Anscombe holds first person ascriptions to be publicly verifiable by oneself and by anyone
else to be about a person, which person? “this one”, surprisingly she answers14 .
I find it difficult to see the appeal of her approach. After all, there are strong reasons to
have in a semantics the referent of a singular term such as ‘I’ neutral among all possible targets
of reference. One reason is that we should try semantically to capture as much as possible of
the thinking of actual speakers and so I believe one should be careful before introducing rather
philosophically “loaded” notions as semantic referents, because referring terms may be used
quite correctly by people who have no use (and no acquaintance with) such sophisticated ideas.
Anscombe’s notion of ‘person’ is a philosophically loaded construction: it does not coincide with
the body physically associated to it, nor it coincides with a ghostly soul, spirit, or what have
you.
There are criticisms to be noticed.
First, the embedding in indirect discourse of first person utterances becomes nearly nonsensical.
For suppose one wishes to report:
I am standing
uttered by E.A., naturally one would have to say:
22
E.A. says she is standing
The token of ‘she’ in the report is a quasi-indicator in Castañeda’s terminology or in Anscombe’s
terminology is an indirect reflexive pronoun. What is its referent? E.A., at the level of publicly
identifiable targets of reference (assuming in a sort of meta-language we can treat names as
purely de re devices directly and uniquely getting hold of their referents). But notice that if the
embedding is faithful in its reporting a certain semantical structure has to be preserved. The
correct report in a slightly regimented idiom is
E.A. says [thinks, believes, . . . ] she-herself is standing
The extended regimented form makes it explicit we are reporting a first person utterance made
by E.A.
Now, if in fact the ‘I’ of E.A.’s original utterance or thought is a vanishing pointer
to ‘this object here’, the reporting agent ought to be able to report that as the object whose
thought/utterance is indeed reported. But this is often impossible or not even grammatical. And
for good reasons.
First person utterances do not occur in a vacuum waiting to be filled by an observant
questioning audience (even if the audience is only oneself): they come with a specific reference
to the subject doing the thinking and the uttering. The specific mode of reference is marked
linguistically by the use of the pronoun. That linguistic structure is preserved and embodied in
our grammatical convention to have systematic shifts of expressions in indirect speech. What
is revealed by the grammatical convention is the need to keep at least the capacity to make a
distinction between a reference to oneself, no matter how, and the reference to oneself qua ‘I’.
23
Were one to answer the same ‘whose standing are you talking about’ question for someone who
does not know the surname of Ms. Anscombe, we would have to switch most naturally to a
reply containing the distal demonstrative pronoun, saying ‘the standing of that person there’.
The systematic syntactical links between reporting utterances and reported contents, the general
phenomenon of opacity of indirect discourse, underpin and make explicit what is seldom explicit
in first person utterances: their occurring within the range of a covert indexical operator15 .
The second criticism is more general in scope. Were one to accept Anscombe’s elucidations about the role of prima facie referential acts performed in first person utterances, it would
still not follow that the tokens of ‘I’ are not referential at all, as she wishes to claim.
On her account the thought, or the utterance grammatically attributed by the use of a
token of ‘I’, or by the inflections of the verb in languages other than English, is attributed to ’this
body’ or to ’this person here’, since she has misgivings in identifying bodies with persons. The
token of ‘this’ in the explanatory clause is therefore referential on her own reckoning, because it
is the specific clue to the establishment and fixation of a referent a speaker, if thus queried, can
supply to the audience. The problem is just reproduced at a further distance: ‘this’ is indexical
again and does not have a referent fixed only by its grammatical meaning. It is only a tokened
‘this’ which is referential, and it “gets at” a referent precisely in virtue of being used by a speaker
in a context of which the speaker is part and parcel. The determination of the referent might very
well be shared by others, by ostension, as it were, but it is a reference made by the agent of the
thought/utterance. ‘this’ being indexical in complete isolation is not any more referential than
‘not’ (though for different reasons). Thus the reference made by a tokened ‘this’ is ultimately
24
depending upon being used by an agent. And that agency is ultimately what is captured by the
‘I’ without making any reference to bodies, people, etc. My rough sketch of a rule would state
in first approximation that within the semantics of the hearers:
a tokened ‘this’ refers to an item in the field of awareness of the thinker/speaker
in whose utterance/thought it occurs
If this admittedly coarse rule of reference is anywhere close to correct, Anscombe could
not claim that the agent is vanishing from the semantical features of pronouns: the agent,
precisely what is captured by ‘I’, is always just beneath the grammatical surface of speech and
thought, and whenever we aim at perspicuity the primary task is to bring out from its covert
status the role of the agent in language. And that is nowhere more superficial than in first person
thoughts.
Often, Wittgenstein remarked, the real depth is just right on the surface of language.
25
Chisholm’s self-attributive view
Roderick Chisholm in an intricate treatise devoted to the explanation of first person
sentences makes the claim that all indexical reference is basically understandable in terms of one
simple mechanism of self-attribution of properties16 .
After having assumed the primacy of intentionality, he states:
I shall propose that the primary form of all reference is that reference
to ourselves that we normally express when we use the first person pronoun.
In the case of believing, this reference may be called ‘direct attribution’. Our
reference to all other things is by way of such reference to ourselves. I shall
argue that, although we express ourselves in first person sentences, the reference
to ourselves that we thus express does not involve the acceptance of first person
propositions –for, I shall contend, there is no good reason to assume that there
are such propositions. The primary form of believing is not a matter of accepting
propositions; it is a matter of attributing are to oneself.17
Chisholm states plainly his standpoints to be platonistic with respect to abstract entities
such as propositions and properties: he sees them as timeless entities with which thinkers are
somehow in contact. Thinking is primarily a matter of that ‘contact’, referring intentionally is a
form of thinking. Chisholm is reductionist in a much stronger fashion than other philosophers:
his tests for adequacy for theories countenance as a ground the sheer number of entities involved
in the ontology suggested by the theory. He writes:
The ontology that I shall develop here is Platonistic, since it presupposes
that there are eternal, or abstract, objects. It presupposes, in other words, that
in the strict and philosophical sense of the expression ‘there are’, there are
properties, relations and states of affairs. In this respect, I believe, the present
theory is not unlike that presupposed by most other contemporary theories of
reference. But the present theory is ‘purified’ in that it refuses to countenance
certain nonPlatonic entities which are prominent in almost all other theorie
26
s. These non-Platonic entities may be suggested by the following expressions:
indexical propositions (the propositions said to be expressed by such sentences
as ‘I am sitting’ and ‘That man is standing’); times, considered as particular
things which may be designated by dates and other temporal expressions; and
possible worlds, considered as particular things and many of them such that
they contain the individuals of the actual world.
The present book is intended to show, in part, that such non-Platonic
entities are superfluous.18
The relative reduction he proposes hinges on the primacy of direct self-attribution of
properties. Chisholm introduces two key definitions to implement his program: one defines
within his conceptual scheme the locution ‘I am F’ (where F is a property taken in intension)
and ‘This thing is F’ (with the same proviso). Here are the two definitions:
[DEF.1] The locution ‘I am F’ has as its primary use in English that of expressing
the following property of its utterer: directly attributing the property of being
F to itself.
[DEF.2] ‘This thing is F’ is used in English to express the following property of
its utterer: believing himself to be such that the thing he is calling attention to
is F.19
The direction of the reduction then ought to go from the second definition to the first:
primacy is given to self-attribution of properties, given the mechanism of self-attribution we can
“define out” other indexicals by showing them to be shortwinded expressions for special cases of
self-attribution20 .
Chisholm’s definitions stand and fall together with the assumptions which constitute
their background, sharing this trait with the vast majority of philosophical views. He presupposes
two things about the faculties of believers: “First, a believer can take himself as his intentional
27
object; that is to say, he can direct his thoughts upon himself. And, secondly, in so doing, grasps
or conceives a certain property which he attributes to himself”.21
Let’s put aside, for the nonce at least, any misgiving about platonistically conceived
properties, the problem remains to see whether the two definitions are adequate in capturing
the evidence we have about first person utterances. Intuitively I think Chisholm’s requirements
are too strong: they impose the necessity of having the capacity of reflecting upon one’s self and
thereby re-flexing the intentional arrow. This is a capacity some people have, but even them
not necessarily are exercising it all the time22 . But the case is even harder, I think, for those
speakers/thinkers who have no form of self-consciousness, cases Castañeda calls “Externus”-types
of consciousness, instances of which we might all experience when absorbed enough we are fully
attending outer phenomena. Very young children are known to be unable to formulate coherent
first person utterances couched in the idiom of the first person pronoun (up to a certain age).
However one wishes to interpret this fact, children of extremely young age are perfectly able to
use demonstrative utterances (and, I surmise, demonstrative thinking anchored to perceptual
appearances).
Federico, the child, facing the choice between two different pieces of hardware, likes the
monkey wrench, and says pointing to it:
This is good
if Chisholm is correct, by [DEF.2] Federico is believing himself to be such that the thing he is
calling attention to is good. Unpacking the phrase, Federico is self-attributing the property of
being such that the thing he is calling attention to is good. And this seems to a bit too much:
28
Federico does not believe anything of himself, by assumption he does not have the foggiest idea
of what a self is, his or somebody else’s. But I think it would be completely arbitrary to declare
Federico not “really” referring to the monkey wrench he likes.
The first partial conclusion is therefore that Chisholm’s [DEF.2] fails to capture in its
entirety the phenomenon of demonstrative reference: there is a residual which one has to rule
“out of court”, as it were, by brute force.
The case is much less clear cut with regards to the first person utterances. Here the use
of the sentence seems to be what is the focus of the analysis, and insofar as one limits oneself to
the communicational aspect of reference [DEF.1] is correct.
What remains an open issue is whether the structure of mind is such that any first person
thought necessarily needs the prop the consciousness of oneself as a self produces. There seems
to be cases in which we think ‘forgetful’ of our own existence23 . I believe the issue deserves a
richer treatment than what can be given here and I’ll go back to it when dealing directly with
the structure of (self)consciousness.
29
‘Knowing Who’ and the “lazy” indexical
In her recent “Myth of the Essential Indexical”,24 Ruth Millikan after having reviewed
some of the “common wisdom” on the essential character of indexical reference puts forward four
intriguing and bold ideas:
. . . it is not indexical thoughts that serve to engage an agent with the world.
First . . . it is not true for the general case that the relation an indexical bears
to its referent yields a relation that would be relevant to action. Second, it is
not true for the general case that those relations of self to world that one must
take into account in order to act are relations of the sorts that indexicals bear to
their referents. Third, it is no part of the job of an indexical token to signify the
relation either of itself to its referent or of its interpreter to its referent: inner
signs that do signify relations between agent and world as needed for action are
not indexical. Finally, if an agent employs a mental term to represent herself,
this cannot in principle be a mental indexical: there can be no thought that has
the character of ‘I’.25
Millikan’s main contention is that even if one could establish the relevance of indexicality to behaviour, the connection between reference and action has to be independent from the
character of the indexical expressions:
the interpreter of an indexical must already know, must know independently
what item it is that bears the indexical’s adapting relation to the indexical token,
hence must already know independently that it bears the relation to the token
(hence to the interpreter). A token of ‘I’ does not tell me who the originator of
the token is.26
Now, the first remark is that even accepting at first brush that interpersonal behaviour
depends on having independent ways of knowing “who” (or “what”) the indexically referred
object is, it is not very obvious how her approach would deal with the problem of reference from
30
the perspective of speaker’s meaning. While it may be the case that there are situations in which
our being prompted to action relies upon non-indexical mechanisms, it is not at all the case that
when we refer to someone as ‘you’ we need independent ways of knowing who ‘you’ is.
Quite the contrary: one of the key features of indexicals is exactly to permit reference
to items about which nothing whatsoever is known at all.
Millikan’s examples draw from John Perry’s problem with a postcard, containing a sentence such as ‘I am having a great time now’ and a blurry unreadable signature. Her response
is to claim that to be moved to action one has to know in advance who is the author of the
postcard in order to be able to interpret the token of ‘I’ (or of ‘now’ for that matter). I think
this is literally false: I may be moved to action, typically in the case of commands, using only
indexically expressed information. If someone tells me ‘lift that’ and I have no other notion of
that (thing) other than being referred to by my commanding interlocutor, I am moved to action
quite naturally by my will to obey, my understanding of what the imperative verb is saying and
by my perceptual awareness of which thing that is. To be sure, mistakes and misunderstandings
abound (think of a natural command ’lift that’ issued to a blind bodybuilder for a quick example
of mismatch of perceptual fields), but they are not solved by resorting to some non-indexical
referential phrase (it might not be available for one thing), but by some appropriate indexical
reference (sight in the fixation of referent for a token of ‘that’ might have to be replace by ‘this’
and a proper tactile clues). Then I conclude that Millikan’s view has some difficulty handling
speaker’s meaning for indexicals: the reference made by the speaker to (the person “named” as)
‘you’ does not require an independent way of referring to you, much less entails its existence.
31
Millikan has to take into account what looks like a most fundamental indexical expression
whose role is most visibly one in which speaker’s meaning is psychologically predominant: ‘I’.
I think Millikan expresses her position in a rather ambiguous and oblique way. She claims she
does react in a special and different way to the knowledge that she herself, RM, is positioned so
in the world, a condition quite unlike how she would react to anyone (else) positioned thus in the
world, and so forth (this apparently holds for any first person case). She holds we have “inner
names” (within the lexicon of our language of thought, as it were):
. . . But what does that have to do with indexicality? My inner name “RM”
obviously is not like other names in my mental vocabulary. It is a name which
hooks up with my knowhows, with my abilities and dispositions to act, in a
rather special way. Conceivably I might even have other mental names for RM
as well that do not hook up with these knowhows because I do not recognize
them, do not identify them as having the same content as “RM”. . . My inner
“RM” is indeed special. It names a person whom I know, under that name,
how to manipulate directly; I know how to effect her behaviour. But In order to
manipulate this person, why would I need to think indexical thoughts? What
has knowhow to do with indexicals?27
One crucial answer to her last question can be provided along the following lines. Since
very little is known about the overall features of a language of thought, let’s assume that each
person has an inner name, whatever its shape will turn out to be. The connection between the
inner name and indexicality is indeed very visible in the situation she has in mind. To be able
to act intentionally, to use knowhows and ski lls, one is bound to execute a form of indexical
reference. For, suppose I am issued an order, say,
Adriano, please, finish the dissertation!
32
In order to be able to carry that out or not to carry that out I have to make a judgment of the
form ‘Adriano is (the same as) ++++’, where ’++++’ is a standin for my inner name. But if
’++++’ functions similarly to the way ‘Adriano’ does, I would be involved in an odd regress. At
each stage a judgment would be made, but I would never be prompted to act unless I understand
that ’++++ is myself’. Hence it seems to me that either one’s inner name functions indexically,
or it could not be of use in the case of action, when the activation of practical skills and knowhows
is at issue.
As I understand it, Millikan’s position is either holding that each one of us has a deviant inner
name for himself which constitutes the referent of each tokening of ‘I’ or else her claim is highly
implausible no matter how she uses the notion of language of thought. First of all: who is
doing the identifying? I might have several inner names, she claims, as much as I could have
several “outer” names (Adriano, Palma, many others in my personal experience anyway). When
I am moved to action, can I refer to myself by just using ‘I’ and the thought there is just ’I
am (the same as) Adriano and this fellow is telling Adriano to X, means I (should) X’ ? The ‘I’
has a transcendental character, it underlies, if I am correct, all my representations, internal and
external as well. Secondly: if I have an inner name, in the normal sense of ‘name’, its referent is
constantly changing (even if one wants to claim that the only reference is to knowhows). Then
that “name” has at least one key feature in common with indexicals: its referent is determined
not just by my (mental) lexicon, but also by its context of use.
My answer to Millikan’s rhetorical question is the following.
i. Indexicality has a lot to do with action because in any given occasion I have to identify the
33
purported reference of others with me, if need be via my inner name;
ii. Whenever I express an indexical reference I have no need to have an independent way of
referring to the item I am “hooking up” with my indexical expression, and if I do more often
than not that second way of reference will be indexical as well (e.g. referring to a male in my
field of awareness as ‘he’ does not preclude at all that the only “other” way of referring to him
is to “call him” that person over there).
On the more technical side, I see in Millikan’s position the danger of eschewing Kaplan’s
distinction between pure indexicals and demonstratives. Pure indexicals, paradigmatically ‘I’, do
not have reference failures: in every context they are denoting the thinker/speaker. In thought ‘I’
denotes a thinker who’s able to think in first person (excluding fancier cases, in first approximation
those are the cases in which self-consciousness is present). Demonstratives’ referential capabilities
depend both on a speaker/thinker and on a field of awareness around her, whether or not they
will be able to express a the correct, intended reference for the audience is a separate issue in
my view. But even pure indexicals such as the first person pronoun “shift” their referent: the
denotatum of a token of ‘I’ is not an intertemporally stable “object” or capacity. The denotata
of our tokened ‘I’ change over time and what “hold them together”, so to speak, are complex
mechanisms having to do with memory and recognition, with the ability to predicate of them
certain semi-stable relevant attributes, a core of properties, to wax metaphysical, which we hold
dear enough to see them as correctly attributed to our ephemeral ‘I’ repeatedly. Millikan seems
to forget, aside from theories, that her view is apparently unable to handle in any clear way
failures of reference for demonstratives: a token of ‘that’ intentionally referring to a non-existent
34
object slips by her analysis. Nothing can be known about it, or so it seems, there are no internal
names that will guide action in cases like the “drowning man” of Castañeda
28
since there is by
assumption nothi ng to be named. The logic of “confusion” has to be rather more complex to
handle our uses of demonstratives.
Related points on what it is to ‘know who’ are made by Steven Boër and William Lycan29 .
They couched their remarks on the irreducibility of indexicals, in particular on first person
cases within a theory of propositional attitudes which relies heavily on paratactic construals of
opacity.30 Their reduction is absolute with respect to certain uses of ‘I’. Their key analysis can
be put in this way:
[u] John believes that I am in danger
is an utterance named [u], and that is true if and only if John believes-true some N is in danger
for ‘N’ denoting one way or another the utterer of [u]. In the case of [u] the way of denoting N
would be her own self-denoting by way of the token of ‘I’ in [u]. This analysis follows a traditional
way, by now, of reading ascriptions of belief as de re ascription (in a semi-regimented language
[u] can be read off as ‘John believes of-N-(whichever the way N is named in the utterance) that
N is in danger’. But the authors are willing to recognize that their paratactic reading gets into
more difficulties dealing with:
[u.I] I believe I am in danger
for [u.I] Boër and Lycan provide two different readings31 . Their key claim here is that first
person ascriptions are ambiguous between a de re understanding (on the assumption that the
35
pure indexical token of ‘I’ is used in an utterance and it does not allow at all referential failures,
this seems to be correct) and a “quasi-indicator” style of reporting which would be symmetric
to the phenomena noted by Castañeda with uses of ‘he-himself’. They point out that this is not
very apparent because it is couched more deeply than usual in the pragmatics of discourse32 .
The point of this analysis seems to me to be a failed attempt to an absolute reduction:
‘I’ is not, at least not always, readable in the de re mode. There are some different grounds to
be skeptical about the prospects of parataxis, but even leaving those aside for the nonce, what
the authors claim has to be a “surd in nature”,33 I think should rather be seen as part of nature,
a nature which includes first person perspectives.34
36
NOTES
\*/ Ludwig Wittgenstein, Das Blaue Buch, Werkausgabe, Ban d 5, S. 107
\**/ W.V.O. Quine, The Roots of Reference, La Salle, Ill. , 1973, p. 68
\1/ Other notions of contents of thought make a sharp division between objects of practical
thinking and objects of contemplative thinking. See, e.g., CASTAÑEDA {5}. The dividends in
terms of explanations seem very large compared to the breadth of the unity of ontology in theory
like the one in Thinking and Doing . I am going to be only marginally involved with the specifics
of practical thinking.
\2/ See section 50 of REICHENBACH and ch. 7 of An Inquiry into Meaning and Truth (RUSSELL {1}: pp. 102-109)
\3/ for a general discussion of the connection between quotation devices and indexicality see
“Quotation” by Jonathan Bennett, in Nous, XXII, 3, 1988, pp. 399-418.
\4/ See REICHENBACH, p. 287
\5/ ‘Egocentric particulars’ is Russell’s own term for indexical expressions. As it has been noted
by many, the connection between Neurath, Caesar and the physicist is not fully transparent:
in point of fact physics discourse is full of indexical expressions, as much as any other piece of
human discourse, not to mention that specific theories in the physical sciences are built around
the very notion of an active observer, much unlike the remote god-like entity that Russell thinks
can look at the universe from a non-perspectival point of view.
\6/ RUSSELL {1}: p. 107
\6/ RUSSELL {1}: pp. 105-106
It is interesting to note that Russell kept struggling with indexicality, tying it to his more general
epistemological views and the distinction between knowledge by acquaintance and knowledge by
description.
In Ch. IV of Human Knowledge (see RUSSELL {2}), he uses an example about Mrs. B about
whom I say “Mrs. B is rich”– According to Russell, I intend to say something about Mrs. B
herself, but I say something about ‘Mrs B’. In the example Mrs. B has a daughter, Mrs. A: “If
I say ‘Mrs. B is rich’, I intend something about Mrs. B herself but what I actually assert is
that Mrs. A has a rich mother” (see RUSSELL {2}: 88). Russell concludes the passage with the
idea that only self-knowledge is direct knowledge by acquaintance: “. . . in fact everybody except
myself is to me in the position of Mrs. B; so are the sun and the moon, my house and garden,
my dog and my cat, Stalin and the King” (ibidem)
\8/ This is the case emphasized by Castañeda in his example of the war hero, who forgetful of
his being the war hero, learns a lot in his hospital bed about the biography of the person who is,
indeed, himself without ever coming to believe that he-himself is the war hero. Thus his stock
of beliefs is very large about someone who is himself, but his beliefs are not in first person mode
37
at all (see CASTAÑEDA {13}).
\9/ See CASTAÑEDA {2}. Patently I am accepting his arguments sharing the assumption that
truth values should be stable when coreferring terms are intersubstitutable and some level of
compositionality between semantical units, both insights we owe to Frege in the first place. The
paper cited presents an array of cases in which truth values are not stable when quasi-indicators
are the correct way of reporting indexical utterances. The discussions about the so-called Frege’s
puzzle are endemic in the whole of Anglo-Saxon philosophy: there is no agreement though on
the diagnosis Frege himself proposed. A very different view of the matter (bearing on proper
names as well as on reference in general) is discussed and presented by Nathan Salmon in his
Frege’s Puzzle.
\10/ I say somehow because Russell’s proposal strikes me as unnecessarily extreme: to impute
to thinkers that kind of translation seem to require from them already some sort of higher level
mental life. To have a biography is not given to everyone at every moment they utter a first
person sentence. As I see it, to claim that first person self-attributions are attributions to a biography (to a narrative construction of events, or to some sort of abstract time-slice metaphysical
compound) is already a move in philosophy and I do not think there is any reason to hold that
thinkers in general ought to have in mind some one specific theory of personal identity to be able
to use correctly ‘I’ (viz. children or brain-damaged people, or quite simply people who do not
have any philosophical position, naive or sophisticated).
\11/ ANSCOMBE: p.60
\12/ ibidem, p. 62
\13/ ibidem
\14/ ibidem, p. 63
\15/ The idea here is to accept Kant’s idea that the transcendental ‘I think’ precedes every
thought. Pushed a step further ‘I think’ is an indexical operator whose scope is any utterance,
no matter how primitive or otherwise uninformed the utterer is about anything else. If the ‘I
think’ operator is tacitly prefixing any utterance then any theory which wants to reduce ‘I’ to
something else, is bound to fail, if for no other reason than that any piece of knowledge or belief
will carry indexicality on its sleeves. Both Castañeda (see, in particular the collection {10}) and,
after him, Clark (see CLARK {2}) have made the point w.r.t. covert indexicals.
\16/ See his The First Person, (CHISHOLM).
\17/ CHISHOLM: p. 1
\18/ CHISHOLM: 4-5. It has to be noted that Chisholm’s definition of ‘property’ has no settheoretical (or extensional) underpinning. He defines ‘property’ thus:
x is a property = Df x is possibly such that there is something that exemplifies it
38
where exemplification and de re possibility are taken as undefined primitive terms of the theory
itself. I take it that for Chisholm any Platonic entity, if it exists (i.e. bar inconsistencies, as
in paradoxical Russellian predicates), then it exists necessarily. Then it is not contingent on
anything else (time, the existence of speakers etc.). In other ontologies such as the explicit one in
Castañeda or the implicit one in Kaplan there exist some abstract entities which are contingent
on time and the existence of thinkers/speakers, e.g. the proposition expressed by an utterance
of the sentence ‘I am alive’ is contingent upon my existence, and given that I utter it or think
it, then it is necessarily true. But in no sense there is a metaphysical necessity of my existence.
\19/ CHISHOLM, pp. 42-43 and 46
\20/ It has to be noted Chisholm in some respect is an absolute reductionist (my terminology):
dealing with temporal indexicals such as ‘now’, ‘then’ and their ilk the direction of his reduction
goes toward the replacement of them by non-indexical pieces of his ontology, such as states of
affairs and the notion of ‘obtaining’. –see in particular CHISHOLM: pp. 49-52 and the appendix
on “The ontology of states of affairs”.
\21/ CHISHOLM: p. 28
\22/ See CASTAÑEDA {9}, where the point is brought to bear on several examples of unreflective consciousness, expressible in sentences of a mental language of sorts. I believe that article
does not make clear enough that Chisholm’s point is very limited, but it is so in an explicit way
so that the reduction is much less extensive than one would expect from the initial claims.
\23/ In Castañeda’s article cited in footnote #22, several examples of this kind are discussed
within a conceptual scheme in which it is taken as a datum that consciousness is never a single
phenomenon but a hierarchical build-up whose bottom is purely sensitive/perceptual.
\24/ unpublished ms.
\25/ ibidem (all the quotes explicitly attributed to Millikan are from the cited manuscript, given
as a lecture in 1989).
\26/ ibidem, p. 8 (the adapting relation is in Millikan’s terminology what is character in Kaplan’s).
\27/ ibidem, pp. 11-12
\28/ See his {8} (in particular pp. 122-124) in which a speaker is confronted with different situations (a drowning man, what appears to be a drowning man, a purely hallucinatory experience),
all of which have in common the indexical reference made via ‘that’.
\29/ See their Knowing Who, Cambridge, Mass. 1986.
\30/ See in particular “On Saying That” in DAVIDSON. In general the program seems to me
very difficult to carry out: I tried to explain some of the reasons in “Parataxis”, Lingua e Stile,
1, 1989. On that subject see also CASTAÑEDA {9}.
\31/ Knowing Who: pp. 147-148. In passing it ought to be noted that one of their claims is
39
rather exaggerated: they claim a sentence like
John believes that I myself am in danger
is ungrammatical, awkward as it is, it is not ungrammatical (often the “ –self” expressions are
used for additional emphasis without carrying over any particular sense of reflexivity.
\32/ ibidem. They point out that the split between the two readings can be made more visible
at the surface of sentences when the reporting sentence and the reported content differ in tense.
\33/ This is Boër and Lycan’s term for what they perceive is the danger of allowing irreducible
indexicals in a semantics. They have made the claim first in “Who, me?”, Philosophical Review,
89 (1980), pp. 427-466. They have themselves changed their position in Knowing Who. It is
not clear whether they see the ambiguity they now recognize in their own theory as admission of
“surds” in nature between reference and irreducibly first person referential acts. They pointed
out that their shift in position was brought about by the response Castañeda gave to their first
article. See specifically the appendix of CASTAÑEDA {9} and the fourth chapter of {10}.
\34/ Epistemically, for reasons that have come to be known under the label “problem of other
minds” one might take the view that only one’s own first person perspective is part of nature. I
do not think this is the case, at least insofar as the use of indexicals and quasi-indicators points
to the attribution we make in thinking of first person perspectives to others.
40
CHAPTER 3
Frege’s notion of sense
If every thought requires an owner and belongs to the contents of his
consciousness, then the thought has this owner alone; and there is no science
common to many on which many could work, but perhaps I have my science, a
totality of thoughts whose owner I am, and another person has his.
Gottlob Frege∗
Gottlob Frege, in Ueber Sinn und Bedeutung1 introduced a distinction between the referent of a linguistic expression (an object, if any) and a mode of presentation (“ein Art des
Gegebenseins”, literally a manner of being given) of such an object. His program was basically
to build such a distinction into the semantics for that expression.
The distinction was needed to explain the import in terms of information of true, non
analytic or synthetic statements (of the form ‘John Le Carré is Mister David Cornwell’).
If two identity statements, such as ‘a=a’ and ‘a=b’ are true, he argued, they should have
the same information conveyed to those who understand both. That they do not is puzzling:
as a matter of fact the first one provides nothing that the logically minded ones do not know
already (for, no matter what ‘a’ stands for ‘a=a’ is provably true from a purely general principle
of identity of absolutely every thing with itself). The second statement is seemingly informative
in that it expresses some fact that has to be established by means that are not necessarily only
logical (in the Cornwell case the means is of course the empirical knowledge of the circumstance
that the author, John Le Carré, uses a nom de plume when he writes). Frege wrote thus:
Now if we were to regard equality as a relation between that which the names
‘a’ and ‘b’ designate, it would seem that a=b could not differ from a=a (i.e.
provided a=b is true). A relation would thereby be expressed of a thing with
itself, and indeed one in which each thing stands to itself but to no other thing.
What we apparently want to state by a=b is that the signs or the names ‘a’
42
and ‘b’ designate the same thing, so that those signs themselves would be under
discussion; a relation between them would be asserted. But this relation would
hold between names only in so far as they named or designated something. It
would be mediated by the connection of each of the two signs with the same
designated thing. But this is arbitrary. Nobody can be forbidden to use any
arbitrarily producible event or object as a sign for something. In that case the
sentence a=b would no longer refer to the subject matter, but only to its mode of
designation; we would express no proper knowledge by its means. But in many
cases this is just what we want to do. If the sign ‘a’ is distinguished from the
sign ‘b’ only as an object (here, by means of its shape), not as a sign (i.e. not
by manner in which it designates something, the cognitive value of a=a becomes
essentially equal to that of a=b, provided a=b is true. A difference can arise
only if the difference between the signs corresponds to a difference in the mode
of presentation of the thing designated.2
This was the starting point of the theory of senses. Frege did not explain fully what a
sense is, rather he went about the business of determining which kind of thing the sense of any
expression is. Later he included complete declarative sentences, as referential expressions, and
dealt with their senses.
In an intuitive fashion a Fregean Sinn is what we come to be aware of when we can make
sense of an expression, what we understand by way of it3 . It is helpful to disentangle three (at
least) different strands in the notion of ‘sense’.4
First, taken in one way, a sense for an expression is a psychological notion. Frege would
say it is such only when it is “grasped” by speakers, a sense per se, he maintained, is a non-mental,
non-physical entity, but an abstract one, belonging to a so called “third realm of entities”, albeit
a realm accessible to speakers because of a common heritage5 . This “common heritage” may be
viewed, at least initially, as a collection of properties or conceptualized aspects which a speaker
43
can use, in thinking, to identify the expression’s referent.
Second, a sense is a purely semantical notion: thus a Sinn plays the role of a linguistic
mechanism by means of which the expression’s referent is determined (this is what comes closer
to the linguistic meaning of the expression).
Third, a sense has also in some other respect an information value, an epistemic status
of sort s.
Clearly for Frege, declarative sentences have thoughts as their senses, and only one of
the two objects ‘True’ and ‘False’ or else nothing as reference – a sentence, for instance, lacks a
referent, it has no truth value, when it is used within a fiction. Predicates have concepts as their
senses. It is not immediately clear in which way one is to take the ‘sense’ of terms expressing
singular reference: most likely a conflation of the three senses of ‘sense’ indicated above, although
it is very clear that, for Frege, it is the (or a?) sense of singular terms occurring in sentences,
that is a constituent of a thought, never their reference6 .
It is this part of Frege’s semantics that has come under attack in recent years. Several
philosophers, and not always for the same reasons, have come to take the view that, at least
in certain cases, thoughts have the rather unFregean characteristic of having an individual as a
component part. In the now prevalent terminology these authors are direct reference theorists,
and the thoughts with individuals as constituents are singular propositions. From now on, I refer
to this approach as K-D-R, in honor of David Kaplan, who wrote thus:
I believe that the issue of Haecceitism reappears. . . as the question whether an
individual itself – as opposed to an individual-under-a-concept – can be an immediate constituent of a proposition. Let us adopt the terminology singular
proposition for those (purported) propositions which contain individuals as im44
mediate constituents. . . ‘I am mortal’ and ‘This is blue’ are thought by some to
express singular propositions.7
Nothing could be more foreign to the spirit of Frege’s philosophy of language than this
notion of singular proposition. The main problem seems to me to be that such a proposition is
not fit as an object of thought. But this needs to be substantiated. One of the reasons singular
propositions have been resurrected is to deal with indexicality. Frege himself recognized that the
very presence of indexical, token-reflexive, expressions in natural language posed problems for
his theory of senses.
In Der Gedanke8 , Frege proposed to treat the sense of a sentence as a thought. Primarily
for him the G edanke expressed by an assertoric sentence is that to which the question of truth
value applies:
Without offering this as a definition, I mean by ‘a thought’ something for which
the question of truth can arise at all.9
In the simplest case of direct discourse the sense of a sentence is not affected by the
replacement of terms by other coreferential terms:
It makes no difference to the thought whether I use the word ‘horse’ or ‘steed’
or ‘nag’ or ‘prad’. The assertoric force does not cover the ways in which these
words differ.10
That is also because the referential feature of the sentence is not affected by such changes:
although they have an effect on the style of the expression they do not bring about any difference
in truth value.
45
The theory of senses required the same group of phonemes constituting a sentence to
refer to a truth value, when occurring in direct discourse, and to refer to the thought expressed
when the same sentence is used in an embedded clause in indirect discourse. Quite palpably from
the fact that it is true that ‘Adriano says that Paris is a nice place in which to live’ it does not
follow that ‘Adriano says that the capital of France is a nice place in which to live’ is true as well.
Paris is the French capital, granted, but Adriano might not know it, or he just does not think
so when asked about it. Let us term this phenomenon the divergence between factual harmony
and cognitive dissonance.
For Frege this dissonance in truth value is again puzzling, and principles similar to
the ones involved in his original Sinn/Bedeutung distinction are evoked to have a difference
recognized. Whereas in direct discourse the exchange of coreferential terms, be they names or
descriptions or what have you, does not affect the thought encoded by the sentence, in indirect
discourse that exchange cannot in general take place without altering the whole about which the
question of truth can arise. Frege’s idea was to allow for a multiply-tiered approach to senses.
Expressions in direct discourse have their normal sense and reference, expressions embedded
within oratio obliqua clauses take on an indirect reference11 .
The first tier of indirect reference comes to be equivalent to the normal senses of expressions (where ‘normal’ is understood as ‘occurring in oratio recta’). In a somewhat less cumbersome
idiom the idea is: what we come to understand, when we understand the utterance of a sentence,
is its sense (the Gedanke expressed). In other words we come to cognize what the sentence says:
a propositional content. Michael Dummett puts the matter in terms of the preservation of the
46
information conveyed:
If I say ‘Jones said that Scott wrote Waverley’, I do not purport to be giving his
actual words; he may have said, ‘Sir Walter Scott authored Waverley’ or ‘Scott
hat Waverley geschrieben’, and my statement will still be true.12
This much seems to be phenomenologically apparent just from the usage of oratio obliqua
constructions.
The import of linguistic phenomena for Frege’s theory requires some more contortions
– basically because the intuitive, pretheoretical notion of content is for him already divided in
sense and reference –:
The replacement of a sentence in indirect speech by another with the same truth
value will evidently not in general preserve the truth value of the whole sentence;
so a sentence occurring in such a clause cannot have its ordinary reference. If we
ask what replacements are possible without change of truth value, we discover
what its reference in such context is. We can alter the sentence in the oratio
obliqua clause, without changing the truth value of the whole, just so long as we
do not change (what is ordinarily) the sense of the constituent sentence, so long
as it continues to express the same thought.13
The intuitive idea behind the appeal to indirect reference is a difference, at bottom,
between use and mention: in oratio recta an expression is used, on the other hand what is taken
to be its content is mentioned in oratio obliqua14 .
Frege also held some form of the principle of compositionality15 . The atomic building
blocks of a language as an expressive symbolism are utterances of sentences, the semantic features
of a sentence depend in diverse ways upon its constituents’ semantics. A sentence is the minimal
unit of assertion: the tokening of a thought. In cases of oratio obliqua referential expressions carry
47
as their reference not the primary, direct referents (objects), but the indirect ones: their Sinn. If
each non-syncategorematic term is supposed to have an indirect reference in oratio obliqua, it is
not difficult to understand why indexicals are doubly puzzling for Frege, for it is not obvious at
all what is the “mode of presentation” of tokens of pure indexicals. Furthermore, however one
wishes to spell out the primary sense of a pure indexical, such as ‘I’, it is not clear at all that
the supposed switch to indirect reference in condition of obliqueness takes place. Suffice it to
say that, as it has been often noted, pure indexicals no matter how deeply embedded in indirect
discourse express a speaker reference. They take in this way “primary scope”.
In cases of direct speech we seem to be bound to have a unique correct reference of a
pure indexical, such as ‘I’. This is often observed by noting that there is no referential failure
for tokens of pure indexicals when used qua indexicals, they have a “Cartesian” character, as
it were. In a speech act expressed by a sentence like ‘I am a pumpkin’, no matter which other
characteristics the sentence might have, in terms of the thought tokened or of the truth value it
has – or denotes according to Fregean orthodoxy – the first person pronoun has as its referent,
as its Bedeutung, the speaker. It is not obvious, at first blush at least, what its Fregean Sinn
is, but more about that later. Now, according to Frege’s theory of thoughts, the sense of the
entire sentence is a thought, and the latter should become the referent of the homophonic group
of expression when embedded in indirect speech. In the exchange all the referential terms would
acquire their new Bedeutung, which were the S inn attached to them in direct speech.
Hence in
Raffaella says that I am a pumpkin
48
the occurrence of ‘I’ should have as denotation not me (that is the utterer of the ‘I am a pumpkin’
sentence) but the (or a?) sense possessed by the token of ‘I’ in direct speech. This is false, unless
pure indexicals escape the alleged necessity of switching to indirect reference when used in indirect
discourse.
Frege tried first to “reduce” some indexical sentences to show that the thought expressed
by the sentence is in fact such only in a context of use: for that reason he wants to claim that
in most cases the tense of the verbal phrase conveys a time-indication, and one “must know
when the sentence was uttered in order to grasp the thought correctly. Therefore the time of the
utterance is part of the expression of the thought”.16
He realized then that even this building into the expressed thought features of the context
would not quite do: the occurrence of ‘I’ in a sentence brings about further problems. His
discussion of the first person pronoun is confusedly mixed with many remarks on proper names
(some of them, by themselves, rather insightful nevertheless). The key part of Frege’s view, as I
understand it, is an admission that the general thesis that an expression has a sense is not quite
correct.
Doktor Lauben is wounded and says ‘I was wounded’, thus tokening a thought. The
referent of this occurrence of ‘I’ is doubtlessly Lauben. What is the manner of presentation of
Lauben to himself? Frege says:
. . . everyone is presented to himself in a particular and primitive way, in which
he is presented to no-one else.17
But, if the latter claim is correct, then Lauben’s first person thought is not capable of
49
being grasped by any one other than Lauben himself, for no one literally can grasp the privately
given manner of self-(re)presentation of the speaker. And the latter is the self-destructive reduction of the claim, since it is fairly clear that the vast majority of speakers (very young children
excluded, it seems) can understand and use in innumerable different cases sentences with an
indexically executed first person reference. Hence for sentences containing indexicals to be object of communication, tokens of indexicals have to possess somehow more than just the purely
private sense which is the way of getting at the referents of the indexically referential expression,
as used by the utterer to refer.
50
NOTES
\*/ FREGE {1}: 362
\1/ published in 1892, now in FREGE {1}: 157-177. Sinn is usually translated in English as
‘Sense’ and I’ll keep the tradition. ‘Bedeutung’ though has often been translated as ‘meaning’
and the latter choice is, to me, not quite felicitous. I will translate ‘Bedeutung’ with ‘reference’
when it is to be understood as a feature of an expression (namely its referential capability) and
with ‘referent’ (with an apology to Latin grammar, but ‘referee’ in English has a wholly different
connotation) when the object referred to by an expression is at issue.
\2/ FREGE {1}: 157-158
\3/ “. . . ‘sense’ is first introduced as correlative of ‘understan d’: the sense of an expression is
what we know when we understand it. . . [and] the notion remains schematic until we have a
theory of sense – an account of what for each class of expressions has to be known in order to
know its sense”, see DUMMETT {1}: 293
\4/ See SALMON: 12
\5/ FREGE {1}: 160
\6/ the point was emphasized much more explicitly in the Frege-Russell correspondence, see
FREGE {2}: 130-170, e.g. in a letter dated 13.11.1904 Frege wrote: “The sense of the word
‘moon’ is a component part of the thought that the moon is smaller than the earth. The moon
itself (i.e. the referent of the word ‘moon’) is not part of the sense of the word ‘moon’; for then it
would also be a component part of that thought. We can nevertheless say: ‘The moon is identical
with the heavenly body closest to the earth’. What is identical, however, is not a component
part but the referent of the expressions ‘the moon’ and ‘the heavenly body closest to the earth”’
[modified translation]. That Frege was unconvinced by Russell’s points is apparent from his
letters to Philip E. B. Jourdain, in which he reiterates his thesis about thoughts containing not
mountains (this time, for a change, it is a volcano) but the sense of the phrase (be it a noun
or a description) denoting the mountains: “Now that part of the thought which corresponds to
the name ‘Etna’ cannot be Mount Etna itself; it cannot be the referent of this name. For each
individual piece of frozen, solidified lava which is part of Mount Etna would then also be part of
the thought that Etna is higher than Vesuvius. But it seems to me absurd that pieces of lava,
even pieces of which I had no knowledge, should be parts of my thought.” See FREGE {2}:
80 (the letter to Jourdain was written about ten years after the quoted exchange with Bertrand
Russell).
\7/ See KAPLAN {1}. Singular propositions are often called Russellian propositions because
Bertrand Russell arguing against Frege proposed a similar notion as the content of certain assertions: “I believe that in spite of all its snowfields Mont Blanc itself is a component part of what
is actually asserted in the proposition ‘Mont Blanc is more than 4000 metres high’. We do not
51
assert the thought, for this is a private psychological matter: we assert the object of thought, and
this is, to my mind a certain complex (an objective proposition, one might say) in which Mont
Blanc itself is a component part. If we do not admit this, then we get the conclusion that we
know nothing at all about Mont Blanc.” See Bertrand Russell’s letter of 12.12.1904, in FREGE
{2}: 169.
\8/ The first of his Logical Investigations (1918-1926), see the translation in FREGE {1}: 351406
\9/ FREGE {1}: 353
\10/ ibidem
\11/ For an exegesis and a discussion of Frege’s views and their superficial incoherence see chapter 9 of DUMMETT {1} and chapter 6 of DUMMETT {2}.
\12/ DUMMETT {1}: 265. A fascinating, albeit in my opinion ultimately unsuccessful, different
treatment of opaque contexts, is given in “On Saying That” in DAVIDSON.
\13/ DUMMETT {1}: 266
\14/ “The object about which I am saying something – what I mean, what I understand by the
sign – is always the referent of the sign; but in saying something about it I express a thought,
and the sense of the sign is part of this thought. Thus what I am talking about when I use a sign
is not the sense of the sign. But it can happen that I want to talk about the sense; e.g. about
a certain thought. This happens in indirect speech. In the period ‘Aristotle believed that the
velocity of a falling body was proportional to the time of its fall’ what we have in the subordinate
clause is indirect speech. What would be the sense of this clause if it was the main clause is
now its referent. I can say: here the subordinate clause is the proper name of a thought . . . the
subordinate clause does not here express a thought but designates a thought . . . in indirect speech
every word has not its ordinary (direct) reference, but its indirect reference, which coincides with
what is otherwise its sense” FREGE {2}: 149-153 [modified translation]. That what counts as
‘talking about a thought’ or ‘mention it’ is very flexible is part of the problem. The preservation of thought-contents when expressed in oratio obliqua may require that the embedded clause
be not reported in literal quotation style. Typically in reporting indexical sentences in indirect
speech the very preservation of contents forces one to abandon quotational style.
\15/ “A Fregean theory of reference will observe the principle of compositionality: the reference
of a complex expression is a function of the reference of its parts”, wrote Gareth Evans, see
EVANS {2}: 293.
\16/ FREGE {1}: 358. He went on to claim that this is generally the case for indicators like
‘here’, ‘there’, ‘now’, and ‘today’: “In all such cases the mere wording, as it can be preserved
in writing, is not the complete expression of the thought; the knowledge of certain conditions
accompanying the utterance, which are used as means of expressing the thought, is needed for
52
us to grasp the thought correctly” (ibidem). The exceptions are the sentences Quine termed
eternal, namely either purely logical or mathematical sentences or, if they in fact exist, sentences
like ‘Otto sees the table in Cambridge, Mass. at 12 noon on the first of January 1937 A.D.’; see
Quine, section 40: “An eternal sentence may be expected to be free of indicator words, but there
is no bar to its containing names, however parsed, or other ostensively learned terms. Terms
present may well have been learned with the help of indicator words.” (QUINE: 194).
\17/ FREGE {1}: 359. It is interesting to note that, in his unpublished Logic (probably dated
1897), Frege toyed with ideas coming very close to the intuitions behind the more recent K-D-R
theories. For instance he wrote: “A sentence like ‘I am cold’ may seem a counterexample to our
thesis that a thought is independent of the person thinking it, in so far as it can be true for one
person and false for another, and thus not true in itself. The reason for this is that the sentence
expresses a different thought in the mouth of one person from what it expresses in the mouth
of another. In this case the mere words do not contain the entire sense: we have in
addition to take into account who utters it. There are many cases like this in which the spoken
word has to be word has to be supplemented by the speaker’s gesture and expression, and the
accompanying circumstances. The word ‘I’ simply designates a different person in the mouth of
different people. It is not necessary that the person who feels cold should himself give utterance
to the thought that he feels cold. Another person can do this by using a name to designate the
one who feels cold.” (see FREGE {3}: 134-135, [emphasis mine]. Apparently (and, perhaps, that
this the reason why Logic is in the Nachlaßamong unpublished works) the line taken here is at
odds with Frege’s criterion that a difference in possible or actual attitude by speakers/thinkers
w.r.t. the sense of a sentence (thought) forces the recognition of a difference between expressed
thoughts, in the plural. It is easy to supply very natural situations in which one can entertain,
and take to be true, ‘I am cold’ without having any such attitude w.r.t. the thought that one
bearing his name, or even he-himself-under-that-name, is cold. Like examples are common coin
among the authors who have stressed the non reducible character of indexicals.
53
CHAPTER 4
Direct Reference
An Expedient was therefore offered, that since Words are only Names for Things,
it would be more convenient for all Men to carry about them, such Things as
were necessary to express the particular Business they are to discourse on.
Jonathan Swift∗
The inadequacy of Frege’s original treatment of indexical sentences has prompted a
discussion which has come to coincide with wider philosophical concerns1 .
John Perry, focussing on the communicational function of language, starts out by considering indexical sentences.
In order to be understandable those sentences have to express propositions, publicly
graspable thoughts.
For sentences in subject-predicate form the situation is the following:
I am making a mess
has as one constituent of its sense (the thought express ed) the sense of the unsaturated (incomplete) concept-word ’. . . is making a mess’. To achieve the completion of a thought in Fregean
terms we ought to be able to supply a sense-completer. The sense-completer which comes to mind
as the most natural is the sense of the ‘I’ occurring in the sentence. The move is problematic
in Fregean terms because, Perry argues, tokens of ‘I’ (and of other indexicals) do not yield any
sense to bring about the “saturation of the incomplete thought”, but just what they refer to: the
utterer of the speech act, hardly a sense at all.
Taking Sinn to be the linguistic mechanism that secures reference in communication
(Salmon’s second possibility2 ), Perry states that when we grasp the meaning of the tokening of
55
an indexical
. . . what we seem to know is a rule taking us from an occasion of utterance to
a certain object. ‘Today’ takes us to the very day of the utterance, ‘I’ to the
speaker. . . 3
This meaning-as-machinery-for-securing-referents is what Perry called a role: a role is a
function from characteristics of an event to an object (e.g. from the time at which a speech act
takes place to an instant). This is the reason K-D-R was dubbed by its proponents a theory of
direct reference: there is no intermediate “entity” between referent and referential expression (at
least in the case of expressions which function as indexicals do, perhaps proper names.
Perry by allowing individuals-in-propositions wishes to avoid an infinity of private perspectives creeping into what he calls our common world4 . This is accomplished in turn by
claiming that what is expressed by the speaker and what is understood by the hearer(s) of an
utterance containing (tokens of) indexicals is a proposition construed thus:
S utters ‘I am cold now’
[and expresses that which]
H understands,
namely S [the role of ‘I’ takes the utterance containing its tokening to the utterer
as referent of ‘I’] is cold [the sense of the unsaturated concept-word ‘. . . is cold’]
at the time of the utterance [the role of ‘now’ takes the time of the utterance
containing its tokening as the referent of ‘now’]
Following Kaplan5 , I shall call the propositions which allow individuals as constituentparts $entences. Using a semi-formal language $entences are structured truth-valued sentences6 .
To go from an uttered indexical sentence to the $entence expressed one has to “compute” the
value of a function (or, as the case might be, of several functions). Indexicals’ functions (Perry’s
56
roles) take the hearers and the speakers from event to objects, that is from aspects of the context
of a speech act to individuals (speakers, events, locations in space, locations in time, locations
within the anaphoric environment created by a stretch of discourse, and so forth). I said ‘take the
speakers and hearers’ because it is a common trait of K-D-R theories to focus on words’ meaning
and not on speaker meaning7 .
The K-D-R theory then presents the view that linguistic expressions have a twofold
meaning-structure: a character and a content, or alternatively a role and a content. Basically a
content is a referent, and a character is a conventional prescription (relative to a language) to
determine the content (in this respect, under one interpretation of Frege’s theory of Sinn, direct
reference approaches might be seen less as competitors than notational variants8 ). Indexical-free
expressions have a fixed character (and a possibly varying content, i.e. they can change content
in different circumstances of use), indexical expressions have a fixed content9 .
In spite of all its formal elegance, important though it is if one is concerned with modeltheoretic semantics, the theory of direct reference seems to be in a predicament symmetrical to
the one Frege faced with respect to the phenomenon of factual harmony and cognitive dissonance.
Patently the content of two tokenings of phrases like ‘I am cold’ and ’You are cold’ when
expressed by two communicating speakers, makes their own production, in a way redundant.
Both productions are reiterating the expression of the same content, namely that S is cold,
where S is the content both the token of ‘I’ as used by S and the token of ‘you’ as used by
her interlocutor. It is equally clear, though, that if a semantics has something to do with how
humans think, then sameness of content does not entail identity of psychological attitudes about
57
that very same content, even when the attitudes are held by a single rationally consistent subject
at one moment in time. All the examples showing that indexical expressions are not reducible
to fully indexical-free ones point exactly to this fact. That is
If I see, reflected in a window, the image of a man whose pants appear to be on
fire, my behavior is sensitive to whether I think ‘His pants are on fire’ or ‘My
pants are on fire’, though the object of thought may be the same.10 .
K-D-R theories, at least in Kaplan’s and Perry’s original versions, if they want to be
true to their aims, are somehow forced to reintroduce a distinction between referential aspects of
language and manners of presentation, which though embodied in language itself, play a major
role in thinking.
The tension is detectable between views of semantics as an enterprise descriptive of a
prior domain of objects referred to by speakers and a semantics as an analysis of language as a
means of thinking and communication of thinkable contents11 .
To be sure, Kaplan is aware of the implausibility of the idea that $entences are objects
of proper thought, given, just to mention one reason, the human condition of being bound by
one specific perspective. His answer to the challenge presented by the divergence between factual
harmony and cognitive dissonance is to let characters play the role of thinkable components, and
have contents being the constituents of propositions. Characters – Perry’s roles – now function as
manners of presentation, and those are the avenues to the individuation of psychological states,
to the accounting of the causal role of thoughts in the explanation of actions. The psychological
states of the thinker are “sensitive to how the content corresponding to the time is presented, as
58
‘yesterday’, or as ‘this March 26th’.” .12
To take stock of the findings: the theory of direct reference postulates that tokens of
indexicals yield contents to the propositions expressed by the uttered sentences in which they are
used. Contents are referents, objects referred to by the speaker. The cognitive status enjoyed
by the indexical at the level of semantics involved with the apprehension of meaning from the
perspective of the thinker-speaker is not identical with the content itself. That role is played
by the character of the expression used. A character, in turn, is a rule of language, relative to
particular definite languages. Those rules are rules that specify which function one has to compute
to “get at” the content referred by the expression. Whereas in the case of pure indexicals the
utterance and the appropriate rule will suffice, in the case of demonstratives the rule will specify
that one has to take into account an associated demonstration, some kind of ostension, for the
determination of the demonstratum, the individual referred by means of the tokening of the
demonstrative expression.
My own interest lies on the side of the semantics of language as a vehicle of thought.
From this perspective I think one can ask two (clusters of) questions:
a) Are the points made by K-D-R theories sufficient to support the claim that
indexicals refer regardless of speakers’ intentions?
What exactly is it one thinks when engaged in indexical thoughts?
Are Kaplan’s characters or Perry’s roles objects of thought? – neutral terminology
here, no reference here to Fregean timelessly hypostatized Gedanke –
Are objects on the other side of the spectrum the right kind of thinkable contents
59
in terms of doxastic reference (i.e. in terms of what the speaker purports to refer,
relative of course to the complex web of her beliefs at the time of the mental episode)?
b) Are singular propositions the right kind of entity to be bearers of truth values?
My answers are negative to both a) and b).
As to a), simply put: the requirement that a semantical explanation of indexicality
should eschew the very notion of speaker intention seems to me just a prejudice13 . As far as
I can tell, there is no specific reason supporting the claim, except perhaps the suggestion that
introducing intentional notions would spoil the project of giving a neatly mathematized formal
semantics. As to b), the answer is again negative: a singular proposition lacks the peculiar
“fact-like” complexity needed to be a bearer of truth and falsity14 .
The idea that the cognitive status of an indexical tokening has to be seen as a function
clashes with an intuition to the effect that, in thinking, one deals with ‘contents’, not necessarily
Kaplanian contents. It is not clear to me that in uttering a sentence like ‘I am writing now’ I
have in mind as referent of my used ‘I’ a rule that gives myself as the object referred to by the
term in subject position, nor it is clear that I have in mind the object represented as the value
the function takes in the given context of utterance: which object? my body at the time of the
utterance? We seem to run up against the same kind of problem Frege had with mountains!
The problem seems to me to be summarizable in this way: in thinking the referential
content of an expression like ‘I’, used literally as a first person pronoun, is not an object (the
privileged object commonly perceived by most a s me, i.e. my body). For if it were my body,
then entire stretches of discourse, such as Descartes’ Meditations would be literally nonsensical.
60
But it cannot be the rule by itself: the rule is not a content of my thinking, at best it is a way
of depicting a content.
The most natural answer, to me, is that I have in mind a specific ephemeral content:
myself-as-the-agent/thinker-at-that time. But notice that this content is, at best, a very thin
time-slice of the Kaplanian content of an ‘I’. And it looks suspiciously like an individual-undera-concept, as opposed to the “immediate” individual constituent part of a singular proposition.
Along Kaplanian lines, one could thicken ontologically the very thin ephemeral slices.
One way to do it is to introduce indexically referred essences.
An essence is the fixed referent of a tokened indexical expression. Here by essence I
mean individual essence. I think the remark is needed to point out that in the case I’ll examine
now the essence of an object in the sense of its essence in terms of natural or artificial kind
will not work. What one might want to fix as an indexically referred object across possible
worlds is something identical with the referent in the actual circumstances of the speech act
executing the reference. I view such an individual essence as a set of identifying traits that
are unique to the individual possessing such an essence. Experientially those traits will suffice
to identify an entity and differentiate it from anything else. In the ontological mode of speech
one could say that an individual essence is a set of properties unique to that one individual (it
remains open whether the set itself is to be thought of as the individual or one wants to have
a substratum-holder of the prope rties included in the set). What seems to me problematic, as
I try to show below, is that there are some serious difficulties when one wishes to specify what
those identifying traits/properties are. Some specifications appears to be counterintuitive, but
61
that is not necessarily a terminal trouble for an essentialist view of this kind. It just seems to
me an essentialism of this form poses a theory of reference, qua theory of thought contents, too
removed from the phenomenology of thinking as we experience it. The essentialist move would
eliminate in a full sweep all the Fregean puzzles about mountains, snowfields, and solidified lava.
By pointing to Kaplan and uttering
(E) That is a philosopher driving a Mercedes
Doctor Hammer refers, following the essentialist proposal, with his ‘that’ not to Kaplan and his
kidneys, but to Kaplan’s essence. This essence of Kaplan gives us the right kind of referential
fix, and the right cognitive fix, when we want to have identity of referents in, e.g., evaluating
counterfactual statements. Hammer, presumably, would want to keep a stable referent for his
tokenings of ‘that’, when, pointing to Kaplan, he utters
(E+C) That is a philosopher driving a Mercedes, and if there were better public
transportation between Pacific Palisades and Westwood, that (man) would not
drive at all but use subways and buses
If one wants to have an identical referent across “possible worlds”, in order to make both
tokens of ‘that’ in (E+C) to be about one entity, Kaplan’s essence will do.
For the author of “Opacity” is driving a Mercedes in this world. So (E) is true in virtue of
its demonstrative-cum-associated demonstration successfully “harpooning” David Kaplan. The
counterfactual statement hints at a world different from the actual in which public Los Angeles
public transportation is not in the sorry state it is in actuality. In that world Kaplan would be
differentiable from the fullfledged (actual) Kaplan, for Kaplan-on-a-bus is different from Kaplan62
driving-his-car. Therefore the two demonstrative tokenings do not target the same individual
after all, if one wants to have (E+C) true.
The essentialist gambit in the context described allows one to have one single trans-world
individual as demonstratum. The latter is achieved by having ‘that’ in (E) refer to Kaplan’s
essence and having the demonstrative tokenings in (E+C) doing the same work.
To keep the same referent of ‘that’ in (E) and (C) one then would have to deny that the
demonstrative in (E) hooks onto Kaplan, the experienced Kaplan sitting at the wheel of the car.
Instead one should say that we are getting at his essence by pointing at him. One has to add
that in the case of (E+C) In the (E+C) case one has to add that, at least in Kaplan’s case, the
property of driving that car is not part of his essence. Are properties of this kind constituent
parts of essences? here one’s intuitions, or mine at any rate, falter rapidly: do we individuate
some one’s haecceitas by the car he drives? conceivably yes and no. Some decisions have to be
made by those who have stronger essentialist intuitions. But once those decisions are made one
could claim that Frege’s problems with lava, or with Kaplan’s cars and pencils, fade away: the
constituent part of the singular proposition, the referent is an essence, however one will have it
spelled out.
If I am correct an essence in this sense is not the commonsensical everyday object Kaplan
wants as contents and constituent parts of singular propositions. It is a referential content whose
manner of presentation can be roughly expressed as follows:
an essence is a constructed individual that we want to be stable across an enormous variety of frames of reference and it can be identified within one or more
such frames demonstratively or descriptively
63
If this characterization of essence can be a Kaplanian content, it seems to me a far cry
from the plain individual, ostensibly picked out by ‘that’ in (E). It is again, in a different way, an
individual-under-a-concept. What is particular of its manner of presentation is that the fixation
of a referent can be achieved by ascribing a certain stability to the highly unstable target of a
demonstrative referential intention. It is along those lines that some have resurrected the talk
about “haecceitas”: it looks very much like the picking out of an eternally stable essential “core”
of the experientially given target of a demonstrative.
The above excursus in the strange land of essentialism is for me a symptom of an ambiguity in Kaplan’s theory. The latter (at least in Demonstratives) is moving between the wholly
pedestrian, pretheoretical, notion of object as it is used in natural language to talk about regular
middle-sized physical objects, and a much more sophisticated theoretical “object” which serves
as a bearer of terms used for singular reference within the semantical theory.
To be sure, there are linguistic conventions regulating the use of indexicals and demonstratives. These conventions seem to me, in passing, more complicated than Kaplan allows. They
certainly play a role in communication: in prompting one’s audience to identify the speaker’s
referential targets. Rules roughly analogous to
by ‘I’ one should understand the agent doing the tokeni ng
are properly placed at the crossroad between syntax and semantics15 .
But if one wishes a theory to be of some generality, it can not be restricted to such
grammatical rules.
Thinking, I assume, is a shared trait among communicators, hence insofar as communi64
cation is an exchange of linguistically expressible thoughts, to communicate one must be able to
bring about in one’s audience thoughts related by some degree of similarity with the thoughts
one has . This is, admittedly, a very rough sketch, but at a pretheoretical level it is close enough
to what I think is the case.
The issue can be broken down, then, in two interrelated sub-headings:
(i) which is the thinking process one engages in when using an indexical?
(ii) which is the thinking process that, with all necessary approximations, one
expects from competent speakers-hearers engaged in the interpretation of one’s
utterances?
As to (i), the most immediate case is demonstrative thinking. By ‘demonstrative thinking’ I mean here the tokening of thoughts that can suitably be expressed linguistically by means of
demonstrative pronouns, in primis ‘this’ for the proximal and ‘that’ for the distal. ‘this’ and ‘that’
can be used in thinking without any attached demonstration: it is the thinker’s intentional object,
informally what-the-thinker-has in mind, which suffices (at least for the thinker) for the identification of a referential target for a demonstrative token. I keep stressing token because as many
occurrences of ‘this’ and ‘th at’ can be employed as one pleases, without using them as coreferential terms within the same stretch of discourse at all. This latter feature should be taken more
seriously into account by all who link very closely their referential mechanisms with the proper
names’ ones.
To refer demonstratively is to refer to ephemeral, intrinsically unstable particulars which
are part of a subjective experience: whether each one or all of one’s ‘thats’ can be properly
identified with a more stable, enduringly intersubjective – thus objective – particular is a different
65
question. That question does not call for doubts at all (it could not) the fact that I (I suppose
we) think demonstratively.
The idea here is that demonstrative thinking is the component of our thought processes
that more closely than others takes place within our own experiential field16 . This idea helps
explaining why demonstrative thinking is present even at the most inchoate stage of intellectual
development. The best examples I am able to cite in support of the claim are the wonderfully
“primitive” thoughts extracted by Aleksandr R. Luria from S., the mnemonist17 . S. is capable,
because of his extreme misfortune, to have the most vivid memories of his mental life as an
infant. All those thoughts of his are populated by demonstrative instances not easily replaceable
by any clear non demonstrative linguistic expression at all:
Light is something I remember very clearly. During the day it looked like “this”,
afterwards, like “that” – twilight. Then came the yellow light of the lamp – it
looked like “this”.18
Now, it seems clear to me that situations of this kind can hardly be accounted for by
claiming that what S. the mnemonist had in mind was a rule giving the content of his ‘this’ and
‘that’ by way of an accessory demonstration and the whole apparatus called into play by direct
reference views.
It has to be admitted, on the other hand, that a stretch of discourse, like the one carried
out by S., may be impenetrable for its audience. Again it is to be pointed out that language is a
learned social skill, and that only as a social fact communication takes place. Did S., then, refer
to nothing with his ‘this’ and ‘that’ ? This answer would be forced only upon he who maintains
66
that the one and only function of language is to be the medium of communicational exchange.
A semantics in my view ought to be able to bridge the gap between thinking, expressed
or not, and communication. If such a gap is in fact not bridgeable, theory should have it as
narrow as possible. What is peculiar of demonstratives is that the gap is wider than with other
expressions. The latter peculiarity is one of the reasons why, more often than not, one’s audience
has to rely upon all sorts of contextual clues to (try to) identify one’s demonstrative references.
Those clues can be behavioral (e.g. demonstrations by way of gestures) but some of them need
not be.
Question a) can be answered along the following lines. In thought the use of a demonstrative depends on the ability of the thinker to focus on something. Here ‘focus’ is, in a way,
a piece of jargon: no visual metaphor is intended. To focus on something in thinking is to give
to it, whatever it is, a sharper distinction against the enormous background of the ephemeral
field of experience the thinker has at any given time. A field of experience is always present,
sometimes consciously and sometimes subliminally (hence one can have, I take it, unconscious
experiences – whether those can be verbalized or expressed is another question, and the role of
language in bringing those back to awareness is a vexed problem indeed!19 .
A field of experience need not be perceptual. One can have experiences with nothing to
do with perception at all. Hence I do count hallucinations, reference to fictional entities, delusions
and the like, together with any kind of perceptual “entry” within the possible experiential field
of a subject.
The peculiarity of demonstrative thinking is that the focussing can be exercised on any
67
of the elements of the experiential field, and even on the field as a whole. Examples of the latter
could be taken to be aired by some statement such as ‘That is thoroughly disgusting’, where
the thinker-speaker has in mind nothing less that her global experience of the universe, perhaps
including the thinker – an experiential field can transcend itself, as it were. The referent of the
demonstrative that would be used by the thinker to express a mental episode she is, or was,
engaged in is then that very item zeroed in on thinking. Notice that, at this level of reference,
no question of vagueness can arise: from the standpoint of the thinker to what ‘this’ or ‘that’
refers is perfectly differentiable from anything else in the experiential field compresent with the
demonstrative thought.
A sustained attack against the idea of intentional reference has been recently mounted
by Howard Wettstein20 .
All of his criticisms depend upon one crucial assumption: that reference is a social
phenomenon. Hence the determination of singular terms’ referents depends exclusively on what
an auditor can make out of a stretch of discourse. I disagree. The reasons are in part explicit
in my assumption that language is a vehicle of thought from a first person perspective. And
even besides my general assumption, I think there are rather simple cases of discourse where we
are almost perforce going to resort to an appeal to the intentions of the speaker. Suppose, and
it is not a farfetched case, that one makes a telephone call to Howard. Howard is not home to
answer, but a voice coming from his phone answers: ‘I am not here now to take your call, but if
you want to be called back please leave your name and telephone number after the tone’. Now,
I take it, this is an utterance as good as any. I think it quite unnatural to take the ‘I’ to be
68
referring to the machine, to the tape or whatever. Note, in passing, how difficult it would be,
then, to understand the ‘here’ and ‘now’. The natural way seems to me to take ‘I’ as referring
to Howard, because that was the intention of the speaker when the Ur-utterance was recorded
on tape to be played back21 .
Nevertheless, something has to be said about the other side of the coin of reference: the
communicational side. I take it that in thinking it is the intention of the thinker-speaker which
determines the denotations of demonstrative tokenings. How are those references conveyed to
auditors?
The first cue to the determination of the referent of a demonstrative for an auditor is
not a demonstration at all, for in a host of possible and actual situations there is simply none.
The cue is given by a shared knowledge of the character of demonstratives: if I am correct, that
character lies in the grammatical meaning of the pronoun. Hence we are attending to (listening
to or reading, or whatever) the tokening of a reference to an item in the experiential field of
the tokener. Which one? To answer this question I think the best way for an auditor is to
rely on the predicative content of the utterance as a whole, and on the specific meaning of the
demonstrative used. The performance required for the understanding of a demonstrative varies
with the pronoun used. Using ‘he’ the speaker is asking the auditor to refer in first approximation
to ‘a certain male’ (an indefinite description), but to achieve singular reference (if that is indeed
achievable) the auditor has to rely on the predicative content of the utterance and contextual
and extracontextual cues (the salience of a definite male within that stretch of discourse, the
presence of that certain male, if physical ostension took place, and so forth). Important remark:
69
the degree of definiteness of a demonstrative reference for an auditor decreases as it moves from
‘he’, which just in virtue of its grammatical meaning in English indicates an item believed by the
speaker to be a human male, to ‘this’ and ‘that’. Both are highly indefinite: both can and do refer
to virtually anything, including purely subjective entries (e.g. in the case of S. the mnemonist),
but a higher definition can be restored by attaching to them a sortal term, as in ‘that man’ which
can conceivably in many situations be communicationally equivalent to the demonstrative use of
‘he’. Notice: this does not show that the meaning of the demonstrative is some sort of formula,
computing which we obtain some expression that is ready to replace the demonstrative in the
sentence salva veritate.
Let us test my approach with one example. David says:
(1) That is a great theory
At the level of David’s referential intention, his ‘that’ refers to the picture, in the sense of the
view of the functioning of proper names preferred by Saul Kripke. We might also suppose that
David is uttering (1) while pointing to a copy of Naming and Necessity. What does an auditor
understand? According to my view he can have David’s token of ‘that’ referring to the theory.
The auditor’s understanding would have to make use of the following cues: David is referring
to some item of his experiential field. Such an item is the kind of thing that should fall in the
categories of ‘theories’ and of ‘great’ (the way I see them, those are properties in the specific
sense of forms of categorization of the world). Such an item would have to be in the right kind of
relationship between a book and a theory (a relation that might be fairly complex: it is almost
a prototype of “Platonic” case of exemplification of an abstract entity). I think it is likely that
70
a linguistically competent auditor with all those contextual and extracontextual hints can, at
an illocutionary level, identify the item targeted by the reference David expresses in (1). The
predicative content of the utterance is essential because the pointing as such would never be
sufficient: I am not even clear whether one can point at a theory by ostension. As it was noted
earlier the understanding by the auditor does not produce a substitutable expression:
(1.S) An item in David’s experiential field is a great theory
does not have the same truth conditions (1) has. And
(2) That is a great theory written in Princeton
while it may have the same truth conditions of
(3) There is an item in David’s experiential field which is a theory written in Princeton
does not have the same cognitive significance. So, in terms of meaning, one would have to conclude
that applying Frege’s test (2) and (3) are not synonymous. Similar kind of referential constraints
can be given for other demonstratives22 . Note that on a literal application of a K-D-R approach
(1) would have to come out false, or at least very elliptical: ‘that’ picks out its demonstratum
(the book) and books are not (literally at least) great theories. I would like to stress that, if I
am correct, this kind of reference is doubly direct: the speaker refers directly to the target of his
referential expression (that segment of the field he intends to single out for reference) and the
auditor can have the same referential target, when the clues given are sufficient. At no point does
either one have to rely on descriptive, conceptual analysis (dissolution) of the singular indexical
term into purely non indexical morphemes.
71
Both the speaker and the auditor have no need to resort to Fregean senses to determine
reference. The referential fix can take place at one side at the level of intentions and on the other
side at the level of executing instructions to perform a similar kind of mental act, as it were.
The thinkable referents for both speaker and audience are contents, and if the mechanisms
for referring have to be interpreted as mathematical functions (I am not convinced that this is
what is going on in speakers’ minds at all, but this is not very important) the referents should
be taken to be the values of the functions, given the arguments provided by the environment of
the speech act to be interpreted.23
It is important to see that in my sense there is never a guarantee that the referents of
the speakers and the auditors are the same, self-identical massive chunk of reality “out there”. In
the simplest case what we have are two references to items in two distinct fields of experience. It
is a matter of the stability of intersubjective agreement to arrive at a stronger sense of “talking
about the same thing”. In most cases our references are pretty much public in the sense of being
easily subject to an intersubjective check, and those are the cases in which we can get more
simply at the referents with the relevant degree of “sameness” with no indexical reference at all.
But this is not the guaranteed outcome of an episode of, perhaps purported, communication.
I now turn to my second question about singular propositions. The idea of singular
propositions as contents expressed by some sentences, I believe, is informed by two powerful
(and pretty good) intuitions. One is the anti-nominalistic view that sentences are carriers of
information, that they have a meaning which is not just a faint echo of the noises made and of
the ink marks: it is an idea going back at least to Greek philosophy. The second motivational
72
component is a concern for the stability of communication, and here what plays an important
part is the realistic view that when we express the same propositions, by whatever means, we
are talking about the same fact. I stress the realism implicit in this kind of view because even
to state a semantics with singular propositions one has assumed from scratch a neat division
of the universe into individuals, properties, relations and the like. Propositions are bearers of
truth value and it seems intuitively sound to have them as the right kind of entities between
brute facts and sentences. Tradition has it then that propositions and not utterances are the
objects of psychological attitudes (like believing, guessing, knowing, doubting, etc.). There are
indeed good reasons to think that what is believed is not in general an utterance, but the content
expressed, some say the information carried by the utterance. Very visibly in the case of indexical
utterances the information carried varies depending upon the context of use. ‘I am cold’ uttered
by X and ‘I am cold’ uttered by Y carry quite different information, or so the story goes. By
disanalogy, it seems intuitively clear that ‘I am hungry’ said by X at one specific time carries the
same information that ‘Ho fame’ said by X would yield at the same time (of course relative to the
language in which the utterance is to be understood: if ‘Ho fame’ were an expression in English
(hence being lexically in both languages) meaning what ‘I am scared’ means in English, then the
two graphically and phonetically identical tokens would carry different information depending
upon the language in which each one of them is understood).
Singular propositions introduce a new twist in the traditional doctrine. In them individuals enter as constituents “in toto”, so to speak, not by being represented by ways of thinking,
perceiving or conceiving them. This trait is controversial from the beginning because it is natural
73
to think of propositions as eminently abstract entities, fit for proper apprehension in thought:
how else can one hold psychological attitudes toward propositions without being able to have
them “in mind”? The difficulty is sensed (and quickly disposed of) by John Perry, a proponent
of singular propositions:
Philosophers who are bothered by singular propositions often complain that
individuals can’t be “inside the mind”. But of course the properties and relations
that are constituents of “general propositions” are no more in the mind than
individuals24
My complaints are partially different. A singular proposition is a complex. Some of its
components are individuals: if I am interpreting correctly K-D-R theorists, those individuals are
pieces of a physical universe, such as people, objects, events, instants and their ilk25 .
I start tackling the issue with a very simple case. Suppose X utters:
I love you
directing his statement to Y. Have we generated a singular proposition? The generated singular
proposition could take one of two forms, in so far as I can tell:
(i) [the individual denoted by ‘I’, the property denoted by ‘love’: the denotatum of ‘you’]
or
(ii) [the denotatum of ‘I’, the relation ‘love’, the denotatum of ‘you’]
Given the context of the utterance the competent speaker and the competent auditor are
able to identify, by means of Perryan roles or Kaplanian characters, the two individuals denoted
74
by the two personal pronouns . Suppose X is Bogart and Y is Bacall. The singular propositions
would then be
(iN) [Bogart, love-Bacall]
or
(iiN) [Bogart, love, Bacall]
Thus seen singular propositions then look like n-tuples of elements which are referents of terms.
Whenever the term is a singular term the propositional constituent is an individual, otherwise it
is a more or less complex property or relation.
Seen in the above way singular propositions seem to be the strongest guarantee that we
are able to communicate about individuals.
For suppose that John Huston is present and overhears Bogart saying ‘I love you’ to
Bacall. Huston would be correct then in reporting in oratio recta ‘Bogart loves Bacall’ and in
oratio obliqua ‘Bogart said that he [pointing, intending to refer to Bogart] loves her [pointing,
intending to refer to Bacall] (this is, to my mind, the sound part of the idea that proper names
and indexicals alike are just referring devices, without any descriptive meaning attached to them:
the descriptive meaning, whatever it is, drops out of the picture for want of relevance).
Often singular propositions have been termed de re, as being about things and not
de dicto or about words. Note that in Fregean terms denoting relations and properties have
unsaturated sense, so the propositional components in (i) and (ii) expressed in the utterance by
‘love’ are incomplete senses, as Perry puts it. The saturation, the completion of the proposition
75
is effected by the individuals which fall within the extensions of the incomplete senses.
Now, some want propositions, among other things to be bearers of truth and falsity, and
we want them to be the proper subject matter of logical relations. Singular propositions lack
the proper articulation to be carriers of information, informally what is said by an utterance. If
one were to be given as information (i), (ii), (iN), or (iiN) she would be hard pressed to evaluate
them. If, that is, an auditor was presented with Bogart juxtaposed with the incomplete sense
of ’love-Bacall’, I have my doubts that such an auditor could literally make any sense out of it,
much less understand what the presentation is supposed to accomplish.
The singular proposition looks very much like an unstructured bare fact: it is not [Bogart,
the unsaturated sense of ‘love’, Bacall] which can be true or false.
For if a non ordered n-tuple of elements that gives us the proposition were the bearer of
truth and falsity, then we would be assigning perforce the same truth-value (in the same context)
to ‘Bacall loves Bogart’. Hence it is natural to think of singular propositions as more structured
entities (formally representable as, say, ordered pairs, triples, and so forth). Taking the road
toward the very structured object of thought [that Bogart loves Bacall] pushes the theory back
to the notion of de dicto propositional contents, which are just plain abstract description of actual
or possible facts, thence true or false (given a domain of evaluation). A tension is detectable: the
closer the semantical reconstruction of the proposition’s structure is to the grammatical form of
the sentence expressing the proposition, the further we move away from the intuitions behind
the “singular proposition” strategy, intuitions which amount to the need for having an object,
plain and simple as propositional component.
76
The obvious response to a critique like the one just presented is to build more structure
“from within” the singular propositions26 making them look more and more like the sentences
expressing them.
It has been claimed that the “extra-structure” proposal fails exactly where it is thought
to help, namely in the semantics for ascriptions of belief 27 . It seems to me that the argument,
though very interesting, suffers from an ambiguity, perhaps a fatal one. I reproduce, almost
verbatim, Scott Soames’ examples and then proceed to point out the source of the ambiguity.
Karl sees me in a supermarket and knows me as an Italian graduate student, and Karl
sees me in the university teaching. Then Karl can utter on two different occasions:
(I) He is an Italian graduate student
and
(U) He is a teacher at the university
Later on, he remarks:
(I+U) He is an Italian graduate student and he is a teacher at the university
Now, following Kaplan, the two tokens of ‘he’ are demonstratives, each one of them associated
with a different demonstration (since the occasions are different, the two utterances are taking
place in two different contexts). Same character, different context of evaluation, same content:
myself. Both of the following ascriptions seem to be correct:
(1) Karl said that he [pointing to me in the supermarket] is an Italian graduate
student, and he [pointing to me while I am teaching] is a teacher at the university
77
(2) Karl said that I am a teacher at the University and an Italian graduate student
[uttered by me]
Paraphrasing Soames, (1) has the form
(F1) K says (believes, asserts, thinks, etc.) . . . t. . . t’
On the other hand (2) has the form
(F2) there is an x [K says x is a teacher and x is a graduate student]
This is because, by the lights of K-D-R ‘I’ is always referential, so it seems one is always allowed
to understand a first person utterance as an existential claim. (2) is true and so if the theory of
direct reference does not allow to move from (1) to (2) it seems to miss a truth. If it does allow
the move in question it seems to be committed to the truth of
(3) K says that t* is a teacher and t* is a graduate student
for any directly referential term t* denoting me, such as the indexical ‘he’ in the context, and
so allowing back into the picture the Frege-inspired puzzles of the form ‘that is the evening star
and that is the morning star’. This is because the theory is committed to the claim that
(F3) K says (believes, etc.) . . . t. . . t
can be true if (F1) is false, only in the case in which (F2) is also false. Soames’ “counterexample”
points out that even within direct reference theory one can construct cases in which more than the
bare content comes to be a propositional component. In this case the propositional component is
myself, but if referred to by way of ‘I’ or by way of ‘he’ the very same content generates disparities
in truth values of sentences which intuitively ought not to appear.
The argument is ingenious, but it suffers from an ambiguity which makes one of the
78
premisses rather doubtful. While attribution (2) seems indeed to be correct if uttered by myself
(it is a case of belief-ascriptions de re), attribution (1) is ambiguous. It is not obvious what the
tokens of ‘he’ do there. If they are straight demonstratives (Kaplan-style), as Soames seems to
imply in his comments, then sentence (2) does not appear to be well-formed. On the one hand
demonstratives have, as many have remarked often, the “largest scope”, they express a speaker
reference. If the “pointing” indicated in the bracketed portions of (2) is done by its utterer, then
the ascription is not correct, after all. The tokens would be different tokens and (2) would not
be a faithful report of what Karl said. If on the other hand the tokens of ‘he’ and the associated
demonstrations are indeed Karl’s, then the sentence is a strange mix of direct speech and indirect
speech (it would make perfect sense to utter: “Karl said: ‘he [pointing, if you wish] is an Italian
graduate student, and he [second pointing] is a teacher at the university”). Another source of
possible ambiguity lies in the possibility of wholly different reading of (1). The tokens of ‘he’ there
could be taken as Castañeda’s quasi-indicators, in which case though (2) would rather obviously
false, in that case Karl would be doing the pointing to he-himself and that does not seem to be
what was intended at all. Suffering from this ambiguity I am unconvinced that Soames’ point
has exactly the force he attributes to it.
Coming to points where my disagreements with K-D-R theorists are more radical, I want
to point out why I am ill at ease with the strong realistic assumptions of K-D-R. The whole formal
apparatus relies heavily on a fixed domain of interpretation. By fixed domain of interpretation I
mean the so called world out there which is supposed to come ready-made for our apprehension
with a crisp division between individuals, predicates, relations, and so forth. I have to admit this
79
form of realism is a deep seated inclination of many, and of many philosophers. I just do not
find any reason to rule out radically different forms of categorization of whichever experiential
entries one wants to categorize.
This antirealist position is a prejudiced opinion on my part. There are reasons though,
why I want to hold it. I am reluctant to see those reasons as arguments in any formal way
because I do not think that arguments will settle disagreements of this kind: at best reasons will
serve as paths to see the line of thought involved in my philosophical position.
I have an adversion to claiming that reference, and even more perspicuously demonstrative reference, is a relational phenomenon between minds and the world as it is, whichever the
ultimately correct understanding for such locution will turn out to be. K-D-R theorists take
the referents of demonstratives to be massive chunks of reality, individuals as they call them. It
seems to me to be a fact that we can demonstratively refer to a variety of entities having very
little to do, if anything, with the cozy familiar world of trees, glasses, and umbrellas.
We can and do refer to dreams, hallucinations, and their internal characters, fictional
figures, and. . . all sorts of entities that do not qualify for the high standard of “reality”. I do not
think it correct to claim in all those situations that we are actually referring to “things-in-thehead” in the sense of cerebral events. With Ray Jackendoff I believe
One is speaking of brain-events no more when discussing one’s dreams than when
discussing one’s perceptions28
To keep the phenomenon of indexicality as unitary as possible and to keep its understanding as unified as possible, one has to postulate and eventually countenance in one’s ontological
80
commitments – if one wishes to have them – internal contents, mental objects of reference, which
may in the last analysis be identified with “real” world counterparts, when there exist indeed
public, intersubjective counterparts of those internal contents. But the latter are the primary
referents of referential attitudes. Language as a medium of communication certainly provides
for an immense background of intersubjective objectivity in the referential relation, for, by and
large, it is language itself that shapes our very capacity to think in terms of singular reference,
and we learn language in communicational exchange. But indexicality is, if one wishes to put the
matter in terms of development, more primitive and more simple: it points directly to contents
as they are experienced by the subject who is doing the “pointing”.29 Demonstrative thinking
constitutes ephemeral targets of reference, ephemeral because we do not have an infinite number
of non indexical naming devices. We do have in natural language a limited stock of singular terms
and, to put the matter jokingly, most of those internal targets do not deserve to be baptismally
named: they flee out of sight and out of existence too quickly.
Internally experienced contents are the proper targets of demonstratives and those need
not be shown to be “in the mind” or “out of it”: they are the ways in which minds organize and
categorize whatever signals impinge on them30 .
The same holds true of propositional contents. While K-D-R theorists’ notion of proposition will work, plus or minus some itches, whenever there exists the external, objective individual
to be the denotatum of demonstrative tokenings and the physical constituent of the proposition,
the same notion will break down the fundamental unity of thinking in any non veridical case,
in which the targeted individual does not exist except from within the experiential field of the
81
tokener.
For take the simple circumstance motivating a course of action prompted by a thought like:
(D) That’s a drowning man31
It makes no difference in this case whether the pointing (the referring) occurs ostensively as a
movement of arms or not: (D) is what I think (perhaps falsely, depending on whether the man
is there or not, and whether the man is drowning or not). (D) together with other premisses can
lead me to try to swim to save the man. In the veridical case the man is out there gasping to
make my thought a thought-of Kaplanian content (a singular proposition). In the non veridical
hallucinatory case there seems to be no content whichsoever. And yet in both cases, all else being
equal, it would lead to the same course of action on my part.
If my interpretation is correct, it is the internally given propositional content that I am
evaluating in my practical inference. But this could not be under K-D-R constraints. The point
is that my deduction is not taking as a premise anything like the character of ‘that’, or the
manner of presentation of my (possibly only purported, be that as it may) referential target. If
I want to save anything I want to save that man, the unfortunate possible counterpart of my
experienced content of ‘that’, or the luckier non-existent (by everybody else’s standards of reality
if I am indeed hallucinating, or even just mistaking the motion of the dolphin for a drowning
man) and nonetheless quite real man in my experience.
The realization of how crucial the causal role of demonstrative and indexical thoughts
is, helps us to see more clearly the need for internal propositional contents. Whereas in all
cases of veridical experience we can, I think correctly, interpret in Kaplan’s way a demonstrative
82
proposition, we can not easily do it in the non veridical one.
In terms of truth conditions, (D) is true if and only if there is something which is that one man
who is the referent of my tokening of ‘that’ in (D), (by means of its grammatical meaning and
my pointing, be that physical or purely intentional) and that man is drowning at the time (D) is
vocally expressed or mentally entertained. Therefore, in the non veridical case, I take it, (D) has
to be false. And yet, for me, to be able to use it as a premiss in my little exercise in practical
reasoning it has to be taken to be true. Were I not to take it to be true I would never be moved
to action. I am disregarding the possibility that I might actually be a sadist and be amused by
the spectacle of the man who dies before me: it is immaterial – even to experience the pleasure
of the man’s painful dying I still have to take (D) to be true.
I am aware of a prima facie destructive objection to the idea of internally construed
propositions as thought contents: it breaks down truth values in a myriad of subjectively accessible evaluations, thereby losing the world for the theory which countenances them.
I am not sure there is such a danger, and I am unpersuaded that it is a danger. I am
also persuaded, however, that at this juncture our ability to communicate plays the key role in
establishing the very idea of a common, intersubjectively available world.
¿From the indexical viewpoint, as I understand it, the experiential environment is subjectively apprehended, my ‘this’ and ‘that’ and even more so for my ‘I’ have as referents internal
construals32 . Those internal construals are not necessarily private. By ‘private’ here I mean even
in principle not open to others. Effable thoughts are transparent. The so called ontological relativity, which is only a term of art for the ability of humans to carve up the experienced world in
83
many different ways, thereby having different constructions for, say, ’individuals’ in their frame
of reference, does not block any one of those ontologies from being linguistically expressible. And
the very possibility of radically divergent ontological assignments to the symbols representing “individuals” in a formal semantics indicates that there are no univocal, absolutely objective values
for functions mapping singular terms like demonstratives into objects. We can both make and
attribute demonstrative/indexical references. Here at work is in part an assumption about the
existence of minds other than one’s own, and a stronger assumption to the effect that there are
no radically diverse conceptual schemes33 . Armed with these two assumptions we can interpret
within the range of our own internal construals the propositional contents others express.
If indexical reference is subjective, ephemeral and experiential, (I agree with Castañeda
in that it is), the purely linguistic fact that it is attributable to others is evidence that our system
of communication is at least able to accommodate the dreaded myriad of private perspectives34 .
Whether or not prima facie perspectivalism has to be a final point of rest in metaphysics
is the question I will deal with in the fifth chapter.35
84
A more radical break with tradition: Guise Theory.
In a long series of works Hector-Neri Castañeda has argued for a radical break with the
semantic tradition. Guise theory, his view, is a grand theory in formal ontology, meant to shed
light on a number of different issues in comprehensive terms. I’ll try to summarize his view first,
and then suggest some ground of possible disagreement.
Castañeda’s views could be described as a doubly tiered structure, one tier is more strictly
linguistic, or phenomenologically linguistic, the second tier develops a philosophical treatment of
indexicality which is, in turn, embedded in a theory of much larger scope, a general ontology, as
it were.
The key ingredient of the phenomenological linguistic tier is the view that indexical terms
express references made by speakers to items present in their experience. As I mentioned earlier
a development in analysis initiated by Castañeda is the finding that along indexical references
natural languages, at least many of them, make room for quasi-indexical references. Quasiindexical references are made by speakers who attribute to others the (capacity of) making
indexical references, whether the original reference was made indexically or not.
Castañeda states the main linguistic properties of indicators thus:
Let S be a sentence containing and indicator i (not in quotation marks). Let S be
used on some occasion o by a thinker-speaker H to formulate a thought content, e.g.,
a statement, a question, a conjecture, a request, a petition, that H thinks out loud
by proffering S. Then on occasion o indicator i:
(i) expresses a thinking reference by H
85
(ii) to an entity determined by o;
(iii) expresses a reference by H that does not depend on H referring to the referent
with another expression;
(iv) expresses no attribution of reference or mechanism of reference to anyone -even
if i appears in a clause of indirect speech in S36 .
Principle (iv) indicates the important fact that the embedding of indexicals within oratio
obliqua does not affect their status of expressing a reference made by the speaker, while (iii) reckons with the absolute character of indexical references as they are made by the speaker, namely
that they do not depend upon alternative non-indexical ways of referring. Thus Castañeda’s
position seems to be diametrically opposed to Millikan’s.
Castañeda holds that all indexical expressions follow laws (in the descripti ve sense)
of experiential reference, and differently from Kaplan he does not make a distinction between
pure and demonstrative indexicals, since for him the character of each indexical plays a role in
determining reference.
There remains a distinction between the levels of certainty attained: whereas ‘I’ for
instance whenever used has a doubly successful reference (double because it has an internal, or
thinking, guaranteed referent and an external, metaphysically certain, referent) ‘that’ has a singly
succesful reference, there being no certainty of a correspondence between the internal, thinking
referent and an external “harpooned” target which might not exist at all. the character of each
indexical plays a role in determining reference.
Setting them apart from other terms indicators undergo a “semantical shift” when con86
tained within the scope of so-called opaque-context forming operators (psychological, modal
prefixes for instance, such as “. . . believe that” or “possibly. . . ”). The shift is needed to have the
indexical reference “jump out” as it were of the scope of opacity.
The cluster of distinctions which operates at the level of phenomenological linguistics
between oratio recta and oratio obliqua constructions will take a radically different aspect in the
ontological aspect of Castañeda’s view, but more about that later.
Two important notes.
a) One claim originally made was to the effect that two sentences, one indexically expressed (such
as ‘I am blue’) and the other quasi-indexically attributed to (such as ‘AP thinks he is blue’),
denote one and the selfsame propositional content. More recently Castañeda claims two such
sentences express two propositions, intimately related one to another, nonetheless not related by
strict identity.
b) Indexical reference is reference to items present in the experience of its maker, qua present in
experience and as long as present in experience, those references have no temporal “stability”,
they are ephemeral references:
Every indexical reference is both personal and ephemeral. . . indicators are expressions used to make immediate and strict references to items presented in
one’s experiences in so far as they are present to such experiences. Even though
Hector-Neri Castañeda has lived for fifty-six years now, his first-person references
are always personal -his and nobody else’s- they are always also ephemeral. The
self to which each of his first-person references refer is the synchronic present self
who owns the experience that prompts the first-person reference in question. In
general, one attributes to oneself adventures in the past only because one identifies oneself (i.e. the current experiencing self) with a previously existing person.
Similarly, one’s this’s, that’s, now’s, here’s and there’s are all ephemeral and
personal.37
87
As he expresses it, the claim that singular thinking references are made towards individuals cries out for an explanation of what such an individual is supposed to be ontologically.
That is the task Castañeda assigns to Guise Theory, which has at least two different
strands. One is the elucidation of what has to be taken as individual to be a fit target of thinking
reference. And the seco nd one is an elucidation of the family of relationships which connects, in
the successful cases, our thinking with the structure of the world. The answer to the first problem
is the introduction of a new kind of individual, while the answer to the second is the introduction
of a family of relations of sameness, one of which is the traditional unrestricted identity.
As a theory of thinking reference, Guise Theory posits a new kind of referents. Whereas
in more “realistic” approaches referents of terms which are singularly referring (indexicals, descriptions, possibly proper names) are what Castañeda calls macro-objects, the thinking referents
in his view are Guises. Guises are structured particulars: the core of a Guise is a set-theoretical
entity, whose members are properties. A sui generis operator particularizes the set of properties,
yielding an individual. Informally the basic idea, I believe, is best understood thinking of a Guise
as a facet of an object in perception. A facet has the following characteristics:
a) it is a kind of particular
b) it is fully characterized by a set of properties
c) it is not closed under implication, i.e. a facet is characterized by certain
properties without being characterized by properties implied by them
d) (perceptual) facets are finite, in the sense of being fully characterized by a
finite set of properties.38
The theory of Guises formalizes with a set of principles the kinds of operators which
particularize set of properties. One advantage has to be noted at the outset: Guises are directly
88
thinkable, there are no further cognitive intermediaries. Hence there is no categorial difference
in the theory between referring to an object of thought and referring to (what I would call) a
perceptual object in my field of awareness. The differentiation would come about by way of the
different properties in the core of the guise thought of. In the case of demonstrative thinking the
core of the guise referred to has to contain irreducibly demonstrative properties (expressible by
locution such as “being there” or “being that (thing) which makes noise”).
Indeed a specific view of perception is one of the motivating grounds for Guise Theory.
The view of perception expounded by Castañeda is best understood, as he himself said, as a
generalized and sophisticated from of phenomenalism, without the classical “sense-data”. Perceptual awareness is a awareness of a private perceptual field: “To perceive is not . . . to perceive
this or that, but to perceive a perceptual field in which there is, often, this or that”.39 The
dwellers of the perceptual spaces (the “origin point” of which is a thinking mind with perceptual
capabilities) are those facets (of ordinary objects), and they can be related in perceptual judgments such as “the tallest evergreen tree on my left is the pine tree with a good smell”. Now,
the perceptual judgments can be veridical. And in that case Guise theory interprets the copula
in the example not as a statement of identity, but a statement expressing a consubstantiation.
As one can gather etymologically, the consubstantion operator brings together the being of two
facets. It is important to notice that there is no a priori guarantee of the consubstantiation
expressed by a statement such as the one in the example of the tree above. The person making
the statement might be hallucinating, or be wrong. Here comes to fruition the remark on the
non-differentiability of veridical from non-veridical perception40 . Within a mind’s perceptual
89
space those facets (guises) are the primary referents of demonstrative expressions, they point as
well to a purported reference to a physical object, very often at least, but the purpose might
be defeated. What seems to be missing are the macroobjects of ordinary perceptual awareness.
Those are ontologically described as systems of Guises, with one specific particularizer, an operator expressing consubstantiation.41 Linguistically one may take the consubstantiation operator
as expressive within the theory of relationships of sameness which ordinarily are thought of as
cases of “contingent identity”, as in my example above “John Le Carre’ is David Cornwell”.
Such a sentence states a true proposition which is not in Guise Theory a proposition of identity,
a weaker notion of sameness is needed to account of Frege-style puzzles.
The key similarity of Guise theory with Frege’s semantics lies in the similarity between an
individual guise and a Fregean sense for a singular referring term. There are two key differences
though. One lies in the direct character of an act of reference: the Fregean sense of a term is (also)
an intermediate entity, it is also a way for the speaker to get at the referent of the expression.
On the contrary in Guise theory the (guise-theoretical) referents are directly, in propria persona,
present before the mind. Secondly, whereas Frege held we have to make a distinction within the
theory of reference between the referents of the same terms when appearing in direct speech vs.
their occurrence in indirect speech, Castañeda’s view is that there is no instance of direct speech,
and thence there are no first-order Fregean Bedeutungen:
With Kant’s assistance, the destruction of Fregean referents is immediately established. Kant observed, and posited as his major principle about the unity
of consciousness, that all representations can be subordinated to I think. From
this we can extract the following important syntactico-semantical datum: All
thought-expressing utterances are implicitly or explicitly embedded in an indi90
rect speech construction of the form “I think that”
. . . Hence every speaker’s use of singular terms, even if ostensibly in direct speech,
must by Frege’s own lights, have as referents what he calls their primary senses42 .
The object Frege took to be the referent of an expression is then not part of the stock of
semantics, but part of the stock of doxastics. Fregean referents are, for Castañeda, transcendent
objects of belief and/or quantification, but not proper subject matter of a referential attitude
from a speaker.
The key distinction is between a semantics for doxastics reference and a semantics for
thinking reference.
So, one distinguishing trait of Guise theory is that there is no direct speech, in the
traditional grammarians’ sense. It is here one can appreciate more clearly the distinction between
a phenomenological linguistic level of analysis and an ontological one. As pointed out before,
the linguistic analysis make use of the distinction between direct and indirect discourse. At the
(perhaps deeper) ontological stage of analysis such a distinction turns out to be in a sense illusory.
While I believe the suggestions of Castañeda’s linguistic analysis ought to be incorporated in any
view of indexicality, as it surfaces in language, the ontological side of his analysis is much more
contentious, in particular for the thorough destruction of Fregean referents (which seem, at least
at first blush, much closer than Guises to what, commonsensically , many take to be objects
of reference). The latter trait I see as an advantage in purely philosophical terms, but as a
problem in terms of expressibility of the theory. It seems to follow that any expressed thought,
or sentence, is within the scope of the Kantian prefix, and so the only “genuine” case of direct
91
speech is a case of non speech (the adjective genuine is Castañeda’s own characterization of the
transcendental prefix ‘I think here now’), since every time the ‘I’ of the prefix is expressed is just
embedded inside another clause in indirect discourse. If one accepts Castañeda’s point on the ‘I
think’ and wants to hold a Fregean view of the semantics shift between sense/reference between
direct and indirect discourse, the almost paradoxical result would be that there can not be any
instance of direct speech.
It seems to me that given the sweeping power of ‘I think here now’ prefixing all of
our representations, there is no reason to think philosophical statement to be excluded, from
which it should follow that transcendental prefixes are absolutely ineffable. I do think there is an
advantage in the generality of Guise theory, and that generality consist in having a unified theory
of reference for all kinds of syncategorematic terms. This is not the only advantage, but is the
one but it is the one I am concerned with now. What I would like to explore in the next chapter is
the possibility of a mixed theory in which the “reach” of a transcendental prefix affects indexical
terms, but not all terms. One of the reasons that pushes me in such a direction is a more deeply
embedded reluctance in taking properties as fundamental ontological components. Guise theory
is a theory firmly steeped in a Platonistic framework. What is appealing of Guise theory is that
it has room to accommodate my intuition43 that we have private contents of thought, encased
linguistically
Castañeda held, in the past, that a quasi-indexical proposition depicting an indexical
one expresses exactly one and the same propositional content. After his exchange with Robert
M. Adams he came to give a much fuller sense of privacy to indexical propositions: indeed the
92
quasi-indexical ones only give a picture of their indexical “cousins”. The relation, intimate as it
may be, is not a relation of identity. In fact in indexical propositions and the quasi-indexical ones
depicting them the indexical contents would be fully shareable by different thinkers. Since the
latter is not the case, a stronger sense of privacy is appropriate for indexical contents, as contents
they are not shareable at all, although the structures of the propositions containing them is .44
Summarizing, Guise Theory is a theory of direct reference which does not countenance
any psychological entity mediating between the thinker and referents. The Fregean denotata are
reinterpreted as systems of Guises, those denotata aren’t thinkable referents but are doxastic
referents45 . Doxastic referents are not directly thinkable, but are reachable in thinking and belief
through references to guises and/or through forms of general reference (e.g. quantificational).
While I have been learning from Castañeda’s analysis and I adopted in much agreement
many of his syntactic and semantic insights, I think it is not useless to explore an alternative
theory, leaving the ontology of objects as it stands, roughly in common sense, but trying to
provide a different framework for indexical expressions. I will go back as well to the effect of the
Kantian ‘I think’ on the ‘I’ in it, and in any first-person utterance in the fifth chapter.
There are two lines of difference I wish to develop, increasingly important I believe.
For one thing I am not convinced that the fuller “Kantian” prefix (‘I think here-now’)
is the best way to characterize our linguistic and thinking practices. In one section of the next
chapter I develop an argument which seems to me to indicate we do not have a ground for the
extension from ‘I think’ to ‘I think here-now’ which is evenly solid for both the temporal and the
spatial indexical. But this might be a minor problem.
93
What I do not think is a minor problem is that despite the complete destruction of
Fregean “primary referents” executed by Castañeda there is one fact which I keep finding compelling: we do use direct speech, and I find convincing the Fregean arguments on the need of
keeping a sense/reference distinction between direct and indirect discourse. One can always discount the surface of linguistic practice when a stronger theoretical pressure imposes the move,
for reasons of economy or of explanatory power. I think it is possible, or minimally it is worth to
try, to save as much we can of the Fregean framework, while accommodating the phenomena of
indexicality without undergoing a complete ontological overhaul. This is the motivating intuition
behind the sketch of a theory I present next.
Eventually theories have to be incorporated in larger bodies of thought, and I have to
admit I do not have a complete alternative to Guise Theory catering to exactly the same (or
possibly even more) corpus of data. That, by itself, does not seem a good reason not to develop
alternatives.
94
NOTES
\*/ Jonathan Swift, “A Voyage to Laputa, Balnirabi, Luggnagg, Glubbdubdrib, and Japan” in
Gulliver’s Travels.
\1/ Large part of the recent literature on indexicality has to do with attacks and counterattacks
aimed at Frege’s remarks. Singular propositions as “a way out from Frege” however came about
from related although different corners (worries about the proper interpretations for systems
of modal logical calculi and, in part, from concerns for the so called metaphysics of individual
essence, often dubbed haecceitas – literally this-ness or this-ity – after Duns Scotus). See, inter
alia, CASTAÑEDA {8}, EVANS {1} and {2}, GALE, KAPLAN {2} and {4}, and th e, by now
classic papers by John Perry, (PERRY {1} and {2}. For a different view of the connections
between indexicality and “possible worlds” semantics, see “Indexical Belief” by Robert C. Stalnaker, in SAARINEN.
\2/ See footnote \1/ and the reference in footnote \5/.
\3/ PERRY {1}: 479
\4/ In an explicit reference to Guise Theory he wrote: “. . . a theory of propositions of limited
accessibility seems acceptable, even attractive, to some philosophers. Its acceptability or attractiveness will depend on other parts of one’s metaphysics; if one finds plausible reasons elsewhere
for believing in a universe that has, in addition to our common world, myriads of private perspectives, the idea of propositions of limited accessibility will fit right in. I have no knockdown
argument against such propositions, or the metaphysical schemes that find room for them. But
I believe only in a common actual world” (see PERRY {2}, p.16.
\5/ KAPLAN {5}: 245
\6/ See KAPLAN {5}, where the method of $entences is brought to bear on a host of issues
stemming from intensional contexts. In turn this method is a development coming from Kaplan’s notion of proposition, for him one should not think of “propositions as set of possible
worlds, but rather as structured entities looking something like the sentences which express
them” (KAPLAN {2}: 17, not too similar though, see his caveat in footnote #41 of KAPLAN
{5}: 281). Valuated sentences ($entences “are virtually the singular propositions they express.
They give us structure. They give us individuals. They bear truth (with respect to a language)”
see KAPLAN {5}: 245).
\7/ “My semantical theory is a theory of word meaning, not speaker’s meaning. It is based on
linguistic rules known, implicitly or explicitly, by all competent users of the language” writes for
instance Kaplan, see KAPLAN {2}: 101, note #4.
\8/ This was Gareth Evans’ position, see EVANS {2}: 314.
\9/ KAPLAN {2}: 21-26. Eternal sentences, in the Quinean sense, are good examples of expressions with a fixed character. While their truth value may be different in diverse circumstances of
95
evaluation, their content (as Kaplan puts it what is said) does not change at all no matter which
utterance comes to express them (and no matter who is the speaker, which is the “possible”
world in which the utterance takes place, and so forth).
\10/ KAPLAN {2}: 64. That the conclusion might be that semantics is not about psychological
attitudes is suggested by Howard Wettstein. He takes the position that semantics for a language
has to zero in to language as it is used in public: a semanticist should be looking at linguistic
practices as an alien visitor would, namely as instances of noise/inscription production with some
degree of regularity. Whether or not his claims about the messiness and intrinsic vagueness of
our use of proper names are correct, I have my doubts that his remarks apply naturally to our use
of indexicals. Part of the problem is that a truly alien visitor, doing anthropological semantics,
would have nearly unsurmountable difficulties (if his, the alien’s, experiential field is markedly
different from humans’) in understanding demonstrative sentences: it is not obvious to me that,
for instance, he could have any grasp of tokens of ‘That is a table’ or, even worse, of ‘This is
a mistake’. See WETTSTEIN {2} and {3}; and for a response from the perspective of K-D-R
theorists who want to hold on to the cognitive significance of language see PERRY {4}.
\11/ This is a tension Kaplan is well aware of: “Frege’s incredulity surely stems from the point
that for an ‘object of thought’ to be an object of thought, all of its parts have to be thinkable.
According to Frege, material objects are not, in this sense, thinkable. They are presented to us
only indirectly, being represented by some concept. It is these representations that are to be parts
of an object of thought. There is an asymmetry in intelligibility here: one which I have observed
in myself and others. From Russell’s point of view, Frege’s theory looks like a subtheory of his
own in which the singular propositions are excluded, . . . From Frege’s point of view, Russell’s
way of ‘extending’ his (Frege’s) idea is utterly baffling because it seems to miss the point (as well
as the method) of the whole enterprise.” see KAPLAN {5}: 280, footnote #34.
\12/ KAPLAN {2}: 63-64
\13/ Interestingly enough, John Perry in a recent paper says that what is said is a “rather complex notion that needs to be explained in terms of intentions to communicate”, see PERRY {4}:
17, footnote #10.
\14/ A similar objection is pressed in GALE, even though I think he did not take into consideration the possibility of a Kaplanian “more structure” move, namely that a singular proposition
may be conceived as a structured entity – as opposed to a simple sequence – somehow mimicking
the grammatical structure of the sentence expressing it. I pursue this counterobjection in what
follows in the main text.
\15/ Rules regulating communication have, I think, to be more complicated than the simple
one just given because there is always an interplay between the context of an utterance, and
the “correct” way of understanding it. It is mistaken to limit one’s scope of interpretation to
96
isolated utterances: the context, linguistic and otherwise, can radically change what is said by
the utterance. Typically this is the case in every non literal use of terms, even pronouns like ‘I’.
There are obvious cases in which ‘I’ does not refer to the utterer in any straightforward fashion.
For instance in: “Suppose I am Mozart . . . and we are in Prague . . . and . . . I am stuck with
debts . . . ”. It seems to me clear that the token of ‘I’ in the above stretch of discourse has to
be understood not so much as myself, tokening it, but as myself under the description ‘Mozart’
commanding the entire piece, whichever the correct reading of myself-qua-Mozart will turn out to
be. On the difficulties of fixing, once and for all, referential constraints for pronouns see BACH,
chapter 9. On uses of ‘I’ non denoting the speaker see also CASTAÑEDA {1} and his “Reply to
Perry” in TOMBERLIN.
\16/ That this is the trait of indexicality philosophically most relevant has been emphasized by
Castañeda in all his work on the topic.
\17/ See LURIA, in particular chapter 4, aptly titled “His world”. nl \18/ LURIA: 77 (record of
August 1934).
\19/ The topic is huge. At least according to some linguistic traces are all we are left with in the
case of unconscious experience. The point is that the resurfacing of what has been submerged
can be accomplished only by a (psycho)analytical interpretation which is operating only on the
expressed thoughts of the subject during the process of interpretation itself.
\20/ WETTSTEIN {1}
\21/ Answering machines, and any form of delayed discourse, bring about strange problems for
theories of reference. An alternative approach could be to treat the ‘I’ as a standard pure indexical, in Kaplan’s sense, and then treat ‘here’ and ‘now’ as covert demonstratives, pointing
to the place and the interval within which the tape is producing the right kind of noises. Still
the ‘I’ would be difficult to interpret as ‘the agent of the context’ (which context? the time the
recording was made? each time the tape plays itself aloud? in the first case we would get Howard
as referent, in the second case . . . the tape – are tapes agents?). It seems to me more natural
to understand such a message by referring back to the intention of Howard’s (admittedly, this
is guesswork on the side of the caller: Howard might be dead, we might have got the wrong
number, and . . . ). That K-D-R theories are not successful in excluding categorically speakers’
intention is also pointed out in BACH.
\22/ Thus in ‘He is a great swimmer’, the referential constraint on the token of ‘he’ would be
of the form ‘a certain male’. David does not say that a certain male is a great swimmer. But
the constraint helps the auditor in delimiting the zone of David’s experiential field most salient
w.r.t. that utterance. As Kent Bach noted, this kind of rules for referring are relative to languages: there are languages in which gender has nothing to do with the sex of the bearer of the
grammatically gendered expression. In German the sun is a grammatical female and the moon a
97
grammatical male, in Italian the opposite is the case. It should also be noted that the referential
constraints for ‘this’ and ‘that’ are limit cases, since they just indicate that the referent is within
the experiential field of the speaker construed as broadly as possible. The ‘as broadly as possible’
modifier is the source of the very many misunderstandings that are generated by demonstratives.
\23/ This is not a crucial point, I believe. Whatever the actual psychological events are they can
be represented within the theory by mathematical functions – see e.g. functions in economics
which serve well the purpose of the theory at hand, but hardly, if ever, are in the mind of the
consumer deciding what to buy.
On the differences between functions, their values within formal theories and thinking realities a
valuable discussion is the appendix to CASTAÑEDA { 8}.
\24/ PERRY {4}: 4. He continues to note that “Minds evolved in a very Strawsonian world,
where the ability to reidentify individuals, and to use information picked up in one encounter
to guide action in a later encounter is crucial. That we can describe minds by reference to the
individuals they have acquired information about, and that our concepts of belief and the other
attitudes embody such a way of describing minds, should not be expecially perplexing.”.
\25/ There are some difficulties here already: do we have actually ways to individuate things like
instants that do not depend on our perspective? I grant the point that instants are individuals,
and that we can refer to them by using, for example, ‘now’. I do not think we do in fact refer
to instants using ‘now’: I am more inclined to view those referents as vague intervals. I will
expound my own on the proper referent of ‘now’ elsewhere.
\26/ This seems to be Kaplan’s present position, if I understand correctly. See the passage cited
in note #6.
\27/ The refurbished version of the singular proposition approach has been dubbed ‘extrastructure’ by Scott Soames. I am here basically following his criticism. I am not convinced
by his move of declaring the matter altogether a subject for pragmatics because we disagree – I
believe – on the range of semantical phenomena.
\28/ JACKENDOFF {2}: 128. If I understand correctly Situations and Attitudes, Barwise and
Perry in keeping a semantical distinction between the veridical case (paradigmatically ’see that’)
and the non-veridical (paradigmatically ‘believe that’) are inclined to treat hallucinations as having as contents brain states and cerebral events. I am disinclined to take such a position because
I am more interested in keeping as unified as possible the understanding of the phenomenon of
reference to experienced contents.
\29/ I am not here talking about the ostensive demonstration of objects at all, rather the pointing
is the directedness of intentionality.
\30/ Jackendoff puts the same point in terms of mental representation: “People have things to
talk about only in virtue of having mentally represented them” JACKENDOFF, ibidem.
98
\31/ The example is from CASTAÑEDA {8}. The importance of demonstrative reference for
guiding action has been pointed out by many authors. It was Castañeda though that brought it
into focus as a motivational factor which does not depend upon external targets of reference for
demonstrative tokens.
\32/ I deal specifically with the first person pronoun, qua indexical expression, in another section.
\33/ I owe the idea of course to Donald Davidson. See his “On the Very Idea of a Conceptual
Scheme”, in DAVIDSON.
\34/ The spotlight on the importance of the attribution of indexical reference to others has been
first switched on in CASTAÑEDA {2}.
\35/ While already working on this piece I came to realize that within the K-D-R proponents
there exists a tendency to allow internal contents, which may or may not come to be (identified
with) identical massive chunks of reality. John Perry, in his most recent article, couches the point
within communication: “Suppose two children are looking out from different windows [of a car]
neither of them is paying any attention to that fact. The discussion goes something like this:
“ ‘That’s Wendy’s. Let’s stop here’ says one child, looking in one direction. ‘No, it’s not, you
idiot. Can’t you see that it is McDonald’s. Who wants to eat there?’, says the other looking
in the other direction. The first child use of ‘that’ and the second child use of ‘it’ are not coreferential -there is not some thing they both refer to. The use of ‘that’ refers to one restaurant,
the use of ‘it’ to another. But to understand the internal structure of the discourse, and the
emotions to which it gives rise, one must see that the various referring expressions are supposed
to be about the same thing. The utterances are not really or shall I say “externally” about the
same thing. But they are “internally” about the same thing. That is, the utterances bear the
relationship that is appropriate in discourse, for utterances that are really about the same thing”
(see PERRY {4}: 13). This is a refreshing example because for once the internal content is one,
and the external contents are more than one (the reverse of many Frege-style puzzles): I would
go a step further in recognizing that the two children have certain internal propositional contents.
And I would also read the example as evidence that what is exchanged in communication is not
just information (in the sense of Situations and Attitudes) but also internal contents.
\36/ See his “The Semiotic Profile of Indexical (Experiential) Reference”, in SAARINEN, p. 281.
\37/ ibidem, pp. 311-312.
\38/ See CASTAÑEDA {6}: 311 –but see also CASTAÑEDA {4}.
\39/ ibidem, p. 287 and pp.288-289 for an argument about the impossibility for a visual space
to be (identical to) a physical space. Note: the key premise of the argument is empirical, namely
light moves at a finite speed, hence whatever is in the visual field of an observer at one time
needn’t correspond to anything in the physical space at the same time. As an empirical point I
am not quite sure this is in general the case for all perceptual modalities: is there an olfactory
99
space? and what distinction can one make between it and its physical “counterparts”?
\40/ See in particular the seventh section of CASTAÑEDA {8}.
\41/ See “Reply to Perry”, in TOMBERLIN, p. 318.
\42/ ibidem, p. 322.
\43/ And Frege’s for that matter.
\44/ See in particular the letters contained in the ADAMS-CASTAÑEDA Correspondence (in
TOMBERLIN) where the change has been made explicit and clarified.
\45/ See in particular Castañeda’s “Reply to Perry” in TOMBERLIN.
100
CHAPTER 5
I (and some speculations)
Consciousness in its deepest home seems to oscillate slowly, will-lessly, and reversibly between stillness and sensation. And it seems that only the status of
sensation allows the initial phenomenon of the said transition. This initial phenomenon is a move of time. By a move of time a present sensation gives way to
another present sensation in such a way that consciousness retains the former
one as a past sensation, and moreover, through this distinction between present
and past, recedes from both and from stillness, and becomes mind.
As mind it takes function of a subject experiencing the present as well
as the past sensation as object. And by reiteration of this two-ity phenomenon,
the object can extend to a world of sensations of a motley plurality.
L.E.J. Brouwer∗
OPACITY
Opacity is the best word I find to name the condition I am engulfed in trying to speak
of myself and of my self. Reasons are there for it to be the clearest area: who is more entitled
than I am to say something of myself?
Introspection reveals only a bottomless crevasse. The very standards of evidence are in
dispute: many are willing to claim that introspective claims are baseless, nearly by definition for
there is nothing to introspect in the first place.
The bet of my semantics of thinking is twofold: since language has to be thinkingly
spoken before it can be understood, language is at least one of the external channels of thought;
and whatever can be said of reality has to be said by examining our thinking reality, with the
participle left ambiguous between its two natural readings.
Our indexical thinking has two key features: it can focus on anything demonstratively
and it is reflexive in a way unique to first person thoughts. From now on I use “thoughts” as
102
short for acts of thought, states of mind, entities individuated by their contents (sometimes called
“texts”), “utteranc e” is reserved to public events and sentences will be used as expressions of
contents of both. This is not meant to suggest that thoughts are of their essence, so to speak,
linguistic in character: I am making the assumption only for simplicity, in fact I think it is very
clear there are thinkable contents which are not linguistic at all1 .
I will deal primarily with two problems which are related in my view
i) ‘I’ when actually used in speech and thought marks a reference made by the user
to the user as self (there isn’t any replacement of the first person pronoun which
preserves all its valences). What are the structures of this unbreakable referential
link?
ii) Demonstrative reference points to a key part of our consciousness: what are its
link with our self-consciousness, somehow encased in first person thoughts?
¿From David Kaplan2 , I borrow the distinction pure indexicals and demonstrative indexicals. Both categories of expression refer within the semantics of understanding language
when utterances are anchored to a context of use. I use semantics of understanding to mark the
difference between the hearer’s semantical valuations of utterances and the thinking/speaking
semantics for the intentional attitudes which are characteristic of reference from the viewpoint
of the speaker who is making the reference.
In Kaplan’s terminology, pure indexicals do not have at all failures of reference: ‘now’,
‘here’, ‘I’ are always referring (in a context of use3 ).
On the contrary demonstratives may or may not have referents: one example of failure
103
was that of his own pointing to a wall where a picture of Carnap use to hang, but not any longer,
and he is giving his shoulders to the wall while uttering “That was the greatest philosopher of
the century”. To use a simpler example, the use of “he is good looking” in the context in which
the utterer points unmistakably to a flower is for Kaplan (and for many others) a failure with
regards to the reference of the indexical token in the sentence. I will argue that even though
this might be a impossible speech act to understand for an audience (from the point of view of
what I have termed valuations of sentences made by hearers) it does not constitute in this view
a failure of thinking reference for the speaker.
The key prediction of this view of demonstrative reference is that every natural language
has some device of demonstrative reference with a wide range of variations in lexicon. To use a
simple example at my disposal, the couple of demonstrative pronouns ‘this’ and ‘that’ in English
has to be translated with three distinct expressions in Italian. However the basic mechanism of
reference remains the same.
Demonstrative indexical reference is a symptom of the intrusion of experience in language: it is an experiencing agent who is the user of this kind of reference. It is also an agent
which has a particular kind of experience. It can not be the experience of the world sub specie
aeternitatis, since demonstrative reference isolates particulars, whose boundaries are very much
conventional4 , and those particulars come and go from one’s field of awareness. Experience it is
nonetheless, no matter how much straying away from our common “world”.
104
”. . . what can be expressed, if the intrinsic content of experience is beyond all
descriptions? What remains over, if all experienced qualities, colours, sounds,
feelings, in short all determinations of the content of the stream of consciousness, are ruled out for communication as absolutely subjective and indescribable? We might think at first that nothing whatever remains over, since we
are assuredly unable to free our experiences and thoughts entirely from all content. Or are the relations between contents of consciousness something that is
removed from the subjective sphere and can be therefore be communicated?”
Moritz Schlick∗∗
A reformulated Fregean theory
If I am correct, there are acts of thoughts, states of mind which are subjective and
private. Strictly speaking, then, those are not objects of communication, if communication has
to involve the ability to prompt one’s audience to share refer ences to zero in on one and the
same thought expressed by a sentence uttered or written. This is one consequence of Frege’s
entertaining the possible existence of private sense of expressions5 . Frege, I believe, failed to
realize how far a theory of thoughts would have to go to countenance private senses.
The kernel of my proposal is to postulate as an explanatory tool the presence of an ‘I
think’ preceding every representation6 and test what this will bring about w.r.t. indexicality
within a broadly Fregean semantics7 . This is a theoretical posit, a way of linking indexical
thoughts and the perspectives from they which they are “executed”.
If indexical terms have, at least, one dimension of privacy, the sentences containing them
are bound to a speaker. Indexical referents may not be egocentric, as Russell thought, but rather
logocentric. Not egocentric because their use does not presuppose, though allows, the presence
of an Ego to consciousness: we can use ‘this’ fully lost in thought without any awareness of a self
105
of any kind, like Castañeda’s Externus.
Logocentric indexical terms have to be because their very existence does presuppose an
agent-speaker-thinker: the centre of an ongoing logos.
I propose to consider the possibility that the primary referents for indexical terms in
thought and in direct discourse are senses, “Sinne” along more or less Fregean lines.
The senses of indexicals are subjective: they are constructed entities bound by the
perspective of the agent/speaker/thinker (although in making quasi-indexical references we have
the means to “capture” at least their fundamental structure ). In and by themselves they are
private targets of reference.
Schematically: when the text of a thought (expressible in direct discourse) is f ree from
indexical terms the Fregean scheme of sense and reference stays in place. In this case we are
resorting to the Fregean common heritage of concepts and Sinne. Those are objective at least
enough to yield intersubjectivity of communicational content. To buttress the distinction between
general terms and perspectival, indexical ones, one may also observe that while in learning the
meaning of indexical terms one learns an abstract mechanism which, when put to use, can denote
on particular occasions, learning the meaning of general terms one learns also their reference, in
the vast majority of cases. It is an open question whether issues of “inscrutability of referent”
are resolved by innate mental structures or by what seems to be a fact of psychological reality;
namely that in learning language humans as a species are subject to an extremely vast, common
array of environmental features which prompts, given their common biological make-up, their
reactions in ways which are vastly similar. The Fregean “common heritage” of concepts and
106
meanings would, then, be common by way of a shared physical environment.
The terms I want to be affected directly by the agent/speaker/thinker’s perspective are
indexicals, pure and demonstrative alike.
There are two main reasons for this choice.
I have been stressing all along the subjective character of indexical reference. And that seems
to be a ground for indexicals being bound to a particular speaker, losing as it were the semantic
“innocence” which they appear to have. But there is also another reason, to me more compelling.
If there is a unique ephemeral character of indexical references, and if their referents are, as I
believe, internal to the fields of awareness of thinkers and speakers, then these particulars are
not ever publicly appearing. Their role is to provide a focus to speakers’ references, and to bring
about certain reactions on an audience. In the limit case of soliloquy, or of thought exclusively
entertained by a thinker, one may take the audience to be the thinking subject itself. The role
of indexical thoughts is causal in communication: indexical references, though private, serve the
purpose of indicating to an audience what should be focussed on.
Indexical references, in spite of the privacy I’m underlining, are expressible publicly: the
tokening of indexical terms signals a speech act which I see as mixed in character, part assertion,
part command.
The process of understanding indexical utterances is arduous: a piece of empirical evidence supporting that is the unclarity that indexical utterances suffer: those are the utterances
that are most sensitive to the hospitality of their context to be comprehensible at all, e.g. how
hard it has to be to communicate a sentence containing a demonstrative based on a visual pre107
sentation within the field of awareness of the speaker to someone else who’s blind.
If one takes what I have called the narrower view, changes have to be made in what gets
transmitted in communication, but not to the point of having to replace the ordinary notion of
objects wholesale.
An underlying motivation not to take the broader view is for me an attempt to see how far
one can go keeping one’s ontological commitments as conservative as possible in accommodating
the phenomena indicated by the occurrence of indexical references in thought and the ir use in
communication.
When the text of a thought contains indexicals the very capacity of thinking it requires
the employment of the perspective from which the indexical is tokened. The switch I am proposing
will take as referent of the indexical its sense. Now I am using ‘sense’ as a combination of two
elements: the object internal to the field of awareness of the thinker/speaker and its manner
of presentation. That ‘sense’ is an object qua present in the field of awareness. Its manner of
presentation might be taken as a descriptive component which the speaker might be able, if
so queried, to supply. This descriptive component might be very poor and again might have
irreducible indexical elements itself. A table might indicate a rough division of indexicals in
terms of their senses. Indexical senses have polarities, they are expressive of the isolation of
certain particulars, hence their meanings come into their sense by way of determining differences
in presentation: what can be now a ‘this’ is soon to be a ‘that’, not necessarily in that order,
whereas what is called now ‘now’ is forever afterwards to be called ‘then’.
108
that in opposition to ‘this’, “entificator” par excellence, the less obvious of the items present
in the speaker/thinker’s field of awareness, its mode of presentation is highly nonspecific,
possibly perceptual or anaphorical with respect to a stretch of discourse; very often its
reference is determinable only contextually, causally inducing the right kind of response
this similar to ‘that’ in being highly nonspecific but its grammatical meaning points to the
most obvious item in the field of awareness, obvious in terms of relative spatial proximity,
or of temporal, sequential order
now the adverbial marker (implicitly present in any unmarked sentence) of a fuzzy interval,
whose upper and lower limits are eternity or timelessness and instants at which thoughts
take place. mostly to be placed in opposition to ‘then’, because of the sequential character
of thought (see more below for some remarks on the absolute certainty of the existence of
a referent for ‘now’)
here the marker of a spatial portion in general, at a more metaphorical level a portion of
discourse as well (as in ‘Now; we are talking!’) ; to be placed in a sort of duality with ‘now’
in expressing the boundedness of the agent of the speech-thought-act
I the most basic and most primitive of all indexical senses (not to be confused with the one
indexical term to which all others can be “reduced”). Expressive of agenthood, key opposition with all others indexical terms: at each occasion of use it has a strictly unique referent,
always present by the very force of the agent performing an indexical speech-act in first person, thinking a first person thought and so forth. Explicitly conceived, essential ingredient
109
of the expression of selfconsciousness (on which some more below)-. In polar opposition to
all other indexical terms which are expressible by personal pronouns. The use of ‘you’ constitutes the possibly highly metaphorical attribution of an Ego to a conversant (as in ‘Now,
you let me down!, uttered by my to the computer which is misbehaving). When I term ‘I’
the most basic indexical, the claim has not to be understood as saying that ‘I’ is genetically
prior to other indexicals, or even somehow grounding them. “Externus”-type episodes of
consciousness may and do take place without an‘I’. ‘I’ is basic just in the pedantic sense it
is the base on which you can attribute self-consciousness to others, hearing them using it.
This is not an exhaustive list, and I am not sure there is in fact such a list, in part because the
phenomenon of indexicality is more widespread than for the extent of indexical terms, tenses
in verbs (for the languages which allow them) are indexical markers as well, but they aren’t
necessarily to be reduced to ‘now’ and ‘then’, whereas languages such as English which allow
the construction of complex indexical terms have a potentially infinite repertoire of secondary
indexical terms; e.g. ‘to-day’, ‘to-morrow’, ‘yester-year’.
110
Now, this would hold for each speaker. Hence there remains the need of providing an
account of how this kind of reference can be transmitted in communication. I say transmitted
and not shared, because indexical references are not directly shareable in my view.
One initial element has to be noted: indexicals, on the surface of speech, are always
indicating references made by the speaker even in sentences construed as reports in indirect
discourse. I take this to be a bit of evidence favoring the view that indexical references are not
common in the same way others are.
I try to illustrate the view using an example involving a demonstrative pronoun.
Suppose a speaker says:
(L) That is a lime tree
(L) is bound by the ‘I think’ of a speaker, I claim, and the indexical term in it takes as primary
referents its sense. The referent of the token of ‘that’ is the presented object targeted by the
speaker within her field of awareness, and the referent of ‘lime tree’ can be taken as the descriptive
component of the sense ‘lime tree’ the speaker has in mind. What are the truth conditions of
(L)? One way to characterize them is in terms of “fit” between the descriptive component of the
sense of ‘that’ and the characteristic properties of the sense of ‘lime tree’, or at least “enough of
them”.
A descriptive component of ‘that’ in (L) can be seen as a series of properties, doxastically
attributed to it. If enough of them are also in the series of properties characterizing the stereotype
of ‘lime-tree’ (for instance ‘being a tree’, ‘being a tree with green fruits’ etc.), then we may talk
of a “fit” between the sense of ‘that’ and the sense of ‘lime-tree’. The “enough” caveat is here
111
by and large because the salient features of ‘lime-tree’ are socially determined, so that what can
count as “fit” for a botanist is generally very different from what will count for the layman in
botany. I have indicated above that I propose to have senses of indexicals as their referents, but
I have not proposed to eliminate outright the terms’ Fregean Bedeutungen. Which role Fregean
Bedeutungen play? In thought they remain intentional objects, something we do not have as
propositional components. But they retain a role in communication as the external anchoring,
the background against which the directing intentions of the speaker can be fulfilled. I believe
that role is best understood shifting the focus to communication.
Before we discuss this I think a summary is not amiss.
In terms of referents, my view is that the referents of indexicals are senses, in a quasiFregean acception. They are not objective in Frege’s terminology, they are subjective and “private”. I put private in quote because although there is one dimension of privacy to them, they
are communicable. In communication, in the interplay between the semantics of thinking, on the
speaker’s side, and the semantics of understanding on the hearers’ side, Fregean Bedeutungen
play a central role. One of my criticisms of Frege’s original approach lies in that he collapsed
the two aspects of thought: the mental act with a content and the communication of such a content to others. In a slogan like fashion Frege thought that the only guarantee of intersubjective
communication is the existence of timeless, spaceless pseudo-Platonic entities he thought Sinne
were.
An indexical judgment when true or false constitutes in my view a form of “mixed”
predication. On the one hand the indexical referent (which might be in subject position as in
112
(L)) is a private object, whereas in the predicate position is a Fregean concept-word or some such.
The proper fit of these two kinds of entities (which makes the judgment itself true or false, as the
case might be) requires a different theory of predication. One option is to be excluded, it seems
to me, the relationship between ‘that’ and ‘is-a-lime-tree’ in L is not a relation of identity, of full,
unrestricted identity obeying to Leibnitzian strictures. On the other hand the conception of such
a relationship along the lines of a standard extensional membership (L is true iff the referent of
‘that’, on the given occasion of use is a member of the set determined by the property ‘being a
lime-tree’) would bring about a clash of intuitions about what is predicable of what, since on the
one side is a “mental” object of sorts, and on the other is a purely public (ontologically speaking)
property.
As Professors Clark and Castañeda have made me realize, the construction of a theory
of predication in mixed form, allowing the compresence of a mental object and a “non-mental”
property in one and the same judgment is what I see as the most urgent task in pursuing the theory
of indexicality which seems to save their privacy and their functional role in communication.
As I have noticed above one such theory of predication is part and parcel of Guise
Theory, where though the relationship to be split is the relation of identity, weakening the full
Leibnitzian one to a potentially growing list of different relationship which picture different kinds
of judgments (of consubstantiation, of consociation and so forth). But the theoretical price of
Guise theory is to relegate Fregean referents to the realm of doxastics, a price I am not willin g
to pay, or, at least, a price which does seem to high for the bill presented by indexicality. The
recognition of the truth of (L) performed by the thinker or speaker of it is not a matter of logical
113
deduction, independent from experience. It is a recognition prompted by a reference to an object
of experience. Indeed the very proposition the judgment expresses is a product of experience:
it is not a Platonic proposition, in Chisholm’s acception, but a contingent one, created by the
episode of thought generating it. The “facts of the matter” contribute to the truth-conditions
of an indexical judgment by providing (or failing to provide as the case might be) the grounding
of the senses of general, non-indexical, terms. The senses of general, non indexical, terms have
a double status: they are objective and subjective, in virtue of their intersubjectivity. In a way,
then, they are within and without minds, if one grants that their very acquisition is constrained
enough.
Which role are Fregean Bedeutungen playing? That role, I think, is best understood
shifting the focus to the scene of communication. ¿From the viewpoint of the speaker an indexical
sentence such as (L) expresses a quasi-Fregean thought. The thought lacks the kind of objectivity
attributed by Frege to Gedanke: it is bound, so to speak, to the spe aker’s field of awareness since
the component of the thought referred to by the indexical grammatically in subject position is a
private sense.
Now, if (L) is actually uttered, at the limit even in soliloquy, a further dimension has
to be added. Indexicals’ grammatical meanings are built-in directing intentions. One way to
understand these, the one I prefer, is to see them as pointers given to an audience8 . Some of the
pointers might be demonstrations, though there does not have to be any. Some of the directing
intentions are hinted at by referential constraints: using ‘she’, in E nglish, the speaker tries to
direct the hearer to a female, using ‘he’ to a male, using ‘that’ to a salient item in the hearer’s
114
field of awareness. ‘that’ and ‘this’, it might be noted, has one of the lowest level of directing
specificity: it points to an item which is somewhat more salient and/or closer in time or in space
than a ‘this’.
If I am correct, we are often unconsciously misrepresenting our references. At the simplest
level the speaker/thinker of (L) is understood, quite likely by himself as well by others, to refer
to a tree by means of ‘that’. That he is referring to an item within her field of awareness does not
have to be something of which she is conscious: the very construction of a field of awareness is
an unconscious process9 . Note too that when the covert ‘I think’ is actually verbalized it carries
a connotation of agnosticism about the correspondence between what one is aware of and the
environment.
What, then, is communicated by (L)? Not the speaker’s thought, but rather an abstract
propositional structure, not a Kaplanian singular proposition at all10 . In the case of (L) the
propositional structure expresses the inclusion of the referent of ‘that’ in the category ‘lime tree’.
A hearer of the utterance of (L) to go from this communicated structure to the formation of her
thought has to use her understanding of the clues given by the speaker, namely the interpretation
of the directing intentions and her knowledge, if any, of the sense attached to ‘lime tree’.
I’d like to note that a thought such as (L), mental entity as it is, is not an incorrigible
statement (if there are any). The Fregean Bedeutung of the token of ‘that’, what I think is only
an intentional object in thinking, has not vanished from the scene of communication. The hearer
might very well interpret the directing intentions of the speaker in such a way that she takes
‘that’ to be what she sees as a cardboard replica of a lime tree and reply with ‘no, that is a
115
cardboard sculpture mimicking a lime tree’. And the speaker, revising her statement (L) can
come to recognize ‘oh, what I thought to be a lime tree is actually a mock-up of a lime tree’. We
might understand the revision implying:
(L.rev.) ‘that, which I thought was a lime tree, is a mock-up’
What are the relationships between the referent of the token of ‘that’ in (L) and the ‘that’ in
(L.rev.)? I do not think it is reasonable to say they are one and the same. There is one level
of analysis (the level I have called semantics of thinking, following Castañeda) in which the two
tokens refer to the same presented object within the field of awareness of the speaker of both
sentences. For otherwise it will not make too much sense even for her to see (L.rev.) as a revision
of (L). But the intended external Fregean referents are not two particulars identical to each
other. Identity is too strong a notion: if one accepts Leibniz’s interpretation of identity, identical
particulars share all properties they have. And at least one property is not shared, namely being
a mock-up. I take it ‘being a mock-up of a lime tree’ entails ‘being not a lime tree’.
One way to characterize the relationship between the two senses is to view the speaker
as indicating an intentional similarity between them.
Intentional in two ways:
i) The speaker asserts of them a similarity, strong enough to revise (L); and
both are intended to be about a Fregean Bedeutung, an object, in this case the lime tree. It
gives the necessary background of the directing intentions of the speaker and the potential target
of the mobilization of the referring intentions of the hearer. It is in the interaction between the
referring intentions of speaker and hearer that a complete judgment such as (L) can come to be
116
revised. And this can happen also in the limit case of soliloquy, for there is no reason not to
suppose that I can see my thoughts as uttered by another. In this way, what Kaplan takes as a
failure of reference, I see as a split phenomenon. ¿From the standpoint of the speaker there is
no referential failure for demonstrative indexicals. What might “misfire” is the mobilization of
the referring attitudes of the hearer(s): they might be unable to interpret the speaker’s directing
intentions, due to the unavailability of any background to the identification within their field of
awareness of items displaying the right fit between an external targeting and an accordance with
their understanding of the speech-act performed by the speaker.
117
Some more content-less structures
Quite differently from Frege11 , the indexical propositions I am envisaging are actual
creations of the speaker/thinker. They are not at all timeless, mind-independent objects. Thus
their existence is contingent upon the existence of the speaker in the first place. Such a creation
though brings into existence a web of structures more complex than it would appear at first
blush.
Let’s suppose, then, (L) comes to be uttered by a speaker and is heard (and understood)
by an audience which may, although need not, consist of several different hearers.
Each one of the hearers can formulate two thoughts:
(A.L) That is a lime tree
and
(Ind.A.L) She says (that) that is a lime tree
According to Castañeda and me, both tokens of ‘that’ in (A.L) and (Ind.A.L) have subjective,
perspectival senses12 . Their contribution to the contents of thought is not the same as the one
made by ‘that’ in (L). And it shouldn’t, because to target the lime tree each hearer has to use her
own sense of ‘that’ in this occasion: different thinkers, different speakers, different perspectives,
hence different presented objects. Spatially and temporally there is an asymmetry between the
perspective from which (L) has been thought of, and (Ind.A.L.) the perspective(s) of (A.L.) and
(Ind.A.L.) Note: in (Ind.A.L) the demonstrative token of ‘that’ takes the “largest scope”, i.e. it
is a token of the thinker of (Ind.A.L), not the ‘that’ of (L). Demonstratives keep, so to speak,
118
their sense bound to the actual user.
But then what is it (Ind.A.L.) attributes to the utterer of (L)?
My proposal is to view that attribution in a twofold manner. On the one hand there
is an indexical proposition entertained by the hearer. This is subjective in the sense specified
by the hearer perspective. On the other hand this proposition comes with a potentially infinite
“cohort” of other propositions. The members of this cohort are propositions which may be the
content of other references made from different perspectives.
In case of successful communicational speaker reference, the success consists in eliciting
in one’s audience the entertaining of one proposition which is in the cohort of the proposition
one wishes to communicate.
Strictly speaking, then, the content of communication and the content of thought are
not (in the indexical case) identical. There is a specific feature of indexical reference which has
to do with the need of causing in the audience’s thought the proposition which has in its cohort
the perspectival proposition entertained by the speaker.
In the case of successful communication I have been using a tacit assumption, namely
that the audience can pick up through its own directing intentions the proper indexical or quasiindexical proposition expressed by (A.L) or (Ind.A.L) respectively.
As I have remarked before, one should not focus exclusively on successful communication.
Whether I am able in fact to prompt the entertaining the “right” proposition or not, it remains
that in thinking I am entertaining an object of thought. Let’s go back to (L). Suppose though
that there is no lime tree, due to hallucinations, my mistaking a bush for a lime tree or whatever
119
other situation would make (L) non veridical. The veridicality of (L) consists in there being an x,
which is the Fregean denotatum of token of ‘that’ and which is identical to a lime tree. If (L) is
non veridical, and by hypothesis it is, I still want to hold the tokening of ‘that’ to be referential.
First of all, there is a datum which I would call, following Castañeda, homogeneity of
thought: from the perspective of the thinker there is no difference between (L) when thought in
the presence of the tree and in the presence of nothing. If it were not so, hallucinations (and
dreams) would not be understandable. There is a level of cognitive significance which is carried
by thought contents alone: we do have reactions of fear, of excitation, and so forth in dreams and
hallucinations as much as in experiences of purely external objects. And as Castañeda argues,
there is also a level of thinking in which we have to take contents of thought to be truth-valued,
at least13 .
This homogeneous character of thought is widespread. We do refer to fictional entities,
whether knowing they are fictional or not. And the case of fictions is interesting because we
do often refer to fictional entities not from within the fiction in which they are embedded, but
from without. Typically this is the case when one wishes to express something like “Diotima is
a woman who is very seductive”.14
As noted before we can refer demonstratively to items in our field of awareness which are
not present and attended at at the time the reference is made. Our field of awareness does include
conscious memories and lots of artifacts of thought, artifacts like fictional characters which have
as background only other objects of thought.
There is a distinction to be kept in mind. Communicationally to prompt in one’s audience
120
the production of what I have called a proposition in the right “cohort” is much more difficult.
One’s directing intentions are much harder to figure out. As I see it the case of demonstrative
communicational reference presents a continuum of cases: from the successful reference made
by the speaker in a perceptual judgment and so well understood by the audience t hat they
are able to form the perspectival proposition which converges exactly on the external target of
the speaker’s reference, to the case of moot or only purported communication, in which one’s
audience is left completely in the dark as of the directing intentions of the speaker. The latter is
probably very often the situation we encounter when an hallucination takes hold of the mind of
the speaker.
What looks like a failure of reference is then a pragmatic failure: the speaker is unable
to cause the right kind of directing intentions in her audience. This shows neither that a target
of reference does not exist, nor that it does. A pragmatic failure of this kind can be brought
about by the difficulty for the audience of interpreting an utterance. In such a case most likely
(Ind.A.L) would be meant by a member of the audience to be a quote, possibly expressed in
speech by adding a shrug.
I’ll call the indexical propositions with their whole cohort of indexical propositions others
can use to target the same referent, successfully or unsuccessfully, imperial propositions. They
bring about a different adjustment in Frege’s notion of thought . The centre of the empire, its
emperor – my indexical proposition – is available to me directly and to others vicariously through
their own targeting one of its attached members of the cohort.
Now, the indexical proposition is contingent upon the existence of its creating mind. But
121
there is a sense in which it can be attributed independently even from the existence of a creating
mind.
Louis Antoine Leon de Saint-Just, before being executed, thought that he was immortal
(in one very idiosyncratic sense). Since he wrote thus, it is a reasonable presumption to think
that he entertained in thought:
I am immortal
and we can attribute to him that proposition, expressing the attribution through
Saint-Just thought he was immortal
The token of ‘he’ here is not a demonstrative pronoun but the quasi-indicator ‘he-himself 15 .
The problem here can be put thus: given what Saint-Just thought, we know beforehand
there is an imperial proposition we are capturing16 . But there is no guarantee that this is the
case. For we might by same token attribute a first person proposition in this way:
The first woman born after the implosion of the universe will think she is pretty
And I take it to be clear that the alleged first person proposition ‘I am pretty’ might never
come to be. First person proposition are contingent upon the existence of a thinker who indeed
thinks them. And in this situation that thinker might simply never exist, a fortiori her indexical
propositions do not exist either.
Therefore the more serious revision for Frege’s scheme has to allow what he calls the
“third realm of thoughts” to be a realm of thinkable thoughts, some of them occurring at one time
or another, some of them never occurring. Note that this has nothing to do with the “grasping”
122
of a thought: if there is (tenselessly) no indexical proposition, the quasi-indexical thoughts which
purport to report it are, at best, what Frege would term “fictional thinking”.17 The difference
between the purported targeting of an indexical proposition via a quasi-indexical sentence and
sentences like (Ind.A.L) lies in the lack of directing intentions. When a demonstrative sentence
is uttered the creation of the indexical proposition expressed by it is the creation of an imperial
proposition, in my terminology. When a quasi indexical proposition is used to attribute to others
an indexical reference no indication is given as to how others could make the attributed reference.
This principle underlies the difference between the direct character of indexical reference to an
item in a field of awareness which is perceptual (or present to the attention of the speaker) and
the reference made in indirect discourse such as the (Ind.A.L) to the content attributed. In the
quasi indexical case we are making a reference to a sense of a sentence. Disagreeing with Frege I
am pointing out how there might be thinkable sense associated with the content sentence while
there is no indexical proposition matching it.
Hence a further distinction has to be introduced in the Fregean frame. In his theory the
content sentence of a clause in indirect discourse does not refer to its ordinary denotatum, but
to its sense (a thought). The introduction of indexical propositions forces to distinguish between
three different situations: a report sentence whose content sentence is indexical free, a report
sentence whose content is indexical, and, a report sentence whose content is quasi indexical. For
the first case there is no need to modify the framework. For the case in which the content sentence
is indexical, the proposition or the sense it expresses is speaker’s sense of it. That it is so is shown
by the phenomenon of the “largest scope” of indexicals: no matter how deeply embedded they
123
express a speaker thinking reference. In the third case the content sentence has a quasi indexical
character: this purports to attribute an indexical proposition to another thinker (who might be
the speaker herself 18 ). In the quasi-indexical case the very attribution is making as assumption.
Namely it is assuming the subject to which the proposition is attributed to be indeed capable of
indexical thinking and referring, possibly of first person referring. This assumption can be highly
metaphorical as in the recent phenomenon of attributing to a computer faced with a mistaken
command “He believed that he was asked nonsense and replied with a stark ‘syntax error!”’. The
attribution may be null and void when the indexical proposition attributed is not thought by the
thinker to whom it has been attributed at all. In my idiom, we may be targeting a member of a
cohort without an emperor, as in the case of the first woman after the implosion of the universe.
The latter is the case I find more interesting: when one and only one subject is considered the phenomenon of attributed indexical propositions raises issues of continuity of selfconsciousness. Attributions are all self-attributions, with an added apparent dimension of certainty in targeting. And whether there is or not continuity of self-consciousness seems to me very
mysterious.
To that mystery, now I turn19 .
124
What is a self? How is a self possible? A self stands in a peculiarly intimate
relationship to itself, being aware or conscious of itself, of itself as itself –or, at
least, it has the capacity for this. How is such knowledge possible, is the capacity
for it part of the essence of selves, an essential feature of what is a self?
Robert Nozick∗∗∗
Awareness and self-awareness
Demonstratives, I have stressed, do not fail to refer because their referents are within
our fields of awareness. That we have a field of awareness, a motley plurality of experiences, I
take for granted.
That we have awareness of ourselves is not to be taken for granted. I begin with one
example.
Suppose we observe a child, maybe mumbling to herself, or thinking out loud for our purpose:
that is a toy, a pretty one, and that is a store, and there now they sell toys. . .
Our child is named Kathleen, and her mumbling goes on:
that is the toy Kathleen wants, and that is why Kathleen will go there, to that store,
with this bunch of coins. . .
So far all she thought can be captured by a description of her field of awareness. In it we find
stores, that one store where they sell that one toy she wants. We might even find Kathleen
herself. Within her linguistic awareness is not only the expression ‘Kathleen’ (if she responds,
however mechanically to it, it has got to be stored somewhat in her mental repertoire of terms),
but there might also be what we refer to by using it, Kathleen herself, of course, our familiar
125
Kathleen20 .
What we do not find is Kathleen referring to herself as herself 21 . Note that it is a fact
that we do have self-reference, but it is not very obvious what “survival value” it has, so to speak.
Kathleen is perfectly capable of surviving, in fact she engages in actions such as observing her
prized toy store, she is able to reason in rather sophisticated manners (she understands that toy
stores will exchange toys for coins). On the face of it, from what we can infer from her thoughts,
she lacks reflective awareness of herself.
126
A genetic account: Theory I
Kathleen notices first that one area of her field of awareness is singled out as the easiest
to affect causally: her body. Her pains are in her field of awareness but she can do nothing about
them. Her hands are there as well and she can move them: in fact there is no phenomenological
distinction from her standpoint between the motion of her hands and the motions of mine. She
can move hers at will, no matter what she thinks she does not move mine. She feels her pains,
but she does not have any similar sensation with respect to anything else. But for all she knows
felt pains might have their seat in the universe at large she faces. Segments of that universe do
not respond to her commands though, by conscious will she has no telekinetic causal powers22 .
Theory I consists in the following hypothesis: the mechanism which triggers the birth
of self-awareness is a psychological mechanism which recognizes certain boundaries within the
subject’s field of awareness. Those boundaries are determined by the reach of the subject’s causal
powers.
At this stage there is no referent of ‘I’: the first person pronoun is not used at all, even
though someone like Kathleen might be able to refer to herself in a third-person fashion as in
“that is the toy Kathleen wants”.
I believe this hypothesis to be somewhat confirmed by the observation of children. It
seems to an established fact that very young children do not express anything like self-awareness
for years, after they have acquired expressive capacities with language which indicate an ability
to refer even to themselves. The linguistic phenomenon I have most prominently in mind consists
in children learning their (given) name and use it to refer to themselves in third-person mode,
127
as it were, before they start using the first person pronoun and the first person inflections of the
verbs in sentences.
128
Theory I is not satisfactory: Theory II
Theory I suggests a close connection between certain physical powers and the referent
of ‘I’.
Can we move ahead and identify those powers with it?
A first move, intuitively plausible, is to answer positively and view the referent of a
tokened ‘I’ as the holder of those powers we experience. The holder is a body: that one body we
learn to single out as a body we can most intimately affect.
We have reached one provisional result: the referent of ‘I’ is a body cum its causal
powers. But this is not enough vis-a-vis the cache of the data we have available from our
thinking activities.
“Cartesian” sentences show a residual which is not captured by the reference to our
body. Thinking is a multilayered phenomenon: we can step back, as it were – no motion of
physical bodies is involved at all –, and look from another first person perspective at our own
body. Brouwer’s “motion of consciousness” is well represented by our capacity of externalize
and reify our empirical ‘I’. At any stage in which we come to refer to a particular in our field of
awareness as ‘I’, we are also able to refer to it demonstratively.
Let’s review the linguistic evidence.
Suppose I were to utter:
(D) That is a computer
The token of ‘that’ in (D) is a demonstrative used by me to refer to an item in my field of
awareness, and (D) predicates of it that it is a computer.
129
If I am able to articulate in the first person mode, making thereby explicit the ‘I think’
prefix, I can think (or utter)
(D.I) I think (that) (D)
According to the first tentative theory of reference for ‘I’, (D.I) ought to be understood as stating
that a particular item in my field of awareness (that one with body with those specific cau sal
powers) thinks that (D). Here comes into play a phenomenon which I think underlies all our
thinking activities: the multiplicity of our empirical referents of ‘I’. Sentences such as (D.I) can
be embedded in indirect discourse ad infinitum: assuming I am able to articulate my first person
perspective I can think (and utter)
(D.I’) I think (that (D.I)
Theory I forces on us a regress. The regress is not vicious if we allow that there is one ‘I’ which
refers to itself as self. We make room, then, for self-awareness proper: the awareness of self as
self.
There is a tensions here between the being-in-the-world of each referent of the tokened ‘I’
and the capacity of the self to refer to each one of those demonstratively. At least one empirical
referent of ‘I’ has to be able to enter the world of the causal network: I have to mobilize it
when engaging in intentional action. And I can also objectify it, witness my ability to refer to it
demonstratively as in ‘that thinks that is a computer’. But the reiteration (explicitation) of the
‘I think’ prefix leaves us with a residual ‘I’ and its referent.
The self seems to be a vanishing point of view, unlocalizable in our field of awareness:
130
“The I is not an object. I objectively confront every object. But not the I.” wrote Wittgenstein23 ,
and, I would add, I can refer to every object, including the ‘I’ insofar as it is seen as an object.
I think now I can make clearer my dissatisfaction with rules such as
(K-I) In any statement in which it occurs, I designates the speaker of the statement24
These rules describe the capacity of the audience of identify as referents of the heard token of
‘I’ the proper objects in their fields of awareness. They do not describe first person thinking
referential uses of ‘I’, insofar as they fail to capture the implicit (though explicitable) ‘I’ in the ‘I
think’ prefix underlying my thoughts and utterances. A residual non-objectual ‘I’ remains. But,
if I am correct, the residual non-objectual ‘I’ is never captured by K − I25 .
And still, I believe there are deeper problems.
I am, from now on, assuming that there is a referent for the tokening of ‘I’. The rules for the
meaning of ‘I’ can not be stated by avoiding a form of circularity: indexical terms are not
reducible. There is a key difference between the reference mechanism of demonstratives and that
of pure indexical: whereas the former can be explained in terms of referent in the projected
world, the latter cannot, at least exhaustively.
According to Kaplan, ‘I’, ‘now’, and ‘here’ are pure indexicals: their reference is achieved
without any need of demonstration (or of directing intentions). And certainly in the case of ‘I’
he is right: I am not pointing to anything (literally or metaphorically) when saying ‘I’. Each
token’s reference is as direct as it can be. It has been noted by many authors that each and
every (indexical) tokening of ‘I’ has a dimension of certainty against which any skeptical wave
has to break. Beginning with Descartes and arriving at Jerrold Katz and Ernst Tugendhat26 ,
131
the remark has been made that ‘I’ indexically used does not and could not fail to refer to an
existent entity, impervious to the most radical forms of skepticism. There are two issues here.
One could be termed the uniqueness of ‘I’, namely is that the only term which self-p, as
it were, its existential commitment, or there are other terms which share this feature?
The second issue could be termed a question of substantiality: what is the (certainly)
existing “object” we refer to with ‘I’ ?
On the first issue Castañeda writes:
“. . . as Descartes himself had causally shown, the Kantian prefix is only a fragment of the relevant prefix, to wit: I think here now. This is to be sure at the
heart of Descartes’ claim that thinking and extension are two distinct attributes
characteristic, respectively, of the mental and the material substances. A comprehensive study of indexical reference suggests that the true transcendental
prefix is, therefore, the extended one: I think here now .27
Tugendhat, following him, elaborates along Strawsonian lines, that while a tokened ‘this’
might fail to refer to any spatio-temporal object “objectively” (i.e. not from the internal perspective of the tokener of it), there is no failing of having space and time as conditions of the
very possibility of identifying anything. While in ‘this beetle in my hand is red’ the ‘this beetle’
token might fail to refer to anything, there is no such phenomenon for ‘here’ and ‘now’: “It may
be false that a beetle is now here at the place at which I am pointing, but it is not possible that
the place is not there”.28 .
As I have noted earlier, I believe some distinctions have to be kept here. There is indeed
an unshakable truth in the idea that ‘I’ does not fail to refer to a real entity ever. But ‘here’ and
‘now’, I wish to maintain, have a dimension of similar certainty coupled with a difference. Not
132
only for the sake of philosophical chicanery one can consider examples fashioned (partially) after
Avicenna’s “suspended man”.
This is a man who has no contact whichsoever with anything, in fact he might be the one
and only existent. The only way I have to make the example vivid to my mind is to see that man
as a god before creation (or after destruction?). Notice that I am assuming there is absolutely
nothing but the man of the example, and following Avicenna the man is “suspended” in the sense
that he can not even touch, say, his feet. No motion is taking place at all, no perception in the
normal sense and so forth (there is nothing to perceive after all).
First question: if the “suspended subject” thinks can he refer to himself as himself in the indexical
fashion? my answer is positive. His thoughts would be highly Cartesian in a sense, but they
would never move beyond the ‘I think’ (plus the contents of these thoughts). And the tokenings
of ‘I’ would surely not fail to refer to the man as much as the ‘I’ used by all non-suspended
thinkers.
Second question: if this person were to think ‘Here now I am rather bored with this nothingness
surrounding me’, would ‘here’ and ‘now’ provide the same dimension of certainty? There is
no place “in” which the occurrence of the thought takes place, by hypothesis. And there is
no “objective identifying time frame” to use Tugendhat’s terminology, there are no external
coordinates of time, there being nothing to coordinate in the first place. My first intuitive
response to a case like this is that while ‘I’ keeps functioning in the normal way, ‘here’ and ‘now’
do not.
The mental life of our “god” is boring indeed, but it isn’t necessarily nonexistent. The
133
strength of this example depends upon the conceivability of the situation. I, for one, finds it
conceivable and think that it also illustrates a further point of difference between indexical expressions. In the example in question ‘here’ fails to refer to any externally conceivable space, what
Strawson and Tugendhat call objective space. It may though still refer in a quasi-pathological
way to the whole in which the suspended man is hanging. There is no “whole” though, so ‘here’
would refer back to its tokener! It may be the way in which some might conceive of the ‘here’
of a god faced with an entire universe. That would be a different situation altogether, although
even the the ‘here’ still would fail to segregate a particular in the normal way.
The ‘now’ of the suspended man, I would maintain, fails to refer to any interval or
portion of time in terms of “objective time-frames”. It does refer though to internal “durations”.
As long as the “suspended man” is capable of retaining any form of memory and has the ability
to keep track of the difference between separate episodes of thought, his tokenings of ‘now’ (and
of ‘then’) would keep providing a dimension of metaphysical certainty in referring beyond that
one given by the ‘I’.
The point made by the argument indicates a difficulty in a fullfledged Frego-Kantian extension of the ‘I think’ prefix of Kant to Castañeda’s ‘I think here now’ as a trascendental phrase,
which unthought itself accompanies all thoughts and representations. Whereas Castañeda’s point
seems to me to hold in straightforward manner for ‘now’, it seems that, given the differences indicated above, there isn’t a basis of equal force for ‘here’.
134
A substantial ‘I’ ?
If ‘I’ has a referent, it has be somehow mentally encoded, if it is to be thinkable at all. ‘I’
has that particular meaning I have tried to illustrate and so we have the inescapable conclusion
that minds are somehow able to reach beyond the experiential world, if I am correct in identifying
a part of the meaning of ‘I’ with a non-worldly, transcendental condition of experience, and
transcendent referent, ‘I’ which is implicit in all our representations, and in all our utterances29 .
The problem is that even accepted the existence of such a transcendental referent, if
referring to it I use any form of demonstrative reference, by doing so I am making it an object of
my awareness. The non reducibility of ‘I’ comes back twice reinforced. It comes back because at
each stage, if I wish to have some substantial claim on my tokenings of ‘I’ then I have to be able
at each point to make a judgment of identity between my occurrent token and the prior ones. I
can always refer to myself (in the past or in the future) as ‘that person who. . . ’ but if I want
an Ego-connection, I have also to add ‘. . . and that person is I’. Robert Nozick puts the problem
thus:
If I always knew something of myself via a term or a referring token, there would
be needed the additional (unexplained) fact of my knowing the term referred to
me. . .
Therefore it seems we must have an access to ourselves which is not via a term
or referring expression, not via knowing that a term holds true (of something or
other)30 .
Trying to make the referent of my tokenings of ‘I’ an entity reiterates the need for
identification: even assuming I know that my ‘I’ tokenings hark back towards my noumenal ‘I’,
I still have to know that ‘I am (the same as) the noumenal I’. Nozick’s answer relies on what he
135
calls self-synthesis: a Fichtian philosophical notion of the ‘I’ positing itself as an ‘I’.31
A self-synthesis is a creative process: ex nihilo an I comes to be referred to in this
peculiar first person irreducible way. This process can be repeated as many times as necessary:
it constitutes the raw material which we utilize when we uncover the covert indexical prefix of
our utterances. Nozick also claims that more often than not we are relying on past synthesis to
establish the reference of the unthought-of I. We can rely on past occurrences of self-synthesis
insofar as we can refer demonstratively to the selves established then. I think this happens, and
it constitutes the source of a sense we have of across times identity.
But logically the priority should be given to the occurring act of self-synthesis. It is by
way of this act of reference to the self as our self that we are put in the condition of being able
to make those judgments. Judgments on past occurrences of self-synthesis have to be in the end
statements of sameness. To have an enduring identity I have to be able to recognize past acts
of self-synthesis as mine, and so I have to perform an occurring synthesis logically prior to my
looking back in memory, to search for my former selves. Quite evidently this is not always a
simple process. Aside from the pathologies of the occurrent multiple selves, even in the normal
case we find much more difficult to recognize our selves as ourselves when the span of time
between the act of recognition and the memorized content. An act of recognition, of reliance
on past self-synthesis, presupposes an occurrent self-synthesis. Thence it is always an occurrent
self-synthesis which grounds our referring to our self.
Castañeda after having indicated that self-reference in origin an anchoring of a quasiindexical reference32 , locates the nature of the synthesis of a self in a system of negative polarities:
136
The internal I-structures are I-strands hinging on contrasts between what one
is qua oneself* and something one is not. Alternatively phrased, an I-strand
is the polar negation of something intrinsically non-I . . . Thus an I-schema is a
complex of negativities.33
Those negativities are lines of contrast, between e.g. ‘I’ and ‘they’, ‘I’ and ‘this’, ‘I’ and
‘we’, each contrast being the block on which an awareness of oneself as ‘I’ is possible. Castañeda
takes the line that at bottom the ‘I’ is a purely primitive phenomenon, something of which
any thinker (who is an ‘I’) is aware: “There is just no criterion one can apply to determine
whether one is an I or not. One simply is an I. This primitive fact is primitively and immediately
apprehended by a thinker who is an I”. Both Nozick and Castañeda have a view in which there is
no substantial ‘I’; that is, there isn’t in their view any enduring entity to which one might refer
with different occurrences of ‘I’. For Nozick one solution to traditional perplexities on the nature
of the self (perplexities started by Hume) is to view it as a Fregean unsaturated property, but he
dismisses it as “too much froth and too little substance”.34
Castañeda on the other hand denies that there is any enduring substance named by
the tokens of ‘I’, even when used by one speaker diachronically. For him the sense of enduring
selfhood some experience is given in judgments, the subject matter of which are I-Guises.
I think there is a real insight in the denial of any one enduring entity captured by ‘I’.
There is a persistent nagging reckoning from the deepest recesses of “common sense” which
resists the full force of this denial. A conception of self-synthesis based on occurrent episodes of
synthesis helps explaining such a resistance. It is the very synthesis which generates an entity. It
seems to me that is an entity which stands in a polar macro-opposition, borrowing Castañeda’s
137
terminology: the opposition ‘I’ vs. everything else. To let go such an opposition literally on the
spur of the moment is, perhaps, too difficult. It is a “small death” after all, and we seem to
have, most of the times, some sense that to exist is “better” than not to. Prior to some sense of
philosophical reflection which might make room for judgments of similarities between episodes
of self-synthesis, or some sense of “fully lived biography” in Castañeda’s sense, it might be too
hard to bear psychologically to die at each moment, too hard psychologically for each ‘I”s sense
of self-worth. That indeed seems to me the price of not identifying tout court the ‘I’ with one’s
own body or the ‘I’ with one’s person, the social construct most commonsensically associated
with the notion of self. But it seems also that looking at the way we talk about ourselves, ‘I’
captures something different each time is used, even by one and the same speaker.
William M. Richards in a recent article35 brings up a difficulty for a theory which allows
reflexive self-reference in Nozick’s fashion. The key criticism consists in a charge of circularity:
reflexive reference of this kind presupposes a unitary self, which is referred to. Nozick’s explanation makes appeal to an individual which is a “sum-individual X+Y+Z”.36 We can take a
sum-individual to be something like Kathleen in my Theory I, where X, Y, and Z could be a
body, certain causal powers, and, say, the repertoire of words she remembers. There is no underlying “I-unity”; that is there isn’t any self-awareness proper, with the sensation of certainty
of attribution which comes with it. Nozick writes:
“Suppose the sum-individual X+Y+Z produces the sum-token x+y+z which
reflexively refers to X+Y+Z. This is sufficient for the closest relation schema
to be applied; the entity X+Y+Z thereby may delineate itself as the entity
X+Y+Z, correctly applying the predicate to itself”.37
138
Richards notes that to refer to oneself qua oneself is an eminently intentional act and to
be in accord with the intention an act would have to be somehow keyed to the content of the
intention itself. But then
“Either exactly one of X, Y, or Z of the sum-individual X+Y+Z was keyed to
the content of the intention [to refer], in which case the operative individual was
not a genuine sum-individual; or X, Y, and Z each were cognizant of different
parts of the intention and they communicated among one another to produce
to produce the unified intentional result. But ‘communication’ is an intentional
unifier par excellence. . .
[or] at least one of X, Y, and Z would have to be cognizant of the subject
component of the intention that guides the token production, and if intentions
are essentially first-person, this means that one of the X, Y, and Z must have
been aware of himself in the first-person way”.38
While I agree that Richards in that there is an intentional element in self-reference, I
do not think that is enough to doom any self-reflexive approach. A Nozickian “sum-individual”
is, I believe, at some stage of development each one of “us”, developed selves of now. Our first
tokenings of ‘I’ might well be not self-reflexive. Our utterances though produce reactions from
others: our own unaware creation of imperial propositions with their cohorts of propositions
others can attribute to us bounces back on us.
Attribution of selfhood to myself, I think, is better placed finally in communication.
We learn to be a self, and at least part of learning depends on a mechanism of attributions of
selfhood to others, before than to ourselves. An occurrent self-synthesis is initially prompted by
our contemplation of the ways others react to our acts, linguistic or not, and by our capacity of
targeting the cohorts of imperial first person propositions created by others.
139
It is this occurrent self-synthesis which I think is the real primitive of the theory. Not
because it is utterly clear intuitively nor because it is so obscure it defies any attempt of explanation. It is primitive because it is a just a label for creative process we all perform, attributing
to others first-personhood, and then speaking in first person.
Perhaps, in the beginning was the word.
140
NOTES
\*/ “Consciousness, Philosophy, and Mathematics” (1948), now in L. E.J.Brouwer,
Collected Works, vol.1, Amsterdam 1975, pp. 480-494.
\1/ I have in mind in particular the thinking of music, perhaps of certain parts of mathematics.
I am not assuming that music or mathematics have to be non representational systems. The
explanations of this statement would take me too far away. It will have to wait for some other
occasion.
\2/ KAPLAN {2}
\3/ A word of caution: each one of these expressions can (and does) have uses which are not
purely indexical. The trivial ones are exemplified by a sentence like “I is a letter in the French
alphabet”. The not so trivial are the ones in which they can be used demonstratively, e.g. in
“Here you see Kaplan at his best” pointing to a passage of his “Demonstratives”, or in “Now
Webern breaks away from seriality” pointing to a section of a score of his (notice the difficulty in
accepting this is the difficulty in having a perceptually clear way to pinpoint a moment in time,
that is why I used a musical example in which the perception on the piece is so closely linked to
our perception of time).
\4/ On the ways in which unconscious mental processes to a large extent determine what our
field of awareness looks like see JACKENDOFF {1} and {2}.
\**/ Moritz Schlick, “Erleben, Erkennen, Metaphysik”, now in Philosophical Papers, Dordrecht,
1979, vol. II, p. 100.
A fascinating attempt to trace the consequences of a Leibnizian approach to thinking is “Monadology” by Montgomery Furth, Philosoph ical Review, Vol. LXXXVI, No.2, (1967), pp. 169-200.
Furth notices that even aside from qualms about distinctions between indexical and non-indexical
references it is not easy even to distinguish between myself and my world, a tail to Lichtenberg’s
point that the only “logical” conclusion of Descartes’ First and Second Meditation is not so much
that I am thinking, but rather that ‘Thinking is happening’.
\5/ Speaking again of Doktor Lauben Frege writes: “when Dr. Lauben has the thought that
he was wounded, he will probably be basing it on this primitive way in which he is presented
to himself. And only Dr. Lauben himself can grasp thoughts specified in this way. But now he
may want to communicate with others. He cannot communicate a thought he alone can grasp.
Therefore, if he now says ‘I was wounded’, he must use ‘I’ in a sense which can be grasped by
others, perhaps in the sense of ‘he who is speaking to you at this moment’; by doing this he
makes the conditions accompanying his utterance serve towards the expression of a thought” (in
FREGE {1}: p. 359-360). Frege is way ahead of his time. His “secondary sense” of ‘I’ (the
“public” sense) is basically the one encased in rules such as K-I (see Perry, in TOMBERLIN: p.
18). What he did not develop is a correspondingly elaborated theory of thoughts: for him those
141
remained paradigmatically entities very similar to classical propositions, hence tenselessly true
or false.
\6/ “It must be possible for the ‘I think’ to accompany all my representations; for otherwise
something would be represented in me which could not be thought at all, and that is equivalent
to saying that the representation would be impossible, or at least would be nothing to me.”
Immanuel Kant, Critique of Pure Reason, A 108, B 131-132
I am adopting the Kantian prefix with rather divergent aims from Castañeda. The prefix, intuitively, is a marker of the agency which binds indexical sentences, giving them private senses.
\7/ A very similar approach has been taken by Castañeda in his “The Self and its Guises”.
\8/ The correct orientation of the audience’s directing intentions is in fact greatly facilitated by
the predicative content of a sentence like (L). The right neighborhood is delimited by the set of
object in the audience’s field of awareness that conceivably can be a lime tree. And this often
makes it superfluous to pay attention to the physical motions, if any, accompanying the utterance
of (L). Together with the predicative content of the sentence in which an indexical token occurs,
another help in semantics of understanding is provided to the audience by what Kent Bach called
“referential constraint” for a pronoun. These kinds of constraints are relative to a language (in
particular where issues of grammatical gender and/or sex are involved). For a detailed treatment
of referential constraints, see BACH, in particular ch. 9 “Reference and Pronouns”. Before Bach,
Castañeda pointed out the same phenomenon, without calling it a constraint: “Each indicator
has a general sense, which may be construed as a range of possible denotations”, he also pointed
out that the predicative part of a noun phrase containing and indexical might function as a guide
to the establishment of a reference from the perspective of an audience, as in ‘this red chair’ –
see in particular his “The Semiotic Profile of Indexical (Experiential) Reference” in SAARINEN,
especially sections 8 and 9.
Even though I have been insisting on an intentional component in reference which is more prominent in indexical cases than in non-indexical ones, I have found to my surprise that David Kaplan
now takes directing intentions to be determinant of referents, while abandoning his prior notion
of “demonstration”. I have indicated some of the problems with that notion in the fourth chapter
myself, for Kaplan’s own statement of his change of heart on the matter, see his “Afterthoughts”
in Themes from Kaplan.
\9/ See JACKENDOFF {2}.
\10/ I owe the idea of communication of content-less structures to Moritz Schlick. See in particular Ch. 17 of The Problems of Philosophy in their Interconnection.
\11/ “When he grasps or think a thought he does not create it but only comes to stand in a
certain relation to what already existed”, writes Frege (see FREGE {1}: p. 363, fn. 7). I believe
that a distortion of the problem in part was brought about by an extreme concentration on the
142
range of mathematical Gedanke Frege took as paradigmatically central. Mathematical thoughts
seem indeed to be the ones where the perspectival ingredient is minimal, if not non altogether
non existent.
\12/ I am disregarding the peculiarity of English in which to indicate the beginning of an indirect
discourse clause ‘that’ is used to conjoin the two clauses. It is the demonstrative use of ‘that’ I
have in mind.
\13/ See, in particular CASTAÑEDA {8}. But the connections between contemplative thoughts
and practical thinking are the centre of his whole work.
The notion of PROPOSITION of CASTAÑEDA {15} is analogous to my notion of “imperial”
proposition: the difference between our two approaches is slight enough to make them parallel
in many respects, although I believe my proposal to be a possibly preferable, less ontologically
burdened alternative .
\14/ I have in mind Diotima in Der Mann ohne Eigenschaften. I take it as evident that Diotima
in-Der Mann ohne Eigenschaften is not a fictional woman who is seductive, but a woman who is
seductive .
\15/ See CASTAÑEDA {2}.
\16/ This can not be the first person indexical proposition Saint-Just had as his referent.
First person propositions are private, as much as indexical referents. For some of the
reasons why there is no identity between indexical propositions and “matching” quasi-indexical
ones, see the “Adams-Castañeda Correspondence” in TOMBERLIN: pp. 293-309.
\17/ He introduces the term while explaining why in the non-veridical case of a demonstrative
sentence such as (L), the lack of an external referent would make not true-and-not false. I have
indicated why I take that not to be correct.
\18/ It might be the speaker in Castañeda’s examples such as ‘The Editor of Soul believes he is
rich’.
\19/ In margin to my approach, I should like to notice that one prediction follows from it. Suppose speaker 1 says: ‘I think that is a flower and now, from here it looks red’. If speaker 2 has to
report the utterance of 1, in a situation in which they have no common background, hence the
directing intentions of 1 would not be much help, he should not be inclined to use a demonstrative to capture the token of ‘that’ in 1’s utterance, whereas he could capture all of the other 1’s
content by way of quasi-indicators. The reporting, I think, should be: ‘He said that he* thought
what he saw was a flower and then from there it looked red’. ‘He*’ is the ‘he-himself as himself’
quasi-indicator of Castañeda, and the tokens of ‘then’ and ‘there’ are quasi-indicators as well.
My point is that speaker 2, say after a period of time, could not have access to the salient item
of 1’s field of awareness which, I surmise, was the reference of the demonstrative ‘that’ used by
1. Quite visibly in 2’s utterance the first occurrence of ‘he’ is demonstrative.
143
\***/ NOZICK: 71
\20/ Were I to be picky on terminology, our familiar Kathleen ought to read #Kathleen# a la’
Jackendoff. But I take it for evident that what we refer to by way of her name is the item in our
field of awareness.
Kathleen’s story is a somewhat re-elaborated story accounting of the empirical results (some of
them of difficult interpretation, once one discounts a behavioristic model of stimulus-response)
collected and examined in KAGAN. See, in particular ch.3: “Signs of self-awareness”.
\21/ This is a case of what Castañeda calls a “Externus”- type of consciousness.
See, in particular, CASTAÑEDA {12} and {13}.
\22/ If we were equipped with telekinetic powers, this genetic account would look wildly implausible. I take it to be a fact that we have no ability to affect causally in direct fashion objects
other than our bodies.
\23/ in Notebooks 1914-1916, notes of 7.8.1916 and 11.8.1 916
\24/ K-I has been proposed by John Perry (and thus named after Kaplan ). The criticisms made
here is originally Castañeda’s. See PERRY {3} and Castañeda’s “Reply to Perry” in TOMBERLIN.
\25/ Castañeda (in “Metaphysical Internalism, Selves, and the Wholistic Indivisible Noumenon
(A Frego-Kantian Reflection on Descartes’s Cogito)” (CASTAÑEDA {14}) writes in a similar
vein: “In a sense, the true Thinking I is the unthought of I, that thinks (D), (D.I), (D.I’). . . “ –
I have adapted the passage by changing the examples to my examples in the text.
\26/ See Jerrold Katz, Cogitationes, Oxford 1986,
and Ernst Tugendhat, Self-Consciousness and Self-Determination, Cambridge, Mass. 1986.
Katz takes the line that the ‘I’ of the Cartesian Cogito is referring to an existing entity in virtue
of analytical entailments built-in the conceptual structure of language. Tugendhat follows a more
Kantian line of viewing ‘I’ (and ‘here’ and ‘now’) as conditions of any possible experience, and
in this “external” sense not open to skeptical doubts, since the very doubting does in fact entail
that a) some one is doing the doubting, b) the doubting takes a certain interval of time, and c)
the doubting takes place somewhere. Tugendhat in passing seems to be much closer than Katz
to Castañeda’s extension of the ‘I think’ to ‘I think here now’. I try to pursue this issue a little
further in the main text.
\27/ See his “Metaphysical Internalism, Selves, and the Wholistic Indivisible Noumenon (A
Frego-Kantian Reflection on Descartes’ Cogito), (CASTAÑEDA {14}).
\28/ See Tugendhat, op. cit., pp.62-66.
\29/ From a different perspective John Perry approximates the view I am trying to spell out.
In recent work he adds to his theory unarticulated constituents of propositions. What I am
indicating is that, at least in the case of the utterer, the unarticulated constituent can be ar144
ticulated, as my sequence of (D), (D.I). . . tries to show. What Perry denies is that one of the
unarticulated constituent of proposition might have an irreducible perspectival aspect built into
it (the ‘I’ has such a character, I think). See Perry’s “Thought without Representation”, in
Proceedings of the Aristotelian Society, (1986), 60: 137-151; Mark Crimmins and John Perry,
“The Prince and the Phone Booth”, Stanford, CSLI, 1988. For some critical responses see Simon
Blackburn’s comments on the first paper in the same issue of the Proceedings and Jon Barwise’s
“Situations, Facts, and True Propositions” in his The Situation in Logic, Stanford, CSLI 1989.
\30/ NOZICK: 81
\31/ ibidem, p. 89. A little taken aback by the very “idealistic” flavour of the idea, Nozick
tries then to have only some acts of self-reference as creators ipso facto of a self. As I have
stressed earlier, it is certainly true that not all consciousness is self-consciousness. The hierarchy
of consciousness, in Castañeda’s terms, begins at the bottom with a perceptual consciousness
which, as in Kathleen’s case, does not need an ‘I’, much less a transcendental one. But I do
believe that each act of self-reference creates a self: the one referred to by the covert indexical
operator [I think here now] accompanying each one of our utterances. I’d like to stress that this
self-referentiality of some stratum of our consciousness might very well be causally dependent
(though not identical with, if this argument is sound) on the matter and the structure of our
nervous system. In that respect one of the key test to decide whether “machines can think” in
some sense close to our sense of thinking, is to test their self-referential capacity.
\32/ In “The reflexivity of Self-consciousness, Sameness/Identity, Data for artificial intelligence”,
forthcoming. There he writes: “First-person reference is necessarily reflexive. This feature is signaled by our formula ‘ONE refers to ONEself as oneself’. The expression one in oneself does
co-refer with ’ONE’. This reflexivity forces first-person reference to harpoon an external reality”
The externality of the reference is what gives the necessary existence of a target for each token
of ‘I’. (Most of the citations in this section attributed to Castañeda are from this work.
\33/ ibidem
\34/ NOZICK: 114
\35/ William M. Richards, “Self-Consciousness and Agency”, Synthese 61, (1984) 149-171.
\36/ NOZICK: 665-666
\37/ ibidem. The closest relation schema is Nozick’s preferred schema for an enduring identity
across times.
\38/ William M. Richards, op. cit., pp. 157-158.
145
BIBLIOGRAPHY
ADAMS, Robert M. ADAMS-CASTAÑEDA Correspondence,
in TOMBERLIN
ANSCOMBE, G.E.M. “The First Person”, in S Guttenplan (ed.)
Mind and Language, Oxford 1975
BACH, Kent Thought and Reference, Oxford, 1987
BARWISE, Jon and PERRY John, Situations and Attitudes, Cambridge, Mass.:1983
BOËR S. and LYCAN W., Knowing Who, Cambridge, Mass.:1986
CASTAÑEDA, Hector-Neri
{1} ‘He’: On the Logic of Self-Consciousness, Ratio, 8, 1 30-157
{2} Indicators and Quasi-Indicators, Amer. Phil. Quarterly, 4, 1967,130-157
{3} On the Phenomeno-logic of the ‘I’, Proceedings of XIV Int. Cong. of Philos ophy,
vol.3, 1968
{4} Thinking and the Structure of the World, Philosophia, 1974, 4, 4-40
{5} Thinking and Doing, Dordrecht, 1975
{6} Perception, Belief, and the Structure of Physical Objects and Consciousnes s,
Synthese, 35, 1977, 315-351
{7} The Semiotic Profile of Indexical (Experiential) Reference,
in SAARINEN, pp.275-316
{8} Direct Reference, the Semantics of Thinking, and Guise Theory (Constructive Reflections on David Kaplan’s Theory of Indexical Reference), in Themes from Kaplan,
146
Oxford, 1989
{9} Self-Consciousness, Demonstrative Reference, and the Self-ascription view of Believing, Philosophical Perspectives, 1, Metaphysics: 1987
{10} Thinking, Language, and Experience, Minnesota University Press, Minneapolis
1989
{11} ADAMS-CASTAÑEDA Correspondence,
in TOMBERLIN
{12} “On the Phenomeno-logic of the I”
in Akten des XIV Internationalen Kongresses für Philosophie, 1968
{13} “On the Logic of Attributions of Self-knowledge to Others”
Journal of Philosophy, LXV, 1968, pp. 439-456
{14} “Metaphysical Internalism, Selves, and the Wholistic Indivisible Noumenon (A
Frego-Kantian Reflection on Descartes’ Cogito)”
in Midwest Studies in Philosophy, 1988
{15} “On the Philosophical Foundations of Communication: I. Reference”
in Midwest Studies in Philosophy, 1977
CHISHOLM, Roderick The First Person, Minneapolis, 1981
CLARK, Romane
{1} “Old Foundations for a Logic of Perception”, Synthese, 33, 1976, 75-99
{2} “Self Knowledge and Self Consciousness: Thoughts about Oneself” Topoi, 6, 1987
DAVIDSON, Donald Inquiries into Truth and Interpretation, Oxford, 1984
147
DUMMETT, Michael
{1} Frege: Philosophy of Language, Cambridge, Mass. 1981 nl {2} The Interpretation of Frege’s philosoph
Cambridge, Mass. 1981
EVANS, Gareth
{1} The Varieties of Reference Oxford 1982
{2} Collected Papers, Oxford 1985
FREGE, Gottlob
{1} Collected Papers, Oxford 1984
{2} Philosophical and Mathematical Correspondence, Chicago 1980
{3} Posthumous Writings, Chicago 1979
GALE, Richard “Omniscience-Immutability Arguments”, Amer. Phil. Quart., 23, 4, 1986, 315335
GRICE, Paul Studies in the Way of Words, Cambridge, Mass. 1989
JACKENDOFF, Ray
{1} Semantics and Cognition, Cambridge, Mass. 1983
{2} Consciousness and the Computational Mind, Cambridge, Mass. 1987
KAGAN, Jerome The Second Year, Cambridge, Mass. 1981
KAPLAN, David B.
148
{1} “How to Russell a Frege-Church” (1975), now in LOUX
{2} “Demonstratives”, Draft no.2, UCLA Phil. Dept., 1977,
now also in Almog, et al. (eds) Themes from Kaplan, Oxford, 1989
{3} “Dthat”, (1978) now in French, et al.
Contemporary Perspectives in the Philosophy of Language, Minneapolis 1979
{4} “On the logic of demonstratives”, ibidem
{5} “Opacity”, in Hahn and Schilpp (eds.) The Philosophy of W.V.O. Quine, La
Salle, Ill. 1986
KATZ, Jerrold, Cogitationes, Oxford 1986
LOUX, Michael J., (ed) The Possible and the Actual, Ithaca, N.Y., 1979
LURIA, Aleksandr Romanovich The Mind of a Mnemonist, Cambridge, Mass., 1987
NOZICK, Robert Philosophical Explanations, Cambridge, Mass. 1981
PERRY, John
{1} “Frege on demonstratives” The Philosophical Review, 1977, 86, 464-497
{2} “The problem of the essential indexical”, Nous, 1979, 13, 3-21
{3} “Castañeda on ‘he’ and ‘I”’, in TOMBERLIN
{4} “Cognitive significance and new theories of reference”, Nous, 1988, 22, 1-19
QUINE, W.V.O. Word and Object, Cambridge, Mass. 1960
REICHENBACH, Hans Elements of Symbolic Logic, New York 1974
RUSSELL, Bertrand
149
{1} An Inquiry into Meaning and Truth, London 1940
{2} Human Knowledge, Its scope and Limits, New York, 1948
SAARINEN, Esa (ed) Synthese, issue on demonstrative and indexical reference, vol. 49, nos.
1-2, 1981
SALMON, Nathan U., Reference and Essence, Princeton 1981
SOAMES, Scott, “Substitutivity”, in On Being and Saying, Judith J. Thomson (ed.), Cambridge, Mass. 1987
SCHLICK, Moritz
{1} Philosophical Papers, Dordrecht, 1979
{2} The Problems of Philosophy in their Interconnection
(winter lectures 1933-1934), Dordrecht, 1987
TOMBERLIN, James E., (ed.) Agent, Language, and the Structure of the world, Indianapolis
1983
TUGENDHAT, Ernst Self-consciousness and Self-determination, Cambridge, Mass. 1986
WETTSTEIN, Howard K.
{1} “How to bridge the Gap between Meaning and Reference”, Synthese 58, 1984,
63-84
{2} “Has Semantics rested on a Mistake?”, Journal of Philosophy, LXXXIII, 1986,
185-209
{3} “Cognitive Significance and Cognitive Content”, Mind, 385, 1988, 1-31
150
Vita
Adriano Palma was born in 1956.
151