City University of New York (CUNY)

CUNY Academic Works

Dissertations, Theses, and Capstone Projects Graduate Center

5-2018

The Two Inexical Uses Theory of Proper Names

and Frege's Puzzle
Daniel S. Shabasson
The Graduate Center, City University of New York

How does access to this work benefit you? Let us know!

Follow this and additional works at: https://academicworks.cuny.edu/gc_etds
Part of the Philosophy of Language Commons

Recommended Citation
Shabasson, Daniel S., "The Two Inexical Uses Theory of Proper Names and Frege's Puzzle" (2018). CUNY Academic Works.
https://academicworks.cuny.edu/gc_etds/2605

This Dissertation is brought to you by CUNY Academic Works. It has been accepted for inclusion in All Dissertations, Theses, and Capstone Projects
by an authorized administrator of CUNY Academic Works. For more information, please contact [email protected].
 THE TWO INDEXICAL USES THEORY OF PROPER NAMES

AND FREGE’S PUZZLE

By

DANIEL SHABASSON

A dissertation submitted to the Graduate Faculty in Philosophy in partial fulfillment of the

requirements for the degree of Doctor of Philosophy, The City University of New York

2018
 ©2018
DANIEL SHABASSON
All Rights Reserved

ii
 THE TWO INDEXICAL USES THEORY OF PROPER NAMES

AND FREGE’S PUZZLE

by

DANIEL SHABASSON

This manuscript has been read and accepted for the Graduate Faculty in Philosophy
in satisfaction of the dissertation requirement for the degree of Doctor of
Philosophy.

Date _________________ _________________________________________

Barbara Montero
Chair of Examining Committee

Date _________________ _________________________________________

Nickolas Pappas
Executive Officer

Supervisory Committee:

David Papineau

Heimir Geirrson

THE CITY UNIVERSITY OF NEW YORK

iii
 ABSTRACT

THE TWO INDEXICAL USES THEORY OF PROPER NAMES

AND FREGE’S PUZZLE

By

DANIEL SHABASSON

Adviser: Professor Barbara Montero

The debate between Millians and Fregeans over proper names has resulted in a stalemate

after roughly forty years. Neither Millianism nor Fregeanism is plausible. Millianism clashes

with our intuitions about cognitive value and truth-value with respect to Frege’s puzzle. Millians

have elaborated interesting theories aiming to show why this clash does not constitute a reductio

of Millianism, but they are not plausible. Fregeanism fares no better, as it was refuted by Kripke

in Naming and Necessity. Following Kripke, a viable theory of proper names must, at a

minimum, respect the constraints that: (a) that names are rigid designators; and (b) if names have

any descriptive or conceptual contents at all, these cannot be reference-determining.

I formulate a non-Fregean, non-Millian theory of proper names that respects the above-

mentioned Kripkean constraints as well as our Fregean intuitions with respect to Frege’s puzzle.

I explain the cognitive value difference between ‘Clark Kent is Clark Kent’ and ‘Clark Kent is

Superman’ by showing that they express different propositions. The propositional attitude

ascription ‘Lois Lane does not realize that Clark Kent is Superman’ comes out as true, not false

as Millians maintain. I propose solutions to both versions of Frege’s puzzle, the ‘Problem of

Rational Inconsistency’ (the question how expressing and believing inconsistent singular

propositions can be compatible with an agent’s rationality), and Kripke’s puzzle about belief.

iv
 I argue that proper names are used as two kinds of indexicals. Sometimes a name is used

indexically just to refer to its bearer. I call this a ‘Millian use,’ and say that a name is used ‘in a

Millian way.’ Other times, a name is used indexically to refer to its bearer and to contribute the

speaker’s descriptive conception of that bearer to the proposition. I call this a ‘Conception-

indicating use’ and say a name is used ‘in a Conception-indicating way.’ Therefore, I call the

theory ‘The Two Indexical Uses Theory of Proper Names’, or ‘TIUT.’ I intend this difference in

use to mark a semantic, and not merely pragmatic, distinction. When used in a Millian way, a

name is directly referential. Its content is just its bearer. When used in a Conception-indicating

way, a name’s content, which is determined by its character in the context of utterance, is not an

object, but rather a meaning that mediates reference. Thus, a name used in a Conception-

indicating way falls outside of David Kaplan’s (1989) paradigm of indexicality, according to

which indexicals are directly referential. When used in a Conception-indicating way, ‘Clark

Kent’ and ‘Superman’ designate rigidly, but they have different contents that are partially

constituted by different descriptive elements. Crucially, these descriptive elements of content are

referentially inert, playing no role in determining reference. Reference is instead a function of

causal-historical factors following Kripke’s causal-historical picture of reference.

v
 Acknowledgments

I wish to thank my adviser, Barbara Montero, as well as my committee members, David

Papineau and Heimir Geirsson. Thanks also to my mother, Susanne Shabasson, and my brother,

Andy Shabasson.

vi
 TABLE OF CONTENTS

Title Page…………………………………………………………………………………. i

Copyright Page ……………………………………………………………………………. ii

Approval Page……………………………………………………………………………... iii

Abstract……………………………………………………………………………………. iv

Acknowledgments…………………………………………………………………………..vi

Table of Contents…………………………………………………………………………... vii-viii

INTRODUCTION – OUTLINE OF DISSERATION……………………………………... ix

1. THE PUZZLES
1.1 Frege’s Puzzle(s)……………………………………....................................1

1.2 The Problem of Rational Inconsistency………………….…………………13

2. THE TWO INDEXICAL USES THEORY OF PROPER

NAMES AND ITS PROPOSED SOLUTION TO THE PUZZLES

2.1 Two Uses of Proper Names………………………………………............... 16

2.2 Definitions: “Dossier Tokens,” “Dossier Types,” “Subjects” …….............. 25

2.3 Names Used in a ‘Millian Way’…………………………………................ 33

2.4 Names Used in a ‘Conception-Indicating Way’…………………………… 35

2.5 Proposed Solution to Frege’s Puzzle

About Informative Identity Sentences……………………………………... 43

2.6 Proposed Solution to Frege’s Puzzle

About Propositional Attitude Ascriptions…..................................................47

2.6.1 Conception-Indicating Attitude Ascriptions….................................. 49

2.6.2 Millian Attitude Ascriptions….……………………………………. 54

2.7 Quantifying in……………………………………………………………… 62

vii
 2.8 Proposed Solution to the Problem of Rational Inconsistency….................... 64

2.9 Informative Identities without Substitution………………………………... 72

2.10 Informative Identities: A Priori or A Posteriori?........................................... 78

2.11 Ascribing Belief to Non-Verbal Agents…………………………………… 92

2.12 Proposed Solution to Kripke’s Puzzle………………................................... 96

2.13 The Distinction between Millian and

Conception-Indicating Names: Pragmatic or Semantic?............................... 106

2.14 Conclusion…………………………………………………………..............108

3. PELCZAR AND RAINSBURY’S ‘SINGLE

INDEXICAL’ THEORY OF PROPER NAMES…………………………………. 114

4. DESCRIPTIVISM…………………………………………………………………. 135

5. MILLIANISM………………………………………………….…………………...163

6. THE HIDDEN INDEXICAL THEORY……………………………………………215

7. FORBES’ THEORY OF PROPER NAMES……………………………………….222

8. THE ‘SAUL CASES’……………………………………………………………… 226

9. ANTI-FUNCTIONALIST THEORIES …………………………………………… 233

10. RECANATI’S THEORY OF MENTAL FILES……………………………………254

REFERENCES…………………………………………………………………………….. 259

viii
 INTRODUCTION – OUTLINE OF DISSERTATION

The debate between Millians and Fregeans over proper names has resulted in a stalemate

after roughly forty years. The problem is that neither Millianism nor Fregeanism is plausible.

Millianism clashes with our intuitions about cognitive value and truth-value with respect to

Frege’s puzzle. Millians have elaborated interesting theories aiming to show why this clash does

not constitute a reductio of Millianism. However, along with many philosophers, I do not find

these theories plausible, although I do not claim to offer any knockdown refutation. Fregeanism

fares no better. If we construe Fregeanism as a species of Descriptivism, which is the standard

interpretation, then it was refuted by Kripke in Naming and Necessity. Following Kripke, a

viable theory of proper names must, at a minimum, respect the constraints that: (a) that names

are rigid designators; and (b) if names have any descriptive or conceptual contents at all, these

cannot be reference-determining.

My purpose is to formulate a theory of proper names that is neither Fregean nor Millian.

It should respect the above-mentioned Kripkean constraints as well as our Fregean intuitions

with respect to Frege’s puzzle. It should explain the cognitive value difference between ‘Clark

Kent is Clark Kent’ and ‘Clark Kent is Superman’ by showing that they express different

propositions. The propositional attitude ascription ‘Lois Lane does not realize that Clark Kent is

Superman’ should come out as true, and not false as Millians maintain. The theory I elaborate,

‘The Two Indexical Uses Theory of Proper Names,’ or ‘TIUT,’ comes as close as possible to

such a theory.

In Chapter 1, section 1.1, I set out the two versions of Frege’s puzzle—the identity

sentences puzzle and the propositional attitude ascriptions puzzle. I conclude the section with a

ix
list of seven constraints that I believe any theory of proper names must respect. In section 1.2, I

set out a puzzle I call the ‘Problem of Rational Inconsistency.’ Along with Millians, I hold that

agents sometimes believe inconsistent singular propositions; moreover, they sometimes utter

sentences that express inconsistent propositions, thus contradicting themselves. Lois Lane, for

example, has inconsistent beliefs with respect to Kent/Superman, both believing and disbelieving

that he can fly when she conceives him in different ways, and she is disposed to contradict

herself by uttering sentences such as ‘Superman flies’ and ‘Clark Kent does not fly.’ Yet her

inconsistency, both in terms of what she believes and what she says about Kent/Superman, is not

due to any irrationality on her part. It must be explained instead by her ignorance of some fact or

set of facts. To solve the problem, we must elaborate a theory of names that elucidates what she

is ignorant of. What proposition does she fail to realize about the Kent/Superman identity and

why does she fail to realize it? Furthermore, why is she incapable of realizing that her

statements ‘Superman flies’ and ‘Clark Kent does not fly’ are inconsistent?

Chapter 2 is the heart of the dissertation, where I set out the TIUT and its proposed

solutions to the puzzles. I argue that proper names are used as indexicals that come in two

fundamentally different varieties. Sometimes a name is used indexically just to refer to its bearer.

I call this a ‘Millian use,’ and say that a name is used ‘in a Millian way.’ Other times, a name is

used indexically to refer to its bearer and to contribute the speaker’s descriptive conception of

that bearer to the proposition. I call this a ‘Conception-indicating use’ and say a name is used ‘in

a Conception-indicating way.’ Therefore, I call the theory ‘The Two Indexical Uses Theory of

Proper Names’, or ‘TIUT.’ I intend this difference in use to mark a semantic, and not merely

pragmatic, distinction. When used in a Millian way, a name is directly referential (though not in

the strictest sense, since, given the indexical use of the name, there is a character level of

x
meaning that determines content). 1 Its content is just its bearer. When used in a Conception-

indicating way, a name’s content, which is determined by character in the context of utterance, is

not an object, but rather a meaning that mediates reference. Thus, a name used in a Conception-

indicating way falls outside of David Kaplan’s (1989) paradigm of indexicality, according to

which indexicals are directly referential. When used in a Conception-indicating way, ‘Clark

Kent’ and ‘Superman’ are designate rigidly, but they have different contents that are partially

constituted by different descriptive elements. Crucially, these descriptive elements of content are

referentially inert, playing no role in determining reference. Instead, reference is a function of

causal-historical factors following Kripke’s picture of reference. After explaining the character

and content of names used in Millian and Conception-indicating ways, I go on to propose

solutions to the two versions of Frege’s puzzle, the Problem of Rational Inconsistency, and

Kripke’s puzzle. I also address the problem of quantifying into propositional attitude contexts,

as well as the epistemological status of informative identity sentences such as ‘Clark Kent is

Superman.’

In Chapter 3, I examine what indexicals are in general. I then go on to discuss one theory

of proper names, the theory of Pelczar and Rainsbury (1998), according to which proper names

are indexicals. The chapter is broken down into the following sections: in section 3.1, I address

the question: what are indexicals? In section 3.2, I discuss the debate about whether proper

names with multiple-bearers, such as ‘John,’ are multiply ambiguous expressions or indexicals.

According to the multiple-ambiguity view, each bearer of ‘John’ has a different name that

1
Kaplan (1989) holds that indexicals are directly referential, but Perry (1997) has clarified that
they are not so in the strictest sense because character mediates content. Kaplan does not think
that proper names are indexicals but rather paradigm cases of genuinely directly referential
expression with no mediating character meaning. He wrote in his 1989: “ ...in the case of proper
name words, all three kinds of meaning—referent, content, and character—collapse. In this,
proper name words are unique.” (562)
xi
happens to be spelled and pronounced the same in English, much as ‘bank’ referring alternately

to a financial institution or the edge of a river are distinct expressions that happen to be spelled

and pronounced the same way. By contrast, according to the indexical view, ‘John’ is one single

unambiguous expression that can refer to different individuals depending on contextual factors,

and is therefore used as a species of indexical. Finally, in section 3.3, I discuss the theory of

Pelczar and Rainsbury (1998), according to which names are used as indexicals (as on the TIUT)

but they come in only one variety. The theory is species of Millianism because the contents of

‘Clark Kent’ and ‘Superman’ are identical—just the name’s bearer. However, unlike standard

forms of Millianism, according to which the bearer fully exhausts the meaning of a name, Pelczar

and Rainsbury maintain that names, being indexicals, also have a character meaning. Thus, there

is more to meaning of a proper name than just its bearer on the view. I argue that Pelczar and

Rainsbury’s theory is well motivated and goes a long way towards solving the Problem of

Rational Inconsistency. However, it does not solve Frege’s puzzles. Two indexical uses, one on

which a descriptive conception is contributed to content, as the TIUT posits, are required to solve

them.

In Chapter 4, I discuss Descriptivism, a category into which I group together the views of

Frege and Russell (and to which I refer as ‘classic descriptivism’), prescinding from the

significant differences between their views. I discuss Kripke’s critique of Descriptivism, and

characterize Kripke’s alternate picture of reference, the causal-historical picture. I then point out

some potential worries for the causal-historical picture, although I go on to endorse Kripke’s

view and incorporate it directly into the TIUT (according to which the reference of a name is a

function only of causal-historical factors, and the descriptive elements belonging to content play

no role in determining reference). Finally, I discuss three modern varieties of Descriptivism—

xii
Rigidified Descriptivism, Causal Descriptivism, and Metalinguistic Descriptivism, and show that

each of them is vulnerable to Kripke’s arguments against Descriptivism in Naming and

Necessity and/or inadequate to solve the puzzles for other reasons.

In Chapter 5, I discuss Millianism, the view that the meaning of a name is exhausted by

its bearer. Alternatively, it can be described as the view that a proper name contributes its bearer

and nothing else to the proposition expressed by the sentence in which it occurs. 2 3
In the

introduction to the chapter, I discuss the two main Millian strategies to solve the puzzles:

(a) To solve Frege’s puzzles: the claim that ordinary speakers do not carefully distinguish

between the propositions semantically expressed by sentences and the propositions they

are used to pragmatically communicate, and this explains ordinary speakers’ erroneous

Fregean intuitions about cognitive value and truth-value; and

(b) To solve the Problem of Rational Inconsistency: the claim an agent can rationally

express and believe inconsistent propositions as long as he or she takes them under

different propositional guises that he or she does not realize such are guises of the same

proposition.

I then set out the four worries for these two Millians strategies (listed a – d) that beset Millianism

in addition to the most commonly known problem (i.e., that Millianism is inconsistent with

2
These definitions of Millianism are not exactly equivalent. The view of Pelczar and Rainsbury
(examined in chapter 3), on which a name contributes its bearer alone to content, but has a
character meaning in addition to content, would be classified as a species of Millianism only on
the second definition.
3
It is common to distinguish the terms “Millian” and “Naïve-Russellian” in the literature, with
the former term concerned with a theory as to the meaning or content of proper names, and the
latter concerned with a particular view of propositions (favoring “singular propositions”).
Typically, the views overlap, since most Millians tend to be Naïve-Russellians. For the sake of
simplicity, I shall elide over this distinction and use “Millian” throughout with the underlying
presumption that most Millians accept singular propositions as the content of sentences.

xiii
strong intuitions about cognitive value and truth-value with respect to Frege’s puzzle). Finally, I

examine the most prominent Millian theories, those of Nathan Salmon (1986) and Scott Soames

(2002), and address how the four worries I identity (a – d) affect their theories. 4

In Chapter 6, I examine the Hidden Indexical Theory (“HIT”), first proposed by Schiffer

in his 1977. Problematically, the HIT proposes a solution to the propositional attitudes puzzle

but not the identity sentences puzzle. I argue that the identity sentences puzzle is the more

fundamental puzzle and it should be solved first. Once we explain why ‘Clark Kent is

Superman’ and ‘Clark Kent is Clark Kent’ differ in cognitive value, we can explain why the

propositional attitude ascriptions in which those identity sentences are embedded in the ‘that’-

clauses differ in truth-value.

In Chapter 7, I look at Graeme Forbes’ theory of proper names (1990), which, like the

TIUT, invokes mental dossiers in defining the meaning of proper names. I argue that Forbes’

theory is overly metalinguistic and cannot handle attitude ascriptions in which the ascribee of the

ascription is a non-verbal agent.

In Chapter 8, I discuss the Saul Cases, which are due to Jennifer Saul (1997). Saul

presents an additional puzzle. She noticed that sentence pairs such as ‘Superman is more

successful with women than Clark Kent’ and ‘Superman is more successful with women than

Superman’ seem to differ in truth-value even though the names do not occur within opaque

contexts, e.g., inside the ‘that’-clause of a propositional attitude ascription. However, in

agreement with Moore (1999) and Pitt (2001), I argue that Saul’s examples are not genuine cases

of substitution failure of co-referential names because the names ‘Clark Kent’ and ‘Superman’

4
In his 2002, Soames does not mention propositional guises. Subsequently, however, after
criticism of his view by David Braun and Nathan Salmon, he wrote in his 2006 (5) that he did
not mean to exclude propositional guises from his theory.
xiv
are not used co-referentially in Saul’s examples, each name instead referring to a different aspect

of Kent/Superman.

In Chapter 9, I touch upon Anti-functionalist theories of proper names, including

Variablism and Predicativism.

In Chapter 10, I briefly discuss Recanati’s Theory of Mental Files, distinguishing the

aims of his theory from those of the Two Indexicals Theory.

xv
 CHAPTER 1

THE PUZZLES

1.1 Frege’s Puzzle

The philosopher and mathematician Gottlob Frege most clearly stated the eponymous

“Frege’s puzzle” in his seminal 1892 paper On Sense and Reference. There are two versions of

the puzzle: the puzzle about identity sentences and the puzzle about propositional attitude

ascriptions.

First, let us briefly examine the puzzle about identity sentences. Treat the Superman

story as non-fictional and consider the following identity sentences, (1) and (2), which differ in

one respect only: one co-referential name has been substituted for another.

(1) Clark Kent is Clark Kent

(2) Clark Kent is Superman

In (2), ‘Superman’ is substituted for the second occurrence of ‘Clark Kent’ in (1). Consider now

the simplest theory of proper names—a theory attributed to the nineteenth-century English

philosopher John Stuart Mill and called “Millianism” in his honor: a proper name always

contributes its bearer only and nothing more to the proposition expressed by a sentence in which

it occurs. Frege argued that Millianism was false by the following sort of reductio. Suppose that

Millianism were true. ‘Clark Kent’ and ‘Superman’ would each contribute their common bearer,

the flesh-and-blood man himself—to whom I’ll refer throughout this dissertation as ‘Kent-

Super’—to the proposition(s) expressed by sentences (1) and (2). Hence, (1) and (2) would

express the same proposition—the singular proposition that Kent-Super is Kent-Super—which

we may schematize as PROP-1.

1
 PROP-1 << Kent-Super, Kent-Super >, identity >>

However, the notion that (1) and (2) express the same proposition is highly counterintuitive.

Sentence (2) is interesting and informative to Lois Lane, who does not realize that Clark Kent is

Superman. Sentence (1), by contrast, merely states the obvious—that a certain man is identical to

himself. It is uninteresting, uninformative, and trivial. Lois might learn something from (2) but

not from (1). A rational agent might harbor doubts about the truth of (2) but not about (1). 5 Frege

summed up these salient differences between sentences such as (1) and (2), “uninformative” and

“informative” identity sentences respectively, by characterizing them as differing in

Erkenntniswert or ‘cognitive value.’ It would be difficult, if not impossible, to explain the

difference in cognitive value between (1) and (2) if they expressed the same proposition, so (1)

and (2) must express different propositions. If they express different propositions, the names

‘Clark Kent’ and ‘Superman’ must make different contributions to them, and therefore

5
Here is an important qualification. Sentence (1) not subject to rational doubt as long as a
speaker utters it to express the trivial proposition that Clark Kent is self-identical. As I touch
upon in section 2.1 and discuss at further length in section 2.9, natural language sentences of the
syntactic form a=a, such as sentence (1), are not always uninformative identities. They do not
always express trivial propositions about self-identity whose truth-value any rational agent may
ascertain merely by inspecting the syntactic form of the sentence. In fact, a speaker might utter
sentence (1) to express an informative identity. For example, a man—call him ‘Tom’—might
utter (1) to himself to express his belief that the Clark Kent he meets at a party is the same Clark
Kent he went to Kindergarten with in Smallville. A rational agent could very well wonder
whether Tom expressed a true proposition in uttering (1) and could not tell whether (1), as
uttered by Tom, was true or false merely by considering the syntactic form of the sentence.
Alternatively, consider Kripke’s Paderewski case. A man named ‘Peter’ might utter ‘Paderewski
has musical talent’ when thinking about Paderewski the famous pianist, but also might utter
‘Paderewski does not have musical talent’ when thinking about Paderewski the politician. Peter
fails to realize that the pianist and the politician are the same person. Peter’s friend might utter
‘But Peter, Paderewski is Paderewski’ to Peter in an attempt to convince him that Paderewski the
politician is the same person as Paderewski the musician. No one can tell, based on syntactic
form alone, whether in uttering the sentence ‘… Paderewski is Paderewski’ Peter’s friend
expressed a true or false proposition. This information cannot be read off the tautological
syntactic form of the sentence uttered.
2
Millianism must be false.

In this dissertation, I am going to take it as a working assumption, with Frege, that this

reductio in fact refutes Millianism and I shall argue for a non-Millian theory of proper names.

Here then is Frege’s puzzle about identity sentences: if Millianism is false, what then is the

correct theory of proper names? Just what are the different contributions of the proper names

‘Clark Kent’ and ‘Superman’ to the propositions expressed by sentences (1) and (2) such that we

may explain why these sentences differ in cognitive value?

Now let us briefly examine the puzzle about propositional attitude ascriptions. Frege

noticed that the substitution of one co-referential name in place of another inside the ‘that’-clause

of a propositional attitude ascription sentence might change its truth-value, rather than its

cognitive value (as in the identity sentences puzzle described above). Consider propositional

attitude ascription sentences (3) and (4).

(3) Lois Lane believes that Clark Kent is Clark Kent

(4) Lois Lane believes that Clark Kent is Superman

In (4) ‘Superman’ is substituted for the second occurrence of ‘Clark Kent’ in (3), inside the

‘that’-clause. Intuitively, sentence (3) is true and sentence (4) is false. Sentence (3) is true

because Lois Lane is acquainted with Clark Kent and, being a rational person, she realizes that

he is identical to himself. Sentence (4) is false because, according to the Superman story, Lois

Lane does not realize that Clark Kent is the same person as Superman. A speaker might indeed

utter the negation of (4), either sentence ¬ (4) or (4n), to characterize her ignorance of the

identity.

¬ (4) Lois Lane disbelieves that Clark Kent is Superman

3
 (4n) Lois Lane does not believe that Clark Kent is Superman

Differences in truth-value resulting from the substitution of co-referential names also may occur

where the ‘that’-clause is not about identity. For example, intuitively (5) is false and (6) true.

(5) Lois Lane believes that Clark Kent flies

(6) Lois Lane believes that Superman flies

However, Millianism entails that our intuitions about the truth-value of sentence pairs (3)-(4) and

(5)-(6) are erroneous. According to Millianism, both (3) and (4) express the proposition that

Lois Lane believes PROP-1, i.e., the singular proposition that Kent-Super is identical to himself.

Since Lois believes this, 6 (3) and (4) are both true. Contrary to our intuitions about the matter,

Millianism has it that (4) is true and therefore Lois Lane believes that Clark Kent is Superman.

According to Millianism, (5) and (6) both express the proposition that Lois Lane believes that

Kent-Super flies. Since Lois believes that proposition, 7 (5) and (6) are both true. Contrary to

our intuitions about the matter, Millianism has it that (5) is true and Lois Lane believes that Clark

Kent flies. However, the simplest explanation for our intuitions that there are truth-value

differences is the supposition that our intuitions are correct and the members of these sentence

pairs express different propositions with different truth-values, and therefore Millianism must be

false.

6
She believes that Kent-Super is Kent-Super both when she conceives him as Superman and
when she conceives him as Clark Kent, for she would assent to both ‘Clark Kent is Clark Kent’
and ‘Superman is Superman.’
7
She believes that Kent-Super flies when she conceives him as Superman. According to
Millianism, (5) is true because it expresses the proposition that Lois believes that Kent-Super
flies, full stop, without respect to how she conceives Kent-Super when she judges that he flies.

4
 As with the identity sentences puzzle, I am going to take it as a working assumption that

this reductio refutes Millianism as well. I do not think that modern Millian theories elaborated

over the past forty years, which attempt to rescue Millianism from the force of these arguments,

are plausible. My task is to formulate an theory of proper names that respects speaker intuitions

about cognitive and truth-value. The two versions of Frege’s puzzle, taken together, pose the

following question: what does a proper name contribute to the proposition expressed by a

sentence in which it occurs such that we may explain the cognitive and truth-value differences

illustrated above? To solve the puzzle, I develop theory of proper names according to which co-

referential names such as ‘Clark Kent’ and ‘Superman’ may, at least on some tokenings, have

different contents and make different semantic contributions to the propositions expressed by

sentence pairs (1)-(2), (3)-(4), and (5)-(6), such that the members of these sentence pairs would

express different propositions.

In his seminal 1892 paper On Sense and Reference, Frege proposed a solution to the

problem he had raised for Millianism. He denied the Millian thesis that a proper name

contributes its bearer to the proposition. In very rough sketch, Frege proposed instead that a

proper name contributes its Sinn or “sense.” Frege was not entirely clear about what he meant by

sense. Most philosophers have interpreted a sense to be equivalent to the meaning of a definite

description specifying a “uniqueness condition,” i.e., a condition uniquely satisfied by an

individual or object (and therefore Frege’s theory is standardly classified as a species of

Descriptivism). The referent of a name 8 would be the object or individual satisfying the

8
Frege uses the term Bedeutung to mean ‘reference.’ This is somewhat unfortunate and
confusing, given that this German word simply means ‘meaning.’ In current German, the terms
Bezugsgegenstand or Referenzobjekt would better express what Frege meant by Bedeutung.

5
uniqueness condition, i.e., the object or individual uniquely denoted by the definite description. 9

In other words, the definite description expressing the sense of a name determines its reference.

For example, the sense of the name ‘Clark Kent’ might be expressed by the definite description

‘the mild-mannered reporter from Smallville working for the Daily Planet.’ The sense of the

name ‘Superman’ might be expressed by the definite description ‘the caped superhero that

protects Metropolis.’ The names ‘Clark Kent’ and ‘Superman’ would co-refer because these

definite descriptions determine the same reference, i.e., they denote the same individual. 10 We

can now see why (1) and (2) differ in cognitive value according to Descriptivism. Sentence (1)

9
In his Stanford Encyclopedia of Philosophy article on Meaning
(http://plato.stanford.edu/entries/meaning/#FrePro), Jeff Speaks explains why it is reasonable to
interpret Frege as intending senses to be equivalent to the meaning of definite descriptions:

Here is one initially plausible way of explaining what the sense of a name is. We know
that, whatever the content of a name is, it must be something that determines as a
reference the object for which the name stands; and we know that, if Fregeanism is true,
this must be something other than the object itself. A natural thought, then, is that the
content of a name—its sense—is some condition that the referent of the name uniquely
satisfies. Co-referential names can differ in sense because there is always more than one
condition that a given object uniquely satisfies. (For example, Superman/Clark Kent
uniquely satisfies both the condition of being the superhero Lois most admires, and the
newspaperman she least admires.) Given this view, it is natural to then hold that names
have the same meanings as definite descriptions—phrases of the form ‘the so-and-so.’
After all, phrases of this sort seem to be designed to pick out the unique object, if any,
which satisfies the condition following the ‘the.’
10
However, in propositional attitude contexts the names ‘Clark Kent’ and ‘Superman’ would not
co-refer, according to Frege’s theory. In propositional attitude contexts, ‘Clark Kent’ and
‘Superman’ would refer to their senses, rather than their usual referent, Kent-Super. Thus,
‘Clark Kent’ would refer to the sense of ‘Clark Kent’ instead of referring to Kent-Super, and
‘Superman’ would refer to the sense of ‘Superman’ instead of referring to Kent-Super. Note also
that in these propositional attitude contexts all of the words inside ‘that’-clauses refer to their
senses instead of their usual referents. In essence, therefore, by claiming this reference shift,
Frege maintained that the words embedded in ‘that’-clauses of propositional attitude ascriptions
refer to the propositions they would ordinarily express were the words not so embedded. Thus,
Frege endorsed the view, which Schiffer (2003) has called the “face value” analysis of
propositional attitude reports, that belief ascriptions report the existence of the belief relation
between an agent and a proposition, to which the ‘that’-clause refer. ‘That’-clauses are, on this
view, singular terms referring to propositions.

6
would be uninteresting and uninformative because it would express the trivial and obvious

proposition that the mild-mannered reporter from Smallville working for the Daily Planet is the

mild-mannered reporter from Smallville working for the Daily Planet. Sentence (2) would be

interesting and informative because it would express the non-trivial and non-obvious proposition

that the mild-mannered reporter from Smallville working for the Daily Planet is the caped

superhero that protects Metropolis. 11 Bertrand Russell (1905) proposed a superficially similar

theory according to which proper names abbreviate reference-determining definite descriptions.

Despite important differences between Frege and Russell’s theories, both are often referred to as

the “Frege-Russell theory” of proper names (Kripke, 1980) or as “Descriptivism,” because both

views propose that the meaning of a proper name is equivalent to the meaning of a reference-

determining definite description and that the meaning of a name can be expressed by this definite

description.

In his seminal work Naming and Necessity (1980), Saul Kripke presented powerful

arguments against Descriptivism that thoroughly undermined it in the view of most philosophers,

myself included. In the process, he made the case for (at least) two important theses about

proper names that are widely accepted nowadays. First, proper names are rigid designators.

Thus, ‘Clark Kent’ and ‘Superman’ refer to the same individual, Kent-Super, in the actual world

and in every possible world, 12 which entails that the propositions expressed by sentences (1) and

11
The received view is that Frege was a Descriptivist and that he would have endorsed this sort
of solution to the puzzle about identity sentences. However, some philosophers have argued that
Frege was not in fact a Descriptivist. See, e.g., Burge, T. Sinning Against Frege, in Philosophical
Review 88, 1979, pp. 398-432. I recognize that Frege’s views are subject to various
interpretations. The genuine nature of Frege’s views is, however, orthogonal to the primary
purpose of this dissertation (which is to present a novel theory of proper names, not to carry out
an exegesis of Frege’s views) so I shall not be concerned with the issue here.
12
I write about proper names here as if they were obstinately rigid designators, as opposed to
7
(2) have the same modal profile. Problematically for Descriptivism, (according to most versions

of it) proper names are not rigid designators and (1) and (2) would express propositions with

different modal profiles. 13 Second, Kripke made the case that proper names lack reference-

determining descriptive meanings. Thus, even if a large percentage of a language community

were regularly to associate the name ‘Superman’ with the definite description ‘the superhero that

protects Metropolis,’ the name ‘Superman’ would not be synonymous with it, and the reference

of ‘Superman’ would not be determined by that definite description. 14 It is my working

assumption in this dissertation that these Kripkean theses—that proper names are rigid

designators and lack reference-determining descriptive meanings—are correct.

Subsequent to Naming and Necessity and the widespread rejection of Descriptivism

persistently rigid designators, although the arguments I make in this paper do not ride on which
notion of rigidity is correct. A designator is obstinately rigid if it designates the same object in
every possible world, even in those worlds in which that object does not exist. By contrast, an
expression is persistently rigid if it designates the same object in every possible world in which
that object exists and designates nothing in those worlds in which that object does not exist.
Salmon expressly drew the distinction in his 1981. Kripke alternates between these two
conceptions of rigidity in his writings, although more often than not he gives the impression that
he favors persistent rigidity. Many philosophers have argued that the notion of obstinate rigidity
is the better one. See, e.g., Branquinho, João. 2003. “In Defense of Obstinacy.” Nous-
Supplement: Phil. Perspectives 17: 1-23.
13
“Rigidified Descriptivism” proposes that the associated definite descriptions are to be
understood as rigidified with the actually operator (often written is ‘@’) or dthat operator so that
the name ‘Superman’ would mean, e.g., the actual superhero that protects Metropolis and the
name ‘Clark Kent’ would mean the actual mild-mannered reporter from Smallville who works
for the Daily Planet. On this account, the names ‘Clark Kent’ and ‘Superman’ would indeed be
rigid designators, and sentences (1) and (2) would express propositions with the same modal
profile. While Rigidified Descriptivism deals nicely with Kripke’s rigidity concerns, it is
nevertheless vulnerable to his semantic and epistemic arguments against Descriptivism.
14
Kripke argued that reference is instead a function of causal-historical factors. See, e.g.,
https://plato.stanford.edu/entries/reference/#CauThe for discussion of Kripke’s causal theory of
reference.
8
amongst philosophers, many philosophers were attracted back to Millianism. 15 They developed

modern versions of it that aimed to explain away its counterintuitive implications. 16 Neither

Descriptivism nor Millianism is adequate, and it is my purpose to formulate a theory of proper

names that is neither Millian nor Descriptivist, a theory with all of the virtues of each and none

of the drawbacks of either. The revival of Millianism beginning in the 1970’s was a step away

from Descriptivism too far in the opposite direction. Many philosophers concluded that because

Kripke showed that proper names lack reference-determining descriptive meanings, they must

altogether lack any sort of meaning. They must be meaningless “tags” (Barcan Marcus, 1961) or

“labels” with only one semantic role—to refer to their bearers. Against Millianism, I argue that

names have character meanings, the sort of meaning borne by indexicals. Furthermore, I argue

that on some tokenings, names have contents partially constituted by contextually determined

descriptive elements, although these are not reference-determining as they would be on

Descriptivism.

Although proper names do not have reference-determining descriptive meanings, it is a

datum that speakers frequently associate descriptive conceptions with proper names and

regularly and purposefully use proper names to communicate these descriptive conceptions to

their audience. For example, we typically associate different descriptive conceptions with the

names ‘Clark Kent’ and ‘Superman,’ ‘Clark Kent’ with a mild-mannered reporter conception and

‘Superman’ with a strong superhero conception. We would typically utter sentence (2)

15
Millianism had been out of fashion for most of the 20th century prior to its revival in the
1970’s and 1980’s.
16
A slim plurality of philosophers today are Millians or sympathetic towards Millianism. See:
https://philpapers.org/surveys/results.pl?affil=All+respondents&areas0=15&areas_max=1&grain
=coarse

9
 (2) Clark Kent is Superman

to pick out two conceptions of Kent-Super—a ‘Clark Kent-y’ mild-mannered reporter

conception and a ‘Superman-y’ strong superhero conception—and say that both conceptions

relate to the same individual, Kent-Super.

We would typically utter sentence ¬ (4)

¬ (4) Lois Lane disbelieves that Clark Kent is Superman

to say that Lois Lane disbelieves that Kent-Super thought of under a Clark Kent-y conception is

the same person as Kent-Super thought of under a Superman-y conception. We would utter ¬ (5)

¬ (5) Lois Lane disbelieves that Clark Kent flies

to say that Lois Lane disbelieves that Kent-Super flies when she thinks of him under a

Clark Kent-y conception. In these cases, speakers regularly and quite purposefully use proper

names both to refer to an individual and to convey a descriptive conception of that individual.

This undercuts the Millian claim that the sole semantic function of proper names is to refer. 17

The Kent/Superman case is somewhat misleading because it lends the false impression

that all speakers associate the same descriptive conceptions with proper names, as if these

conceptions were invariable and fixed to names by convention. However, in real life, most

Frege’s puzzle cases involve idiosyncratic conceptions associated with names by particular

individuals, and said conceptions may vary from one conversational context to another. Proper

names can be used to communicate these idiosyncratic conceptions as well. For example, my

17
Many Millians would claim that descriptive conceptions are conveyed by pragmatic, rather
than semantic mechanisms. I shall argue the mechanism is semantic.

10
eleven-year-old niece Emma does not realize that Mrs. Green, her sixth-grade teacher, is a

former lead Rockette who performed at Radio City Music Hall in the late 1960’s under the stage

name ‘Roxanne Rockets.’ She associates very different conceptions with these names ‘Mrs.

Green’ and ‘Roxanne Rockets.’ I can inform her of the identity by uttering ‘Mrs. Green is

Roxanne Rockets.’ Here, the conceptions associated with these names are particular to Emma

and not shared across the wider language community.

Although these Frege’s puzzle cases arise from time to time—in which I shall be arguing

that names semantically contribute non-reference determining descriptive conceptions to

content—nevertheless speakers use proper names just as Millians claim most of the time—

merely to refer, just to call their audience’s attention to the right individual or object. In other

words, the Millian picture of proper names is right most of the time. We use proper names to

communicate conceptions less frequently, where there are two (or more) salient conceptions of

an object or individual in a conversational context and we wish to distinguish between or draw a

contrast between them, as in the puzzle cases illustrated above (involving Lois Lane, Clark

Kent/Superman; or Mrs. Green/Roxanne Rockets). See section 2.1, infra, for discussion of my

claim that we regularly and purposely use proper names in these two fundamental sorts of ways:

to refer and nothing else (which I call ‘Millian uses’ of names), and to refer and convey

descriptive conceptions as well (which I call ‘Conception-indicating uses’ of names).

In light of the foregoing considerations, to solve Frege’s puzzle I propose that we need a

theory of proper names that, at a minimum, respects the following seven constraints:

1. Proper names are rigid designators.

2. Proper names do not have reference-determining descriptive meanings.

3. Identity sentences like (1)-(2) express different propositions with the same modal

11
 profile.

4. Propositional attitude ascriptions like (3)-(4) and (5)-(6) express different

propositions differing in truth-value.

5. Sometimes, we use a proper name as Millians claim, merely to refer, in which

case the name semantically contributes its bearer only to the proposition
expressed.

6. Sometimes, we use a proper name both to refer and to convey a descriptive

conception, in which case the name semantically contributes its bearer plus a
descriptive conception to the proposition expressed.

7. The descriptive conceptions we use names to communicate are not conventionally

“built into” names and may be idiosyncratic and variable, differing from speaker
to speaker and from one conversational context to another. A theory of proper
names must provide that names are contextually sensitive in such a way that they
load the relevant descriptive conceptions into propositional content.

With respect to a pair of co-referential names such as, e.g., ‘Clark Kent’ and

‘Superman’ 18 we therefore need a theory on which these names:

(a) Rigidly designate Kent-Super in all possible worlds, regardless of the different

descriptive conceptions a speaker might associate with the names or intend to

communicate to his/her audience when tokening either name;

(b) Differ in semantic content when we purposely use them to communicate different

descriptive conceptions, which descriptive conceptions constitute a non-reference-

determining element of semantic content;

18
I leave open the possibility that ‘Clark Kent’ and ‘Superman’ can be used non-co-referentially.
Consider the sentence ‘Clark Kent went into the phone booth and Superman came out,’ which
example is due to Jennifer Saul (1998). (See Chapter 8, infra, for more detailed discussion of this
case). I would contend that ‘Clark Kent’ and ‘Superman’ are not co-referential in this case, as
they refer to different aspects of a single individual, or perhaps to as an individual dressed in
different ways. See Moore (1999) and Pitt (2001), who support this view. However, I take it that
the names ‘Clark Kent’ and ‘Superman’ in (both versions of) Frege’s puzzle are co-referentially
used, since Frege’s puzzle is based on the presupposition that Clark Kent is Superman. If ‘Clark
Kent’ and ‘Superman’ did not co-refer, sentence (2) (‘Clark Kent is Superman’) would be false,
which it is not.

12
 (c) Are identical in semantic content when we use them merely to refer to Kent-Super, in

which case the semantic content of either of the names is just their bearer, Kent-

Super, the flesh-and-blood man himself.

1.2 The Problem of Rational Inconsistency

The ‘Problem of Rational Inconsistency’ is an important companion puzzle to the two

versions of Frege’s puzzle. 19 20

The capacity to solve this puzzle should be considered an eighth

constraint for any minimally adequate theory of proper names, to be added to the list of seven

elaborated at the end of the previous section. Philosophers often discuss this puzzle in the same

breath as discussing Frege’s puzzles without separately identifying it, suggesting they may not be

properly distinguishing the puzzles. Indeed, they should be distinguished. The Problem of

Rational Inconsistency challenges us to explain how an agent’s expressing and believing

inconsistent propositions can be compatible with his or her rationality. Ordinarily, we would say

that believing inconsistent propositions is incompatible with an agent’s rationality. For example,

if an agent were to claim (and not in jest) that it is raining outside now and that it is not raining

19
This puzzle is similar to but not identical to Kripke’s Paderewski puzzle (Kripke, 1979).
There, Kripke poses the question whether Peter holds inconsistent beliefs with respect to
Paderewski, depending on whether he conceives Paderewski as a musician or a politician. Over
and above this, Kripke addresses wider issues about our practices of attitude attribution.
Supposing that we decide that Peter does in fact hold inconsistent beliefs (which I do), why is it
that we intuit that there is something wrong with belief ascriptions of the form ‘Peter believes
Paderewski has musical talent and believes that he does not have musical talent’? Why do most
ordinary speakers take this belief ascription to be incoherent or at least highly unidiomatic? As
Mark Richard put it, Kripke raises the question of how to express what Peter believes “in the
idiom for belief ascription provided by English, if we limit ourselves to identifying the object of
his beliefs [with the name ‘Paderewski’].” To address Kripke’s puzzle, I will need to employ a
bit of pragmatics. I address Kripke’s puzzle in section 2.12.
20
Pelczar and Rainsbury (1998) call this puzzle ‘The Problem of Coherent Rationality.’ I look at
their theory in section 3.3, infra.
13
outside now, we would judge that agent irrational. 21 However, according to Millianism, Lois

Lane, who we may presume is a rational and reflective agent, both believes and expresses

inconsistent singular propositions with respect to Kent-Super. Suppose someone asked her to

name a non-avian flying being and in response, she uttered sentence (8).

(8) Superman flies

According to Millianism, (8) would express the singular proposition that Kent-Super flies.

Suppose someone then asked Lois to name a non-avian being incapable of flight and she uttered

¬(7) in response.

¬(7) Clark Kent does not fly

According to Millianism, ¬(7) would express the negation of the singular proposition expressed

by sentence (8). To wit, ¬(7) would express the singular proposition that it is not the case that

Kent-Super flies. In uttering (8) and ¬(7), Lois would contradict herself, according to

Millianism. She would say that Kent-Super flies and that he does not. Assuming that Lois

believes the propositions expressed by sentence (8) and ¬(7)—sentences she understands and

accepts—then she would believe inconsistent propositions, 22 a proposition and its negation.

Although not a Millian theory of proper names, the TIUT concurs with Millianism that Lois

Lane would indeed express and believe inconsistent singular propositions in uttering sentences

21
Assume here that the agent means that it is both raining and not raining in the same place at the
same time in the same way.
22
That Lois Lane believes the propositions expressed by sentences ¬(7) and (8) follows from the
“weak disquotation principle” (Kripke, 1979), which says: if a competent, sincere, reflective, and
rational speaker s who understands a sentence S is disposed to accept S, and believes S to be
true, then s believes the proposition semantically expressed by S.
14
(8) and ¬(7). The TIUT takes it as a datum that proper names are often used in a Millian way,

just to refer, and that in this case Lois would indeed use the names ‘Clark Kent’ and ‘Superman’

in a Millian way, thus expressing inconsistent singular propositions (she is after all not intending

to distinguish between differing conceptions of Kent-Super and communicate them to her

audience when she utters (8) and ¬(7), given that she does realize she is speaking about one

individual conceived in two ways; see section 2.1, infra, for discussion). The problem of rational

inconsistency challenges us to answer the following two questions:

THE PROBLEM OF RATIONAL INCONSISTENCY

(1) How do we square Lois believing inconsistent singular propositions with her
being a rational agent?

(2) Why is Lois incapable of realizing that she expresses inconsistent propositions
and contradicts herself in uttering sentences (8) and ¬(7), sentences she
understands and uses competently?

To solve the problem—which can be divided into two sub-problems, as indicated

above—we will have to claim that Lois has these inconsistent beliefs because of ignorance, not

irrationality. Furthermore, her ignorance of some fact or facts renders her incapable of seeing

that the propositions expressed by (8) and ¬(7) are inconsistent. She is rational despite believing

inconsistent propositions, entertaining what I shall call ‘rationally inconsistent beliefs.’ Most

Millians appeal to the notion of propositional guises to solve the problem (see Chapter 5, where I

discuss modern Millian theories). The TIUT (section 2.8) instead appeals to the claim that

proper names are used as indexicals to solve the problem—especially with respect to addressing

prong (2) of the problem.

15
 CHAPTER 2

THE TWO INDEXICALS THEORY OF PROPER NAMES

AND ITS SOLUTION TO THE PUZZLES

According to the TIUT, proper names have two possible semantic uses. When we use a

proper name in a ‘Millian way,’ it merely contributes its bearer to the proposition expressed by a

sentence in which it occurs. When we use a proper name in a ‘Conception-indicating way,’ it

contributes its bearer as well as the descriptive conception of the bearer that the speaker has in

mind when he or she utters the name. (N.B.: I claim this descriptive conception is part of content

but is not reference determining.) Names function as indexicals and are always rigid designators

whether we use them in a Millian or a Conception-indicating way. I discuss reasons for

supposing that names have these two uses in section 2.1, below. In section 2.2, I introduce

several terms of art I use to characterize the character and content of proper names on their two

indexical uses, which I then set out in sections 2.3 (for Millian uses) and 2.4 (for Conception-

indicating uses). In the sections thereafter, I show how the TIUT solves the puzzles, including

Frege’s puzzle (both versions), the Problem of Rational Inconsistent Belief, and Kripke’s

Paderewski Puzzle (which encompasses what I have termed the ‘Problem of Rational

Inconsistent Belief’ but raises additional issues about our practices of belief attribution).

2.1 Two Uses of Proper Names

As discussed above in section 1.1, Frege’s puzzles—both the puzzle about identity

sentences and the puzzle about propositional attitude ascriptions—focus a spotlight on situations

in which the descriptive conceptions that speakers associate with proper names are important

pieces of information they convey when they utter them. In these case, these descriptive

conceptions may even be the most important information communicated. I will be claiming that
16
these descriptive conceptions constitute part of the semantic content of the names (rather than

being merely pragmatically conveyed, as many Millians would claim). Although these

conceptions are part of content, they are not reference-determining.

Speakers would typically utter sentence (2)

(2) Clark Kent is Superman

to communicate two descriptive conceptions of Kent-Super—a ‘Clark Kent-y’ mild-mannered

reporter conception and a ‘Superman-y’ strong superhero conception—and say that both

conceptions relate to the same individual, Kent-Super. Speakers typically utter sentence ¬ (4)

¬ (4) Lois Lane disbelieves that Clark Kent is Superman

to say that Lois Lane disbelieves that Kent-Super conceived in a Clark Kent-y way is the same

person as Kent-Super conceived in a Superman-y way. They use sentence ¬ (5)

¬ (5) Lois Lane disbelieves that Clark Kent flies

to say that Lois Lane disbelieves that Kent-Super flies when she conceives him in a Clark Kent-y

way.

Even the two occurrences of ‘Clark Kent’ in sentence (1) could be used to pick out

different conceptions of the same individual.

(1) Clark Kent is Clark Kent

For example, consider again the example discussed in footnote 5, supra: a man named ‘Tom’

meets Clark Kent, the reporter for the Daily Planet, at a party and suspects that he is the Clark

17
Kent he went to Kindergarten with in Smallville. Tom utters (1) to himself to express his belief

that the Clark Kent from the party is the same Clark Kent he went to Kindergarten with in

Smallville. Here, (1) is used not to express a trivial self-identity—that a man is identical to

himself—but rather to say something informative and interesting. The two tokenings of ‘Clark

Kent’ in (1) allude to different conceptions of Kent-Super—one the conception of an adult man

met at a party and the other of a Kindergartner. Alternatively, consider Kripke’s Paderewski case

(1979). A man named ‘Peter’ might say ‘Paderewski the pianist has musical talent, but

Paderewski the politician surely does not,’ failing to realize that the pianist and the politician are

the same person. A friend of Peter’s who realizes the pianist is the politician might correct

Peter’s confusion by saying ‘But Peter, Paderewski is Paderewski,’ tokening ‘Paderewski’ twice

to allude to different conceptions of the man—one the conception of a pianist and the other of a

politician.

At the same time, in some conversational contexts (perhaps in most), speakers lack the

intent to convey conceptions when they utter a name. In most cases, speakers do not use names

to draw contrasts between what is or is not the case depending on how an object or individual is

conceived (e.g., contrasting Lois Lane’s varying beliefs with respect to Kent-Super depending on

how she conceives him). Instead, they use names merely to direct the audience’s attention to the

right person/object so that the speaker can say something about that person/object. Such speakers

use names in what I call a ‘Millian way.’

I take Frege’s puzzle itself to constitute solid evidence that proper names are sometimes

used to communicate descriptive conceptions. There is also ample evidence, which I shall now

discuss, that we regularly use proper names in a Millian way, merely to pick out a name’s bearer,

both in simple sentences and propositional attitude ascriptions.

18
 Let us first look at the case of names occurring inside the ‘that’-clause of propositional

attitude ascriptions. According to Lycan (2000), it is “undeniable” that a proper name may occur

there “just to refer to its bearer, without any further suggestion about the way in which the

subject of the belief sentence would have represented the bearer.” Lycan offers the following

example:

“Suppose that Smith and Jones are among the few people who know that their
acquaintance Jacques is in fact the notorious jewel thief that has been terrorizing Paris’
wealthy set, called “Le Chat” in the popular press and by the gendarmes. Smith and
Jones read in the newspaper after a particularly daring but flawed robbery that the
gendarmes believe that “Le Chat dropped the fistful of anchovies as he or she ran.”
Smith and Jones say to each other, “The gendarmes think Jacques dropped the anchovies
as he ran.””

Clearly, Smith and Jones’ use of the name ‘Jacques’ is not intended to suggest that the

gendarmes would use that name to refer to the thief. Smith and Jones are aware that the

gendarmes, not realizing that Jacques is Le Chat, would refer to the thief only using the name

‘Le Chat.’ However, here, Smith and Jones are indifferent to the way that the gendarmes

conceive the thief and their sole concern in uttering the name ‘Jacques’ is pick out a certain

individual, Jacques/Le Chat, to say of him that the gendarmes believe he dropped the anchovies.

Any name that referred to their acquaintance Jacques/Le Chat (any name with which both Smith

and Jones were mutually familiar) would serve their communicative ends—to pick out a specific

individual. 23 24

23
Interestingly, Lycan also points out that we also use definite descriptions inside the ‘that’-
clause of a propositional attitude ascription just to refer and not to suggest anything about the
way in which the ascribee conceives of the object or individual denoted by that definite
description. He writes:

“Consider (3):

19
 Suppose I utter, “Everyone who has ever heard him sing believes that Elton John has a

great voice.” ‘Elton John’ is the stage name of the famous pop star whose birth name was

‘Reginald Dwight.’ By using the name ‘Elton John’ in this ascription sentence, am I restricting

myself to saying that those who have heard Elton John sing qua famous pop singer named ‘Elton

John’ thinks he has a good voice? Or could I very well intend my utterance to mean that

everyone who has heard Elton John sing thought he had a good voice, whether they knew him as

‘Reginald Dwight’ or ‘Elton John,’ or whether they heard him sing before or after he became a

famous pop star? I think the latter is the case. I could very well intend a purely

Millian/referential reading of ‘Elton John,’ uttering the name just to refer to that individual qua

individual regardless of how conceived, and say that any person who ever heard him sing, from

his childhood music teacher up to one of his biggest fans as a pop star, thought he had a great

voice. I would intend that my audience, in interpreting my utterance, abstract away from any

conceptions they might associate with the name ‘Elton John.’ Of course, I could use ‘Elton John’

(3) Columbus reckoned that Castro’s island was only a few miles from India.

We all know what one would mean in asserting (3); the speaker would mean that when
Columbus sighted Cuba he thought that he was already in the East Indies and was
approaching India proper. Of course, being 450 years early, Columbus did not know
anything about Fidel Castro; yet we can assert (3) with no presumption that its
complement clause represents things in the way that Columbus himself represented them.
The speaker makes this reference to Cuba without … assuming that Columbus would
have referred to Cuba in that way or in any parallel or analogous way. So it seems
undeniable that there are transparent positions inside belief sentences, in which the
referring expression does just refer to its bearer, without any further suggestion about the
way in which the subject of the belief sentence would have represented the bearer.
Singular terms can be and are often understood transparently.”
24
They would likely use the name ‘Jacques’ because that is the man’s “real” name and they are
accustomed to using that name to refer to him in everyday situations. The point is that in this
sentence the conception that the gendarmes associate with the name ‘Jacques’ and ‘Le Chat’ is
not a factor for Smith and Jones in deciding which name to utter, and their choice of name is not
intended to suggest anything about the conception under which the gendarmes conceive their
friend.

20
to refer to the man qua famous pop star, but my point is that there is also a plausible reading on

which I intend the name just to refer to that individual qua individual. 25

Suppose that Jimmy Olson, Perry White, and Perry White's secretary, Janet Smith, were

all “enlightened” about the identity of Clark Kent and Superman (i.e., they all realize that Clark

Kent is Superman). Furthermore, they all mutually know that each of them is enlightened. One

morning, Olson says to White “Clark Kent is so tall, he could probably play in the NBA. He's

over seven feet tall!” White is somewhat skeptical about Olson’s judgment of Clark Kent’s

height. That afternoon, White remarks to Smith, his secretary: “Jimmy Olson believes that Clark

Kent is over seven feet tall. But I think he’s more like 6′ 9″.” Because White knows that Smith

knows that Clark Kent is Superman and he knows that Smith knows that Olson knows this fact

as well, White could just as well have said to Smith “Jimmy Olson believes that Superman is

over seven feet tall.” Either of these propositional attitude ascriptions would, under the

circumstances described, communicate the same information. White could use either of the

names of Kent-Super, ‘Clark Kent’ or ‘Superman,’ to refer to him, unconcerned about any

conceptions that he, Olson, or Smith (his audience) might associate with those names. In this

conversational context, conceptions associated with names do not affect the information

communicated. White utters the name with the sole intent of referring to its bearer, and given

the audience he is addressing, any name that refers to Kent-Super (with which his audience is

familiar) will communicate the same information.

An expressly de re propositional attitude ascription constitutes a further example of the

25
My point here is that it seems undeniable that a speaker could utter this sentence intending a
purely referential/Millian reading. There are two plausible readings of the sentence, and which
reading is correct depends on what the speaker had in mind. There is of course no guarantee that
the audience to this utterance will take it in the way the speaker intended.

21
use of a proper name in a Millian way in a propositional attitude ascription. Compare (5) and

(5)de re:

(5) Lois Lane believes that Clark Kent flies

(5)de re Clark Kent is such that Lois Lane believes he flies

The non-expressly de re (5) is intuitively false (although there is a de re reading on which (5) is

true, it is the less intuitive reading). Sentence (5)de re, unlike (5), is unambiguously true, for the

‘is such that’ language makes the ascription expressly de re, indicating that ‘Clark Kent’ just

refers to its bearer and any conceptions that Lois Lane may associate with the name ‘Clark Kent’

or whether Lois even knows Kent-Super under the name ‘Clark Kent’ has no bearing on the

truth-conditions of the sentence.

I shall now turn to simple sentences (i.e., non-propositional attitude ascriptions) in which

proper names are used just to refer. Consider again what Olson said to White two paragraphs

above: “Clark Kent is so tall that he could probably play in the NBA. He's over seven feet tall!”

Surely Olson uttered the name ‘Clark Kent’ here just to refer to Kent-Super and not to

communicate a Clark Kent-y conception of him to Olson. After all, White and Olson mutually

know that they are both enlightened about the identity of Clark Kent and Superman. Olson and

White both know that if ‘Clark Kent is over seven feet tall’ is true, then ‘Superman is over seven

feet tall’ must be true as well. Olson could have made the same claim about Kent-Super’s stature

using either the name ‘Clark Kent’ or ‘Superman,’ and either way he would have made the very

same claim and White would have understood him as making the same claim. Differing

conceptions of Kent-Super are irrelevant to whether what Olson said is true or false, and both

Olson and White are aware of this.

22
 Unenlightened speakers such as Lois Lane can (and often do) use the names ‘Clark Kent’

and ‘Superman’ in Millian ways in simple sentences. For example, suppose that someone asks

Lois Lane to name a non-avian flying being and she responds by uttering (8).

(8) Superman flies

Here, she uses the name ‘Superman’ just to refer to Kent-Super. She would not use the name

‘Superman’ to make her Superman-y way of conceiving him salient and contrast this with a

Clark Kent-y way of conceiving him. She has no idea that these are different conceptions of the

very same individual. When she utters (8), Kent-Super’s Clark Kent persona is not on her mind,

nor does she intend to distinguish between Kent-Super’s Clark Kent and Superman personae

with respect to his ability to fly. This is not to say that she attaches the same conceptions to the

names ‘Clark Kent’ and ‘Superman.’ On the contrary, she has very different conceptions of

Kent-Super when she thinks of him under a Clark Kent-y conception and when she thinks of him

under a Superman-y conception. However, my point is that she would not use the names with

the intention of making those conceptions salient or to refer to them, and thus communicate them

to her audience, because she is not drawing any contrast between Kent-Super based on the

various ways she conceives him. She thinks that these names refer to distinct individuals, rather

than to the same individual conceived in different ways.

Now suppose someone asks Lois to name a non-avian being incapable of flight and she

responds by uttering ¬(7) in response.

¬ (7) Clark Kent does not fly

Here again, Lois utters ‘Clark Kent’ just to refer to Kent-Super. She would not use ‘Clark Kent’

23
to call attention to or make salient a Clark Kent-y conception, contrast this with a Superman-y

conception, and communicate it to her audience, since she has no idea that these are conceptions

of the same individual. When she utters ¬(7), a Superman-y conception of Kent-Super is not on

her mind.

Finally, we have the utterance of sentence (1) where a speaker intends to express a trivial

self-identity—to say that a man is identical to himself. 26 Such a speaker would use the name

‘Clark Kent’ twice over just refer to the same individual twice over, intending each tokening of

the name to have the identical semantic content—the name’s bearer. The speaker would not

attach different conceptions to the two occurrences of the name intending to express an

informative identity sentence, as he or she would if she uttered sentence (2) to say that Clark

Kent and Superman are the same person, or if he, like Tom (see footnote 5, supra), uttered (1) to

say that the Clark Kent from the party is the Clark Kent with whom he attended Kindergarten as

a child in Smallville.

In the examples above, speakers use names just to refer to their bearers, full stop. Indeed,

we use proper names most frequently just to refer to their bearers, i.e., in a ‘Millian way’.

However, the very existence of Frege’s puzzle undercuts the Millian claim that merely referring

is the only semantic function of proper names. In the puzzle cases, speakers purposefully use

proper names to convey conceptions as well as to refer, i.e., they use names in ‘Conception-

indicating’ ways. Hence, the TIUT posits that proper names have two uses: to refer only (names

used in a Millian way), and to refer and to convey conceptions as well (names used in a

26
Practically no one would ever say such a thing with the exception of a philosopher or
logician—given that an individual’s self-identity is too obvious to need pointing out—but it is
nevertheless possible to meaningfully say this.

24
Conception-indicating way). 27 Given that speakers purposefully use names in this conception-

indicating way regularly and without extensive stage-setting (Devitt 2004), I will claim that the

conception-indicating use constitutes a standard linguistic convention. Propositional attitude

reports such as ‘Lois Lane does not realize that Clark Kent is Superman’ are the only idiomatic

means we have in our language to ascribe ignorance of identities, suggesting that such

constructions are standard conventional devices for communicating such ignorance (no Millian

theory suggests any better idiomatic, non-technical way to state such ascriptions in English).

The distinction between Millian and conception-indicating uses should be deemed semantic, and

not pragmatic, in nature. (See discussion in section 2.13, infra, where I invoke Michael Devitt’s

“argument from convention” (2004) to claim that the distinction should be classified as semantic

rather than pragmatic).

2.2 Definitions: “Dossier Tokens,” “Dossier Types,” “Subjects”

In explaining the character and content of names used in Millian and Conception-

indicating ways (in sections 2.2 and 2.3), I use the following terms of art: “dossier tokens,”

“dossier types,” and “subjects.” Hence, I will define these terms of art here.

27
I leave open the possibility that there are other uses of proper names. For example, sometimes
we use proper names to refer to aspects or time-slices of individuals or objects. In the sentence,
‘I have been to Chemnitz but never to Karl-Marx-Stadt,’ these names refer to different (thick)
time-slices of a single German city, rather than the city as a whole. Or consider ‘Clark Kent went
into the phone both and Superman came out’, which example is due to Jennifer Saul (1997). See
chapter 8 for discussion of Saul’s cases. I contend that ‘Clark Kent’ and ‘Superman’ are not co-
referential in Saul’s example, as they refer to different aspects of an individual, not to a whole
individual qua individual. Clark Kent and Superman, as used in that sentence, are distinct
entities, even if they inhere within the same person. See Moore (1999) and Pitt (2001). Names
can be used meta-linguistically, as in ‘There are many Smiths in the phone book’; here ‘Smiths’
means ‘bearers of “Smith”’. See my discussion of Predicativism in Chapter 9, section 9.2.
Finally, there is a reasonable argument to be made that names can be used non-referentially as
variables bound to an indefinite antecedent, as is claimed by adherents of Variablism, notably
Sam Cumming in his 2008. See chapter 9, section 9.1 for discussion.
25
 Dossier tokens and their subjects

The notion of dossier plays a key role in Graeme Forbes’ theory of proper names (1990).

According to Forbes, “when we receive what we take to be de re information which we have an

interest in retaining, our [mental] operating system may create a locus, or dossier, where such

information is held; and any further information which we take to be about the same object can

be filed along with information about it we already possess… The role of a name is to identify a

dossier for a particular object… (Forbes, 538).” I adopt this basic characterization of dossiers

from Forbes. Dossiers are encyclopedia entries or files in the mind of an agent: they have

“subjects”—they contain de re information about persons or objects, and they contain stores of

descriptive representations associated with and thought by an agent to characterize the subject of

the dossier, i.e., what I refer to as conceptions. Simply put, each dossier is about a particular

individual/object (a subject) and contains descriptive representations about what the owner of the

dossier thinks that individual/object is like (a descriptive conception). Lois Lane has one dossier

that is about Kent-Super (Kent-Super is its subject) that presents him conceptually in a “Clark

Kent-y” way—as a mild-mannered reporter. It contains as part of its descriptive conception the

representation that its subject bears the name ‘Clark Kent.’ Lois also has another dossier that is

about Kent-Super (it likewise has Kent-Super as its subject) that presents him conceptually in a

“Superman-y” way—as a strong superhero. It contains as part of its descriptive conception the

representation that its subject bears the name ‘Superman.’

On Forbes’ theory, as on the TIUT, the subject of a dossier token is not necessarily the

individual that the descriptive representations that make up the conception in the dossier best

‘fit.’ In the terminology of Kent Bach (1987, 12) the subject of a dossier is determined

26
relationally, not satisfactionally. 28 Whereas the denotation of a definite description is determined

satisfactionally because its denotation is just whatever object satisfies or best fits it, the

subjecthood of a dossier depends on the existence of an appropriate causal-historical relation

between the dossier and its subject that explains the dossier’s coming into being, roughly along

the lines of Kripke’s causal-historical account of reference. 29 (See

http://plato.stanford.edu/entries/reference/#CauThe for discussion of Kripke’s causal-historical

picture of reference). The subjecthood or ‘aboutness’ of a dossier can be analogized to the

aboutness of a photograph. Suppose that President Obama sits for a photograph and because of

an odd camera angle and lighting, he strongly resembles Malcolm X. Despite the greater

resemblance to Malcolm X, the resulting photograph is nevertheless a photograph of Obama, and

not of Malcolm X, because there is a causal relation obtaining between Obama and the

photograph that does not obtain between Malcolm-X and the photograph. After all, the photo

came into being because Obama, and not Malcolm-X, sat for it. The ‘aboutness’ or ‘of-ness’ or

subjecthood of photographs is determined relationally rather than satisfactionally, and the same

is the case with dossiers. In other words, externalist causal-historical considerations largely (or

28
Bach has characterized the distinction between “relational” and “satisfactional” properties as
follows (Bach, 1987, 12): “If all your thoughts about things could only be descriptive, your total
conception of the world would be merely qualitative. You would never be related in thought to
anything in particular. Thinking of something would never be a case of having it ‘in mind,’ as we
say colloquially, or as some philosophers have said, of being ‘en rapport,’ in ‘cognitive contact,’
or ‘epistemically intimate’ with it. But picturesque phrases aside, just what is this special
relation? Whatever it is, it is different from that involved in thinking of something under a
description. If we can even speak of a relation in the latter case, it is surely not a real (or natural)
relation. Since the object of a descriptive thought is determined SATISFACTIONALLY, the fact
that the thought is of that object does not require any connection between thought and object.
However, the object of a de re thought is determined RELATIONALLY. For something to be
the object of a de re thought, it must stand in a certain kind of relation to that very thought.”
29
However, I would not rule out the possibility that subjecthood might turn out to be a hybrid
satisfactional-relational property. See section 4.3 for discussion.

27
perhaps fully) determine the aboutness or subjecthood of dossiers, so that the subjecthood of a

dossier is not a function of what is “in speaker’s head” (as the reference of a proper name would

be according to Descriptivism).

Lois Lane’s ‘Clark Kent’ dossier has Kent-Super as its subject because she causally

interacted with Kent-Super (in his Clark Kent guise) and this resulted in the coming into being in

her mind of her ‘Clark Kent’ dossier. Kent-Super is also the subject of Lois’ ‘Superman’ dossier

for the same reason, i.e., she causally interacted with Kent-Super (in his Superman guise) and

this led to the creation of her ‘Superman’ dossier in her mind. Dossiers may also be generated in

the mind in the absence of direct causal contact between an agent and an individual, as when an

agent hears about an individual by name. The subject of the dossier that comes into being upon

hearing the name for the first time would be the individual causally-historically linked to the

name roughly along the lines of the causal-historical picture of reference borrowing (Kripke,

1980; Devitt, 1981).

Dossier types and their individuation criteria

I have characterized dossier tokens above. The TIUT distinguishes between dossier

tokens and dossier types (with dossier types being fundamental to the statement of the character

and content of proper names when used in a Conception-indicating way; see section 2.4).

Dossier types are abstract objects individuated by their subject and their conception. Dossier

tokens with the same subject and the same (or relevantly similar) conception of their subject

instantiate the same dossier type. For example, if two dossier tokens, d1 and d2, each have Kent-

Super as their subject and contain a “Clark Kent-y” mild-mannered reporter conception, d1 and d2

are dossier tokens instantiating the same dossier type.

28
 More about Dossiers

Agents both enlightened and unenlightened may have multiple dossiers for the same

subject. Suppose that Jimmy Olson is ‘enlightened’—he realizes that Clark Kent is Superman.

He may maintain two separate dossiers, one ‘Clark Kent’ dossier and one ‘Superman’ dossier,

each dossier containing different conceptions of their common subject, Kent-Super. There

would be some overlapping representations in his two dossiers reflecting the fact that Olson

believes that the dossiers have the same subject. Perhaps the dossiers are ‘linked’ in his cognitive

architecture, reflecting the fact that he knows that they have the same subject. The enlightened

Olson’s ‘Clark Kent’ dossier and his ‘Superman’ dossier cannot match the ‘Clark Kent’ and

‘Superman’ dossiers of the unenlightened Lois Lane perfectly. Olson’s dossiers contain

representations reflecting his awareness that the dossiers are about the same subject, whereas

Lois’ corresponding dossiers lack that representation. Nevertheless, by having two dossiers,

Olson can conceive Kent-Super in either a Clark Kent-y or Superman-y way, thus mirroring

Lois’ distinct conceptions of Kent-Super. Olson’s ‘Clark Kent’ dossier and his ‘Superman’

dossier are sufficiently relevantly like Lois’ such that his dossiers are of the same dossier types

as hers despite his added knowledge of Clark Kent and Superman’s identity.

The structure of an agent’s dossiers and the representations contained in them—the

agent’s “mental architecture”—is dynamic, such that agents regularly (and rapidly) restructure

the data in their dossiers, split one dossier into two, and merge dossiers. Much of this

restructuring involves the creation of temporary dossiers. For example, suppose that instead of

two linked dossiers about Kent-Super, one ‘Clark Kent’ dossier and one ‘Superman’ dossier, the

enlightened Jimmy Olson maintained one single dossier on Kent-Super. Inside of this single

dossier are all the descriptive conceptual representations relating to both of Kent-Super’s

29
personas and the names he bears. I posit that Olson can temporarily split this unified dossier into

two separate temporary dossiers, one containing a Clark Kent-y conception and the other

containing a Superman-y conception. This would not entail the actual breaking up of the

original, unified, “mother dossier” covering both of Kent-Super’s personas, but rather the

creation of two temporary dossiers. Olson’s operating system loads the Clark Kent-y conceptual

representations into one temporary dossier and the Superman-y conceptual representations into

another temporary dossier. Each of these temporary dossiers would continue to have the same

subject, Kent-Super, as that of the mother dossier from which they were created. Importantly, the

temporary dossiers would “borrow” the subjecthood of the dossiers from which they are

generated. 30 Thus, Olson would have two temporary dossiers that would be of the same types as

Lois Lane’s two dossiers, one for Clark Kent and one for Superman, having the same

subjecthood and conceptions similar to those in her respective dossiers. His cognitive

architecture with respect to this individual, Kent-Super, mirrors that of Lois.

Whether Olson has maintained two separately dossiers or has a unified mother dossier for

Kent-Super, he can entertain dossiers, via the dynamic creation of temporary dossiers, that are of

the same or similar types as those possessed by Lois Lane in her mental architecture. He thereby

has the ability to mirror the epistemic structure of Lois’ mental architecture vis-à-vis Kent-Super

within his own dossier organization. He can, as it were, adopt the point of view of Lois and see

the world from her perspective by modeling her mental architecture/dossier organization. Where

I discuss propositional attitude ascriptions in section 2.6 infra, I shall urge that understanding an

agent’s propositional attitudes (and speaking about them via propositional attitude ascription)

30
Much as Kripke (1980) and Devitt (1981) propose that the reference of a proper name can be
borrowed downstream in the chain of communication by speakers who have not been in direct
causal contact with its bearer.
30
involves us in cognitively mirroring the mental state of that agent and referring to our own

mirroring states as if they were the states of ascribing agent. Speaking about Lois’ confusion

about who Clark Kent and Superman are involves us having (even if just temporarily) distinct

dossiers relevantly like hers, one for Clark Kent and one for Superman. Olson can refer to Lois’

confusion by “pointing” to his own temporary dossiers and saying that Lois does not realize that

dossiers like those, those instantiating those types, have the same subject. 31

The dossier metaphor is a fruitful device for theorizing about the structure of an agent’s

beliefs. Of course, considering the current nascent state of cognitive science, the dossier

metaphor will have to remain metaphorical. A dossier should be thought of as a pool of

information in the mind about an object (in virtue of some causal-historical relation to the object)

that represents it as being a certain way, not literally as a discrete location in the mind or region

of the brain devoted storing information about one individual or object.

In addition to Forbes and me, numerous philosophers have taken the dossier metaphor to

be an illuminating tool for reflecting upon de re belief. The idea of a mental file or dossier has

been around for almost fifty years. According to Francois Recanati (2013, 3), who writes about

mental files in his 1993 and 2013 (see Chapter 10, infra, for brief discussion of Recanati’s

31
The enlightened Jimmy Olson, who knows that Clark Kent is Superman, can conceive Kent-
Super in a Clark Kent-y way or in a Superman-y way just as Lois does. However, there are some
differences. Lois’ ‘Clark Kent’ dossier represents him as weak and indecisive. However,
Olson’s ‘Clark Kent’ dossier does not represent Clark Kent as genuinely weak and indecisive,
since Olson realizes that Kent-Super is in fact strong and decisive and is putting on an act when
he is being Clark Kent. Whereas Lois’ dossier represents him as weak and indecisive, Olson’s
dossier represents him as weak acting or seemingly weak. Alternatively, perhaps his dossier
contains the representation weak and indecisive, but Olson treats this representation as
fictional—with a healthy dose of suspension of disbelief. So Olson is able to conceptualize
Clark Kent as weak, despite knowing, at some level, that it is only as if he were weak. Olson,
being enlightened, cannot associate certain Clark Kent-y properties with Kent-Super in quite the
same way as the unenlightened Lois, yet there is sufficient similarity between their
representations in their respective ‘Clark Kent’ and ‘Superman’ dossiers such that we may say
their dossier tokens are of the same dossier type.

31
Theory of Mental Files), the notion of mental files or dossiers “was introduced by several

philosophers in the late sixties or early seventies, in connection with the referential use of

definite descriptions (Grice 1969, 140-44) or with identity statements (Lockwood 1971, 208-11;

Strawson 1974, 54-56). Several authors subsequently exploited the notion, including Evans

(1973, 199; 1982, 276), Bach (1987, 34-37), Devitt (1989, 227-31), Forbes (1989, 1990, 538-45),

Crimmins (1992, 87-92)…” and Perry (1980). Most recently, Peter Cave (2014) makes

extensive mention of dossiers in his discussion of Frege and Kripke’s puzzles.

Forbes (1990) claims that proper names ‘label’ dossiers. Unlike Forbes, I do not claim

that proper names label dossiers or that dossiers have labels. I claim instead that dossiers

typically have representations in them about what name or names the subject of the dossier bears.

That is, a typical dossier contains the representation bears the name ‘NN’ inside of it, i.e., the

representation that the subject of the dossier bears the name ‘NN’ is a part of the dossier’s

descriptive conception. However, some dossiers lack representations about the names of their

subjects. For example, suppose a man named “Bob” meets a woman at a party who makes a

strong impression on him, but he fails to learn her name. Within Bob’s mental architecture a

dossier is created when the encounter takes place to collect and store information about that

woman, but the dossier does not represent its subject as bearing a name. Bob can refer to the

subject of this dossier by introspecting on the conceptual representations within it and picking

out a description of the woman to refer to her. Using a name for the subject of the dossier is of

course the most efficient way of referring to that subject (in situations where one’s conversation

partner is familiar with the name and its bearer) and therefore we generally aim to learn the

32
names of people and objects in order to refer to them more effectively and efficiently. 32

2.3 Names Used in a ‘Millian’ Way

When employed merely to refer, the TIUT posits that we use names as indexicals with a

two-tiered semantic structure, consisting of the following character and content:

Character of ‘NN’

The subject of the dossier token from which the speaker draws ‘NN’

Content of ‘NN’

< the subject >

By way of clarification, here is what I mean by the term of art “the speaker draws ‘NN’

from the dossier token” in the schema of character above. Suppose a dossier token a speaker

mentally entertains represents its bearer as named ‘NN.’ The speaker has introspective access to

the descriptive representations in the dossier, including what name or names the subject of the

dossier is represented as bearing. By introspecting on his dossier, picking out this information

and then uttering ‘NN,’ the speaker “draws the name from the dossier token.”

I state the content above within angular brackets “< >”. This is a standard convention to

indicate that what is inside of the brackets is an object. In the schema of content above, ‘< the

subject >’ means that the content of the name is the subject of the dossier token, the flesh-and-

blood individual himself, qua object.

32
Names have a special status/role in communication. We use them as a conventional
established shortcut means of referring when speaking with others. We prefer using a name, if
we know a name and it is unambiguous in the context to whom we refer with it, over using a
definite description or a demonstration.
33
 To illustrate the content and character of names used in a Millian way, consider for

example Lois’ Millian uses of the names ‘Clark Kent’ and ‘Superman’ when she utters (8) and ¬

(7). 33

(8) Superman flies

¬(7) Clark Kent does not fly

At the moment she utters ¬(7), Lois mentally entertains her ‘Clark Kent’ dossier—a

dossier that has Kent-Super as its subject with a Clark Kent-y conception. She draws the name

‘Clark Kent’ from of the dossier, i.e., she introspects the dossier, sees that the subject is

represented as bearing the name ‘Clark Kent,’ and utters the name. By the Millian character, the

content of her utterance is the subject of the dossier token from which she drew the name—Kent-

Super. When she utters (8), Lois entertains her ‘Superman’ dossier—a dossier that has Kent-

Super as its subject with a Superman-y conception. She draws the name ‘Superman’ from of the

dossier, i.e., she introspects, sees that the subject is represented as bearing the name ‘Superman’

and utters the name. By the Millian character, the content of her utterance is the subject of the

dossier token from which she drew the name—again, Kent-Super.

Both of her utterances of ‘Superman’ and ‘Clark Kent’ refer to Kent-Super (whether Lois

realizes it or not) because in both cases the Millian character maps the utterances to Kent-Super.

That is, Kent-Super is the subject of both dossiers entertained and it is from these dossiers that

Lois has drawn the names ‘Clark Kent’ and ‘Superman.’

33
As mentioned above, Lois Lane would use the name ‘Superman’ in a Millian way—just to
refer to him, not to call attention to her “Superman-y” way of conceiving him and communicate
it, for as far as she knows she only has one way of conceiving him. She has no reason or ability
to contrast her Superman-y conception of Kent-Super with her Clark Kent-y conception of him
because she has no idea that these are different conceptions of the same individual.
34
2.4 Names Used in a ‘Conception-Indicating’ Way

When used in a ‘Conception-indicating’ way, a name functions as an indexical that

contributes both the bearer and a descriptive conception of it to the proposition. Unlike a name

used in a Millian way, which has a two-tiered semantics (character and content/referent), a name

used in a Conception-indicating way has a three-tiered semantic structure (character, content,

and referent). The character of a name first determines its content, which is the meaning of a

definite description. The referent of the name is the denotation of the definite description. Thus,

we do not have direct reference here, but mediated reference, with the character determining

content, a meaning, and then that meaning determining reference, an object.

Character of ‘NN’

Ḿ [ the subject individuating dthat {{the dossier type D instantiated by the dossier
token from which the speaker has drawn ‘NN’ 34}} ]

Content of ‘NN’

Ḿ [the subject individuating D]

Referent of ‘NN’

< the subject individuating D >

I shall now explain the above schema, including special terminology/symbols used.

Dthat is Kaplan’s operator (1989) used to convert a definite description into a directly

referential expression. It functions according to the following schema:

34
See section 2.3, second paragraph, supra, for meaning of “draws a name.”
35
 Dthat Operator Where δ is a definite description, ‘dthat [δ]’ is a directly
referential expression whose referent is the denotation of δ.

For example, ‘dthat [The president of the United States in 2014]’ is a directly referential

expression that refers to Barack Obama, the denotation of the definite description inside the

operator’s scope. ‘dthat [The president of the United States in 2014]’ is semantically equivalent

to any directly referential expression referring to Obama, including any directly referential

proper name referring to him.

Ḿ (for “meaning”) is an operator functioning according to the following schema:

Ḿ Operator Where δ is a definite description, ‘Ḿ [δ]’ is an expression whose

referent is the meaning of δ.

For example, the expression ‘Ḿ [the tallest man ever to live]’ refers to the meaning of the

definite description ‘the tallest man ever to live.’ It also would refer to the meaning of the

definite description ‘der größte Mann, der je gelebt hat,’ since this German definite description

has the same meaning as ‘the tallest man ever to live.’

The double curly brackets indicate the order of operations. The first operation involves

determining the value only of the part inside these double curly brackets. In this first operation,

neither ‘Ḿ’ nor the language ‘the subject individuating’ operates on what is inside the brackets.

Hence, the first operation involves determining the dossier type instantiated by the dossier token

from which the speaker has drawn the name uttered. Thus, after the first operation we have a

content that looks the same as character but for having the language inside the double curly

brackets replaced by a name for a dossier type (here represented by the metavariable D). This

name, D, is directly referential with respect to that dossier type in virtue of the dthat operator in

36
front of the double brackets. 35

Thus, the first operation, in which the dossier type of the token from which the speaker

has drawn the name in the utterance context is ascertained/determined, yields the content of the

name. The content is a meaning, as indicated by the Ḿ operator. It is the meaning of the definite

description ‘the subject individuating D.’ The referent of the name is then the denotation of this

definite description.

Alternatively, one can state the character/content/referent in the following three ways:

(1) The character of a name used in a Conception-indicating way takes us from a

context in which a speaker draws a name ‘NN’ from a dossier token to the
meaning of the definite description ‘the subject individuating D’, where ‘D’ is a
metavariable for any expression referring directly to the dossier type instantiated
by that dossier token. The referent of ‘NN’ is the denotation of the definite
description ‘the subject individuating D’.

(2) The character of a conception-indicating name ‘NN’ is such its content in the
context is the meaning of the definite description ‘The subject individuating D’,
where D directly refers to the dossier type instantiated by the dossier token from
which the speaker has drawn ‘NN’. The referent of ‘NN’ is the denotation of this
definite description.

Here, the schema is stated in the form of a step-by-step algorithm:

(3) (a) Consider the dossier type instantiated by the dossier token from which the

35
A directly referential name appears (in content) because the dthat operator (in character) gives
rise to direct reference—direct reference to whichever dossier type is instantiated by the
speaker’s token in the context. I do not mean to suggest that dossier types bear names. We do not
name dossier types (although we could). In this schema, ‘D’ is a metavariable standing in for any
(of potentially infinite number of) directly referential names or other directly referential
expressions that would pick out the dossier type instantiated by the dossier token entertained by
the speaker in the context. It makes no difference which directly referential name for a particular
a dossier type appears in this content schema because every directly referential expression
referring to an object o has the same semantic value—its bearer o. Hence if a and b are two
directly referential names of a dossier type, Ḿ [the subject individuating a] = Ḿ [the subject
individuating b]; moreover, the definite descriptions ‘the subject individuating a’ and ‘the
subject individuating b’ have the same meaning.

37
 speaker has drawn the name ‘NN’ in the utterance context.
(b) Call that dossier type ‘D.’
(c) The content of ‘NN’ is the meaning of the definite description ‘The
subject individuating D’ (which meaning could also be stated as: Ḿ [The
subject individuating D])
(d) The referent of ‘NN’ is the denotation of that definite description.

Let us consider a concrete example to illustrate this three-part schema. Consider sentence

(2) as uttered by the enlightened Jimmy Olson when he is trying to convince Lois Lane that

Clark Kent is Superman.

(2) Clark Kent is Superman

Olson’s ‘Clark Kent’ dossier token has Kent-Super as subject and a Clark Kent-y

conception. He mentally introspects the dossier, realizes that its subject bears ‘Clark Kent’, and

draws the name from the dossier (he utters ‘Clark Kent’). Let ‘DCK’ be a metavariable

representing the dossier type instantiated by the dossier token from which Olson draws ‘Clark

Kent,’ which is the dossier type individuated by Kent-Super as subject and a Clark Kent-y

conception. The schematization of the semantic structure of Olson’s utterance of ‘Clark Kent’ is

as follows:

Character of ‘Clark Kent’

Ḿ [the subject individuating dthat {{the dossier type instantiated by the dossier
token from which the speaker has drawn ‘Clark Kent’}}]

Content of ‘Clark Kent’

Ḿ [the subject individuating DCK]

38
 Referent of ‘Clark Kent’

< Kent-Super >

In the first operation, the language inside the double curly brackets is evaluated in the

context of utterance. A directly referential name referring to the dossier type instantiated by the

speaker’s (Olson’s) dossier token is loaded into the content, replacing the part in double curly

brackets. Thus, after the first operation, the content is: Ḿ [the subject individuating DCK].

“DCK” has replaced “dthat {{the dossier type instantiated by the dossier token from which the

speaker in the context draws ‘Clark Kent’}}.”

Kent-Super is the referent of ‘Clark Kent’ as uttered by Olson because its content, Ḿ [the

subject individuating DCK], can be expressed by the definite description ‘the subject

individuating DCK’, which denotes Kent-Super (since dossier type DCK is individuated by Kent-

Super as subject).

Olson’s ‘Superman’ dossier token has Kent-Super as subject and a Superman-y

conception. He introspects the dossier, realizes that its subject bears ‘Superman’ and draws the

name from the dossier (he utters ‘Superman’). Let ‘DSM’ be a metavariable for any directly

referential name for the dossier type instantiated by the dossier token from which Olson draws

‘Superman’, which is the dossier type individuated by Kent-Super as subject and a Superman-y

conception.

The schematization of the semantic structure of Olson’s utterance of ‘Superman’ is as

follows:

Character of ‘Superman’

Ḿ [the subject individuating dthat {{the dossier type instantiated by the dossier
token from which the speaker has drawn ‘Superman’ }} ]

39
 Content of ‘Superman’

Ḿ [the subject individuating DSM]

Referent of ‘Superman’

< Kent-Super >

In the first operation, the language inside the double curly brackets is evaluated in the

context of utterance. A directly referential name referring to the dossier type instantiated by

Olson’s dossier token is loaded into the content, replacing the part in double curly brackets.

Thus, after the first operation, the content is: Ḿ [the subject individuating DSM]. “DSM” has

replaced “dthat {{the dossier type instantiated by the dossier token from which the speaker in the

context draws ‘Superman’}}.”

Kent-Super is the referent of ‘Superman’ as uttered by Olson because this content, Ḿ [the

subject individuating DSM], can be expressed by the definite description ‘the subject

individuating DSM’, which denotes Kent-Super (since dossier type DSM is individuated by Kent-

Super as subject).

Notice the TIUT’s similarity to Descriptivism, which also posits that the content/meaning

of a name is the meaning of a definite description. (As mentioned, in section 1.1., according to

Descriptivism the content of a proper name is not a definite description, but the meaning of a

definite description. That meaning is expressible by any definite description with that meaning,

and the referent of the name is the denotation of any definite description with that meaning.) As

on Descriptivism, on the TIUT the referent of a name is the denotation of the definite description

whose meaning is the name’s content. As on Descriptivism, ‘Clark Kent’ and ‘Superman’ differ

in meaning but still co-refer because the definite descriptions expressing their meanings denote

40
the same individual. However, the TIUT is not a form of Descriptivism. Firstly, the TIUT has

Kripke’s causal-historical theory of reference built right into it, as the subjecthood of a dossier is

a causal-historical property. Reference is determined externalistically/relationally. It is a function

only of the subjecthood of a speaker’s dossier token. The descriptive conception in the dossier

token is inert with respect to determining reference. By contrast, Descriptivism claims that the

descriptive conceptions associated with proper names wholly determine their reference.

Secondly, for this same reason, ‘Clark Kent’ and ‘Superman’ are rigid on the TIUT, just as they

would be according to any theory of proper names on which reference is causal-historically

based, whereas on classical forms of Descriptivism these names would not be rigid because

reference is determined by the speaker’s descriptive conception. Because of these sharp

differences with Descriptivism, the TIUT cannot be classified as a Descriptivist theory or proper

names and is therefore not vulnerable to Kripke’s arguments against Descriptivism, despite the

TIUT sharing a superficially similar approach to the identity sentences puzzle.

The similarity of approach is, to be clear, that both Descriptivism and the TIUT

distinguish between content (a meaning) and reference, and posit that reference is determined by

content via denotation (when a name is used in Conception-indicating way). Furthermore, the

speaker’s descriptive conception is an element of the content of the name. The difference

between the theories is that the descriptive conception is, on the TIUT, not reference-

determining. Moreover, the TIUT posits a prior level of meaning—character—that determines

the content of a name based on contextual factors. This level of character is essential because

both whom the speaker is referring to and the way the speaker represents that person is

contextually variable. An utterance of ‘John’ can refer to different bearers of ‘John’, and can

present that bearer under different descriptive conceptions. To whom one refers with a name and

41
how that person is conceived descriptively is a function of situational/contextual factors—which

dossier of the speaker was causally responsible for the utterance, what sort of descriptive

representations are in that dossier, and how that dossier is externalistically connected to the

outside world. Whether a speaker uttering ‘Aristotle’ refers to the philosopher or the shipping

magnate is a function of facts about the speaker’s dossier from which the name was drawn—how

that dossier is causally related to the one Aristotle or the other. The descriptive conception

communicated is a function of what representations the speaker associates with the name when

he utters it and what descriptive conception he or she intends to communicate. When used in a

Conception-indicating way, a name is sensitive to these contextually variable factors and loads

them into the proposition.

On the TIUT, both bearer and conception are “in” content (albeit in an indirect way) by

being the individuation criteria of the dossier type referenced in the content. Thus, with Olson’s

utterance of ‘Superman,’ both Kent-Super and a Superman-y conception are in content by virtue

of dossier type DSM being individuated both by Kent-Super and by a Superman-y conception.

And with Olson’s utterance of ‘Clark Kent,’ both Kent-Super and a Clark Kent-y conception are

in content by virtue of dossier type DCK being individuated both by Kent-Super and a Clark

Kent-y conception.

Note the deviation here from Kaplan’s notion of indexicals, on which:

● the content of an indexical is always an object, and

● the indexical expression directly refers to that object.

Neither is the case here. When used with a Conception-indicating character, the content of a

name is a meaning, not an object. That meaning then determines an object. Thus, names used

with a Conception-indicating character are not directly referential because the name’s content

42
mediates reference. That is, character determines the name’s content in the context, and then this

content determines reference. Although this indexical use deviates from Kaplan’s notion of

indexicals, there is no reason to think that all indexicals must follow Kaplan’s notion.

Importantly, there is rigid designation on the TIUT (despite lack of direct reference) because

reference is a function solely of causal-historical factors (given that subjecthood is a causal-

historical property modeled on Kripke’s causal-historical picture of reference). It is important

that a theory of proper names respect rigidity, not direct reference.

2.5 Proposed Solution to Frege’s Puzzle about Informative Identity Sentences

As discussed in section 2.4 supra, on the TIUT the contents/meanings of ‘Clark Kent’

and ‘Superman’ in the context of Olson’s utterance of (2) are different. The content of ‘Clark

Kent’ is Ḿ [the subject individuating DCK], which is expressible by the definite description ‘The

subject individuating DCK’. The content/meaning of ‘Superman’ is Ḿ [the subject individuating

DSM], which is expressible by the definite description ‘The subject individuating DSM’.

Substituting these definite descriptions expressing the meanings of the names ‘Clark Kent’ and

‘Superman’ into (2) in place of the proper names occurring there yields sentence (2)†, which I

claim means the same thing as (2) in the context of Olson’s utterance of (2).

(2) Clark Kent is Superman

(2)† The subject individuating DCK is the subject individuating DSM

Sentence (2)† reveals to us on its face that the proposition expressed by (2) in the given

context, which I’ll refer to as ‘PROP-2’, is the proposition that the subject individuating dossier

type DCK is the subject individuating dossier type DSM.

43
 PROP-2 The subject individuating dossier type DCK is the subject individuating
dossier type DSM

By contrast, were Olson to utter (1) to express an uninformative identity, tokening the

name ‘Clark Kent’ in Millian way twice over (i.e., merely referring twice to Kent-Super), each

utterance of ‘Clark Kent’ would have the very same content, Kent-Super, the flesh-and-blood

man himself. Olson would express PROP-1, or the trivial singular proposition that Kent-Super is

identical to himself.

PROP-1 << Kent-Super, Kent-Super >, identity >>

This reveals that PROP-1, expressed by sentence (1), and PROP-2, expressed by sentence

(2), are different propositions. Sentence (1) expresses a singular proposition of which Kent-

Super, the man flesh and blood-man-himself, is a constituent. Sentence (2) expresses a

proposition about the common subjecthood of two different dossier types. The sentences express

different propositions and that is why they differ in cognitive value.

It could be objected here that sentences (2) and (2)† cannot be synonymous, or express

the same proposition, because no ordinary speaker of English would recognize them as meaning

the same thing. Ordinary speakers of English would not even understand sentence (2)† because

no ordinary speaker would know what a dossier type is. Therefore, runs the objection, (2) and

(2)† cannot mean the same thing, as I am claiming, and therefore the content of the names ‘Clark

Kent’ and ‘Superman’ cannot be what the TIUT proposes. However, this objection mistakenly

relies on the strong disquotation principle (a.k.a. ‘converse disquotation’), which most

philosophers have rightly recognized to be false. The strong disquotation principle states that if a

speaker s believes a proposition p, then s will be disposed to assent to every sentence he or she

44
understands that expresses p. As Scott Soames put the principle somewhat differently, the

principle says that “in order to believe a proposition, one must be disposed to accept every

sentence one understands that expresses that proposition (2002: 11).” But the strong

disquotation principle (a.k.a., converse disquotation) is widely considered false (see Salmon,

1986; Soames, 2002). (Cf. the weak disquotation principle, which is widely seen as

uncontroversial). 36 Hence, the fact that many English speakers will not agree that (2) and (2)†

are synonymous—even thinking (2) might be true and (2)† false—does not entail that they are

not synonymous. Intuitions about synonymy can be unreliable, especially in the realm of

philosophical or linguistic theory employed to solve challenging semantic puzzles. We should

evaluate our theory according to whether it does the work we want it to do: to account for the

ordinary intuitions of language users in the puzzle cases.

By way of clarification, I do not claim that ordinary speakers have explicit mental

representations of the formalized schemas of character and content of proper names, or any

common indexical expressions (e.g., I, today, here, etc.) for that matter. Rather, these schemas

are embodied in speakers’ linguistic practices and thus understood by ordinary speakers only

tacitly. The evidence that they are so embodied is that supposing that they are so-embodied

explains speaker intuitions in the puzzle cases. Similarly, the extremely complex rules for the

use of the subjunctive mood in Spanish are embodied in the minds of native Spanish speakers

and only understood by them tacitly, and the evidence for this embodiment is the linguistic

competence of native Spanish speakers with the subjunctive mood.

I think it is now evident from the above discussion why I claim that ‘Clark Kent’ and

‘Superman’ have different meanings as they occur in sentence (2) and why I claim sentences (1)

36
I will also be denying strong disquotation in conjunction with my solution to the puzzle about
rational inconsistent belief. See section 2.8.

45
and (2) express different propositions. It is evident why (1) is uninformative, as it expresses a

trivial singular proposition about self-identity.

Here is (a somewhat metaphorical account of) how and why Olson’s utterance of (2) is

informative to Lois Lane. Upon hearing Olson utter (2), Lois Lane will tacitly understand that

Olson has drawn these names from dossier tokens representing their subject as bearing these

names, and that Olson is asserting that that those dossier tokens have the same subject, i.e., they

are about the same individual. Given that Olson and Lois both have similar background

experiences, both being familiar with the reporter known as ‘Clark Kent’ and the superhero

known as ‘Superman,’ Lois will presume that she and Olson’s ‘Clark Kent’ and ‘Superman’

dossier tokens are likely to be of the same or similar types. If what Olson asserted is true, this

entails that Lois’ respective ‘Clark Kent’ and ‘Superman’ dossiers must also have the same

subject, just as Olson’s do. If Lois believes that Olson intends to speak the truth (and is not

pulling her leg) and is a reliable and authoritative source of information, she will come to accept

that her own ‘Clark Kent’ and ‘Superman’ dossier tokens have the same subject, just as Olson’s

do. She would accept thereby, as a matter of logic, that the dossier types her tokens instantiate,

DCK and DSM, are individuated by the same subject. Hence, she will have come to believe PROP-

2, that Clark Kent is Superman, since the proposition that Clark Kent is Superman just is the

proposition that dossier types DCK and DSM are individuated by the same subject.

This proposed solution to the identity sentences puzzle avoids the pitfalls of both

Millianism and Descriptivism. Millianism gets the modal profile issue right (sentence (1) and (2)

express propositions with the same modal profile—both necessary) but clashes with the intuition

that sentences (1) and (2) express different propositions and fails to explain the difference in

their cognitive value. Descriptivism explains the difference in cognitive value but has the

46
unwanted consequence that (1) and (2) have different modal profiles, 37 and implausibly posits

that names have reference-determining descriptive meanings. The TIUT provides instead that

causal-historical factors determine reference, given that the subjecthood of dossiers is (largely, if

not wholly) a causal-historical property. 38 The TIUT gives us what we seek from our theory of

proper names according to the seven constraints set out in section 1.1. We have an explanation of

the cognitive value differences between uninformative and informative identity sentences based

on their semantically expressing different propositions (with identical modal profiles). ‘Clark

Kent’ and ‘Superman’ lack reference-determining descriptive meanings, both rigidly refer to

Kent-Super, and at the same time differ in semantic content (by differing with respect to

descriptive conceptions), thus explaining the difference in cognitive value between (1) and (2).

2.6 Proposed Solution to Frege’s Puzzle about Propositional Attitude Ascriptions

According to the TIUT, there are two sorts of propositional attitude ascriptions: those in

which names used in a Millian way occur within the scope of the ‘that’-clause, and those in

which names used in a Conception-indicating way occur in the scope of the ‘that’-clause. In the

former case, we have ‘Millian Ascriptions.’ A Millian ascription relates the ascribee to a

singular proposition and is silent about how either the ascriber or the ascribee conceives the

referent of the name in the ‘that’-clause. In the latter case, we have ‘Conception-indicating

Ascriptions.’ A Conception-indicating ascription relates the ascribee to a singular proposition

37
They would have different modal profiles on a non-rigidified Descriptivism. However, they
would have the same modal profiles according to Rigidified Descriptivism. See 4.5 for
discussion of Rigidified Descriptivism.
38
It may turn out to be a partly descriptive property (a hybrid property), but it certainly will not
be entirely descriptive. See my discussion of Kripke’s causal-historical picture of reference,
section 4.3 infra.
47
and, additionally, conveys the way in which the ascribee conceives the referent of the proper

name inside the ‘that’-clause. Conception-indicating ascriptions express a more finely grained

picture of the proposition the agent believes than Millian ascriptions.

Both Conception-indicating and Millian ascriptions can be used to ascribe belief, the

salient difference between them being their levels of grain. Millian ascriptions are coarser-

grained than Conception-indicating ascriptions because they merely indicate the singular

propositions the ascribee believes without indicating the ascribee’s conception of the name’s

referent. Conception-indicating ascriptions are finer-grained because they also convey the

speaker’s descriptive conception of / way of taking the referent of the proper name(s) inside the

‘that’-clause.

Whether a propositional attitude ascription is Millian or Conception-indicating is a

function of the speaker’s expressive intent in uttering it (although, as I elaborate below in section

2.6.2 infra, there are important pragmatic constraints on what the speaker can rationally intend to

communicate to his audience via his or her utterance given what the audience is likely to take the

utterance to mean in the context). That is, propositional attitude ascriptions are semantically

ambiguous between Millian and Conception-indicating readings. To determine whether a

propositional attitude ascription is due a Millian or Conception-indicating reading, we need to

ascertain the speaker’s expressive intent (although, again, contextual clues will generally serve as

reliable evidence about this).

The TIUT’s proposed solution to Frege’s puzzle about propositional attitude ascriptions

is found in the next section, 2.6.1, below, concerning Conception-indicating ascriptions. In the

subsequent section, 2.6.2, I also characterize Millian ascriptions.

48
2.6.1 Conception-Indicating Ascriptions

Consider propositional attitude ascription sentence (4), in which a speaker, Olson, uses

the names ‘Clark Kent’ and ‘Superman’ in a Conception-indicating way.

(4) Lois Lane believes that Clark Kent is Superman

According to the TIUT, sentence (4), if uttered by Olson (supposing he uttered the sentence

knowing it to express a falsehood), would express the false proposition that would be expressed

by (4)†:

(4)† Lois believes that the subject individuating DCK is the subject individuating DSM

where DCK is the dossier type instantiated by Olson’s dossier token that has Kent-Super as

subject and a Clark Kent-y conception and from which he draws ‘Clark Kent,’ and DSM is the

dossier type instantiated by the Olson’s dossier token that has Kent-Super as subject and a

Superman-y conception and from which Olson draws ‘Superman.’

Likewise, suppose that Olson used the names ‘Clark Kent’ and ‘Superman’ in a

Conception-indicating way in ascription sentences (5) and (6).

(5) Lois Lane believes that Clark Kent flies

(6) Lois Lane believes that Superman flies

In uttering sentence (5), Olson would express the false proposition that would be expressed by

(5)†, and in uttering (6) he would express the true proposition that would be expressed by (6)†. 39

39
Of course, Olson would not ordinarily utter either sentence (4) or (5), unless he intends to lie
or mislead, because he realizes that both are false. Nevertheless, I am supposing here that Olson
utters (4) and (5) in order to examine their semantics. By examining the semantics of the true (3)
49
 (5)† Lois believes that the subject individuating DCK flies

(6)† Lois believes that the subject individuating DSM flies

4-s, 5-s, and 6-s, below, schematize the semantics (i.e., state the truth conditions) of the

propositions Olson expresses by uttering sentence (4)-(6) (indicated de re vis-à-vis Kent-Super):

4-s Kent-Super is such that he is the subject of Lois’ dossier token d1 instantiating
the same dossier type, DCK, as instantiated by Olson’s ‘Clark Kent’ dossier token d2 from
which Olson draws ‘Clark Kent,’ and Kent-Super is such that he is the subject of Lois’
dossier token d3 instantiating the same dossier type, DSM, as instantiated by the Olson’s
‘Superman’ dossier token d4 from which Olson draws ‘Superman,’ and Believes (Lois,
<<the subject of d1, the subject of d3>, identity >)

5-s Kent-Super is such that he is the subject of Lois’ dossier token d1 instantiating
the same dossier type, DCK, as instantiated by Olson’s ‘Clark Kent’ dossier token d2 from
which Olson draws ‘Clark Kent,’ and Believes (Lois, <the subject of d1, flies>)

6-s Kent-Super is such that he is the subject of Lois’ dossier token d3 instantiating
the same dossier type, DSM, as instantiated by Olson’s ‘Superman’ dossier token d4 from
which Olson draws ‘Superman,’ and Believes (Lois, <the subject of d3, flies>)

The above schematization of the semantics of the proposition expressed by sentence (4) makes

plain why (4) would express a false proposition. Lois does not realize that her two dossiers

tokens, d1 and d3, have the same subject. In fact, she disbelieves that they have same subject.

Hence, she believes (erroneously) that the dossier types, DCK and DSM, instantiated by her dossier

tokens, d1 and d3, are individuated by different subjects.

The above schematization of the semantics of (5) and (6) reveal that they genuinely differ

in truth-value. Sentence (5) expresses a false proposition and (6) a true proposition. (5) and (6)

do not merely appear differ in truth-value as Millians maintain. Whereas Lois does not believe

and contrasting it with the semantics of false (4), and by examining the semantics of false (5) and
contrasting it with the semantics of the true (6), I aim to show that the Millians’ claim that (3)
expresses the same proposition as (4), and (5) the same proposition as (6), is erroneous.
50
that the subject of her dossier d1 flies, she clearly does believe that the subject of her dossier d3

flies. Hence, (5) expresses the false proposition that Lois believes that the subject individuating

dossier type DCK, which is the dossier type instantiated by her dossier token d1, flies. And (6)

expresses the true proposition that Lois believes that the subject individuating dossier type DSM,

which is the dossier type instantiated by her dossier token d3, flies.

In ascribing beliefs (or other propositional attitudes) to Lois, Olson has in his mind a

model of Lois’ mental architecture. He aims to have a ‘Clark Kent’ dossier token and a

‘Superman’ dossier token that are of the same or similar dossier types as Lois’s ‘Clark Kent’ and

‘Superman’ dossier tokens respectively. Olson ascribes beliefs to Lois by referencing his own

dossier tokens qua tokens of the same types as Lois’s tokens. Below is a (metaphorical)

description of the nature of Olson’s ascription sentence (5n)

(5n) Lois Lane does not believe that Clark Kent flies

from Olson’s first-person perspective:

There is a dossier token in Lois’ mental architecture which is an instance of the dossier
type of which this dossier token is a token [Olson makes his ‘Clark Kent’ dossier token
salient by drawing ‘Clark Kent’ from it]; Lois does NOT believe that the subject of her
dossier token, instantiating the same type as the token from which I am now drawing the
name ‘Clark Kent,’ flies.

Below is a (metaphorical) description of the nature of Olson’s ascription sentence (6)

(6) Lois Lane believes that Superman flies

from Olson’s first-person perspective:

There is a dossier token in Lois’ mental architecture which is an instance of the dossier

51
 type of which this dossier token is a token [Olson makes his ‘Superman’ dossier token
salient by drawing ‘Superman’ from it]; Lois believes that the subject of her dossier
token, instantiating the same dossier type as the token from which I am now drawing the
name ‘Superman,’ flies.

Propositional attitude ascriptions involve the ascriber imagining the way that the ascribee

sees the world and creating within his own mind a temporary model of the ascribee’s worldview,

temporarily forming a dossier structure that mimics and mirrors that of the ascribee. Then the

ascriber can speak about his own dossier tokens as if they were those of the ascribee. He may do

so by relying on the fact that, if he has understood the ascribee’s cognitive state well enough, his

own dossier tokens are of the same or similar type as those of the ascribee. His own tokens and

Lois’ (or anyone’s who is confused about the identity of Clark Kent and Superman) all belong to

the same or similar types, and therefore, the ascriber can point to his own dossier tokens as

instantiating that type and say that all those who do not realize the identity fail to realize that

these dossier types (the one’s his own tokens instantiate) are individuated by the same subject,

Kent-Super, the subject of his own dossier tokens. In other words, the ascriber ascribes belief to

the ascribee by him- or herself “mock” entertaining a similar mental state and then referring to

that mental state as if they belonged to the ascribee. On this view, ascribing belief or other

attitudes involves cognitively “putting one’s self into another’s shoes.”

On the TIUT, (3)-(4) and (5)-(6) express different propositions with different truth-

values. Sentence (3) expresses the trivial singular proposition that Lois believes that Kent-Super

is identical to himself due to the Millian use of both proper names in the ‘that’-clause of (3). A

rational Lois Lane could never disbelieve the proposition referred to by the ‘that’-clause of (3).

However, the semantics of (4), as I have schematized them above, make clear, the ‘that’-clause

of (4) refers to a different proposition from that of (3). Lois Lane can rationally disbelieve the

52
proposition referred to by the ‘that’-clause in (4). She may rationally do so because she fails to

realize that her two dossier tokens, d1 and d3, have the same subject. Furthermore, she can

rationally disbelieve the proposition referred to by the ‘that’-clause in (5) and believe the

proposition referred to by ‘that’-clause in (6), because she is ignorant of the fact that her dossier

tokens d1 and d3 have the same subject. She may rationally believe that the subject of d1 does not

fly and believe that the subject of d3 does fly.

One might wonder why I schematize the Conception-indicating character in terms of

dossier types, rather than dossier tokens. To see why types, rather than tokens, are crucial,

consider the following example. Suppose a man named ‘Tony’ knows a lot about Lois Lane, but

Lois has never met him or heard of him. His existence is completely unknown to her. Now

suppose Tony were to utter propositional attitude ascription ¬ (5)

¬ (5) Lois Lane disbelieves that Clark Kent flies

and suppose further that the content of ‘Clark Kent’ were given by the definite description the

subject of dossier token dCK,Tony, where dCK,Tony is a dossier token in Tony’s head that has Kent-

Super as subject and a Clark Kent-y conception. Sentence ¬(5), as uttered by Tony, would

therefore express the same proposition as the sentence ‘Lois Lane disbelieves the subject of

dossier token dCK,Tony flies.’ This sentence would make a claim about Lois’ belief about the

properties of a particular dossier token in Tony’s head, dCK,Tony. The crux of the difficulty is that

Lois’ belief that Clark Kent does not fly is not a belief about Tony’s dossier token dCK,Tony, and

Lois Lane does not even know who Tony is, so she can hardly entertain beliefs about his dossier

tokens. However, the TIUT avoids this problem because the Conception-indicating character

references dossier types: Lois Lane does indeed have beliefs about dossier types in virtue of

53
having beliefs about her own dossier tokens, which instantiate dossier types. Because Tony can

have a dossier token of the same type as Lois’ token, he can “point to” his token as an instance of

the type and say that Lois has a belief about the type by her having a belief about a token of that

type in her head. The truth-value of the proposition Tony expresses does not depend in any way

on whether Lois has any beliefs about Tony’s dossier tokens. It depends only on her having

tokens of the same type as his tokens and having beliefs about her own tokens. She need not be

familiar with Tony or in an epistemic position to entertain a thought about his dossier tokens in

order for Tony accurately to ascribe beliefs to Lois with respect to Clark Kent’s flying ability (or

lack thereof).

In sum, if the Conception-indicting character were phrased in terms of tokens rather than

types, propositional attitude ascriptions would be overly “autobiographical,” as it were: a

speaker, such as Tony, in ascribing belief to an ascribee such as Lois Lane, would be making a

claim about his own dossier tokens—asserting that Lois has beliefs about his tokens. But when

Tony ascribes belief to Lois about whether she thinks Clark Kent flies, he is not talking about

himself or referring to his dossier tokens nor asserting that Lois has any beliefs about his tokens.

Therefore, one must schematize the Conception-indicating character of names in terms of dossier

types, as the TIUT does.

2.6.2 Millian Ascriptions

Some propositional attitude ascriptions contain Millian names and such ascriptions report

belief between agents and singular propositions. If, when someone uttered sentence (3),

(3) Lois Lane believes that Clark Kent is Clark Kent

54
s/he were to use the name ‘Clark Kent’ in a Millian way (for both occurrences in (3)), s/he would

express the proposition that Lois believes the singular proposition that Kent-Super is Kent-Super.

For the content of each occurrence of ‘Clark Kent’ would, according to the Millian character, be

Kent-Super himself.

If, when someone uttered sentence (6), she were to use the name ‘Superman’ in a Millian

way, she would express a true proposition.

(6) Lois Lane believes that Superman flies

Lois believes the singular proposition referred to by the ‘that’-clause of (6)—that Kent-Super

flies, for she believes Kent-Super that he flies when she thinks of him under a Superman-y

conception. So (6) is true when ‘Superman’ is used in it in a Millian way. As I argued in

section 2.6.1 supra, (6) would also be true if the speaker used the name ‘Superman’ in (6) in a

Conception-indicating way, except the proposition expressed would be different, as set out in

section 2.6.1 (it would be a finer-grained ascription, indicating the ascribee’s Superman-y

conception of Kent-Super.)

With sentence (5), however, truth-value depends crucially on whether the ascriber intends

the name ‘Clark Kent’ to be understood in a Millian or Conception-indicating way.

(5) Lois Lane believes that Clark Kent flies

As I argued in section 2.6.1, consistent with intuitions about the matter, (5) is false when ‘Clark

Kent’ is used in a Conception-indicating way to say that Lois Lane believes that Kent-Super flies

when she conceives him in a Clark Kent-y way. When she conceives Kent-Super in a Clark

Kent-y way, she very much is convinced that he cannot fly. But suppose that someone uttered

55
sentence (5) intending to use the name ‘Clark Kent’ in a Millian way, just to refer Kent-Super

qua individual. The content of ‘Clark Kent’ would be Kent-Super, the man himself. Hence, the

proposition referred to by the ‘that’-clause of the ascription, ‘that Clark Kent flies,’ would be the

singular proposition that Kent-Super flies. Lois does in fact believe this singular proposition (for

she believes it when she conceives Kent-Super as Superman), so (5) is true when ‘Clark Kent’ is

used in a Millian way, even though it is false when used in Conception-indicating way.

It might seem quite surprising that I would maintain that (5) could be used to express a

true proposition, since we tend to have the strong intuition that (5) is false, period. 40 However, I

think this intuition is unreliable and (5) in fact admits of two readings. We have this intuition

because neither we, nor any enlightened speaker (such as Olson), would typically utter (5) qua

Millian ascription because we know that our audience would likely misinterpret it. An utterance

of (5) as a Millian ascription (to express the proposition that Lois believes that Kent-Super flies,

i.e., that Kent-Super is such that Lois believes he flies) would be highly misleading. Upon

hearing a speaker uttering (5), our target audience would presume that we were saying that Lois

Lane believes that Kent-Super can fly when she thinks of him under a Clark Kent-y conception,

as a mild-mannered reporter (i.e., the false proposition that (5) would express were it a

conception-indicating ascription). Thus, we would not typically utter (5) as a Millian ascription

because our audience would misunderstand it, interpreting it in a way inconsistent with our

expressive intent. However, the fact that we would not typically utter a given sentence because

we think we would be misunderstood does not mean that sentence cannot have two legitimate

readings. It is just that as speakers who wish to be understood, we value knowing our audience

40
The Millian claims that (5) only has a Millian reading and it is always true. The TIUT, by
contrast claims that (5) has two readings, one on which it is true (qua Millian ascription) and one
on which it is false (qua Conception-indicating ascription).
56
and make true and maximally comprehensible statements, i.e., statements that can easily be

decoded by our audience in the context.

Some speakers who utter (5) may not know their audience or the nature of Lois Lane’s

confusion. Consider a scenario in which an enlightened speaker—call him “Kurt”—is unaware

that Lois Lane, or anyone in Metropolis, is unenlightened about the identity of Clark Kent and

Superman. Kurt knows that ‘Clark Kent’ and ‘Superman’ are both names for Kent-Super, but he

does not realize that different conceptions are commonly associated with the names by people in

Metropolis. Kurt might overhear Lois Lane utter “Superman sure can fly fast” and utter sentence

(5) merely to report that Lois Lane believes of Kent-Super, de re, that he can fly, without concern

for (or knowledge of) the different ways in which she conceives him. Has Kurt said something

false in uttering (5)? I doubt it. He surely has said something deficient given the context. Kurt

has expressed, via uttering (5), the true proposition that Lois Lane believes the singular

proposition that Kent-Super flies, but his utterance is defective because it is highly misleading

given Lois’ confused state of mind and given the audience he is addressing, which, unlike Kurt,

knows about Lois’ confusion. That Kurt has not spoken falsely, strictly speaking, is evidenced

by the fact that ordinary enlightened speakers aware of Kurt’s ignorance of Lois’ confused state

would not correct Kurt by uttering merely ‘that is false; she does not believe he can fly.’

Realizing that Kurt intends to say something true—that Lois believes of Clark Kent, de re, that

he can fly— but is saying it in a misleading way, the audience would instead correct him by

saying something like: ‘Well Kurt, she does believe the guy can fly, but not when she thinks of

him as Clark Kent, but only when she thinks of him as Superman, for she does not realize Clark

Kent and is Superman.’ In other words, Kurt’s audience can understand Kurt’s utterance as a

(technically true) Millian ascription if they are able to put themselves into Kurt’s shoes and

57
interpret his utterance from his epistemic perspective. They can read his utterance as aiming to

communicate the proposition that Clark Kent, qua individual, is such that Lois believes he can

fly.

We may have difficulty interpreting ‘Clark Kent’ as a name used in a Millian way—just

to refer to Kent-Super and not to pick out Kent-Super under a mild-mannered reporter

conception—precisely because it is so commonly used as a Conception-indicating name in the

Superman story. The names ‘Clark Kent’ and ‘Superman’ have, almost by convention, come to

mean a certain individual under a certain conception. ‘Clark Kent’ has come to mean Kent-

Super dressed as an ordinary civilian and acting a like a regular weak earthling, and ‘Superman’

Kent-Super dressed in a super-hero costume acting like a superhero. Each name is so strongly

associated with a specific well-defined conception from the repetition of this iconic fictional

narrative that our ability to use these names without immediately calling to mind the intimately

associated conceptions has become highly diminished. Therefore, we have difficulty seeing that

we might use either of those names in a strictly Millian way, i.e., just to refer to Kent-Super qua

individual without calling to mind a way of conceiving him. However, bad cases make bad law,

for in real life few Frege’s puzzle cases involve conceptions conventionally associated with

proper names. 41 Let us consider a Frege’s puzzle case in which we do not strongly

conventionally associate certain conceptions with a pair of co-referential proper names, where

the conceptions are instead associated with the names only in the confines of a specific

idiosyncratic conversational context. Suppose that you are standing on the street with your

41
One of the great didactic advantages of using the using the Superman story to explain Frege’s
puzzle is that almost everyone understands the different conceptions commonly associated with
these different names, and these conceptions differ from one another starkly. On the other hand,
this feature of the Superman story can generate intuitions about truth value inapplicable to those
examples of Frege’s puzzle where the conceptions associated with names are idiosyncratic or not
widely known in the wider language community.
58
friend, Bob. Barack Obama walks by, and Bob utters, “Wow, I haven’t seen Barry in ages. He

sure is looking handsome.” Later that day, you report to me, “Bob thinks Obama is looking

handsome.” This propositional attitude ascription seems clearly true. Now, suppose you find

out the following day that that Bob is a childhood friend of Barack Obama, but he only knew

him by the name “Barry” as a child. Bob is unaware that Barry, his childhood friend, is Barack

Obama, the president of the Unites States. Bob knows that the president of the US is named

“Barack Obama” and knows a lot about his policies but somehow does not know what Obama

looks like. When Obama walked by, Bob only recognized him as his childhood friend, Barry,

not as the president of the US, Barack Obama. Query: now that you have found out about Bob’s

confusion, do you deem the propositional attitude ascription you uttered the previous day, “Bob

thinks Obama is handsome,” to have been true or false? My intuition is that the ascription was

true and remains so despite what you learn about Bob’s idiosyncratic confused state and despite

Bob possibly dissenting from the sentence “Obama is handsome,” since Bob has never seen

Obama under the guise of president of the US. When you uttered the ascription sentence, you

intended merely to refer to Obama and describe Bob’s opinion of the way he looks. You were

not concerned with Bob’s various conceptions of him.

In sum, I think the intuition that sentence (5) expresses a false proposition is erroneous.

The intuition is generated the peculiarity of the Kent/Superman case, i.e., the fact that

conceptions have become ossified, strongly conventionally associated with the names. In Frege’s

puzzle cases like the Barry/Obama case, where different conceptions are associated with the

names ‘Barry’ and ‘Obama’ only within the idiosyncratic confines of a particularized

conversational context by a particular individual, we are more inclined to accept that the name

‘Obama’ could be used in a Millian way just to refer to the man himself qua individual.

59
 I have claimed that the speaker’s expressive intent is the principal factor determining

whether the tokening of a name in a propositional attitude ascription is owed a Millian or

Conception-indicating reading. One might then wonder therefore whether a speaker could token

the names ‘Clark Kent’ and ‘Superman’ side by side inside the ‘that’-clause of a propositional

attitude ascription intending a Millian reading of both names. For example, could a speaker use

the names ‘Clark Kent’ and ‘Superman’ in a Millian way inside the ‘that’-clause of ascription

sentence (4) to express the true but trivial singular proposition that Lois believes PROP-1, the

singular proposition that Kent-Super is Kent-Super, i.e., the Kent-Super is self-identical?

(4) Lois Lane believes that Clark Kent is Superman

I do not believe so. What I shall call the ‘Meaning Consistency Principle,’ a pragmatic principle

of conversation explains why no ascriber would use the names ‘Clark Kent’ and ‘Superman’ in

(4) in a Millian way to express the trivial (and true) proposition that Lois Lane believes that

Kent-Super is Kent-Super.

Meaning Consistency: Multiple occurrences of the same syntactic string in a sentence

generally entail that the speaker meant the same thing by each occurrence. 42 Occurrences
of different syntactic strings entail a difference in meaning.

When any audience hears an utterance of (4), it presumes, according the Meaning Consistency

Principle, that the expressions ‘Clark Kent’ and ‘Superman’ are non-synonymous. After all, the

speaker surely must have a reason for using two different names. The audience would never

42
By generally, I mean that there are exceptions when the conversational context makes it clear
that the speaker means different things by the same expression (see Bill/Liz example, section 2.9
infra), or where the grammar of the sentence entails a difference in meaning (e.g., in the sentence
“Rose rose in popularity beginning in high school” we infer from the grammar that the two
occurrences of “rose” must have different meanings).

60
understand both the names to be used in a Millian way, because this would involve using the

names synonymously (since each of their semantic contents would be the same—Kent-Super

himself and nothing more). Every rational speaker would realize that his audience would not

read the names as Millian names, and hence he would understand that his audience would

misunderstand him if he uttered (4) with the intention to use both the names in a Millian way to

say that Lois Lane believes the trivial proposition that Kent-Super is Kent-Super. Perhaps I

would go as far as to say that no rational ascriber could utter (4) to express this proposition,

knowing that the proposition would be universally misunderstood. For what a speaker may

express via an utterance is not merely a function of the expressive intent of the speaker, but in

addition, it is constrained by what the speaker can reasonably expect his audience to take it to

mean. 43 Pragmatic principles of conversation, such as the meaning consistency principle, control

here. Hence, whether a proper name is due a Millian or Conception-indicating reading is a

function of speaker expressive intent constrained by pragmatic principles that guide speakers in

their choice of words with respect to what the audience will understand them to mean. Without

doing violence to the language, speakers cannot intend to mean something by their words when

they know that pragmatic guiding principles of language interpretation will prevent their

message from coming across. 44

43
It is also unlikely that the audience would read the sentence as a Millian ascription (to say that
Lois Lane believes that Kent-Super is self-identical) simply because practically no one except a
logician or a philosopher would ever discuss self-identity or any agent’s beliefs about self-
identity.
44
I do not mean to suggest that it would be impossible to create an artificial language that did not
follow the Meaning Consistency rule. In such a language, (4) would have a possible reading on
which it merely stated that Lois believes that Kent-Super is Kent-Super. However, my concern
here is to characterize natural language in actual use. Such an artificial language would be far
more unwieldy and full of ambiguity than natural language, since the audience would lack
syntactic clues enabling him easily to tell whether the speaker uttered (4) to say that Lois
61
2.7 Quantifying in

Fregeans notoriously have problems with quantifying in to propositional attitude

contexts. Quine first raised the problem in his 1956. To see what the problem is about, consider

the difference between the following propositional attitude ascriptions:

(5) Lois Lane believes that Clark Kent can fly

(5)de re Clark Kent is such that Lois Lane believes he can fly

Ascription (5)de re differs from (5) in that it is expressly syntactically de re with respect to Kent-

Super. (5), by contrast, is ambiguous, having two possible readings, one de re and one non-de

re. On a non-de re reading of (5), it is intuitively false. Lois would never assent to the sentence

“Clark Kent can fly.” However (5)de re is true, for Clark Kent is the same person as Superman,

and Lois believes of him that he can fly under some way of conceiving him.

(5)de re has the following formal analysis:

Θ ∃x (Lois Lane believes that x can fly and x = Clark Kent)

Here, we have quantification into a belief ascription, as the variable ‘x’, which is bound by the

existential quantifier, is within the scope of the propositional attitude belief, inside the ‘that’-

clause of the belief ascription. Here is the problem for the Fregean. As Jeff Speaks points out,

“the standard semantics for the existential quantifier says that a formula ∃xфx is true if there's

believes someone is self-identical or to say that Clark Kent and Superman are the same person.
62
some object o in the domain such that x is true relative to an assignment of o to x.” 45 With

respect to (5)de re , we would say that Θ is true if there is some object o such that o = Clark Kent

and Lois Lane believes the proposition which attributes flying ability to o. Note that this

proposition contains Clark Kent, the flesh-and-blood man as a constituent, and is therefore a

singular proposition. For a Fregean, however, objects cannot be constituents of propositions.

Fregean propositions are composed wholly of senses, not objects. Only a Fregean proposition,

and not a singular proposition, can be the object of propositional attitudes for a Fregean, and thus

the Fregean must consider Θ incoherent if ‘x’ ranges over objects rather than senses. However,

it is not clear what to make of Θ if ‘x’ ranges over senses. A Fregean could say that Θ is true if

there is some sense s such that s = Clark Kent and Lois Lane believes the proposition which

attributes flying ability to s. The problem is that Clark Kent is not a sense, and senses cannot

fly. Perhaps the Fregean can solve the problem, but suffice to say that quantifying in is a

significant problem for the Fregean, whereas it is no problem at all for Millians, who expressly

posit that objects or individuals such as Kent-Super, the flesh-and-blood man, are constituents of

propositions. The Fregean must be able to deal effectively with this problem, and cannot dismiss

it, for there is nothing problematic about (5)de re—it is a well-formed English sentence and

intuitively it is true. Kaplan (1968) proposed an interesting solution to the problem of

quantifying-in for the Fregean, but it is controversial whether he was successful and Kaplan

himself embraced Millianism shortly after publication of his paper.

By contrast with Fregeanism or Descriptivism in general, the TIUT has no difficulty with

quantifying in because it provides that we may use proper names in a Millian way, that is, in a

purely referential way in which the content of a name just is its bearer, just as on Millianism

45
https://www3.nd.edu/~jspeaks/courses/2007-8/93914/_HANDOUTS/fregeanism-5-
problems.pdf
63
itself. Hence, on the TIUT the name ‘Clark Kent’ is ambiguous between Millian and

conception-indicating readings, with the Millian reading corresponding to the de re reading. The

TIUT, in the same way as Millianism, can handle the de re reading, for the TIUT provides that

objects/individuals can be direct constituents of propositions, and singular propositions can be

and often are the objects of propositional attitudes (see section 2.6.2 supra, on Millian

ascriptions).

2.8 Solution to the Problem of Rational Inconsistent Belief

In uttering (8) and ¬(7), Lois Lane would express and believe inconsistent singular

propositions, for she uses the proper names in a Millian way in her utterances.

(8) Superman flies

¬ (7) Clark Kent does not fly

As I argued in section 2.1, she would use the names in a Millian way since she does not realize

that she has two different conceptions of the one Kent-Super, and so she would have no reason to

refer to those conceptions, make them salient, or communicate them in uttering (8) and ¬(7) by

using names in a Conception-indicating way. Sentences (8) and ¬(7), where the names are used

in a Millian way, would express the inconsistent singular propositions that Kent-Super flies and

that it is not the case that Kent-Super flies.

As described above in section 1.2, the Problem of Rational Inconsistent Belief challenges

us to answer two questions.

THE PROBLEM OF RATIONAL INCONSISTENCY

(1) How do we square Lois believing inconsistent singular propositions with her
being a rational agent?

64
 (2) Why is Lois incapable of realizing that she expresses inconsistent propositions
and contradicts herself in uttering sentences (8) and ¬(7), sentences she
understands and uses competently?

The first question can be answered by appealing to dossiers. Lois can believe that Superman

flies, but that Clark Kent does not, because she has two dossiers, one for each persona, and she

does not realize that these dossiers have the same subject, i.e., are about the same individual.

Lois, who believes inconsistent singular propositions, would be irrational only if believed the

inconsistent propositions knowingly, or if she were in an epistemic position such that she could

infer (based on reason alone) that the propositions are inconsistent, and believes them anyway.

The dossier model explains why Lois is not irrational: Lois having two distinct dossiers, and her

inability to find out that they have the same subject based on pure reason, explains why she is not

irrational. The fact that positing dossiers so neatly solves this problem strongly militates in favor

of the view that they must be included in an adequate account of belief ascription.

To solve the second question, the TIUT posits that Lois cannot tell that (8) and ¬(7) are

inconsistent because, according to the Millian character, she has drawn the names ‘Clark Kent’

and ‘Superman’ from dossiers that she does not know have the same subject, and each of those

names refers to the subject of the dossier from which it was drawn—Kent-Super. Thus, on the

TIUT, the key to explaining why Lois is not in an epistemic position to infer that she expresses

inconsistent singular propositions in uttering sentence (8) and ¬(7) is to be found in the indexical

nature of proper names. She does not know that the names ‘Clark Kent’ and ‘Superman’ have

the same bearer because she does not know that the dossiers from which she draws them have

the same subject. Because of this ignorance, she cannot infer that the propositions expressed by

(8) and ¬(7) are inconsistent.

65
 The indexical nature of proper names explains how an agent can utter a sentence and

express a proposition while being partially ignorant of the content of a proper name in a sentence

he or she utters. I want to say that proper names should be seen as closer to pure or automatic

indexicals, rather than true demonstratives. (However, at base Kaplan’s division of indexicals

into two classes—automatic indexicals and true demonstratives—is not a sharp distinction, one

of degree and not in kind. See e.g., Mount (2008), for an argument criticizing the distinction

between automatic indexicals and true demonstratives). The “automatic” part signals that the

content of the indexical is loaded into the proposition automatically upon utterance of the

indexical—i.e., with minimal cognitive demands placed on the part of speaker, for the speaker

need only utter the name to load the content. The content is loaded even where the speaker is

ignorant about or has erroneous beliefs about the content. For example, suppose I look out my

window on April 1, 2014, see that it is sunny out, and utter, “today is a sunny day.” The

indexical expression ‘today’ automatically loads April 1, 2014 into the proposition. I have

expressed the proposition that April 1, 2014 is a sunny day, even if I am under the mistaken

belief that the date is April 2, 2014 when I utter the sentence. Features of the context in which

the utterance occurred, rather the speaker’s belief about the date, determine the content of a

speaker’s utterance. Likewise, features of context, causal-historical factors linking the name to

its bearer, determine the reference of a name, not the agent’s descriptive beliefs about the bearer.

An agent using a proper name as an indexical in a sentence can express a proposition even if

partially ignorant about or under mistaken beliefs about the object picked out by the indexical,

and therefore, does not fully grasp the nature of the proposition he or she has expressed in

uttering the sentence. 46 Lois Lane, for example, does not realize that her ‘Clark Kent’ and

46
Nathan Salmon (1986) also stresses this point: agents may express and believe propositions
66
‘Superman’ dossier tokens have the same subject, so she is partially ignorant about the

propositions she expresses when she utters ‘Superman flies’ and ‘Clark Kent does not fly,’ since

the names ‘Clark Kent’ and ‘Superman’ are drawn from these dossiers and refer indexically, via

the names’ Millian character, to whatever is the subject of the dossier from which drawn. Due to

this ignorance, she cannot tell that the sentences express inconsistent proposition. It is this

ignorance, not irrationality, which explains her inability to recognize the inconsistency of her

utterances. 47

On this view of proper names as indexicals, it is quite easy for a speaker to count as

understanding a name or being competent with a name. This is consistent with Kripke’s

admonition in Naming and Necessity that a speaker requires very little information about the

bearer of a name to be competent with it. A speaker need not have much (or perhaps any)

substantive descriptive information about the bearer of the name and can even be mistaken about

the bearer’s descriptive properties, and still count as understanding the name, as being competent

with it. Contextual, causal-historical factors surrounding the speaker’s utterance, and not what

beliefs the speaker has about the bearer, determine what proposition the speaker has expressed.

while at the same time being partially ignorant of their constituents. They may therefore
simultaneously believe a proposition and its negation without being irrational, since their
ignorance about the nature of the constituents of the propositions blinds them to their
inconsistency.
47
An enlightened agent, by contrast with an unenlightened agent such as Lois Lane, cannot
rationally accept (7) and ¬ (8) or (1) and ¬ (2), for an enlightened agent would realize that the
names co-refer, and s/he either draws both names from the same dossier’s label (a dossier labeled
with both names) or draws the names from distinct dossiers representing their subjects as bearing
different names, but realizes that the dossiers have the same subject. The use of the proper
names as indexicals does not mask the structure or nature of the proposition for an enlightened
agent because the enlightened agent’s different mental architecture (i.e., different dossier
structures) places him/her in a distinct epistemic position giving him/her access to information
about the nature and structure of the propositions expressed that the unenlightened agent lacks.

67
The speaker merely needs to have in his or her head is the intention to use a name with a certain

character meaning (which the speaker knows in a tacit, infra-conscious way) to be a competent

user of it. Thus, Lois Lane can competently use the names ‘Superman’ and ‘Clark Kent’ even

though she is very much in the dark about some very important facts about the bearer(s) of those

names, to wit, that the names have one single bearer who leads a double life with two identities.

Millians cannot appeal to the indexical nature of proper names to solve the problem of

rational inconsistent belief, as the TIUT does. According to standard Millianism (which

excludes single-indexical Millian theories such as that of Pelczar and Rainsbury—see chapter 3),

there is nothing more to the meaning of a proper name apart from its bearer. The standard

Millian view denies that proper names have a character meaning, as the TIUT maintains. The

bearer fully exhausts the meaning of the name. 48 If Lois understands ‘Clark Kent’ and

‘Superman,’ she knows who their bearers are, and she should therefore capable of realizing,

given enough time to reflect and introspect and reflect, that she expresses inconsistent

propositions when she says that Superman can fly but Clark Kent cannot. However, she cannot

discover she is inconsistent, no matter how much time she is given, and Millians cannot easily

explain why. Some Millians (“guise Millians”, most notable Nathan Salmon) appeal to the

48
See Pelczar and Rainsbury (1998, 308-310) for an argument that construing proper names as
indexicals would solve the problem of rational inconsistent belief. Their theory is discussed in
Chapter 3. Their theory of names as indexicals is substantially different from that of the TIUT,
for theirs does not invoke mental dossiers, nor does it provide that names can be used in ways in
which they communicate conceptions. On their theory, proper names have character meanings,
and their contents are always their referents, full stop. Therefore, their theory is a species of
Millianism that differs from standard Millianism owing to their positing that proper names have
character meanings. As a species of Millianism, the theory has the same difficulty (as other
Millian theories) in explaining the cognitive values and truth-value differences arising in Frege’s
puzzles, even if it has an advantage over other forms of Millianism in addressing the Problem of
Rational Inconsistent Belief in a way similar to the TIUT. Francois Recanati (1993) also claims
that proper names have indexical-like properties, although he stops short of claiming that they
are indexicals.

68
notion of propositional guises or ways of taking propositions to explain how agents can be

(partially) ignorant of the contents of the propositions they believe, and thus incapable of

recognizing the inconsistency between some of the propositions they believe and express. A

guise is supposed to be something like a mask covering a proposition in such a way that an agent

cannot see the proposition he believes behind the mask, only the mask itself. An agent could

believe a proposition when it wears one mask and suspend belief (or even disbelieve) that very

proposition when it wears a different mask. The notorious problem is that guise Millians,

including Salmon, have never specified what these propositional guises or ways of taking

propositions are supposed to be. Furthermore, guise Millians have not specified the exact

relation between the propositional guise and the proposition it masks. We know the relation

between a mask and a face (the “wearing-on relation”), but what is relation between a guise and

a proposition? (See Section 5 on Millianism for discussion). The TIUT need not appeal to

propositional guises, since the dossier model combined with the indexical nature of names

explains Lois’ ignorance, essentially playing an analogous role in the TIUT as guises play in

guise Millianism. On the TIUT, the character of a proper name is the guise behind which its

content hides. The TIUT’s indexical proposal, by contrast with guise Millianism, gives us a

concrete and specific account explaining how an agent might believe a proposition while being

partially ignorant about it, and why she could rationally utter sentences expressing inconsistent

propositions. The character of the name is in Lois’ head but a part of the content (the subject) is

not—it is determined externalistically. All Lois has cognitive access to is the character meaning

of the names she utters (which she knows tacitly) and the different descriptive conceptions in her

‘Clark Kent’ and ‘Superman’ dossiers. She would need outside empirical evidence to find out

that the subjects of these dossiers are identical.

69
 Besides the problem of propositional guises being poorly explicated, Millians face

another serious problem with respect to the Problem of Rational Inconsistent Belief, the

challenge of explaining why both enlightened speakers (such as Jimmy Olson, who knows about

the identity) and unenlightened speakers (e.g., Lois Lane, who does not) are inconsistent. By

contrast, on the TIUT only unenlightened speakers such as Lois Lane are inconsistent (e.g., when

she expresses the inconsistent singular propositions that Kent-Super flies and that he does not

fly). Unlike the TIUT, which claims (in accord with common-sense intuitions) that propositional

attitude ascriptions (3) and (4) express different propositions, Millians posit that they express the

same proposition. Therefore, Millians must claim that an enlightened speaker such as Jimmy

Olson who knows that Clark Kent is Superman 49 would contradict himself if he uttered the

seemingly true propositional attitude ascriptions (3) and (4n).

(3) Lois Lane believes that Clark Kent is Clark Kent

(4n) Lois Lane does not believe that Clark Kent is Superman

According to Millianism, (3) and (4n) would express inconsistent propositions—the respective

propositions that Lois believes that Kent-Super is Kent-Super and that Lois fails to believe that

Kent-Super is Kent-Super. That is, in uttering (3) and (4n), Olson would express the proposition

that it is the case that Lois Lane believes Kent-Super is Kent-Super and that it is not the case that

she believes that Kent-Super is Kent-Super. Olson would contradict himself and express

inconsistent propositions in uttering (3) and (4n). The problem of rational inconsistent belief is

hence a deeper problem for Millianism than for the TIUT, since Millians must explain why both

49
Presume throughout this dissertation that Jimmy Olson is aware of the identity of Clark Kent
and Superman, i.e., he is “enlightened.” This may conflict factually with various versions of the
Superman legend.
70
rational unenlightened speakers (such as Lois Lane) and enlightened speakers (such as Jimmy

Olson) contradict themselves. For the TIUT the problem is limited only to unenlightened

speakers such as Lois Lane who do not realize that Clark Kent is Superman.

I do not think that the problem of rational inconsistent belief could be solved by Millians

for enlightened speakers, were it to be the case that enlightened speakers such as Olson really

contradicted themselves in uttering (3) and (4n) as Millians claim. The unenlightened Lois Lane

is ignorant of that her ‘Clark Kent’ and ‘Superman’ dossiers have the same subject. This

ignorance, rather than irrationality, explains why she has inconsistent beliefs with respect to

Kent-Super, such as both believing and disbelieving that he flies. By contrast, the enlightened

Jimmy Olson fully realizes that Clark Kent is Superman and he is fully informed about Lois’

confused and erroneous beliefs about the identity. Olson is not ignorant of any facts whatsoever

about Kent-Super or about Lois Lane that would explain why he makes (allegedly) inconsistent

statements when he when he utters (3) and (4n), leaving irrationality as the only possible

explanation. However, Olson is rational. Any rational enlightened speaker (perhaps even a die-

hard Millian philosopher in his or her moments of inattention to his or her philosophical

commitments) would unhesitatingly utter (3) and (4n) to ascribe beliefs to Lois. Thus, we have

no plausible explanation of Olson’s inconsistency—neither ignorance nor irrationality explains

it. This casts doubt on the Millian claim that that (3) and (4n) express inconsistent propositions

(and, by implication, the claim that (3) and (4) express the same proposition), and thereby

constitutes a reductio of Millianism.

The only possible move for the Millian is to claim that the enlightened Olson utters (3)

and (4n) without realizing they are inconsistent because Olson takes the propositions expressed

by these sentences under different propositional guises such that he does not realize that they

71
express inconsistent propositions. Salmon in fact makes this move, arguing that Olson’s

misconception about the propositions (3) and (4n) express, due to his conflation of pragmatics

and semantics, constitutes a guise under which he erroneously takes the proposition expressed by

(3) and (4) under different guises, such that he has propositional recognition failure with respect

to that proposition, failing to realize that the same proposition is presented in two different ways.

Salmon’s claim on this point is further discussed in section 5.3.6, infra.

2.9 Informative Identities without Substitution

In the literature on Frege’s puzzle about informative identity sentences, it is sometimes

assumed sub silentio that tautological-looking sentences of the syntactic form a=a such as (1)

always express uninformative identities.

(1) Clark Kent is Clark Kent

This assumption goes back to Frege’s characterization of the identity sentences puzzle as he

stated in On Sense and Reference. There, Frege posed the question why sentences of the form

a=a were uninformative and one could tell them to be true just by inspection of their syntactic

form, whereas sentences of the form a=b were informative and one could not tell them to be true

or false merely by inspecting their syntactic form. The rule that sentences of the form a=a can

be determined to be true merely by inspection is true for logic, but not for natural language

sentences. The TIUT takes account of the fact that a natural language sentence of the form a=a,

such as sentence (1), could express either an informative or uninformative identity. It depends on

the speaker’s expressive intent. Sentences do not express propositions on their own; rather,

speakers use sentences to express propositions. On the TIUT, sentence (1) would express an

72
informative identity if the first and second occurrences of ‘Clark Kent’ corresponded to two

distinct ‘Clark Kent’ dossier tokens in the speaker’s mind containing different conceptions, and

the speaker intended the name ‘Clark Kent’ in a Conception-indicating way for both occurrences.

I have discussed an example of (1) as an informative identity supra in section 2.1 and in footnote

5—the case of Tom who meets a man named Clark Kent at a party and suspects he is the same

Clark Kent with whom he went to Kindergarten in Smallville, and utters ‘Clark Kent is Clark

Kent’. Alternatively, consider a variant of Kripke’s Paderewski case discussed above in section

2.1. A man named ‘Peter’ might say ‘Paderewski the pianist has musical talent, but Paderewski

the politician surely does not,’ failing to realize that the pianist and the politician are the same

person. Peter’s friend might say to Peter ‘But Peter, Paderewski is Paderewski’ to inform him

that Paderewski the politician is the same person as Paderewski the musician. Here, Peter’s

friend utters ‘… Paderewski is Paderewski’ to express an informative proposition and not to

point out the trivial fact that Paderewski is self-identical. The tokenings of ‘Paderewski’ allude

to different conceptions of the man—one the conception of a pianist and the other of a

politician. 50 For another example of (1) as an informative identity, consider the following

scenario: Bill and Liz, a married couple, are Clark Kent’s (Kent-Super’s) next door neighbors in

Brooklyn. Bill and Liz have met Clark Kent on several occasions. On each occasion, Bill and

50
Kit Fine’s Semantic Relationism is an example of theory designed to solve Frege’s puzzle that
fails to take seriously the datum that Frege’s puzzle is not about the difference of
informativeness between identity sentences with the syntactic form a=a or a=b. We want to
explain rather the difference between uninformative and informative identities regardless of their
syntactic form—regardless of whether the same or different names appears on either side of the
‘=’ sign. Fine’s theory is built on explaining the difference between sentences of the form a=a
and a=b, with Fine claiming that the names are “coordinated” in the former but not in the latter.
Kripke’s Paderewski puzzle shows that Fine’s undertaking is wrongheaded, for the same puzzle
is generated whether we consider sentences such as ‘Clark Kent is Superman’ or ‘Paderewski is
Paderewski’, or the attitude ascriptions in which these identity sentences are embedded in the
‘that’-clauses.
73
Liz have conceived Clark Kent as being their next-door neighbor and a reporter by profession.

One day, Bill and Liz are at a party and meet a man who introduces himself as ‘Clark Kent,’ but

he claims that he is a resident of Hoboken and a mechanic by trade. He denies ever having lived

in Brooklyn, denies ever having worked as a reporter, or being Bill and Liz’ neighbor. Yet this

Clark Kent (from the party) looks and sounds identical to the other Clark Kent (the neighbor).

Bill and Liz suspect that this Clark Kent at the party is in fact the neighbor pretending to be

someone else. Bill now wants to convey to Liz his belief that the Brooklyn resident reporter,

their neighbor, is the Hoboken mechanic. To do this, Bill utters sentence (1) to Liz, ‘Clark Kent

is Clark Kent.’ Here, Bill uses (1) as an informative identity sentence to inform Liz that he

believes that the Clark they have met at the party is Clark Kent, their neighbor. Here is what is

going on, according to the TIUT, when Bill utters (1). Bill first entertains a dossier token that

contains the conceptual representations named Clark Kent and lives in Hoboken, is a mechanic,

met him at a party. Let us call the dossier type instantiated by Bill’s dossier token ‘DCK1.’ He

introspects the dossier and draws the name ‘Clark Kent’ from it (he utters ‘Clark Kent’). With

the second occurrence of ‘Clark Kent,’ he entertains a different dossier token, also containing the

representations named Clark Kent and lives in Brooklyn, is a reporter, is our next-door neighbor.

Let us call the dossier type instantiated by Bill’s dossier token ‘DCK2.’ He draws the name ‘Clark

Kent’ from the dossier (he utters ‘Clark Kent’). Bill has thereby expressed the informative

identity proposition whose structure is revealed by sentence (1)INF†:

(1)INF† The subject individuating DCK1 is the subject individuating DCK2

Liz may then report Bill’s belief back to him by uttering (3^) as a Conception-indicating

ascription.

74
 (3^) You believe that Clark Kent is Clark Kent

(3^)† below offers us a window onto the proposition expressed by Liz’ propositional attitude

ascription (3^) (reporting that Bill believes (1)INF†):

(3^)† You (Bill) believe that the subject individuating DCK1 is the subject
individuating DCK2

The semantics of the proposition expressed by (3^)†, schematized de re vis-à-vis Kent-Super, is

set out in 3^-s below:

3^-s Kent-Super is such that he is the subject of Bill’s dossier token d1 instantiating
the same dossier type, DCK1, as instantiated by Liz’ ‘Clark Kent’ dossier token d2 from
which Liz has drawn ‘Clark Kent,’ and Kent-Super is such that he is the subject of
Bill’s dossier token d3 instantiating the same dossier type, DCK2, as instantiated by Liz’
‘Clark Kent’ dossier token d4 from which Liz has drawn ‘Clark Kent,’ and Believes
(Bill, <<the subject of d1, the subject of d3>, identity >)

Uttering sentence (1) is an effective way to express an informative identity, and uttering

(3) is an effective way to express a Conception-indicating ascription, given the facts of this

specific conversational setting described here where Bill and Liz are each aware of the different

conceptions they each associate with the name ‘Clark Kent.’ In conversational contexts in which

the different conceptions are not as salient to the conversational participants as in the one

described above, uttering (1) would not be effective in communicating an informative identity,

and uttering (3) would not be an effective in communicating a Conception-indicating ascription.

The pragmatic ‘Meaning Consistency Principle’ already discussed supra in section 2.6.2,

explains why.

75
 Meaning Consistency: Multiple occurrences of the same syntactic string in a sentence
generally entail that the speaker meant the same thing by each occurrence. 51 Occurrences
of different syntactic strings entail a difference in meaning.

According to the above Meaning Consistency Principle, speakers avoid using ascriptions such as

(1) or (3) (to express an informative identity or a Conception-indicating ascription, respectively),

in general, because the repetition of the name ‘Clark Kent’ suggests that each occurrence of the

name is synonymous. Here, however, the tokenings of ‘Clark Kent’ are not intended to be

synonymous. The term “generally” in the above definition of Meaning Consistency indicates

that there are exceptions to the general rule: two uses of the same expression might be used to

mean different things if the conversational context in which the sentence uttered would alert the

audience to the fact that the speaker is using them to mean different things. In cases such as Bill

and Liz example, there is strong mutual knowledge between Liz and Bill regarding two different

conceptions associated with the same lexical item—the name ‘Clark Kent,’ and these different

conceptions are salient in the conversational context. Hence, Bill and Liz are aware that two

occurrences of ‘Clark Kent’ might not mean the same thing in a sentence, overriding the general

principle that syntactically identical strings are presumed to be synonymous. Furthermore, it is

highly unlikely that Bill would utter ‘Clark Kent is Clark Kent’ merely to point out the trivial

and obvious fact that Kent-Super is self-identical. (The utterance would communicate no

relevant information, violating Grice’s maxim of relation). Realizing this, Liz is naturally attuned

to the possibility that the two occurrences of ‘Clark Kent’ in Bill’s utterance are meant to be non-

synonymous.

51
By generally, I mean that there are exceptions when the conversational context makes it clear
that the speaker means different things by the same expression (as in the Bill/Liz example), or
where the grammar of the sentence entails a difference in meaning (e.g., in the sentence “Rose
rose in popularity beginning in high school” we infer from the grammar that the two occurrences
of “rose” must have different meanings).
76
 In conversational contexts in which this mutual knowledge is not as strong with respect to

how the difference occurrences of ‘Clark Kent’ could differ in meaning as here, the principle of

meaning consistency would counsel that a speaker ought to use different names to correspond to

the different conceptions to avoid ambiguity. Where there is just one single name, ‘Clark Kent,’

as here, the speaker could coin partially descriptive names to refer to the same individual under

different conceptions of him or her. For example, a speaker might coin the partially descriptive

expressions ‘Clark from the party’ and ‘Clark our neighbor.’ Then the informative identity

sentence would be ‘Clark from the party is Clark our neighbor.’ The use of these partially

descriptive names avoids the potential ambiguity that might be occasioned by tokening ‘Clark

Kent’ twice over and uttering ‘Clark Kent is Clark Kent.’ Furthermore, uttering ‘Clark from the

party is Clark our neighbor’ or ‘Clark the reporter is Clark the mechanic’ to express the identity

has the advantage that the audience is clued into the different conceptions the speaker has in his

dossiers of the individual referred to, for the conceptions the speaker associates with the names

are built directly into the partially descriptive names.

To summarize my discussion in this section, I maintain that (1) could be used to express

an informative identity, and (3) to express a Conception-indicating ascription, where the

conversational participants are aware of the different conceptions attached to the names by the

speaker and the speaker and his audience mutually know that they are aware of the relevant

conceptions in the context. 52 The use of (1) to express an informative identity represents an

exception to the Meaning Consistency Principle because the conversational participants in the

52
And conversely, using sentences like (1) or (3) will be disfavored by speaker on pragmatic
grounds (because of the Meaning Consistency Principle) where there is uncertainty whether the
conversational participants will realize that the name is used to communicate different
conceptions on different occurrences of it within a sentence.

77
conversational context can readily determine from contextual clues that the speaker likely does

not mean the same thing by the two tokenings of ‘Clark Kent’—that the speaker is using the

names in a Conception-indicating way—and can puzzle out the different conceptions associated

with them from contextual clues.

2.10 Informative Identities: A Priori or A Posteriori Propositions?

In Naming and Necessity, Kripke claims that true informative identity sentences such as

‘Hesperus is Phosphorus’ or ‘Clark Kent is Superman’ express necessary a posteriori

propositions. I agree with Kripke about the now rather uncontroversial claim that the

propositions expressed are necessary, 53 but I shall argue, in agreement with most Millians and

contra Kripke, that strictly speaking they are a priori. Then I then go on to argue that there is

also a “looser” or “non-strict” sense in which we may consider such propositions to be a

posteriori.

Kripke’s claim that the propositions expressed by true informative identity sentences are

both necessary and a posteriori remains controversial. It was revolutionary when he made it in

Naming and Necessity in 1980. The standard view from the time of Kant through the time

immediately prior to Kripke’s Naming and Necessity was that the necessary and the a priori

were coextensive categories, i.e., that a proposition being necessary was both necessary and

sufficient for it being a priori.

Kripke supports his claim that true informative identities involving proper names express

a posteriori propositions in two ways. First, he appeals to our raw intuition that they are a

posteriori. Second, he offers a formal argument for his claim. The raw intuition that they a

53
That informative identities express necessary propositions follows from Kripke’s now widely
accepted thesis that proper names are rigid designators.
78
posteriori seems prima facie compelling, but it is ultimately erroneous, or so I shall claim.

Intuitively, it seems to us they are a posteriori, for we cannot discover Hesperus’ identity with

Phosphorus (they are both names of the planet Venus) or Clark Kent’s identity with Superman

by armchair reflection—just by thinking about the matter. We would have to observe something

empirically to find out that these identities are true. The fact that the Astronomer who discovered

that Hesperus and Phosphorus was one single celestial body (the planet Venus) used

astronomical investigation, not pure mathematics or logic, to make the discovery, seems to

militate in favor of the view that the proposition that Hesperus is Phosphorus is a posteriori.

Similarly, Lois Lane does not discover that Clark Kent is Superman merely through reflection,

but only based on empirical evidence such as seeing Clark Kent changing into his Superman

outfit or spying Kent-Super dressed as mild-mannered reporter bending a bar of steel, or being

told about the identity by Olson. As compelling as the intuition may seem that the propositions

expressed by these informative identities are a posteriori, I shall argue that they are in fact a

priori, strictly speaking.

Several philosophers have pointed out that Kripke’s formal argument for the claim that

identity propositions involving proper names are a posteriori, which appears briefly on page 104

of Naming and Necessity, is flawed. The argument relies on the strong disquotation principle

(a.k.a., converse disquotation), which is now widely seen as false (as opposed to the weak

disquotation principle, which is widely regarded as true). 54 I will not examine Kripke’s formal

argument here. See Soames’ 2003, pp. 379-89 for a detailed critique of Kripke’s formal

54
The weak disquotation principle says: if a competent, sincere, reflective, and rational speaker s
who understands a sentence S is disposed to accept S, and believes S to be true, then s believes
the proposition semantically expressed by S. The strong disquotation principle says: If a
competent, sincere, reflective, and rational speaker s believes a proposition P, then s will be
disposed to accept any sentence S that s understands that expresses P.

79
argument for the a posteriori status of informative identities and Kripke’s problematic reliance

on the strong disquotation principle. See also this footnote.→ 55 To my knowledge, no

philosopher other than Kripke has advanced any formal argument for the existence of necessary

a posteriori identities involving proper names. 56

55
Kripke’s formal argument that informative identity propositions are knowable only a
posteriori is found on page 104 of Naming and Necessity:
“... we do not know a priori that Hesperus is Phosphorus, and are in no position to find out the
answer except empirically… this is so because we could have evidence qualitatively
indistinguishable from the evidence we have and determine the reference of the two names by
the positions of the two planets in the sky, without the planets being the same.” (104)
Scott Soames, Jeff Speaks, and others have criticized this argument. According to Jeff Speaks
(www3.nd.edu/~jspeaks/courses/mcgill/415/kripke-identity-sentences.html):

“Kripke’s point seems to be that we could be in a qualitatively identical situation with

respect to the contexts of introduction and use of these names, and yet, in that possible
situation w, the sentence ‘Hesperus is Phosphorus’ could be false. [But] this argument
seems puzzling: the sentence ‘Hesperus is Phosphorus’ expresses a different proposition
as used in w than it does as used in the actual world. So why does the fact that the
proposition expressed by this sentence in w is false show anything about the epistemic
status of the proposition expressed by this sentence in the actual world? A way to fill the
gap in the argument is via principles connecting acceptance of sentences with belief in
the propositions expressed by those sentences. Consider, e.g., the following such
principle:
If an agent understands some sentence S which expresses the proposition p,
then: (the agent is justified in accepting S iff the agent is justified in believing
p)
We can then read Kripke as arguing that agents cannot know a priori that ‘Hesperus is
Phosphorus’ is true, and using the above principle to reach the conclusion that they
cannot know a priori that Hesperus is Phosphorus.”

The problem with Kripke’s argument is that it goes through only if we accept the principle
Speaks refers to, which is in fact the Strong Disquotation Principle, a.k.a. converse disquotation.
However, the Strong Disquotation Principle is highly dubious. See Scott Soames’ 2005, pages
379-89 for a critique of Kripke’s argument and a critique of the strong disquotation principle.
Kripke himself most likely does not endorse the strong disquotation principle, as his Paderewski
Puzzle (1979) can be seen as a reductio of it.
56
I shall have nothing to say here about Kripke’s argument that identity sentences expressing
natural kinds, such as ‘Water = H2O,’ express necessary a posteriori propositions. Kripke’s
argument for the a posteriori status of the propositions is more closely tied to his essentialism.
There are distinct arguments behind the claims about the epistemological status of these sorts of
80
 Before I argue that these identity propositions are a priori, it is necessary first to clarify

what it means for a proposition to be a priori and what means for a proposition to be a

posteriori. Following Fitch (1975, 243) I shall say that a proposition is a priori iff it is knowable

a priori (even if the proposition is also knowable a posteriori), and a proposition is a posteriori

iff it is knowable only a posteriori. Hence, mathematical propositions, (at least some of) which

are knowable both on a priori and a posteriori grounds, are to be classified as a priori

propositions. 57 Propositions such as the proposition that Barack Obama was the president of the

US in 2014, which are knowable only on a posteriori grounds, are to be classified as a posteriori

propositions. Accordingly, I shall construe Kripke’s claim that true identity sentences involving

proper names express a posteriori propositions as tantamount to the claim that these propositions

are knowable only a posteriori. I shall argue that they are strictly speaking a priori because they

are knowable on both a posteriori and a priori grounds. The a posteriori way of knowing them

is much more important and powerful because it enables the knower to draw many further

inferences, whereas the a priori way of knowing them does not. For this reason, that we have the

powerful but erroneous intuition that they are a posteriori.

The A POSTERIORI way of knowing PROP-2 (the informative identity proposition

expressed by ‘Clark Kent is Superman’)

Suppose Lois finds out that her ‘Clark Kent’ dossier token (which we may call ‘dCK’),

which is about Kent-Super and contains a Clark Kent-y conception, has the same subject as her

‘Superman’ dossier token (which we may call ‘dSM’), which is about Kent-Super and contains a

Superman-y conception. It follows as a matter of pure logic that she believes that the dossier

propositions on the one hand and identity propositions involving proper names on the other.
57
For example, the a priori proposition expressed by ‘1+1 = 2’ can be justifiably believed on the
basis of empirical evidence. I can have empirical evidence for it if I, e.g., place one apple into a
basket, another apple, and then count them to see that I end up with two apples.
81
types instantiated by dCK and dSM, DCK and DSM respectively, are individuated by the same

subject. That dossier types DCK and DSM are individuated by the same subject is precisely the

proposition that, according to the TIUT, is expressed by (2)—PROP-2. (See sects. 2.4, 2.5). By

coming to believe or discovering a “local” fact about her own mental architecture, i.e., that her

dossier tokens dCK and dSM have the same subject—Lois thereby believes the general proposition

that dossier types DCK and DSM are individuated by the same subject.

Here, Lois has learned PROP-2 via a posteriori means. Neither Lois nor anyone else can

tell that Lois’ dossier tokens dCK and dSM have the same subject by armchair reflection. To find

out whether Lois’ dossier tokens have the same subject we would have to examine the dossier

tokens in Lois’ mental architecture—the ‘Clark Kent’ and the ‘Superman’ dossier tokens—to

determine whether they have the same subject. This would involve an inquiry into causal-

historical facts; we would need to learn about the events that led to the creation of these dossier

tokens in Lois’ mind. This would be a purely empirical/a posteriori inquiry. As Hume taught

us, we cannot know anything about what-causes-what by pure reason. Lois relies on empirical

evidence to learn that her two dossiers have the same subject. She would need to see Kent-Super

dressed as Clark Kent bending a bar of steel, or catch Clark Kent changing into his Superman

outfit, or notice that Clark Kent and Superman look suspiciously similar, to realize that that her

dossiers have the same subject. These sorts of empirical evidence will make it manifest to her

that her dossier tokens are about the same man. Hence, the discovering of an informative

identity, such as that Clark Kent is Superman, is here an empirical process involving empirical

evidence. The agent discovers empirically that his or her distinct dossier tokens have the same

subject, thereby coming to believe that, in general, the dossiers types instantiated by tokens of

those types are individuated by the same subject.

82
 The A PRIORI way of knowing PROP-2 (the informative identity proposition expressed
by ‘Clark Kent is Superman’)

In addition to being known in an a posteriori way as described above, the proposition

expressed by (2), PROP-2, can also be known in an a priori way. Therefore, it is an a priori

proposition (since a proposition being knowable a priori is sufficient for it being classified as an

a priori proposition).

According to the TIUT’s account of the content of proper names when used in a

Conception-indicating way, as set out in section 2.4 supra, the proposition expressed by (2),

PROP-2, could also be expressed by (2)†.

(2)† The subject individuating dossier type DCK is the subject individuating
dossier type DSM

Importantly, note that sentence (2)† is semantically equivalent to sentence (2)†† below:

(2)†† The subject individuating the dossier type individuated by Kent-Super as

subject and a Clark Kent-y conception is the subject individuating the
dossier type individuated by Kent-Super as subject and a Superman-y
conception.

Sentence (2)†† results from sentence (2)† via substitution of the individuation criteria of the two

dossier types in place of their names, DCK and DSM. (2)† and (2)†† express the same proposition

because dossier types DCK and DSM have these individuation criteria and necessarily so. The

individuation criteria of dossier types are after all essential properties of dossier types.

Note that sentence (2)†† is a logically true and therefore, one may know a priori that it

expresses a true proposition. 58 59

Therefore, any rational agent who read (2)†† and noticed that it

58
Technically, (2)†† does not express an a priori proposition because it is not a priori knowable
whether Kent-Super exists. A sentence that would in fact express an a priori proposition would
be: ‘If Clark Kent exists, then Clark Kent is Superman’, for on the TIUT this would mean the
same as the logically true sentence: ‘If the subject individuating the dossier type individuated by
83
was logically true should assent to it and be willing to assert it. As discussed in section 2.8,

supra, the TIUT posits that the agent would count as understanding sentence (2)†† even if s/he

knew next to nothing about or has erroneous beliefs about the properties of the bearer of ‘Kent-

Super’ and even if she fails to realize that ‘Kent-Super’ refers to the same person as ‘Clark Kent’

or ‘Superman.’ Grasping the character of the names (tacitly and infra-consciously) is sufficient

for understanding them, being a competent user of them, and referring to their bearers, since (I

claim) proper names are automatic indexicals. Via Kripke’s weak disquotation principle, 60 we

may infer that any such agent who is disposed to accept and assert (2)†† believes the proposition

it expresses, PROP-2. Since the agent’s reason for believing PROP-2 is the mere fact that a

sentence that he understands and accepts is logically true, rather than empirical evidence, his or

her grounds for believing PROP-2 are a priori. Hence, PROP-2 is a priori because it’s being

knowable a priori is sufficient for classifying it as an a priori proposition, even if it is also

knowable empirically (via Lois discovering that one’s dossier tokens, dCK and dSM, have the same

subject).

Kent-Super as subject and a Clark Kent-y conception exists, the subject individuating the dossier
type individuated by Kent-Super as subject and a Clark Kent-y conception is the subject
individuating the dossier type individuated by Kent-Super as subject and a Superman-y
conception.’ This sentence is logically true. For the sake of exegetical simplicity, I leave out the
“if...exists” clause and treat (2)†† as if it genuinely expressed an a priori proposition.
59
In order for (2)†† to be logically true, it would have to be uttered in a language in which it is
stipulated by convention that syntactically identical strings are synonymous (such as in formal
logical), so that there can be no doubt that the first and second occurrences of ‘Kent-Super’ in
(2)†† have the same reference. Note that what I call the Meaning Consistency Principle captures
this presumption, which already exists in natural language, that syntactically identical
expressions have the same meaning/semantic value. However, this presumption is not carved in
stone in natural language as in formal logic.
60
According to the weak disquotation principle, if a competent, sincere, reflective, and rational
speaker s who understands a sentence S is disposed to accept S, and believes S to be true, then s
believes the proposition semantically expressed by S.
84
 Learning PROP-2 a posteriori matters more than learning it a priori

When Lois comes to believe the proposition expressed by (2) via believing the a

posteriori proposition that her dossier tokens dCK and dSM have the same subject, she can draw a

host of important and interesting inferences. Upon learning this, she realizes that all the

conceptual representations in her two dossiers, her ‘Clark Kent’ and ‘Superman’ dossiers, relate

to the same individual. To the extent that her conceptions are accurate, she may infer that they

guy she works with is the Superhero she loves, that if her colleague named ‘Clark Kent’ ate eggs

for breakfast then a flying super hero did, that if her mild-mannered colleague is 6’ 9”, so is her

superhero love interest; that the guy she calls ‘Clark Kent’ is amazingly strong, and so forth. By

contrast, when she believes the same proposition merely by accepting (2)†† and seeing that it is

logically true, she can infer no further facts whatsoever from her belief. She cannot even infer

from her acceptance of (2)†† that sentence (2) expresses a true proposition, since she does not

recognize that sentence (2) and (2)†† express the same proposition. Lois might assent to (2)†† and

assert that it expresses a true proposition, but nevertheless insist that the person referred to as

‘Kent-Super’ in sentence (2)††, whoever he may be (and if he in fact exists), is not the individual

she knows as ‘Clark Kent’ or ‘Superman.’ As I claimed in section 2.8, Lois may rationally fail to

realize that the names ‘Clark Kent,’ ‘Superman’ (used in (2)), and ‘Kent-Super’ (used in (2)††)

all co-refer. She grasps the character of the names, and so counts as a competent user of them,

without realizing that they have the same content. Hence, Lois suffers here from a phenomenon

akin to what Nathan Salmon (1986) has called “propositional recognition failure” because she

understands sentences (2) and (2)†† without seeing (and without being able to see no matter how

much she reflects upon the matter) that they express the same proposition. 61

61
However, the TIUT explains propositional recognition failure in terms of Lois having distinct
85
 I’ll say that Lois realizing that PROP-2 is true via learning that (2)†† expresses a true

proposition results in an “inferentially impoverished” belief that PROP-2. By contrast, when she

learns PROP-2 via learning that her dossier tokens dCK and dSM have the same subject, she ends

up with an “inferentially rich” belief that PROP-2. This explains why we erroneously intuit that

PROP-2 must be a posteriori: anyone who learns it merely via a priori means ends up with

“inferentially impoverished” belief—unable infer anything useful or interesting about the world.

By contrast, anyone who learns it via empirical means, by believing that their dossier tokens

have the same subject, ends up with an “inferentially rich” belief. Since we are only interested in

what agents believe in inferentially rich ways and are disposed to ignore situations in which

agents believe propositions in inferentially impoverished ways, we focus only on the case where

Lois learns PROP-2 in an a posteriori way. Nevertheless, PROP-2 is to be counted as a priori

because it can be learned by a priori means, even if learning the proposition this way results in

an inferentially impoverished belief. This claim, that PROP-2 is a necessary a priori

proposition, should not be surprising, given that the very notion that a necessary proposition

could be knowable only a posteriori is puzzling and highly controversial.

We are interested in an agent’s beliefs to be able to make sense of and predict his or her

behavior, both linguistic and otherwise. This involves understanding the sorts of inferences an

agent is likely to draw based on what the agent believes. However, an agent’s inferentially

impoverished beliefs tell us little or nothing about what the agent will do or say. We do not care

about them. Since we care only about inferentially rich beliefs, we ignore them, and we would

not ascribe to Lois Lane the belief that Clark Kent is Superman merely because she reads

sentence (2)†† and recognizes it as logically true and therefore expressive of a true proposition.

dossiers and the indexical nature of proper names, rather than via the Millian notion of guises.
86
Rather, we would only ascribe to her the belief that PROP-2 if she came to believe it via

believing the a posteriori proposition that dCK and dSM have the same subject, for this would

result in her having an inferentially rich belief. The expression ‘believes’ (as well as other

expressions used to ascribe propositional attitudes) are used in propositional attitude ascriptions

to ascribe only inferentially rich beliefs (or other attitudes) to the ascribee, and exclude

inferentially impoverished beliefs. Hence, I claim that sentence (6)

(6) Lois Lane believes that Clark Kent is Superman

would express a literally false proposition even if Lois believed the proposition referred to by the

‘that’-clause in an inferentially impoverished way. Sentence (6) would express a true proposition

only if Lois believed the proposition referred to by the ‘that’-clause in an inferentially rich

way. 62

62
The contrast between inferentially rich and inferentially impoverished belief can be illustrated
by an example from physics. Suppose that a science novice attends a seminar on particle physics
and learns from the seminar that charmed baryons have masses ranging between 2300 and 2700
MeV/c2. Apart from understanding that charmed baryons are some sort of particle, this science
novice does not genuinely understand the function of a particle in the atom and does not grasp
the meaning of “MeV/c2” except insofar as s/he realizes that is refers to a measurement of the
mass of a particle. He does not really understand what mass is as it pertains to a particle, only
possessing the layman’s non-scientific understanding of “mass” as it applies to macroscopic
objects. Are we prepared to say that this science novice really understands the proposition that
charmed baryons have masses ranging between 2300 and 2700 MeV/c2 well enough to say that
s/he really believes it? There is a weak or non-robust sense of understanding/belief according to
which we are disposed to say that s/he does believe this proposition. However, there is also
stronger and more robust sense of understanding/belief on which we would say that this science
novice does not believe this proposition. The science novice can infer no further facts from his
belief. His belief is inferentially impoverished. He cannot infer any further facts about particle
physics from his belief, nor can he say why this fact is important or how it fits in with quantum
theory. Here, we a case where there is a sense in which this person believes that charmed
baryons have masses ranging between 2300 and 2700 MeV/c2 and a clear sense in which his
level of understanding is too rudimentary to say he really believes this proposition in a robust
sense. Does this science novice really believe this proposition or does he not? Can s/he really be
said to believe the proposition given his poor, inferentially impoverished understanding of it? I
87
We can recognize and understand propositions in different ways; ways of knowing propositions
may vary in their inferential strengths

As Nathan Salmon has claimed (1986), we grasp propositions in different ways. I agree

with this claim. An agent may believe a proposition when it is presented one way but fail to

believe it when presented another way, in which case the agent suffers from “propositional

recognition failure” because the agent fails to recognize the proposition as the same one when it

is presented in different ways). Lois may come to believe PROP-2 by believing that (2)††

expresses a true proposition and at the same time believe that (2) expresses a false proposition:

she fails to realize that the same proposition is presented by the two sentences. Her failure to

recognize two sentences as expressing the same proposition should not be surprising once we

reject the strong disquotation principle, which, as I mentioned supra in section 2.5, is widely

regarded as implausible. 63 The TIUT’s account of proper names as indexicals explains

propositional recognition failure: agents may understand two sentences by grasping the character

of the indexicals it contains without realizing that these indexicals have the same content. Lois

Lane may understand both sentence (2)†† and sentence (2) and see that (2)†† must express a true

proposition (because she realizes it is logically true) but nevertheless she fails to realize that the

proposition it expresses is the same one that (2) expresses. She does not realize that the name

‘Kent-Super’ in sentence (2)†† has the same content as either of the names ‘Clark Kent’ or

do not think there is any non-arbitrary answer to the question whether the science novice
understands the proposition well enough to be said to believe it. All we can do is recognize that
there are different answers we can give to this question depending on our standards for what
should count as sufficient understanding when we attribute belief.
63
As stated previously, the strong disquotation principle states that if a speaker s believes a
proposition p, then s will be disposed to assent to every sentence he or she understands that
expresses p. As Scott Soames put the principle somewhat differently, the principle says that “in
order to believe a proposition, one must be disposed to accept every sentence one understands
that expresses that proposition (Soames, Beyond Rigidity 11).”

88
‘Superman’ in sentence (2).

A ‘Looser sense’ in which these propositions are a posteriori

I stated at the beginning of this section that I would argue that identity propositions such

as PROP-2 are a priori strictly speaking. I also stated that there was a “looser” or “non-strict”

sense in which we can say that propositions such as PROP-2 are a posteriori. I shall now

discuss this looser sense on which such propositions are a posteriori.

On the looser sense of “a priori,” we do not count an agent as understanding a proposition

p when reading or hearing a sentence expressing p if the agent understands p in an inferentially

impoverished way. An agent counts as understanding a proposition only if he or she understands

it in an inferentially rich way. Hence, Lois does not count as understanding PROP-2 when she

reads and assents to sentence (2)†† because her grasp of PROP-2 via this sentence is inferentially

impoverished. This is reasonable, since when we attribute belief to an agent, we are interested in

understanding what the agent will infer from his or her beliefs and in predicting how that agent is

likely to behave, linguistically or otherwise, in virtue of holding the belief. Inferentially

impoverished beliefs do not tell us anything useful in this regard.

On this looser sense, Lois Lane does not grasp PROP-2 via assent to (2)††. This carves

out a quasi-exception to Kripke’s weak disquotation principle. Again, the weak disquotation

principle states that if an agent accepts sentence s, a sentence the agent understands, and s

expresses proposition p, then s believes p. The quasi-exception I propose is the following: if the

agent’s grasp of p is inferentially impoverished when she grasps the proposition p via

understanding sentence s, she does not count as genuinely understanding p via s. She does not

understand p via s, and therefore cannot believe p via s, since one cannot believe what one does

not understand. However, note that I am not in fact suggesting a formal exception to weak

89
disquotation, since I am maintaining that strictly speaking an agent reading (2)†† would count as

understanding it and grasping p via it. The exception applies only to the non-strict sense of

aposteriority. Strictly speaking Lois does understand (2)†† because understanding a proper name

requires only grasping its character and does not require having fully accurate or rich descriptive

knowledge about the referent (or content) of the name. For Lois to count as understanding (2)††,

she does not need to know anything substantive about the name’s referent, nor realize that ‘Kent-

Super’ refers to the same individual as the names ‘Clark Kent’ or ‘Superman.’ But the question

at issue is whether there is another sense of understand, a non-strict or looser sense on which we

can say that Lois fails to understand sentence (2)††. And clearly such a sense of ‘understand’

exists.

Understanding is not an all or nothing matter. Suppose I, a non-physicist with limited

understanding of physics, assent to ‘E = mc2’ because I have it on good evidence that all

scientists believe it is true. Furthermore, suppose I know what each of the symbols stands for.

But I do not know what one uses the formula for or how it was derived, and I do not know how

relativistic mass is different from mass in Newtonian mechanics. I think of mass as just ‘the

amount of stuff.’ Do I understand ‘E = mc2’ well enough to count as believing it is true? I doubt

this question has any non-arbitrary answer. We can have different standards for what counts as

sufficient understanding of a proposition to count as believing it, but there is no such thing as the

correct true standard. It is we who decide upon the standard, which may vary from person to

person and context to context. There’s likely no fact of the matter which standard is correct.

Conclusion

Propositions are a posteriori iff they are only knowable a posteriori. Propositions

knowable both a posteriori and a priori are a priori propositions. According to the TIUT,

90
informative identity propositions are knowable both a priori and a posteriori. But it is only

when the proposition is known a posteriori, via learning that her dossier tokens dCK and dSM have

the same subject, that the agent can infer anything interesting. As ascribers, we are only

interested in agents’ beliefs that will allow us to predict their behavior, both verbal and

otherwise, and to predict the further inferences they will draw from what they already believe.

Hence, we are disposed to discount the a priori way of knowing an informative identity

proposition, and therefore, we have the erroneous intuition that informative identity propositions

are knowable only a posteriori. The TIUT explains why we have the strong (but ultimately

erroneous) intuition, emphasized by Kripke (1979, 1980), that the learning of the necessary

propositions such as those expressed by informative identity sentences involves an empirical

discovery. The TIUT offers us a picture of necessary identity propositions (necessary, because

of the rigidity of the names in the identity sentences expressing them) that are discovered to be

true empirically because the discovery process involves the agent learning that her dossier tokens

dCK and dSM have the same subject, i.e., learning empirical facts about his own mental

architecture. The agent empirically finds out that his distinct dossier tokens have the same

subject, thereby coming belief that, in general, the dossiers types instantiated by tokens of those

types are individuated by the same subject. The TIUT agrees with Kripke to the extent that it

posits that the necessary propositions expressed by ‘Clark Kent is Superman’ and ‘Hesperus is

Phosphorus’ are usually learned empirically, and we are mostly interested in the cases in which

agents learn those propositions empirically. However, the TIUT takes issue with Kripke’s claim

that these propositions are therefore a posteriori. There are ways of learning them that are

strictly speaking a priori, and this is sufficient to classify them as a priori propositions. If we

speak non-strictly, there is a sense on which these propositions can be considered a posteriori,

91
and this is because the propositions are inferentially impoverished when learned in an a priori

way and are inferentially rich and interesting only when learned/grasped empirically.

2.11 Ascribing Belief to Non-Verbal Agents

Suppose that Lois Lane were a non-verbal but intelligent deaf-mute or an intelligent non-

human animal incapable of speech. She does not use or recognize proper names because she

incapable of using language. Nevertheless, she might very well believe that Clark Kent and

Superman are two distinct individuals and that Superman flies and Clark Kent does not, despite

her inability to express these beliefs using language. Despite her neither using nor recognizing

the names ‘Clark Kent’ or ‘Superman,’ we may report her beliefs using sentences (5n) and (6).

(5n) Lois Lane does not believe that Clark Kent flies

(6) Lois Lane believes that Superman flies

On the TIUT, the truth of (5n) and (6) does not entail that Lois is disposed to utter either ‘Clark

Kent flies’ or ‘Superman flies,’ or that she would assent to either sentence. Furthermore, the

truth of (5n) and (6) do not entail that Lois has dossiers with representations that their subjects

bear the names ‘Clark Kent’ or ‘Superman.’ The semantics of these sentences (when those

names are used in a Conception-indicating way) reveal why:

5n-s Kent-Super is such that he is the subject of Lois’ dossier token d1 instantiating
the same dossier type, DCK, as instantiated by the ascriber’s ‘Clark Kent’ dossier token d2
from which the ascriber has drawn the name ‘Clark Kent,’ and NOT Believes (Lois, <the
subject of d1, flies>)

6-s Kent-Super is such that he is the subject of Lois’ dossier token d3 instantiating
the same dossier type, DSM, as instantiated by the ascriber’s ‘Superman’ dossier token d4
from which the ascriber has drawn the name ‘Superman,’ and Believes (Lois, <the
subject of d3, flies>)

92
As shown above, the Conception-indicating character does not state that the ascribee’s dossiers

represent theirs subject(s) as bearing any specific names, or any names at all. Rather, it entails

only that the ascriber’s dossiers represent their subjects as bearing the names uttered (since the

names uttered are drawn from the ascriber’s dossiers, which contain the representation that the

subject of the dossiers bears those names). The Conception-indicating character provides that

the ascribee has a dossier token of the same type as the one from which the ascriber has drawn

the name uttered, but not that the ascribee necessarily has a dossier token representing its subject

as bearing the name the ascriber uttered. The ascribee may have a dossier token of the type

instantiated by the dossier token from which the ascriber has drawn the name even if the

ascribee’s dossier token represents its subject as bearing a different name or even no name at all.

Dossier tokens do not have to match in all details to be of the same type. So, when uttering (5n)

and (6), an ascriber may use the names ‘Clark Kent’ and ‘Superman’ to express contrasting

conceptions of Kent-Super (which aim to be like the non-verbal Lois’ contrasting conceptions)

even if Lois does not associate those conceptions with those names (or for that matter, with any

names whatsoever). Lois has two distinct dossiers for Kent-Super, one containing a Clark Kent-

y conception and the other a Superman-y conception, even though her dossiers contain no

representation about what name or names their subject(s) bear because she is nonverbal.

In addition to explaining ascription of propositional attitudes to non-verbal agents, the

TIUT also explains ascription of propositional attitudes where the ascribee is verbal but the

ascriber does not know which contrasting proper names the ascribee would use to communicate

contrasting conceptions. For example, consider the Hesperus/Phosphorus case. For centuries, the

ancient Greeks believed that Hesperus, a celestial body seen at a certain location in the morning

93
sky at certain times of the year, was a distinct object from Phosphorus, a celestial body appearing

in the evening at a different time of the year at a different location in the sky. However, ancient

Babylonian astronomers had previously discovered that Hesperus and Phosphorus were the same

celestial body, which we today refer to as the planet ‘Venus.’ Eventually the Greeks came to

accept the Babylonian view and came to believe that Hesperus was Phosphorus. ‘Hesperus’ and

‘Phosphorus’ were Greek names for this object as seen in the evening and as seen in the

morning, respectively. We do not know what names the Babylonians used prior to the discovery

of the identity of Hesperus and Phosphorus. The ancient Maya also independently discovered

the identity of Hesperus and Phosphorus, but we do not know what names they used either. 64

Nevertheless, we can accurately say, using the Greek names, which neither the Maya nor the

Babylonians used, that the Maya and the Babylonians discovered that Hesperus was Phosphorus.

The TIUT explain why we can use the Greek names to ascribe propositional attitudes to the

Babylonians and Maya. A modern English speaker uttering the Greek names ‘Hesperus’ and

‘Phosphorus’ picks out the dossier types instantiated by his or her own ‘Hesperus’ and

‘Phosphorus’ dossier tokens, and says that the Babylonians and the Maya had dossier tokens of

these types (but not necessarily that they used these names) such that they did not initially realize

that dossier token instantiating these types had the same subject, and then came to later to believe

that they did. Nothing in the semantics of the TIUT entails that the Maya or Babylonians ever

had any representations in their dossiers that their subjects bore the names ‘Hesperus’ or

‘Phosphorus.’

Here, the TIUT has an advantage over Forbes’ dossier based theory of names (1990).

Problematically, on Forbes’ theory, the sentence ‘The Maya believed that Hesperus was

64
The platform of Venus at Chichen Itza was built specifically to track the celestial orbit of
Venus.
94
Phosphorus’ would be glossed as:

Hesperus is such that, for the Maya’s so-labeled way of thinking of it, α, and Phosphorus
is such that, for the Maya’s so-labeled way of thinking of it, β: Believed (The Maya, < α
= β >)

The problem is the “so-labeled” language, which renders Forbes’ theory overly metalinguistic.

According to the theory, the names used in the ‘that’-clause of every propositional attitude

ascription are the labels of the ascribee’s dossiers. Thus, according to Forbes, to say ‘The Maya

believed that Hesperus was Phosphorus’ is to say that the Maya had dossiers that they labeled

‘Hesperus’ and ‘Phosphorus,’ that these dossiers were both about the same object but represented

that object in different ways, α and β, and the Maya believed that these dossiers were about the

same object. The problem is that the Maya did not label their dossiers with these names. Thus,

on Forbes’ theory, it is not possible to ascribe belief where one does not know what names the

ascribee used or uses to label his or her dossiers. The theory is too metalinguistic in its holding

that proper names themselves are constituents of the content of propositional attitude ascriptions.

Plainly, we can ascribe belief to agents when we do not know how they label their dossiers or

which names they would use, as in the case of the Maya/Babylonians and Hesperus/Phosphorus.

We can moreover ascribe belief even when we are dealing with a nonverbal agent (who does not

use names and whose dossiers contain no representation about names), as in the non-verbal Lois

Lane example above. 65 Forbes’ theory is further discussed in chapter 7.

65
Metalinguistic theories of proper names run into similar problems. Metalinguistic theories of
proper names, although they differ from one another in the details, maintain, in rough sketch,
that the meaning of a proper name ‘NN’ is the bearer of ‘NN’ or the individual called ‘NN.’ On
such a theory, the ascription ‘The Maya believes that Hesperus was Phosphorus’ would mean the
same as ‘The Maya believes that the bearer of ‘Hesperus’ was the bearer of ‘Phosphorus.’ The
problem is that the Maya were not familiar with either of these names, so they did not entertain
any beliefs about them.
95
2.12 Proposed Solution to Kripke’s Puzzle

In “A Puzzle about Belief” (1979) Kripke aims to blunt the force of a certain sort of

reductio argument often used against Millianism by neo-Fregeans (although Kripke does not end

up fully endorsing Millianism). Let us call this neo-Fregean argument against Millianism “The

argument from substitution failure.”

The Argument from Substitution Failure

(i) Millianism entails that the following principle, “substitutivity,” is viable:

SUBSTITUTIVITY: because co-referential names do not differ in any semantic

property, they are freely substitutable into propositional attitude and modal contexts
without changing the proposition expressed or its truth-value.

(ii) However, substitutivity entails that agents sometimes believe inconsistent

propositions. For example, Lois Lane would believe both that Kent-Super flies (when
she conceives him in a Superman-y way) and that he does not fly (when she
conceives him in a Superman-y way).

(iii) A rational agent would not, upon reflection, believe inconsistent propositions

(iv) Lois Lane is a rational and reflective agent, yet she believes inconsistent propositions
even after rational reflection upon her beliefs.

(v) Therefore, substitutivity (and Millianism itself) must be false. Co-designative names
are not substitutable in propositional attitude contexts. 66

Kripke aims to show that the argument from substitution failure, sketched above, does not in fact

impugn Millianism. One cannot solve Frege’s puzzle by giving up Millianism, becoming a

Fregean, and barring substitution of co-referential names in propositional attitude contexts that

lack the same sense. Kripke shows that the same sort of puzzle, in which a rational and reflective

66
Neo-Fregeans would allow substitution of co-referential names only if they had the same
sense.
96
agent ends up looking like they believe a contradiction, can be generated whether we are

Millians or Fregeans, and whether we allow substitution of co-designative names or bar it.

Hence, the argument from substitution failure fails to count as a reductio of Millianism or

militate in favor of Fregeanism. Which theory of proper names we hold is irrelevant to the

puzzle. All we need to do to generate the puzzle is to assume the innocuous weak disquotation

principle (which I shall discuss below) and the puzzle is generated independent of which theory

of proper names we adhere to.

There are two versions of Kripke’s puzzle, but here I discuss only the Paderewski version

of the puzzle, the more challenging version. Kripke asks us to consider the case of Ignacy Jan

Paderewski (1860 – 1941), a noted Polish pianist who also a politician—the prime minister of

Poland for most of 1919. Suppose that a man named Peter is familiar with Paderewski the pianist

and with Paderewski the politician, but Peter fails to realize that these are the same person. Peter

thinks the pianist and the politician are two separate people. He assents to the sentence

‘Paderewski had musical talent’ when he takes the speaker to be referring to the pianist, and

assents to the sentence ‘Paderewski did not have musical talent’ when he takes the speaker to be

referring to the politician (believing that politicians rarely have musical talent). Now consider

the Weak Disquotation principle.

WEAK DISQUOTATION
If a competent, sincere, reflective, and rational speaker s who understands a sentence S is
disposed to accept S, and believes S to be true, then s believes the proposition
semantically expressed by S.

Based on Peter’s assent to ‘Paderewski had musical talent,’ a sentence Peter understands, we

may infer that he believes that Paderewski had musical talent. Based on his assent to

‘Paderewski did not have musical talent,’ a sentence Peter understands, we may infer that he

97
believes that Paderewski did not have musical talent. Hence, we may infer that Peter believes

both that Paderewski had musical talent and that he did not have musical talent. Thus, it looks

like Peter believes inconsistent propositions. And Kripke has shown this without making any

assumptions about whether Millianism is true or whether substitutivity is viable. He has merely

assumed the unobjectionable weak disquotation principle. Therefore, a puzzle about belief

ascription can be generated whether we presuppose the truth of Millianism or its falsity. The

puzzle arises merely if we presuppose an unobjectionable principle such as Weak Disquotation,

which seems commonsensically true.

Kripke concludes that Peter’s case “lies in an area where our normal apparatus for the

ascription of belief is placed under the greatest strain and may even break down.” (Kripke 1979:

452). According to Kripke, the problem is that we have no plausible answer to the question:

does Peter believe or does Peter disbelieve that Paderewski had musical talent? We have four

possible answers to this question, none of which Kripke thinks are acceptable:

(a) Peter believes neither that Paderewski had musical talent, nor that he did not have
it.

(b) Peter believes that Paderewski had musical talent.

(c) Peter believes that Paderewski did not have musical talent.

(d) Peter believes that Paderewski had musical talent and that he did not have musical
talent.

But what exactly is wrong with option (d)? Why not accept that Peter both believes that

Paderewski had musical talent and believes that he did not? Why not accept that rational agents

can believe inconsistent propositions? Kripke thinks that a reflective and rational agent should

be able to examine the content of belief and, given enough time to reflect, see that there is an

98
inconsistency and correct him- or herself. Kripke writes: “…surely anyone … is in principle in a

position to notice and correct contradictory beliefs if he has them.” However, the notion that a

rational agent should be able to examine the contents of his beliefs, as if they were perfectly

transparent to him/her, is implausible. In section 2.8 above where I discussed the Problem of

Inconsistent Rational Belief, I argued (in agreement with many Millians such as Nathan Salmon

and other Millians who speak or ways of taking propositions or propositional guises) that agents

may not always have transparent cognitive access to the contents the propositions they believe

because some of those contents are determined externalistically. The TIUT takes the

fundamental motivation behind this Millian approach, known as ‘guise Millianism,’ to be

essentially correct, except that the TIUT explains the non-transparent access to propositions in

terms of indexicality. Proper names are used as indexicals, whose characters but not contents in

the context are transparent to the agent. Peter both believes and disbelieves that Paderewski had

musical talent, but he is not irrational for doing so because he does not realize that his beliefs are

inconsistent and cannot realize this no matter how long he reflects on the matter. 67 The

67
There is a distinction to be drawn between the following ascription sentence forms:
Peter believes P and disbelieves P (= Peter believes P and believes NOT P)
Peter believes P and does not believe P (= Peter believes P and it is not the case that
Peter believes P)
I think that the former ascription could be true, since the TIUT posits (as do Millians) that agents
may rationally believe singular propositions and their negations, i.e., rationally believe and
disbelieve singular propositions. (See the discussion on the Problem of Rational inconsistent
belief, supra at section 2.8). However, the second ascription is necessarily false; this follows
from the law of noncontradiction. If Peter believes P, it is false that he fails to believe P. If it is
the case that Peter believes P, then it is not the case that he does not believe P. There is a
difference in meaning between ‘X disbelieves P’ (which is the same as ‘X believes NOT P’) and
‘X does not believe P’ (which is the same as ‘It is false that X believes P’). Ascribers would be
guilty of contradicting themselves (as a strictly technical matter) if they uttered: ‘Peter believes
that Paderewski was musically gifted and does not believe that Paderewski was musically
gifted,’ but I shall show that ascribers would not contradict themselves if they uttered: ‘Peter
99
information that the propositions are inconsistent is not inside the speaker’s head, nor is it

inferable from information inside the speaker’s head. The reference of both occurrences of

‘Paderewski’ is determined externalistically, for the subjecthood of the dossiers from which the

name was drawn is a causal-historical property.

Mark Richard has argued that Kripke’s puzzle presents a puzzle not so much about the

nature of the propositions that Peter believes but is fundamentally a puzzle about belief

attribution. How are we to express what Peter believes “in the idiom for belief ascription

provided by English, if we limit ourselves to identifying the object of his beliefs [with the name

‘Paderewski’]?” 68 Consider sentence (9), which is unidiomatic in English:

(9) Peter believes that Paderewski had musical talent and disbelieves that he has
musical talent.

Although (9) expresses a true proposition, most speakers would use a more idiomatic sentence to

communicate information about Peter’s confused and inconsistent belief with respect to

Paderewski. Why is sentence (9) unidiomatic? Why would most speakers avoid using it? What

other, more idiomatic or informative ways would ordinary speakers find to ascribe belief to

Peter? I shall now turn to this issue.

First, I shall briefly revisit the issue of substitution in Hesperus/Phosphorus and

Kent/Superman sorts of cases where two names used in the propositional attitude ascription

pairs. Then I will come back to the Paderewski case, which involves just one name.

believes that Paderewski had musical talent and disbelieves that he had musical talent.’
68
Richard, Mark. “Kripke’s Puzzle about Belief,” page 18.
https://markrichardphilosophy.files.wordpress.com/2011/07/kripkes-puzzle-about-belief.pdf
[accessed on May 31, 2015].

100
 As discussed in section 2.6, according to the TIUT, in the Hesperus/Phosphorus and

Kent/Superman cases speakers can utter Millian (coarse-grained) or Conception-indicating

(finer-grained) ascriptions, depending on their expressive purposes. Sentence (10) below would

be a Millian ascription in the Hesperus/Phosphorus case relating to an ancient Greek astronomer

living in an era before it had become known in Greece that Hesperus was identical to

Phosphorus:

(10) The ancient Greek astronomer believed that Venus was visible in the morning but
also disbelieved that Venus was visible in the morning. 69

(10) has the same logical form as (9). The ascription attributes inconsistent beliefs to the ancient

Greek astronomer using one single name, ‘Venus,’ just as (9) attributes inconsistent beliefs to

Peter using one single name, ‘Paderewski.’ The ancient Greek Astronomer believes inconsistent

singular propositions of which Venus is a constituent. The fact that the same name, ‘Venus,’

appears twice over suggests that we are attributing inconsistent beliefs to the ancient Greek

astronomer (pursuant to the Meaning Consistency Principle). Though expressing a true

proposition, (10) is nevertheless unidiomatic. The syntactically de re attitude ascription (11),

below, would be a more usual and idiomatic way to express the same proposition.

(11) The ancient Greek astronomer believed of Venus both that it was visible in the
morning and that it was not visible in the morning.

Consider now the Conception-indicating ascription (12):

69
‘Hesperus’ was the ancient Greek name for the planet Venus when observed in the evening (at
a certain location in the evening sky during at certain portions of the year), and ‘Phosphorus’ the
ancient Greek name for the planet Venus when observed in the morning (at a different location in
the morning sky during certain portions of the year). The Greeks long believed that Hesperus and
Phosphorus were two distinct celestial bodies. At some point, the Greeks learned from the
Babylonians that Hesperus and Phosphorus were the same body.
101
 (12) The ancient Greek astronomer believed that Phosphorus was visible in the
morning but disbelieved that Hesperus was visible in the morning.

In (12), the use of two different co-designative proper names (‘Hesperus’ and ‘Phosphorus’)

alerts the audience, pursuant to the Meaning Consistency Principle (see sections 2.6.2 supra),

that the speaker is drawing a distinction between different conceptions.

Meaning Consistency: Multiple occurrences of the same syntactic string in a sentence

generally entail that the speaker meant the same thing by each occurrence. 70 Occurrences
of different syntactic strings entail a difference in meaning.

The Meaning Consistency principle explains for the differences between (10) and (12). It

explains why we would not utter (10) (or (11)) as a Conception-indicating ascription and would

not utter (12) as a Millian ascription. In (10), the use of the same name twice over, ‘Venus,’

signals to the audience that the speaker means the same thing on each of its occurrences, so the

audience will tend to suppose that the speaker is attributing inconsistent beliefs to the ascribee.

In (12), the use of different names usually conveys that the speaker means something different by

the different names.

There is a pragmatic principle of conversation that I think explains why (12) is a more

idiomatic than (10), the “Attribution of Consistency” principle:

Attribution of Consistency: We tend to attribute beliefs to ascribees in a way that avoids

making them sound irrational.

70
By generally, I mean that there are exceptions when the conversational context makes it clear
that the speaker means different things by the same expression (see Bill/Liz example, section 2.9
supra), or where the grammar of the sentence entails a difference in meaning (e.g., in the
sentence “Rose rose in popularity beginning in high school” we infer from the grammar that the
two occurrences of “rose” must have different meanings).
102
Sentence (10) is somewhat ambiguous because a sentence of this form could be used to state that

a person is irrational and believes a proposition and its negation knowing full well that they are

inconsistent. We would probably not understand it that way, since, on the principle of charity,

we tend to presume that people are not irrational in this sort of way. (11), by contrast with (10),

is not similarly ambiguous—syntactically de re attitude ascriptions with the “of”-locution are

specially tailored so that speakers avoid the seeming attribution of irrationality to the ascribee.

There is a further pragmatic principle of conversation that I think explains why (10)

would be an unusual, somewhat unidiomatic sentence:

Preference for Fine-grained Ascriptions: When ascribing belief, one should give as
much information about conceptions as is possible when two conceptions of the same
object are being contrasted, if one can easily do so and if one is aware of the relevant
conceptions in the conversational context.

Sentence (12) is a more natural and idiomatic attitude ascription than (10) because (12) gives

much more information to the audience than (10) would, consistent with Grice’s maxim of

quantity. It tells the audience about the different ways that the ancient astronomer conceived

Venus. It imparts this additional information in an efficient way, without the use of any

additional words (assuming that the audience knows that ‘Hesperus’ and ‘Phosphorus’ are

different names for Venus and is familiar with the conceptions the names are associated with). If

an ascriber were to utter the coarse-grained Millian ascription (10), the audience would need to

ask the ascriber follow-up questions to elicit additional information regarding contrasting

conceptions to understand why the ancient astronomer entertained inconsistent beliefs (unless

this information about contrasting conceptions was already clear from the conversational

context). A speaker who uttered (10) without following-up by providing details about the

reasons for the ancient Greek astronomer’s inconsistent beliefs vis-à-vis Venus would be flouting

103
cooperative principles of conversation.

There is an important and salient difference between the Hesperus/Phosphorus and the

Kent/Superman cases on the one hand, and the Paderewski case on the other: in the former cases,

two syntactically different co-referential names (Clark Kent/Superman; Hesperus/Phosphorus),

each associated with a different conception by members of the language community, are readily

available to speakers to draw contrasts between different conceptions because of the fact that the

conceptions are commonly associated with the names. In the Paderewski case, by contrast, there

is just one name to which speakers have recourse. Speakers cannot resort to pre-established

names, each typically associated with a different conception, to draw any contrasts when

ascribing beliefs to Peter. The principle of Meaning Consistency admonishes us, ceteris paribus,

to have two expressions to indicate differing conceptions to make a Conception-indicating

ascription with respect to Peter’s beliefs. Arbitrarily inventing a second name for Paderewski

from whole cloth would confuse the audience (especially since the audience, in this case, would

not be familiar in advance with the different conceptions associated with the names by the

speaker). So generally, a speaker would coin context context-derived contrasting names for

Paderewski by combining ‘Paderewski’ with descriptions that key in the audience to the

contrasting conceptions. For example, a speaker could utter (13) as a Conception-indicating

ascription.

(13) Peter believes that Paderewski the pianist had musical talent and he disbelieves
that Paderewski the politician has musical talent

By adding descriptions to the name ‘Paderewski,’ the partially descriptive names ‘Paderewski

the pianist’ and ‘Paderewski the politician’ can be used in (13) to make a Conception-indicating

ascription just as the names ‘Hesperus’ and ‘Phosphorus’ were used in (12) in the

104
Hesperus/Phosphorus case to make a Conception-indicating ascription. In both (12) and in (13)

different conceptions in different dossiers are being referred to—the use of different co-

referential names alerts the audience to this. The salient difference between the cases is that in

(12) the speaker can resort to two different co-referential names conventionally associated with

specific conceptions, whereas with (13) the speaker must invent his own names to get his

audience to understand that the names are co-referential and clue his audience into the

contrasting conceptions each name is associated with.

We would not typically utter the somewhat unidiomatic (9)

(9) Peter believes that Paderewski had musical talent and disbelieves that he had
musical talent.

because the conversational principles of Attribution of Consistency and of Preference for

Fine-grained Ascriptions lead us to prefer ascriptions of the form of (13), (14), or (14b) instead.

(13) Peter believes that Paderewski the pianist had musical talent and he disbelieves
that Paderewski the politician has musical talent

(14) Peter believes of Paderewski both that had musical talent and that he lacked
musical talent.

(14b) Paderewski is such that Peter believes he had musical talent and that he lacked
musical talent.

The syntactically de re ‘of’ locution in (14) and (14b) makes it abundantly clear the

speaker means that Peter’s believing the inconsistent propositions may be unintentional and

unknowing. In most cases, and if it can be imparted without great circumlocution, we prefer

more informative and finer-grained Conception-indicating ascriptions, such as (13), which

provide information about conceptions that indicate to the audience to how Peter believes what

105
he believes, if this can be easily done in the conversational context. To accomplish the latter, we

coin partially descriptive names when do not have recourse suitable distinct co-referential names

available to draw the contrast.

To sum up, the TIUT answers Kripke’s puzzle as follows. Peter both believes the

singular proposition that Paderewski had musical talent and that he did not. His beliefs are

inconsistent. He is not irrational or illogical for entertaining these inconsistent beliefs, and we do

not suggest that he is irrational when we ascribe to him a belief in inconsistent singular

propositions. Nevertheless, we generally have an aversion towards locutions that have the

outward appearance of attributing inconsistent beliefs, and we prefer ascription sentences that

give the audience information about conceptions because this provides a higher level of clarity

about the reasons the inconsistent beliefs are held. Hence, ascription sentences such as (9) sound

unidiomatic and we tend not to utter sentences of this form.

2.13 The Distinction between Millian and Conception-Indicating Names: Pragmatic or

Semantic?

I am proposing that the distinction between names used in a Millian and Conception-

indicating way is a semantic distinction and not a pragmatic one. Some philosophers may object

that I am going too far in positing a semantic distinction—perhaps agreeing with my claim that

proper names, as well as many classes of expressions, can be used in various ways—but

objecting that not all differences in use mark a semantic distinction. However, I do really do

want to posit a semantic distinction. For there is, as far as I can see, no pragmatic mechanism—

neither implicature (Salmon), descriptive enrichment (Soames), nor any other pragmatic

mechanism—that could adequately explain the fundamentally different ways in which proper

names are regularly and consistently used. Unfortunately, there is a dogma in the current

106
philosophy of language according to which the positing of a semantic distinction is strongly

discouraged. Wherever possible, one is to explain away the appearance of a potential semantic

distinction by appealing to pragmatics. However, I think this dogma is too extreme, and some

philosophers have recently attacked this dogma with some measure of success. For example,

Devitt (2004, 2008) and Reimer (1998) argue persuasively that the distinction between

attributive and referential uses of definite descriptions is a semantic distinction (and not merely a

pragmatic difference in use). They employ an argument that has come to be called “the argument

from convention, 71” according to which the fact that a certain expression is used regularly (i.e.,

with high frequency) and without “special stage setting” (Devitt 2004, p. 283) to convey some

content C, constitutes solid evidence that the expression conventionally (i.e., semantically)

means C. As Devitt put it,

“The basis for RD [Devitt’s thesis that the referential uses of definite descriptions
constitute referential meanings] is not simply that we can use a definite [description]
referentially, it is that we regularly do so. When a person has a singular thought, a
thought with a particular F object in mind, there is a regularity of her using ‘the F’ to
express that thought. And there need be no special stage setting enabling her to
conversationally imply what she has not literally said, nor any sign that her audience
needs to use a Gricean derivation to understand what she means. This regularity is
strong evidence that there is a convention of using ‘the F’ to express a thought about a
particular F, that this is a standard use. This convention is semantic, as semantic as the
one for an attributive use. In each case, there is a convention of using ‘the F’ to express a
thought with a certain sort of meaning/content.” (Devitt 2004: 283)

Devitt argues that the fact that definite descriptions are regularly used without special

stage setting—sometimes attributively and sometimes referentially—is solid evidence that the

referential/attributive distinction is semantic, and not pragmatic, in nature.

With respect to proper names and the distinction between Millian and Conception-

indicating uses that I propose here, I urge that the argument from convention militates in favor of
71
This name for Devitt’s argument for Referential Descriptions was coined by Neale in his 2004.
107
a semantic distinction. The existence of two regular uses of proper names without special stage

setting—sometimes just to contribute a name’s referent/bearer to the proposition expressed, and

sometimes to contribute both the referent/bearer and a conception of it—is solid evidence that

we are faced with a genuine semantic distinction. Proper names are semantically ambiguous

between Millian and Conception-indicating readings, and this ambiguity can be resolved only by

looking to the speaker’s expressive intent. 72

This issue is further discussed in Chapter 5. There, I claim that Millians have failed to

show there are alternate conventions to report propositional attitudes (e.g., alternatives to

allegedly false propositional attitude reports such as ‘Lois Lane does not realize that Clark Kent

is Superman’) or alternatives to informative identity sentences such as ‘Clark Kent is Superman’

to express informative identities. We have no viable alternative to using these sorts of

sentences—they are the means par excellence to convey these sorts of propositions, and

therefore, they must be the conventional, idiomatic way to say these sorts of things in our

language. They are essential constructions and we have no recourse to any other efficient

idiomatic way to express such things. No Millian has shown otherwise. This militates, via the

argument from convention, for these sentences semantically expressing, rather than merely

pragmatically conveying, the propositions that ordinary speakers intuit they do.

2.14 Conclusion

The very existence of Frege’s puzzle constitutes evidence that proper names do not

always contribute merely their referents to the propositions expressed by sentences in which they

72
I am not suggesting, of course, that we must be mind-readers to figure out a speaker’s
expressive intent to resolve ambiguities. Utterance interpretation generally proceeds via
contextual clues, which are a highly reliable guide to the expressive intent of the speaker.
108
occur. They sometimes also contribute to the proposition a way of descriptively conceiving or

thinking about the referent of the name. I take this evidence at face value. Common sense tells us

that speakers do not utter ‘Clark Kent is Superman’ merely to say that a person is identical to

himself. We should respect this piece of common sense. However, there are also some sentences

in which proper names play a simpler role: they just refer and do not contributing a conception.

This is not to say that the speaker lacks a conception of the object or individual he is referring to,

but rather than the speaker does not regard the conception as an element of the message he is

conveying. The use of names just to refer to their bearers occurs both in simple sentence and in

propositional attitude ascriptions. Thus, proper names have these two uses: merely to refer, and

to refer and communicate a conception. I call the former ‘Millian’ uses and the latter

‘Conception-indicating’ uses.

From Kripke we have received at least two important truths about proper names. The

first is that they are rigid designators. The second is that there is no reference-determining

descriptive or conceptual element built into their meanings. They are synonymous with neither

ordinary definite descriptions nor rigidified definite descriptions. Since the conceptions

communicated are not part of the fixed conventional meanings of proper names, we must explain

how speakers communicate conceptions. There are two possible routes. The first would be to

propose that conceptions are communicated via pragmatic, rather than semantic, mechanisms,

and that the meaning of a name is exhausted by its referent. This is a route pursued by many

modern Millians. There are many problems associated with taking that route. The second route,

the one taken by the TIUT, is to propose that the mechanism by which conceptions are

communicated is semantic: proper names are sometimes used as indexicals that pick out the

conception that the speaker associates with the name at the time he utters it. The TIUT shows

109
how this occurs consistent with the thesis that proper names are rigid designators. This is where

the TIUT is decisively superior to Descriptivism, on which names are not rigid. It is also

decisively superior to Millianism because of its respect for the intuitive cognitive and truth-

values of sentence like (1)-(6). Millians do violence to the intuitive cognitive and truth-values of

these sentences. The TIUT does what neither Millianism nor Descriptivism is able to do: explain

how names such as ‘Clark Kent’ and ‘Superman’ can rigidly refer to the same individual and at

the same time differ in semantic content.

Once we recognize that speakers can use names just to refer, we must deal with the

Problem of Rational Inconsistent Belief. Rational speakers may entertain inconsistent beliefs and

utter inconsistent sentences to express them. Lois Lane inconsistently believes that Kent-Super

can and cannot fly and is disposed to utter the inconsistent ‘Clark Kent cannot fly’ and

‘Superman can fly.’ Peter both believes and disbelieves that Paderewski had musical

talent. Unless we are willing to say that these speakers are irrational, which we obviously should

not, we must explain their inconsistency as arising out of ignorance of some fact or facts. But

what are these speakers ignorant of? The TIUT and guise Millianism take a somewhat similar

approach here and posit that speakers do not have full cognitive access to the propositions they

believe and express through their utterances. Content is partially externalist. Guise Millians

maintain that speakers grasp the propositions the express/believe indirectly, only via a

propositional guise that may hide part of the content of those propositions. Problematically, all

accounts of propositional guises are not well fleshed out. The TIUT’s approach is like that of the

guise Millians in purpose—to explain how a rational Lois Lane can believe inconsistent

propositions and utter sentences expressing inconsistent propositions without being in a position

to recognize their inconsistency—but the TIUT offers a specific proposal: names are indexicals

110
with a character meaning as well as content. Character is the guise behind which content

hides. Speakers are inconsistent because they understand the character of the propositions they

express and believe, but not the content (or at least, not the full content). They have direct

access only to the descriptive conceptions they associate with the name, which conceptions are in

their dossiers. The causal-historical properties of the dossier in virtue of which the dossier is

about its subject are not apparent to the speaker.

In the introduction, I claimed that to solve Frege’s puzzle, we needed a theory of proper

names that, at a minimum, respects the following seven constraints:

1. Proper names are rigid designators.

2. Proper names do not have reference-determining descriptive meanings.

3. Identity sentences like (1)-(2) express different propositions with the same modal
profile.

4. Propositional attitude ascriptions like (3)-(4) and (5)-(6) express different

propositions differing in truth-value.

5. Sometimes, we use a proper name as Millians claim, merely to refer, in which

case the name semantically contributes its bearer only to the proposition
expressed.

6. Sometimes, we use a proper name both to refer and to convey a descriptive

conception, in which case the name semantically contributes its bearer plus a
descriptive conception to the proposition expressed.

7. The descriptive conceptions we use names to communicate are not conventionally

built into names and may be idiosyncratic and variable, differing from speaker to
speaker and from one conversational context to another. A theory of proper
names must provide that names are contextually sensitive in such a way that they
load the relevant descriptive conceptions into content.

The TIUT respects these constraints. It also offers an answer to the problem of rational

inconsistent belief and Kripke’s puzzle. Hence, the TIUT constitutes a very attractive non-

Millian theory of proper names.

111
 A virtue of the TIUT it meshes well with how one might characterize Lois Lane’s

ignorance of Clark Kent’s identity with Superman in a common sense, pre-theoretical way. Lois

Lane fails to realize that there are two sets of information in her mind about the same person,

Kent-Super. Let us call each of those sets of information a “dossier.” Let us say that each dossier

“is about” Kent-Super in virtue of some property and call this property “subjecthood.” Lois’ two

dossiers have the same subjecthood, so they are about the same person. Lois’ dossiers contain

different representations: one represents its subject as a mild-mannered reporter, the other as a

superhero. Let us call these representations “conceptions.” Lois can introspect and see these

conceptions clearly. She can see that the conceptions in the dossiers represent their subjects in

very different ways, and she believes, quite rationally, that they are about different individuals.

Although both dossiers have the same subjecthood, Lois cannot introspect, no matter how hard

she reflects upon the matter, and determine that the dossiers have the same subjecthood. This is

because the subjecthood is at least in part a relational property, inhering in facts external to Lois’

mind. In much the same way, you could not look at a photo of Obama in which he more closely

resembled Malcolm X and tell, just by the way the photo looks, that this is a photo of Obama and

not Malcolm X. What makes the photo a photo of Obama is relational and causal, and therefore

something not evident within the four corners of the photo. One cannot see the history of the

photo in the image. Likewise, the property of dossier subjecthood is such that Lois does not

have cognitive access to it via armchair reflection. Lois cannot find out that the dossiers are

about the same person just by thinking about it. She would have to look at these extra-mental

facts to find out that these dossiers are about the same individual.

When those in the know about the identity want to describe Lois’ confusion about Clark

Kent’s identity with Superman, they utter ‘Lois does not believe that Clark Kent is Superman.’

112
On the TIUT, this ascription sentence means that Lois has one dossier with which she associates

the name ‘Clark Kent’ and another with which she associates the name ‘Superman,’ and she does

not realize these dossiers are about the same individual. When we want to say, in general, what

anyone who suffers from the sort of confusion that Lois does, i.e., they fail to realize that Clark

Kent is Superman, we want to say that, in general, failing to realize that Clark Kent is Superman

means not realizing that one type of dossier has the same subject as another type of dossier.

Specifically, not knowing that Clark Kent is Superman means not knowing that that the type of

dossier that is about Kent-Super and represents him in one way—a Clark Kent-y way—is about

the same person as a different type of dossier that represents him in a different way—a

Superman-y way. To specify these types of dossiers, ascribers entertain dossiers tokens of these

types and then make them salient by drawing the names from these dossiers. They thereby

contribute the dossier types that their dossier tokens instantiate to content per the Conception-

indicating character.

Propositional attitude ascriptions involve the ascriber imagining the way that the ascribee

sees the world and creating within his own mind a temporary model of the ascribee’s worldview.

The ascriber forms a dossier structure that mimics and mirrors that of the ascribee. Then the

ascriber can speak about his own dossier tokens as if they were those of the ascribee. He may do

so by relying on the fact that, if he has understood the ascribee’s cognitive state well enough, his

own dossier tokens are of the same type as those of the ascribee. His own tokens and Lois’ (or

anyone’s who confused about the identity of Clark Kent and Superman) all belong to the same

type, and therefore, the ascriber can point to his tokens as instating that type and say that all

those who do not realize the identity fail to realize that these dossier types (the one’s his own

tokens instantiate) are individuated by the same subject, Kent-Super, the subject of his own

113
dossier tokens.

CHAPTER THREE

PELCZAR AND RAINSBURY’S ‘SINGLE INDEXICAL’

THEORY OF PROPER NAMES

Pelczar and Rainsbury (1998) argue that proper names are used as indexicals, just as I do.

Their theory is distinct from the TIUT for several reasons. For one, theirs is a Millian theory,

although they diverge from standard Millianism, which claims that the name’s bearer fully

exhausts its meaning. They claim instead that names, being used as indexicals, have character

meanings that determine their bearers in the context. A salient difference between the TIUT and

their theory is that they do not distinguish between two indexical uses, as the TIUT does. Thus, I

refer to their theory as a “single-indexical use theory.” Because Pelczar and Rainsbury do not

distinguish between two Millian and Conception-indicating uses, I maintain that they cannot

solve Frege’s puzzles. However, their theory is well-suited to solving the Problem of Rational

Inconsistent Belief.

In section 3.1, I discuss the nature of indexicals in general, largely based on Kaplan

(1989). In section 3.2, I discuss the debate over whether the fact that names have multiple

bearers is best explained by their being multiply-ambiguous or being indexicals (in agreement

with Pelczar and Rainsbury, I argue for the indexical view). In section 3.3, I look at Pelczar and

Rainsbury’s single-indexical theory. I argue that their theory solves the Problem of Rational

Inconsistent Belief. However, it cannot solve Frege’s puzzle. Two sorts of indexical uses, as the

TIUT proposes, are essential for that.

114
3.1 What are Indexicals?

An indexical is a linguistic expression with the following characteristics:

(a) It is context-sensitive, i.e., its reference shifts from one context to another. For example,

when I utter ‘I am tired’, ‘I’ refers to me. When my friend John utters ‘I am tired,’ ‘I’ refers to

him. The reference of ‘I’ shifts depending on who the speaker is in the context of utterance.

Similarly, the referent of ‘today’ shifts. If ‘today’ is uttered on April 12, 2011 it refers to April

12, 2011. If uttered on April 13, 2011, it refers to April 13, 2011.

(b) The linguistic meaning of an indexical—which Kaplan calls its “character”—is a function

from contexts to contents that delivers the expression's content at each context. For example,

Peter’s utterance of ‘I’ delivers him, Peter, as its content because he is the speaker in the context.

Jose’s utterance of ‘I’ delivers him, Jose, because he is the speaker in the context.

(c) Kaplan (1989) identifies four features by which contexts are individuated and to which

contents are sensitive: utterer, time, spatial position, and possible world. 73

(d) Kaplan’s list of the features by which contexts are individuated is not exhaustive. As

we’ll see in section 3.3 infra, Pelczar and Rainsbury, who argue that proper names are

indexicals, would add “dubbings-in-force” to the list of features by which contexts are

individuated. I also argue that proper names are used as indexicals and I add “subjects,”

“dossiers,” and “dossier types” to the list.

73
For the sake of simplicity, I speak in terms of utterances. Kaplan speaks in terms of sentences
in contexts.
115
(e) The character, or linguistic meaning, of an indexical is invariable—it is the same

character no matter what context it is uttered in. The content of an indexical varies from context

to context.

(f) On Kaplan’s model, the content of an indexical expression (which, for Kaplan, is always

an object) depends on context. Once the referent/content is determined in the context, it is

invariant across circumstances of evaluation, i.e., possible worlds. For example, once the content

of ‘I’ on a tokening is determined, the truth value of the proposition expressed will depend on

that content evaluated at every possible world. Similarly, as we saw in Chapter 2, on the TIUT,

names used in a Conception-indicating way have a content (a meaning, not an object), which,

once determined is the same across all circumstances of evaluation. The salient difference

between Kaplan’s conception and that of the TUIT is, again, that on Kaplan’s theory content is

always an object, whereas on the TIUT content (of a name used in a conception-indicating way)

is a meaning. That meaning (the meaning of a definite description) mediates reference, since the

referent of the name is the denotation of the definite description expressing content.

(g) The reference of an indexical is independent of the speaker’s beliefs about it or his beliefs

about the context in which he finds himself when he utters the indexical, (i.e., the reference of an

indexical is automatic). This is to be contrasted with the reference of proper names on

Descriptivism, where reference depends on which descriptions or descriptions the speaker

associates with the name and what object those descriptions denote. To illustrate the difference,

suppose a man is knocked unconscious, kidnapped, and finds himself in a dark room suffering

from total amnesia. He does not know who he is (he can’t recall his name or other details about

his personal life), where he is (he is locked in a dark room), or what day or time it is (he does not

116
recall what day or year it is and does not know how long he has been unconscious). He may utter

to himself ‘I am here today’ and successfully refer to himself, the place he is in, and the day of

the utterance despite lacking information about these things. He associates no uniquely

identifying definite descriptions with himself, the day it is, or the place he is at. His mere

semantic understanding of the character of the expressions ‘I’, ‘here’, and ‘today,’ are sufficient

to enable him to express a true proposition (not a very useful or informative proposition, but

nonetheless, a true one). (By way of contrast, on Descriptivism, a speaker who uttered

‘Aristotle,’ who associated with the name a definite description failing to uniquely denote

anything or anyone would have failed to refer to anything whatsoever in uttering ‘Aristotle’).

(h) Indexicals contribute only their contents (not their characters) to the proposition

expressed. For example, the character of ‘I’ is the speaker in the context; the character of ‘here’

is: the place of the utterance; the character of ‘now’ is: the time of the utterance. The

proposition expressed by an utterance of ‘I am here now’ by speaker s at time t and location l is a

singular proposition containing s, t, and l as constituents. The proposition does not contain the

characters as constituents—i.e., the proposition expressed is not the descriptive proposition: the

speaker in the context is at the place of the utterance at the time of the utterance. That is the

character of the sentence ‘Í am here now,’ not its content.

(i) Characters are different from the descriptive meanings that Descriptivists ascribe to

proper names—definite descriptions that denote the referent according to a property uniquely

borne by it considered apart from the contextual relation of the utterance of the name. Characters

are also definite descriptions that denote the referent, but not by picking out a property uniquely

borne by the referent in general, but by designating the referent by a relational property borne by

the referent towards a particular utterance in a particular context. Whereas the name ‘Albert

117
Einstein’ has a descriptive meaning (on Descriptivism) that denotes Albert Einstein whenever the

name is uttered, the character of an indexical such as ‘now’ denotes different moments in time

when used at different times. As John Perry put it in his 1997:

“…Indexicals denote, as descriptions do. But indexicals do not describe. They refer. The
individuals that meet the conditions, rather than the conditions themselves, are
contributed to the official content (6).”

Both characters and definite descriptions, such as, e.g., the inventor of the bifocals, denote. But a

description such as the inventor of bifocals picks out its denotation by describing it in a general

way independent of the its relation to the utterance. By contrast, an indexical does not describe

the denotation of its character in the same way. ‘I’ denotes a particular individual in the context

it is uttered, but it does not attribute to that individual any permanent or intrinsic property that

would allow him or her to be picked out apart from its relation to the context of the utterance.

(j) On the view of indexicals outlined here, they are what John Perry calls “referential”—

they just contribute their contents to the proposition, not their characters. Although Kaplan says

that indexicals are “directly referential,” as Perry (1997, 1, footnote 2 74) points out, they are not

directly referential in the strictest sense in that their reference is mediated by a sort of meaning—

character. However, the view of most Millians (N.B.: but not “single-indexical” Millians such

as Pelczar and Rainsbury; see section 3.3) is that proper names, by contrast with indexicals, are

genuinely directly referential in the strict sense, for unlike indexicals, is it assumed that there is

74
Perry writes: “David Kaplan’s term is directly referential. Kaplan has a precise concept of
“directness” in mind, but unless one is focusing on his exact words, the term “directly” is likely
to suggest that there is no semantic mechanism intervening between the expression and its
referent. This is pretty clearly not the case with indexicals, as Kaplan’s own analysis shows; it
may be more plausible for proper names. Using terminology introduced in Section 3 we can say:
Kaplan’s language suggests that directly referential terms name, but what he really says is simply
that they refer.”

118
no sort of meaning at all, neither an associated definite description nor a character meaning

associated with proper names that determines reference. Most Millians hold that the sole

semantic function of proper names is to pick out their referent and they do not carry out this

function via an intermediating meaning of any sort.

(k) Indexicals include all expressions whose content is contextually sensitive, but many

philosophers follow Kaplan in distinguishing between “pure” or “automatic” indexicals (such as

‘I’, ‘here’, and ‘now’) and true demonstratives, such as ‘he’, ‘she’, and ‘that’. In case of pure

indexicals, reference is supposed to be independent of the speaker’s demonstrations or intentions

accompanying his or her utterance (the content of the expression is loaded into the proposition

automatically), so that the reference of ‘here’ is always the place of the utterance and the

reference of ‘now’ the time of the utterance, no matter what place of time the speaker intended to

refer to. By contrast, in the case of a true demonstrative, content depends on the speaker’s

accompanying demonstrations (such as pointing) or intentions (e.g., intending to refer to a

particular object or particular type of thing in his visual field). It is controversial how to draw the

distinction (e.g., there is a debate about how ‘you’ should be classified) and it is controversial

whether the distinction is cogent on a more than superficial level. Mount (2008) argues that so-

called pure indexicals are not in fact distinguishable from demonstratives, as their content is also

a function of speaker intentions.

3.2 Proper names: Multiply ambiguous or indexicals?

Proper names have multiple bearers. There are many people with the name ‘John.’

There are two approaches to explaining this phenomenon: the multiple-ambiguity view and the

indexical view. According to the multiple-ambiguity view, proper names are multiply

119
ambiguous. Just as the word ‘bank’ is ambiguous in English (it can mean the either the edge of a

river or a financial institution), ‘John’ is ambiguous in that it can be used to refer to many

different people. According to the standard view of ambiguity, ‘bank’ is in fact two different

expressions that happen to be syntactically and phonologically identical in English. If we

wanted to, we could adopt a syntactic convention to distinguish between these two expressions—

using ‘bank1’ for the edge of a river and ‘bank2’ for a financial institution. Similarly, we could

use subscripts to distinguish between all the different proper names that happen to have the

syntactic and phonological form ‘John’ that are in fact different names for distinct individuals

who bear the name ‘John.’ ‘John1’ would be the name for one individual named ‘John,’ ‘John2’

for a different individual named ‘John,’ and so on. Each person in the world would have a unique

name shared with no other. His name would belong to him alone. “Two people have the same

name” would mean that their two distinct names share their phonological/syntactical

properties—the names are qualitatively identical in this way, not numerically identical. Different

Johns would share what Kaplan (1990) calls a “generic name”: each John would have his own

proprietary name ‘John,’ which happens to have the same syntactic and phonological properties

as the names of other individuals bearing the name ‘John.’ Analogously, ‘bank’ (to refer to a

financial institution) and ‘bank’ (to refer to the edge of a river) are phonologically/syntactically

identical expressions—they are the same generic expression, but they are not numerically

identical expressions.

On the indexical view of proper names (defended by the TIUT and by Pelczar and

Rainsbury—see section 3.3), by contrast, proper names are not ambiguous. There is just one

name ‘John.’ It can be used to refer to any of the many individuals who bear that name. Names

are used as indexicals, whose content is sensitive to context. Just as we would not say that ‘I’ is

120
ambiguous just because it can be used to refer to many different individuals on different

occasions of utterance, analogously, we should not say that ‘John’ is ambiguous merely because

it can be used to refer to different individuals on different occasions of utterance.

Any indexical theory of proper names must posit features of context beyond the four that

are on Kaplan’s list: utterer, time, location, and world. Those four features of context cannot

account for the difference in content of proper names. As discussed below in section 3.3, Pelczar

and Rainsbury would add what they call “dubbings-in-force” to the list of features of context. I

add “subject,” “dossier(s),” and “dossier types” to the list (see section 2.3).

Perry’s (1997) primary objection to the indexical view of proper names is that proper

names cannot vary in content if we limit context to Kaplan’s four contextual features that

determine content in the case of indexicals such as ‘here,’ ‘I,’, ‘now,’ etc. Perry (1997) writes:

“The role of context in resolving the issues of which naming conventions are being
exploited is quite different from its role with indexicals. In the case of indexicals, the
meaning of a given expression determines that certain specific contextual relationships to
the utterance and utterer—who is speaking, or to whom, or when—determine
designation. Different facts are relevant for different indexicals, and the meaning of the
indexical determines which. Names don’t work like this. The difference between “David”
and “Harold” is not that they are tied, by their meanings, to different relationships to the
utterance or utterer. The role of context is simply to help us narrow down the possibilities
for the permissive conventions that are being exploited.” (7)

Perry’s objection to the thesis that proper names are indexicals could be blunted if one

could show that there are additional features of the context to be added to Kaplan’s list of

environmental features by which contexts of utterances are individuated and to which content is

sensitive that would indeed tie the content of a proper name to features of context surrounding

the utterer and the specific utterance. Specifically, the features of context would have to do with

the causal-historical features of the utterer’s mental state (e.g., the speaker’s dossier) at the very

instant the name is tokened that directly tied the utterer’s mental state (or dossier), and his or her

utterance itself, to the bearer of the name.

121
 A potential objection to the indexical view of names might be that these causal-historical

features linking the speaker’s mental state to the bearer of the name are far more complex and

difficult to ascertain in a conversational setting than in the case of other indexicals such as ‘I’ or

‘now’, where such features are far easier to determine. Causal-historical factors linking a name

via a casual chain back to its bearer are not readily discernible to the participants in a

conversation. However, I think there is no principled place to draw the line with respect to the

level of complexity that would admit some contextually sensitive expressions to the class of

indexicals and exclude others. An expression whose content depends on features of context is an

indexical, full stop, regardless of how complex the features of context to which the indexical are

sensitive.

Moreover, the distinct advantage to considering proper names to be indexicals—that it

offers a ready solution to the Problem of Rational Inconsistent belief, as shown in Chapter 2 (and

also discussed, infra, at 3.3)—militates in favor of the indexical view. Furthermore, the

indexical view is more parsimonious than the view that proper names are multiply-ambiguous

expressions, which would claim that ‘John’ constitutes millions of distinct names (one for each

person named ‘John’) that happen to be spelled (and pronounced) the same way. The more

parsimonious and intuitive view that ‘John’ one single expression that can be used to refer to

many different individuals bearing that name.

Why think that proper names are indexicals, rather than multiply ambiguous expressions?

For one, names have all the features of indexicals (a)-(k) listed above: they are context sensitive,

are functions from context to content, upload content into the proposition, and contribute just

their contents, and not the name’s character, to the proposition. The thesis that names are

indexicals is strengthened by the Kripkean (anti-Descriptivist) notion that a proper name has its

122
referent in virtue of a relational, a causal-historical, property. A speaker refers not in virtue of

what is “in his head,” but rather in virtue of some property of the environment or context in

which he utters the name. Just as an utterer of ‘I’ need not have specific uniquely denoting

information to refer to himself when uttering ‘I,’ according to the Kripkean causal-historical

theory of reference an utterer of ‘Aristotle’ need not have uniquely denoting descriptive

information to refer to the ancient philosopher (rather than the shipping magnate). Externalist

causal-historical features of the environment, about which the utterer of a name need not have

belief or knowledge, determine the reference of his utterance, largely or wholly independent of

the speaker’s intentions and beliefs. If we were to endorse a purely causal theory or reference for

proper names, we certainly we want to claim that names are indexicals. Their character would

be a rule that maps the utterance, based on the causal-historical factors present at the time and

place of the utterance, to the name’s referent. It would do so automatically, as the speaker does

not (and cannot) have in his head the information about the causal-historical links between his

utterance and the referent of the name he utters in virtue of which he refers to one person rather

than another bearing a particular name.

Even if we were to posit a theory of reference that was not purely causal-historical, e.g., a

hybrid theory or reference for proper names that included some descriptive beliefs as reference

determining factors, we would nevertheless want to claim that names are indexicals with

character determining content. We would want a more complex character—perhaps a rule

capturing a complex algorithm that takes causal-historical factors together with facts about some

reference determining descriptive beliefs the agent associates with the name and maps it to the

referent. The mapping of an utterance of ‘Aristotle’ to the philosopher rather than the shipping

magnate would be in virtue of various features of context—both causal-historical features of the

123
utterance, the speaker’s mental state that caused the utterance, and some descriptive beliefs of the

agent (e.g., perhaps beliefs about the general kind of thing, such as the property of being a

human being, he intends to refer to).

To recap, the argument for the indexical view proceeds as follows: Start with the

question—In virtue of what does an utterance of ‘John’ refer to one John rather than a different

John? Certainly, it cannot be in virtue of the orthographic or phonological properties of ‘John,’

for each person named ‘John’ bears an orthographically identical name. It cannot depend on the

context of the conversational setting alone, since a speaker could intend an utterance of ‘John’ to

refer to any individual bearing the name ‘John,’ even to a bearer of ‘John’ who the

conversational participants would not expect to be the referent of ‘John’ in the context of the

conversation. The reference of an utterance of ‘John’ must be a function of the speaker’s intent

to refer to one John rather than another. The speaker has a specific John in mind when he utters

‘John.’ In virtue of what does the speaker have one John in mind rather than another? It must be

in virtue of some property of the speaker’s mind, his/her mental state at the time he utters the

name and which causes the utterance. What sort of property? Possible answers include: a

relational (extrinsic) property such as a causal-historical property of the mental state, an intrinsic

property of the mental state (e.g., a purely general descriptive belief), or a hybrid property—a

function of both causal-historical features and descriptive beliefs. In short, the reference of a

tokening of ‘John’ depends on some property of the speaker's’ mental state at the time of the

tokening. This mental state differs from speaker to speaker and time to time across tokenings of

‘John.’ Different bearers of ‘Johns’ are picked out by utterances of ‘John’ in virtue of the mental

states producing the utterances differing in their properties. So, the reference of ‘John’ is context

dependent. The reference depends on features of the context/environment—properties of the

124
mental state of the utterer at the time of the utterance, just as the reference of ‘I’ or ‘now.’ The

main difference between the case of proper names and indexicals such as ‘I’ or ‘now’ is that the

features of the environment/context upon which the reference of a name depend would be far

more complex than the features of context on which the reference of other sorts of indexicals,

such as ‘I’ or ‘now,’ depend. In the case of proper names, the features of context would consist

extrinsic causal-historical properties in the case of a causal-historical theory or reference, and in

the case of a hybrid theory, they could consist of some complex mix of intrinsic mental

properties and such causal properties. Whereas in the case of ‘I’ or ‘now,’ the features of context

that determine content are relatively straightforward and simple (i.e., for ‘I,’ the only feature of

context one need consider is who the speaker is), there is little reason to think that there is some

threshold of complexity of environmental factors to which it is sensitive above which a

contextually sensitive expression would fail to count as an indexical. Why should we draw any

such line? Drawing such a line is especially unwarranted if not drawing the line would yield a

solution to the problem of Rational Inconsistency, such as the one I outlined in section 2.8. 75 76

3.3 Pelczar and Rainsbury’s “Single-Indexical” Theory of Proper Names

On the theory of Pelczar and Rainsbury (1998) (hereinafter “P & R”), proper names are

indexicals whose content is just the referent of the name. I call this theory a “single-indexical”

75
Specifically, treating the names as indexicals tells us that Lois can have the character meanings
of ‘Clark Kent’ and ‘Superman’ in her mind without realizing that they determine the same
contents.
76
If Millianism is true and the bearer exhausts the meaning of a name, it is mysterious how Lois
Lane could count as understanding the names without being in a position to see they co-refer.
We need propositional guises if we do not have indexicality, but I see no way of making sense of
propositional guises, unless these guises are ultimately explained in terms of some underlying
mechanism, and indexicality seems to me a promising in this regard.

125
theory because it posits that names are used as one type of indexical, rather than two types of

indexicals (between which they are ambiguous), as he TIUT claims. P & R’s theory is a species

of Millianism because proper names always just contribute their bearers and nothing further to

the propositions expressed by the sentences in which they occur. 77 P & R claim that their theory

has the potential to solve Frege puzzles. However, I shall argue that they can solve only the

Problem of Rational Inconsistent Belief (which P & R term ‘the problem of coherent

inconsistency’). Along with other Millian theories, they do not provide for a mechanism by

which proper names can sometimes contribute descriptive conceptions to propositions, and

therefore they cannot solve Frege’s puzzles.

P & R’s theory has Kripke’s causal-historical theory of the reference of proper names as

its touchstone (see 4.3 for discussion of Kripke’s theory). On Kripke’s theory, a naming

convention is established for an individual—it comes into existence for referring to its bearer—

via an act of dubbing (a.k.a., a baptism). You are in the presence of an individual, or perhaps en

rapport with that individual (even if they are not physically present before you), and via a

reference fixing definite description, an act of ostension, or a combination thereof, you fix the

reference of the name. For example, you point to the baby and say to yourself “let that baby bear

the name ‘John.’” The baby has this been dubbed with the name ‘John.’ You have established a

77
Besides Pelczar and Rainsbury (1998), Recanati (1993, 140-145) is another theorist who has
recognized the indexical nature of names. Recanati writes that names are very much like
indexicals, having character and content, although it is not clear whether he has expressly
claimed they are indexicals. According to Powell (2016, 92-93), Recanati maintains that the
character of name ‘NN’ would be the nominal description the bearer of ‘NN’; the content of
‘NN’ would be the bearer of ‘NN,’ i.e., the flesh-and-blood bearer of that name. What separates
Recanati’s view from that of a proponent of metalinguistic descriptivism, e.g., Kent Bach, a
proponent of the “nominal description theory”, is that Bach claims that the meaning of the
nominal description the bearer of ‘NN’ is the content of ‘NN’, not its character, while Recanati, a
Millian, thinks the bearer is the name’s content and the description the bearer of ‘NN’ its
character.
126
convention according to which one may use ‘John’ to refer to that person. There are of course

multiple other co-existing conventions that allow ‘John’ to be used to refer to numerous other

individuals.

P & R’s (1998) key notion for their theory is what they term a “dubbing-in-force”. As

they state it,

“A dubbing is a speech-act whereby a name acquires a referent, and a dubbing is in force

in a given context if in that context the item that was dubbed in that dubbing bears the
name it received in that dubbing.” (294)

Although P & R do not explicitly spell out the character of proper names, on a rational

reconstruction of their theory we may say that the character of any proper name, ‘NN’, is:

CHARACTER OF NAME ‘NN’: The individual bearing the name ‘NN’ who has
been dubbed ‘NN’ in the dubbing-in-force in the
context.

On P & R’s theory, the dubbing-in-force is a feature of the context to which character is

sensitive. They suggest that dubbing-in-force should be added to Kaplan’s list of four features:

utterer, place, time, and possible world. P & R do not elaborate the specifics of how to

determine which dubbings are in force in a given context.

“…the dynamics of dubbings-in-force can be complex and we shall not attempt to

provide a systematic way to decide which dubbings are in force in a given context.” (295)

P & R make it clear that more than one dubbing may be in force for a name in each context. In a

context of token utterance of ‘John,’ there may be numerous dubbings in force, since many

individuals have been dubbed ‘John’ and bear the name. Not every dubbing involving the name

‘John’ is in force in the context. There are two reasons for this. First, not all the individuals

127
dubbed with ‘John’ now bear the name ‘John.’ In Gareth Evans’ (1982) famous example, two

newborn babies lay in the hospital. One had been given the name ‘John’ and the other the name

‘Jim’ by their respective parents. The babies are switched at birth. Unaware of the switch, the

baby dubbed ‘John’ grows up being called ‘Jim’ and the baby dubbed ‘Jim’ grows up being

called ‘John.’ The original dubbings are no longer in force: they have fallen into disuse. The

baby dubbed ‘John’ no longer bears the name ‘John’; the baby dubbed ‘Jim’ no longer bears the

name ‘Jim.’ The original naming conventions have died out. Secondly, according to P & R,

certain dubbings are not in force in a context because the conversational setting automatically

excludes certain dubbings. For example, if I am reminiscing with a high school friend about

people we knew in common in our high school graduation class and I utter the name ‘John,’ the

conversational setting entails that certain dubbings are not in force in the context of the

conversation. The dubbing of John the Baptist with ‘John’ or John Quincy Adams with ‘John’

are not in force in the context. For it is highly improbable that I would be referring to these

individuals, given the setting of the conversation mutually known to the conversational

participants. The relevant conversation is about individuals with whom the conversational

participants attended high school, not about historical figures that died centuries before the

conversational participants attended high school. When there is more than one dubbing in force

in a context,

“…one of the competing dubbings must be brought to prominence to determine a unique

referent of the name (in that use). This might be achieved by a variety of mechanisms.
One important factor in the raising to prominence might be relevant features of the
conversation (if any) of the context of utterance of the name. For example, in the course
of a conversation dealing both with president Bush and his son, someone says: ‘George
Bush occupied the oval office for only one term’, Gricean conversational maxims (and in
particular the maxim of quality) might contribute to raising of the dubbing of the
President with ‘George Bush’ to prominence, over his son.” [N.B. this was written in
1998, before George W. Bush was elected president]

128
In some contexts, there could be staggeringly large numbers of individuals potentially picked out

by a single use of ‘John’ from the hearer’s perspective. Upon an utterance of ‘John’ a hearer

interprets the utterance via pragmatic principles—ascertaining which John was the likely

intended referent of the speaker based on contextual clues. 78

P & R claim their theory can solve Frege’s and Kripke’s puzzles, although I believe that

they can solve only the Problem of Rational Inconsistent Belief (which P & R call ‘the Problem

of Coherent Inconsistency’). They begin their discussion of their solution of this puzzle on the

bottom of page 304:

“We believe that when seen in the right light, most of the well-known puzzles of attitudes
semantics can be explained as cases involving some contextual error on the part of an
attitude ascribee. Some preliminary illumination may be afforded by the observation that
a main goal of attitudes semantics is to explain the possibility of coherent inconsistency
[italics mine]. This is the possibility for a rational individual with complete linguistic
competence to bear incompatible attitudes towards the same object of attitude-bearing.
Classically, the problem is how to explain the consistency of a statement like this:

(17) Thales believes that Hesperus is shining and disbelieves that Phosphorus
is shining.

with a statement like this:

(18) Thales is coherent (i.e., is neither irrational nor in any relevant respect
linguistically incompetent).”

Here, P & R have set up the problem of coherent inconsistency (which I call ‘the Problem of

Rational Inconsistent Belief): to explain how a rational and coherent Thales could have

inconsistent beliefs with respect to Venus, believing both that it is shining and disbelieving that it

78
Where there are multiple dubbings in force, the character of the name alone does not map it to
a unique content/referent, but a range of possible contents/referents. To determine the correct
bearer of a name with multiple dubbings in force requires one to look at speaker intentions,
which suggests that proper names are closer to true demonstratives than automatic indexicals
(although they do not expressly state this, as they seem not to be concerned with this distinction).
129
is shining. By construing proper names as indexicals whose content is a function of externalist,

extra-semantic factors (as opposed to internalist semantic factors inside the speaker’s head), P &

R explain Thales’ rationality/coherence:

“It is a peculiar feature of an indexical that its literal content is partly determined by an
extra-semantic feature of its context of utterance. We call a feature extra-semantic with
respect to an expression just in case it need not be mentioned in a satisfactory
specification of the literal meaning or character of that expression. For example, the
literal content of the present utterance of ‘I’ depends partly on the fact that Joe Rainsbury
uttered it. But this fact need not figure at all in a satisfactory explanation of the meaning
of ‘I’. Therefore, the fact is an extra-semantic feature of the present context of utterance,
with respect to ‘I’.” (305)
…

If 'Hesperus' or 'Phosphorus' or both are indexicals, then the mutual consistency of (17)
and (18) can be explained… [A] rational Thales can know the meanings of both
'Hesperus' and 'Phosphorus' - i.e., he can know that 'Hesperus' refers to whatever was
dubbed in the dubbing-in-force governing 'Hesperus' in his context of utterance, and
mutatis mutandis for 'Phosphorus' - without knowing that both names happen (in his
context) to refer to the same entity. (307)

Thales is coherent and rational because his inconsistency is explained by his ignorance of

the fact that the indexicals ‘Hesperus’ and ‘Phosphorus’ have the same content, and this

ignorance is wholly constituted by ignorance of externalist, extra-semantic information to which

Thales has no cognitive access (without receiving additional empirical evidence). Thales knows

the meanings of ‘Hesperus’ and ‘Phosphorus’, despite his ignorance of their co-reference,

because to know the character of these names is to know their meanings. That is, Thales is a

competent user of ‘Hesperus’ and ‘Phosphorus’ because he knows that they are used to pick out

whatever objects were dubbed with those names in the dubbing in force in the context.

P & R astutely point out that standard Millians, who hold that the meaning of a name is

exhausted by its bearer (and that names have no character meanings) cannot easily explain

Thales’ rationality/coherence.

130
 “… the currently reigning direct reference theory of names seems, at least in its present
form, incapable of explaining the mutual consistency of, e.g., (17) and (18). In fact, the
direct reference theorists' best attempt to solve the problem seems to presuppose the
indexical theory of names. This is their argument that (17) is consistent with (18) because
someone (like Thales) can know what each of two coreferential terms refers to, without
knowing that the terms are coreferential. But it is not clear how a coherent Thales could
know what 'Hesperus' referred to as well as what 'Phosphorus' referred to without
knowing that they co-referred, if the meanings of the names are exhausted by their
referents. What must Thales's knowledge of what these names refer to consist in, for it to
be compossible with his ignorance of their coreference? It is not available to a direct
reference theorist to explain this possibility by construing 'Hesperus' and 'Phosphorus' as
disguised descriptions, for a direct reference theorist maintains that they are rigid. In fact,
when forced to elaborate on the nature of Thales's knowledge of what 'Hesperus' refers to,
a direct reference theorist (especially one with a taste for conceptual economy) will tend
to fall back on indexicality. For example, he will say that a coherent Thales can know that
that object (pointing skyward) is the referent of 'Hesperus', and that that one (indicating a
point in a photo graph of another clear night sky) is the referent of 'Phosphorus', without
knowing that the referent of 'Hesperus' is the referent of 'Phosphorus'. And this is true.
But the reason why it is true is that 'that' is an indexical, owing to which Thales can know
the literal meaning of 'That object is the referent of "Hesperus"' without knowing its
literal content. But on the standard direct reference theory of names, 'Hesperus' is not an
indexical. Therefore, on the direct reference theory it must be in principle possible for
Thales to know the literal meaning - and therefore (since 'Hesperus' is rigid, on the
theory) the referent - of 'Hesperus' without having to employ indexical terms or concepts
(just as it is possible to know or specify the literal meaning - and thereby the reference -
of 'excellence' without recourse to indexicality). But in the face of such a case a direct
reference theorist is left without any conceptually parsimonious characterization of what
Thales's knowledge of what 'Hesperus' refers to consists in, such that it is possible for
Thales coherently to know also what 'Phosphorus' refers to while remaining ignorant of
the coreference of the two names. The only sure way to guard against such a case would
be by claiming that 'Hesperus' is an indexical. For only granted that claim would it not be
in principle possible for Thales to know the literal meaning of 'Hesperus' without having
to employ indexical terms or concepts. Thus, fully developed, the best strategy a direct
reference theorist has to explain the mutual consistency of (17) and (18) is to adopt the
indexical theory of names.” (308-309)

Here, P & R provide a perfect explanation of why Millians should adopt their indexical

view. Millians have trouble explaining how Thales, who knows the meanings of ‘Hesperus’ and

‘Phosphorus’, could fail to realize their co-reference. After all, on standard Millianism, knowing

the referents of those names wholly constitutes understanding those names and being a

competent user of them. If Thales knows the referents of the names, what explains why he fails

131
to recognize them as the same when presented by the different names? The standard Millian

answer is that Thales takes Venus under different guises or modes of presentation when

presented by different names. But the notion of guises is more of a gesture at an explanation

than a genuine one. By contrast, the indexical theory of names gives a straightforward and

simple explanation. Thales knows the character of the names ‘Hesperus’ and ‘Phosphorus’, and

is thus a competent user of the names, but fails to know the externalistic factors in virtue of

which the names co-refer.

As P & R astutely point out above, if pressed to explain in concrete terms how Thales

could be competent with the names ‘Hesperus’ and ‘Phosphorus’ and still fail to realize they co-

refer, standard Millians will typically resort to explanations that bring in indexicality, and this

suggests that the phenomenon of Rational Inconsistent Belief arises because proper names are in

fact indexicals in the first place. As P & R state it:

“… when forced to elaborate on the nature of Thales's knowledge of what 'Hesperus'

refers to, a direct reference theorist (especially one with a taste for conceptual economy)
will tend to fall back on indexicality. For example, he will say that a coherent Thales can
know that that object (pointing skyward) is the referent of 'Hesperus', and that that one
(indicating a point in a photo graph of another clear night sky) is the referent of
'Phosphorus', without knowing that the referent of 'Hesperus' is the referent of
'Phosphorus'. And this is true. But the reason why it is true is that 'that' is an indexical,
owing to which Thales can know the literal meaning of 'That object is the referent of
"Hesperus"' without knowing its literal content.” (309)

Although P & R furnish an elegant solution to the Problem of Rational Inconsistent

Belief (or as they term it, ‘the problem of coherent inconsistency’), I believe that they exaggerate

the capability of their theory to solve the puzzles—claiming that they can solve Frege’s puzzles

as well. This is due to their taking the problem of coherent inconsistency to represent the

essence of Frege’s puzzle. They write:

132
 “Should someone object that we have misrepresented the puzzles by construing them as
so many variants of what we have called the problem of coherent inconsistency, his
objection would be very much to the point. However, we believe that the puzzles are
instances of that more general problem, and we think that the order that begins to emerge
in this field (where puzzles often seem to be a dime-a-dozen) when seen in light of the
problem of coherent inconsistency is itself evidence that we are not misrepresenting
them.” (311)

They claim here that the key to solving Frege’s puzzle begins (and perhaps ends) with a

solution to the Problem of Rational Inconsistency. But here I believe P & R have things wrong

because this puzzle and Frege’s puzzles are distinct. While treating proper names as indexicals

can explain why Lois Lane is not irrational for saying ‘Superman flies’ and ‘Clark Kent does not

fly’, 79 it does not explain why (3) is (or at least seems) true while (4) is (or at least seems) false.

(3) Lois Lane believes that Clark Kent is Clark Kent

(4) Lois Lane believes that Clark Kent is Superman

Since P & R are Millians under one definition of Millianism (because they hold that the content

of either ‘Clark Kent’ or ‘Superman’ is just Kent-Super), they must construe (4) as being literally

true, since the names in the ‘that’-clauses of (3) and (4) have the same content, Kent-Super. To

solve the Frege’s puzzles, we need a non-Millian theory of content that differentiates (3) from (4)

in content and shows why (4) is false. Or, alternatively, we need the sort of Millian theory that

explains why we have the powerful by erroneous intuition that (4) is false, as standard Millians

try to provide (by arguing that ordinary speakers confuse semantics and pragmatics). P & R do

79
Lois Lane cannot see that ‘Clark Kent’ and ‘Superman’ co-refer. What is apparent to her,
what she “sees” is the character of these indexicals, but not their contents. The characters of the
names are in her head, the content is not. Content is a function of dubbings-in-force, which are
not in a speaker’s head. The contents of the names are hidden behind the guise of the character.
P & R’s solution to the Problem of Rational Inconsistent Belief if thus resembles the TIUT’s
proposed solution, as discussed in section 2.8.

133
not provide either of these. We also need a theory that explains the difference in cognitive value

between (1) and (2).

(1) Clark Kent is Clark Kent

(2) Clark Kent is Superman

P & R offer nothing to explain how these express different propositions. At most, they explain

why the unenlightened Lois Lane may take them to express different propositions, why these

have different cognitive value to her, and why she judges them to differ in truth-value. However,

with respect to an enlightened speaker such as Olson who knows the names have the same

referent, we need to explain why he strongly intuits that these sentences express different

propositions. Olson, being enlightened, knows that the character of the indexicals ‘Clark Kent’

and ‘Superman’ determine the same content; yet he still strongly intuits that (1) and (2) say

completely different things. They have different cognitive value to him, even though he is not

confused about their co-reference. Thus, while having solved the problem of Rational

Inconsistent Belief, P & R have not solved Frege’s puzzles.

134
 CHAPTER 4

DESCRIPTIVISM

4.1 Classical Descriptivism: The Descriptivism of Frege and Russell

As I discussed already in section 1.1, classical Descriptivism, the view of Frege (1892)

and Russell (1905), posits that the meaning of a proper name is equivalent to the meaning of a

definite description. Furthermore, Descriptivism holds that these descriptive meanings are

reference-determining. For example, the meaning of the name ‘Clark Kent’ might be expressed

by the definite description the ‘mild-mannered reporter from Smallville working for the Daily

Planet.’ The meaning of ‘Superman’ might be expressed by the definite description ‘the caped

superhero that protects Metropolis’. The names ‘Clark Kent’ and ‘Superman’ would co-refer

because these definite descriptions determine the same reference, i.e., they denote the same

individual. 80 Descriptivism thus explains why (1) and (2) differ in cognitive value in Frege’s

80
Some Descriptivists advocate a weaker form of Descriptivism. They maintain that a
description or set of descriptions fixes the reference of a proper name, but they do not hold that
that description (or set thereof) constitutes the name’s meaning. For example, such a
Descriptivist might stipulate that the name ‘Aristotle’ shall refer to the greatest philosopher of
antiquity and teacher of Alexander the Great. The reference of ‘Aristotle’ to Aristotle would be a
function of Aristotle satisfying this description, and not on a causal-historical relation between
‘Aristotle’ and Aristotle himself as on a Kripkean causal-historical theory of reference. However,
on this weaker Descriptivism, ‘Aristotle’ would not be synonymous with that definite description
(or set thereof) in that speaker’s idiolect. Hence, one could not freely substitute a definite
description into a sentence in place of the name whose reference it fixed without changing the
meaning of a sentence. The sentences ‘Aristotle was fond of dogs’ and ‘The greatest philosopher
of antiquity and teacher of Alexander the Great was fond of dogs’ would not express the same
proposition. Kripke’s modal, epistemic, and semantic arguments against Descriptivism would
not work against this weaker version of Descriptivism in quite the same way as it does against
the stronger versions. This weaker form of Descriptivism is really a theory of reference-fixing
and not a theory of content. In fact, a proponent of this weaker form of Descriptivism could be a
Millian with respect to content, holding that the content of a proper name is whoever the name
refers to. In this section, and in the subsequent sections in which I discuss Descriptivism and
Kripke’s arguments against it, I take Descriptivism to be exclusively the stronger version and set
135
puzzle about identity sentences. Sentence (1) would be uninteresting and uninformative because

it would express the trivial and obviously true proposition that the mild-mannered reporter from

Smallville working for the Daily Planet is the mild-mannered reporter from Smallville working

for the Daily Planet. Sentence (2) would be interesting and informative because it would express

the non-trivial and non-obviously true proposition that the mild-mannered reporter from

Smallville working for the Daily Planet is the caped superhero that protects Metropolis.

Descriptivism also claims to furnish a solution to Frege’s puzzle about propositional

attitude ascriptions, although the details of the Russellian and Fregean solutions differ

significantly. (I’ll prescind from those differences in the discussion that follows and state the

descriptivist solution in a generic and simplified way, given that Kripke’s influential critique of

Descriptivism applies in equal measure to all variants of Descriptivism). According to

Descriptivism, ascription sentences (5) and (6)

(5) Lois Lane believes that Clark Kent can fly

(6) Lois Lane believes that Superman can fly

would mean the same as (5)desc and (6)desc respectively.

(5)desc Lois Lane believes that the mild-mannered reporter from Smallville
working for the Daily Planet can fly.

(6)desc Lois Lane believes that the superhero that protects Metropolis can fly.

aside any discussion of the weaker version. In this dissertation, I am principally concerned with
the content of proper names, not the theory of reference. The TIUT has Kripke’s causal-historical
theory of reference as a touchstone, but it is potentially compatible with various theories of
reference. Furthermore, the weaker reference-fixing version of Descriptivism is uninteresting
because it suggests no solution to Frege’s puzzle.

136
Clearly, (5)desc and (6)desc express different propositions, with the former being false and the

latter true. It follows that (5) and (6), which mean the same as (5)desc and (6)desc, differ in truth-

value as well. Just as (5)desc and (6)desc attribute distinct beliefs to Lois Lane, so do (5) and (6). 81

Descriptivism also offers a solution to the Problem of Rational Inconsistency (see section

1.2). On Descriptivism, (1) expresses a different proposition from (2), and (3) a different

proposition from (4). Hence, in believing that Clark Kent is Clark Kent and disbelieving that

Clark Kent is Superman, Lois Lane does not believe and disbelieve the same proposition. She

does not simultaneously believe the inconsistent propositions PROP-1 and ¬ PROP-1 (i.e., that

Kent-Super is Kent-Super and that Kent-Super is not Kent-Super) as Millians maintain. Instead,

she believes one descriptive proposition and disbelieves a different descriptive proposition.

Likewise, her believing that Superman can fly and Clark Kent cannot fly is tantamount to her

believing that the superhero that protects Metropolis can fly but a mild-mannered reporter for the

Daily Planet cannot.

Until the 1960’s, classical Frege-Russell Descriptivism (modified somewhat by ‘cluster’

theories of Descriptivism 82) was largely unquestionably accepted as providing the gold standard

81
This characterization of Descriptivism’s solution to the propositional attitude ascription puzzle
is simplified and it does not reflect the details of Frege’s solution to the puzzle. As mentioned in
a footnote in section 1.1, Frege has a reference shift theory whereby the words in the ‘that’-
clause refer to their senses (instead of expressing them), so that a propositional attitude ascription
reports the belief relation between an agent and the proposition referred to by the ‘that’-clause.
The idea is that (5) and (6) can differ in truth-value because some of the singular terms occurring
in them—to wit, their ‘that’-clauses, including the proper names occurring within them, differ in
reference.
82
According ‘Cluster Descriptivism’, associated with Wittgenstein (1953), John Searle (1958),
and Peter Strawson (1959), a name would be associated with a cluster of definite descriptions,
rather than a single definite description as on classical Frege-Russell Descriptivism (Cumming, §
2.3). For example, a speaker might associate the name ‘Albert Einstein’ with the cluster of
definite descriptions ‘the discoverer of the theory of relativity’, ‘the most famous Scientist of the
20th Century’, and ‘the winner of the Nobel Prize for explaining the mechanism of Brownian
137
for addressing Frege’s puzzle. However, several philosophers began to criticize it beginning in

the 1960’s. The most significant and comprehensive critique of Descriptivism came in 1970’s

with Kripke’s Naming and Necessity. Many of Kripke’s criticisms of Descriptivism reflected

the influence of ideas discussed by other philosophers in the previous decade, including John

Searle, Hilary Putnam, Keith Donnellan, David Kaplan, Ruth Barcan Marcus, et alia. Kripke

made three principal arguments against Descriptivism that have come to be called the Modal

Argument, the Epistemic argument, and the Semantic argument. I set out these arguments in

brief below. I also touch on Kripke’s alternate theory of reference, the causal-historical picture

of reference, which he intended to replace Descriptivism’s picture of reference. Finally, I discuss

and criticize three modern variants of Descriptivism: Rigidified Descriptivism, Causal

Descriptivism, and Metalinguistic Descriptivism. I argue that these modern versions of

Descriptivism are inadequate to solve the puzzles, just as Frege-Russell classical Descriptivism

was.

4.2 Kripke's Arguments against Descriptivism

Descriptivism has been subjected to withering criticism since the early 1970’s and few

philosophers ascribe to it today. The most comprehensive critique of Descriptivism came in

1970 with the delivery by Saul Kripke of a series of lectures, which were eventually transcribed

and published as the book Naming and Necessity in 1980.

Descriptivism had become the most widely accepted theory of proper names around the

motion’. ‘Albert Einstein’ would refer to whomever is the denotation of the majority (or
weighted majority) of the definite descriptions in the cluster associated with it, even if some of
the associated definite descriptions denoted other individuals or failed to denote any individual at
all. Cluster Descriptivism is vulnerable in equal measure to Kripke’s argument against
Descriptivism as is classical Frege-Russell Descriptivism.
138
turn of the twentieth century, mainly because Frege’s puzzle, first discussed in Frege’s 1892,

seemed to show that Millianism was false. (See the two reductio arguments against Millianism in

the introduction and chapter one, supra). Descriptivism also appeared to provide an answer to

the problem of empty and fictional names. See infra at section 4.4 for brief discussion of the

problem of empty and fictional names.

Despite the promise that Descriptivism seemed to hold to solve these difficult

problems—Frege’s puzzles and the problem of empty and fictional names—Kripke (1980) made

three powerful arguments against Descriptivism—the Modal Argument, the Epistemic Argument,

and the Semantic Argument, a.k.a. ‘the argument from ignorance and error.’ These arguments

demonstrated that Descriptivism was highly implausible. Although Kripke showed that

Descriptivism was highly implausible, he did not thereby show that Millianism was correct.

Kripke himself has never stated that he is a Millian. But in the wake of his successful arguments

against Descriptivism, many philosophers have become Millians.

4.2.1 The Modal Argument (N & N: 48-49, 71-77)

Kripke’s modal argument against Descriptivism (1980, 22 et seq.) can be summarized as

follows:

(a) A proper name is a rigid designator (Kripke, 1980), meaning it designates the same
individual/object in all possible worlds.

(b) An ordinary definite description is not a rigid designator. It does not denote the same
individual/object in every possible world. 83 (For the present, ignore the complication that

83
Ordinary definite descriptions such as ‘the mild-mannered reporter from Smallville who works
for the Daily Planet’ and ‘the superhero that protects Metropolis’ are non-rigid designators. To
see this, suppose that there is an alternate world w2 in which Kent-Super never leaves the planet
Krypton. Some other individual from Krypton, say, Kent-Super’s cousin Ziggy, comes to earth
and protects Metropolis while hiding his identity as a reporter for the Daily Planet. With respect
to w2, those definite descriptions (as used in the actual world) designate to Ziggy, not Kent-
139
 rigidified definite descriptions are rigid designators; this complication is discussed in
section 4.5 infra).

(c) If one expression a is a rigid designator and another expression b is not a rigid designator,
then a and b do not have the same meaning.
______________________________________________________________________________
(d) Therefore, the meaning of a proper name cannot be equivalent to the meaning of a
definite description.

The current philosophical consensus accepts Kripke’s claim that proper names are rigid

designators. Some modern defenders of Descriptivism impugn the modal argument not by

denying that proper are rigid designators, but rather by arguing that the meaning of a proper

name is equivalent to the meaning of a rigidified definite description (or cluster thereof) that

fixes the name’s referent according to properties the referent bears in the actual world. Hence,

they argue that the modal argument does not show that Descriptivism is incorrect, full stop, only

that a Descriptivism on which the definite descriptions are not rigidified is defective. However,

rigidified Descriptivism is nevertheless not an adequate theory of proper names. It is vulnerable

to Kripke’s Epistemic and Semantic arguments. Rigidified Descriptivism discussed infra in

section 4.5.

4.2.2 The Epistemic Argument (N & N: 86-87)

Consider sentence (a), which expresses a proposition knowable a priori:

(a) If Superman exists, then Superman is Superman.

Now suppose that Descriptivism were true and ‘Superman’ meant the same as the definite

description ‘the superhero that protects Metropolis’. We should be able to substitute this definite

Super, in w2. If these definite descriptions were rigid designators, they would designate Kent-
Super in every world.
140
description for any of the occurrences of ‘Superman’ in the sentence above and the new sentence

(containing the substitution) should express the same proposition as before. For example,

substituting this definite description for the third occurrence of ‘Superman’ would yield the

sentence (b), below, which should express the same proposition as (a):

(b) If Superman exists, then Superman is the superhero that protects Metropolis.

If (a) and (b) express the same proposition, as Descriptivism claims they would, then they should

have the same epistemological status—both a priori. However, here is the problem. Whereas (a)

is clearly knowable a priori, (b) is not. Intuitively, sentence (b) expresses an a posteriori

proposition. We find out that Superman is the superhero that protects Metropolis empirically,

not by conceptual reasoning. Because they differ in epistemological status, sentence (a) and (b)

cannot express the same proposition. So, the definite description ‘the superhero that protects

Metropolis’ and the name ‘Superman’ do not mean the same thing, and therefore, Descriptivism

is false.

4.2.3 The Semantic Argument (N & N: 78-86)

The Semantic Argument (Kripke, 1980, 83 et seq.) is also known as ‘the argument from

ignorance and error’ (Devitt and Sterelny 1999, 54). Consider first the argument from error.

Consider a name you are competent with using, and count as understanding, like ‘Albert

Einstein.’ But suppose that you erroneously associated the name ‘Albert Einstein’ with the

definite description ‘the inventor of the atomic bomb.’ The inventor of the atomic bomb (or at

least the person who ran the team that developed it) was in fact J. Robert Oppenheimer. The

definite description ‘the inventor of the atomic bomb’ in fact picks out J. Robert Oppenheimer,

141
not Einstein. Now suppose that Descriptivism were true. According to Descriptivism, the name

‘Albert Einstein,’ tokened on this instance, would refer to Oppenheimer and not Einstein, the

reference of a proper name being determined by the definite description associated with it. But

this is a very counterintuitive consequence of Descriptivism. Intuitively, you refer to Einstein

when you utter ‘Albert Einstein.’ It’s just that you have a false belief about him. Your false

belief does not prevent you from referring to him, or mean that you refer to someone else. Thus,

associating the correct descriptive information with a proper name cannot be required for

competence with the name.

Now I shall briefly discuss the argument from ignorance. Again, consider a name you are

competent with using, and count as understanding, like ‘Albert Einstein.’ Suppose you are

unable to associate a single uniquely identifying definite description with a proper name in his or

her lexicon. Many speakers could at most associate ‘Albert Einstein,’ a name with which they

are familiar (a name in their lexicon), with the indefinite description a scientist. Even the

description scientist named ‘Albert Einstein’ is indefinite (not uniquely denoting) if there is more

than one scientist in the world named ‘Albert Einstein.’ Yet, intuitively, despite you not

associating any uniquely identifying definite descriptions with ‘Albert Einstein,’ you still refer to

Albert Einstein upon uttering ‘Albert Einstein.’ We do not want to say that ‘Albert Einstein’ has

no reference for you just because the descriptions you associate with the name do not pick

anyone out uniquely. After all, a speaker completely ignorant about Albert Einstein might utter

“I wonder who Albert Einstein was” and he or she would refer to Albert Einstein in this

utterance (and say of Einstein that he wishes he knew more about him). If descriptivism were

true, this person would not refer to anyone at all with ‘Albert Einstein’ and would not express a

complete proposition in uttering this sentence (since the name would be empty).

142
 According to Kripke, possession of uniquely identifying accurate descriptive beliefs

about the referent of a proper name is not a requirement for one to be competent to use a name to

refer successfully. Instead, successfully referring by uttering a proper name depends on a causal-

historical link between the utterance of a name token and the referent itself. Kripke thus

develops the causal historical picture of reference, which I characterize briefly below.

4.3 Kripke’s Causal-Historical Picture of Reference (N & N: 88-97)

On Kripke’s view, a speaker refers to Richard Feynman by uttering ‘Richard Feynman’ in

virtue of the existence of a causal-historical chain connecting the utterance of the name with the

actual flesh-and-blood man, Feynman. The reference of the name ‘Richard Feynman’ was fixed

in an act of dubbing or in a “baptismal ceremony” by a speaker (presumably one or both of his

parents) who was in causal contact with him and introduced the name to refer to him. The name

was fixed to its bearer, Richard Feynman, by a reference-fixing description, by an act of

ostension, or a combination thereof. For example, Feynman’s parents might have fixed the name

to him by looking at him as a newborn and thinking or saying aloud something like let the name

‘Richard’ be used for this, our newly born son, in front of us now. Although this description

fixes the referent, attaching the name ‘Richard’ to the individual, the description is not the

meaning of the name. After the name is fixed to its bearer, the name is transmitted by this

original dubber via a causal chain of communication to other individuals who do not need to

know the reference-fixing description or act of ostension that was used to fix the name in order

to acquire competence with the name. Speakers downstream along the chain of communication

acquire the name ‘Richard Feynman’ when they hear utterances of it by other members of the

language community (who in turn had heard the name from others, such that there is a chain of

143
name transmission tied to the original act of dubbing and the individual dubbed with the name),

provided that each speaker in the chain has the intention of using the name with the same

reference as the speaker from whom he picked up the name. None of these speakers need any

information about reference-fixing descriptions or acts of ostension, nor do they need to know

anything about who fixed the name’s reference or how it was fixed to be competent with the

name. Furthermore, no speaker needs to possess any uniquely denoting descriptive information

about Richard Feynman, and he may even associate incorrect descriptions with the name, and he

will nevertheless refer to Feynman when uttering the name. As Kripke states it:

“Someone, let’s say a baby, is born; his parents call him by a certain name. They talk
about him to their friends. Other people meet him. Through various sorts of talk the name
is spread from link to link as if by a chain. A speaker who is on the far end of this chain,
who has heard about, say Richard Feynman in the market place or somewhere, may be
referring to Richard Feynman [through his use of name ‘Richard Feynman’] even though
he can’t remember from whom he first heard of Feynman or from whom he ever heard of
Feynman. ...A certain passage of communication reaching ultimately to the man himself
does reach the speaker. He then is referring to Feynman even though he can’t identify
him uniquely. He doesn’t know what a Feynman diagram is ...Not only that: he’d have
trouble distinguishing between Feynman and Gell-Mann. So he doesn’t have to know
these things, but, instead, a chain of communication going back to Feynman himself has
been established, by virtue of his membership in a community which passed the name on
from link to link, not by a ceremony that he makes in private in his study: ‘By
“Feynman” I shall mean the man who did such and such and such and such’.” (91-2)

...

“A rough statement of a theory might be the following: An initial ‘baptism’ takes place.
Here the object may be named by ostension, or the reference of the name may be fixed by
a description. When the name is ‘passed from link to link’, the receiver of the name must,
I think, intend when he learns it to use it with the same reference as the man from whom
he heard it. If I hear the name ‘Napoleon’ and decide it would be a nice name for my pet
aardvark, I do not satisfy this condition. ...” (97)

A speaker’s utterance ‘Richard Feynman’ or ‘Albert Einstein’ or any name whatsoever is

a link on the causal chain provided that each speaker acquiring the name intends to refer to

whomever the individual from whom he or she learned the name referred to when uttering the

144
name. The speaker acquiring the name ‘borrows the reference’ of the person from whom s/he

acquired the name. Again, no speaker on the chain requires uniquely identifying information

about the name’s bearer to refer to him, provided that the tokening of the name is ultimately

causally linked to bearer in the right sort of way and grounded in the bearer in an act of

reference-fixing.

Kripke recognizes that his proposal is not a full theory of reference. He called it a

“picture” of reference (1980, 93). There are several difficulties with the picture. First, suppose a

speaker—call him “Eugene,” whose use of the name ‘Albert Einstein’ is causally-historically

connected to Albert Einstein in the correct sort of way (e.g., he picked up the name from a friend

from whom he fully intended to borrow the reference), erroneously believes that ‘Albert

Einstein’ is the name of a mountain in Switzerland and utters “I want to climb Albert Einstein

one day.” Has Eugene referred to Albert Einstein here? Has Eugene stated that he wants to

climb a man? One could answer that, yes, Eugene stated he wanted to climb a man, and

specifically, he wanted to climb Albert Einstein, the scientist. One could also answer that, no,

Eugene is not a competent user of ‘Albert Einstein’ because his confusion about Einstein’s

fundamental or essential properties is too extreme. Some philosophers have suggested that

perhaps a speaker needs to conceive of the referent under the right general sortal concept for

reference to go through. That is, to avoid reference failure, a speaker must know the general

kind of object a name designates. For example, you would have to realize Einstein is a human

and not a mountain to refer to him by uttering ‘Einstein.’

Here’s another worry for the causal picture: the reference of a name can sometimes shift

from one object to another, even if it is transmitted in a casual chain with each speaker in the

chain intending to borrow its reference. This should not be possible under Kripke’s theory, since

145
he provides that acquiring a name with the intent to refer to the same object and the person from

whom the name is borrowed is supposed to be sufficient for preservation of reference. Evans

(1973) cites as an example the name ‘Madagascar.’ It originally referred to some part of the

African mainland, but at some point, it shifted its reference to the large island off Africa’s east

Coast we now call ‘Madagascar’. Here, we have an example of a reference shift among an entire

language community. It seems that such a reference shift could also occur for a single speaker.

For example, suppose that the name ‘Albert Einstein’ entered my lexicon after I picked up the

term from another speaker who clearly and unambiguously used the expression to refer to the

Albert Einstein, the scientist who discovered the theory of relativity. I intend to borrow the

reference from this speaker. But I come to believe in error that the speaker was using the name

to refer to the pioneering Soviet Russian film director and film theorist with a somewhat similar

name, ‘Sergei Eisenstein.’ For years after first pick up the name ‘Albert Einstein,’ I absorb

information about Eisenstein and associate this information with the name ‘Albert Einstein.’

Every time I utter ‘Albert Einstein’ I intend to refer to a Russian film director. Meanwhile, I

hear nothing about the scientist and am completely unaware that ‘Albert Einstein’ is standardly

used to refer to a scientist. If I utter one day ‘There is an Albert Einstein festival playing at Film

Forum,’ and there is in fact an Eisenstein festival playing there, have I expressed a true or false

proposition? Have I uttered a false proposition about Albert Einstein, or have I uttered a true

proposition about Sergei Eisenstein and merely mispronounced his name? Or should we say that

‘Albert Einstein’ is a name for ‘Sergei Eisenstein’ in my idiolect and I have referred to

Eisenstein by uttering ‘Albert Einstein’? Or have I referred to both men at the same time and

hence uttered two different propositions simultaneously, one true and one false? Or am I just so

confused that I have failed to express a complete proposition at all?

146
 I am skeptical that the question about Einstein/Eisenstein can be answered on any non-

arbitrary basis, which suggests to me that reference may well be an indeterminate matter under

certain circumstances: to wit, when the causal-historical properties of a name and the descriptive

properties a speaker associates with it, on a particular utterance, clash radically. This is not to

say that there is anything wrong with our notion of reference in general: there is after all no

problem with the notion of gender just because there are cases of inter-gender individuals (where

it is indeterminate whether the individual is male or female). Indeterminacy is a widespread

phenomenon, found wherever and whenever we divide up the world into categories and

classifications. We do not and should not throw out these categories and classifications every

time we encounter indeterminacy, and so likewise we should not abandon the notion of

reference. Reference is clearly a real phenomenon notwithstanding indeterminacy in outlier

cases.

I believe that one should, at a minimum, accept the follow lesson about reference from

Kripke: the reference of proper names is at least partially relational and cannot be purely

satisfactional. These terms are due to Kent Bach (1987, 12). On a purely satisfactional theory of

reference such as Descriptivism, reference via utterance of a name obtains in virtue of the agent

having accurate and uniquely identifying beliefs about the referent (though some subset of the

beliefs might be false on some variants of Descriptivism, such as the cluster theory). By

contrast, according to a purely relational theory of reference (as a purely Kripkean causal-

historical theory would be), reference is not a function of any of the agent’s beliefs about the

properties of the referent. On a purely relational theory of reference, an agent refers by uttering a

proper name without regard as to whether the agent has any beliefs whatsoever about the referent

and without regard to whether any beliefs he may have are true or false. Reference would go

147
through (i.e., not fail) whenever there is the right sort of causal-historical connection between the

referent and the speaker’s utterance, full stop. I take Kripke’s arguments against Descriptivism

to have made an excellent case for the proposition that reference is at least largely relational (i.e.,

not purely satisfactional). On the other hand, I believe that successfully referring requires the

agent to have some accurate beliefs about the referent (and be free of certain grossly inaccurate

beliefs). For example, a condition of successful reference would be for the speaker not to hold

wildly incorrect beliefs about the referent, especially when those beliefs concern its essential

properties. In no possible world is Albert Einstein a mountain in Switzerland or an abstract

object. A speaker who believed that ‘Albert Einstein’ referred to a mountain in Switzerland or

an abstract object would probably fail to refer altogether in uttering ‘Albert Einstein.’ However,

perhaps there is a nearby possible world in which Albert Einstein is female, such that a speaker

who erroneously associated the property of being a woman with ‘Albert Einstein’ would

successfully refer to Albert Einstein in uttering ‘Albert Einstein.’ The boundary between

essential and inessential properties (assuming the distinction is cogent) would be hard to draw,

and any such boundary might be fuzzy. Furthermore, there might be no specific set of descriptive

information that a speaker would need to associate with a name for reference to go through:

whether a speaker refers using a proper name might be highly context sensitive. Moreover, the

predicate ‘refers’ might be vague, such that there would be no sharp boundary between cases of

referring and non-referring, only matters of degree of reference success/failure.

I need not resolve these problems here. I am not developing a theory of reference in this

dissertation but rather a theory of content, which is potentially compatible with many various

theories of reference. In developing this theory of content, I will bracket problems with the

notion of reference and leave them for the theory of reference. I do not commit myself to any

148
theory or reference except insofar as I reject Descriptivism or any purely satisfactional theory of

reference. I presuppose a post-Descriptivist picture of reference that is at least partially relational

along Kripkean lines. Specifically, I make the reference of a proper name a function of the

subjecthood of the dossier from which the name was drawn, and the subjecthood is a (largely)

causal-historical property. If it turned out that environmental factors other than causal-historical

ones were partially determinative of reference, then these could be incorporated into the notion

of dossier subjecthood, and these factors would be contextual factors picked up by names used in

a Millian and Conception-indicating way. In other words, as our understanding of reference

changes and improves, the TIUT can simply incorporate that improved notion of reference into

its notion of dossier subjecthood (as long as that notion were sufficient for rigidity), without

requiring a re-working of the TIUT. 84

4.4 The Problem of Empty and Fictional Names

Another worry for Kripke’s causal-historical picture of reference is the problem of empty

and fictional names. It is not possible to be in causal contact with Santa Claus, since he does not

exist and only existent objects can enter into causal relations. What, according to Kripke, does

‘Santa Claus’ refer to when there no object whatsoever causally linked to the name? If ‘Santa

Claus’ and other empty or fictional names do not refer, then they contribute nothing to the

propositions expressed by the sentences in which they occur. It follows that sentences containing

empty or fictional proper names would fail to express complete propositions, resulting in truth-

value gaps.

The problem of empty and fictional is a problem for any Millian theory (even a Millian

84
If some of the speaker’s beliefs about the bearer were partially determinative of reference, this
would make proper names more like demonstratives, rather than pure indexicals.
149
theory on which the reference of proper names is not explained along the lines of Kripke’s

causal-historical picture). On Millianism, the meaning of a proper name is completely exhausted

by its bearer. A proper name without a bearer would be completely meaningless. On Millianism

the occurrence of the empty/non-referring name ‘Santa Claus’ in the sentence ‘Santa Claus does

not exist’ would seem to entail that the sentence expresses an incomplete proposition and lacks

truth-value. If ‘Santa Claus’ failed to refer, the sentence would express a proposition of the

form: < _____, non-existence > (‘Santa Claus’ has no referent, so it contributes nothing to the

proposition). But obviously—according to the reasoning of Russell—the sentence is both

meaningful and expresses a true proposition. After all, Santa Claus does not exist. (Frege, unlike

Russell, thought the sentence was meaningful, i.e., expressed a sense, but denied that it had a

truth-value). Furthermore, on Millianism, the sentence ‘Pegasus does not exist’ would also

express a proposition of the form: < _____, non-existence >. But clearly, the sentences ‘Santa

Claus does not exist’ and ‘Pegasus does not exist’ do not express the same proposition as

Millianism seems to entail. So, Frege and Russell believed, Millianism had to be false. Both

Frege and Russell developed versions of Descriptivism, which they thought would address the

problem. 85

Some philosophers, e.g., Jerrold Katz (1994), have defended Descriptivism by claiming

that any causal-historical theory of reference along Kripkean lines and/or any Millian theory of

proper names will offer no plausible answer to the problem of empty/fictional names. Like

Frege and Russell, Katz believed that Descriptivism (by contrast with Millianism) would solve

the problem of empty and fictional names. On Frege and Russell’s versions of Descriptivism, the

85
Braun (1993) claims that sentences such as ‘Santa Claus does not exist’ express “gappy
propositions,” propositions of the form: ______ does not exist. A difficulty with the view is that
‘Santa Claus does not exist’ and ‘Pegasus does not exist’ would express the same gappy
proposition. But intuitively, these sentences make different claims.

150
sentence ‘Santa Claus does not exist’ would express a complete proposition. On a Fregean

spirited version of Descriptivism, the sentence would mean something like ‘The toy-giving red-

suit wearing man living at the North Pole does not exist.’ Although the sentence lacks a truth-

value because the name lacks a referent (since the definite description whose meaning is

equivalent to the meaning of the name ‘Santa Claus’ fails to denote an object 86), a Fregean

spirited Descriptivist can at least explain why the sentence is meaningful: the description the toy-

giving red-suit wearing man living at the North Pole has a sense (Sinn); so the sentence ‘The

toy-giving red-suit wearing man living at the North Pole does not exist’ despite being neither

true nor false (Morris, 52), is at least meaningful and expresses a Fregean proposition. Russell’s

version of Descriptivism handles the problem of empty and fictional names somewhat more

neatly than Frege’s, for on Russell’s version there is no truth-value gap (Morris, 60-61). On

Russell’s version, ‘Santa Claus does not exist’ would mean the same (roughly) as the sentence

‘there is nothing that is a man, gives toys, wears a red suit, and lives at the North Pole.’

However, Descriptivism’s proposed solution to the problem of empty names (on either

Frege’s or Russell’s version) fails to solve the problem. To see why, suppose that there were, by

sheer coincidence, two legends that use the name ‘Santa Claus’ to talk about a toy-giving man

living at the North Pole. Two children, Jan and Joan, being only superficially conversant with

the details of the legends, associate identical descriptive information with ‘Santa Claus.’ Jan’s

use of ‘Santa Claus’ stems from hearing one legend, and Joan’s use stems from hearing the other

legend. Jan and Joan both utter: ‘I hope Santa Claus brings me lots of toys this year.’

Intuitively, Jan and Joan are talking about different things, expressing their gift-getting desires

vis-à-vis different mythical characters. However, on both Frege and Russell’s versions of

86
Here, I am assuming Frege’s theory to be a kind of Descriptivism. As noted previously, this is
controversial, but the issue is immaterial to my argument in this dissertation.
151
Descriptivism, Jan and Joan express the same proposition in uttering this sentence, for they

associate identical descriptions with the name ‘Santa Claus.’ Descriptivism predicts incorrectly

that Jan and Joan are speaking about receiving toys from the same mythical character, when they

are not. Consider the fictional detective Sherlock Holmes. We can refer to a character that is not

merely descriptively like Holmes, but to the specific character that appears in a certain literary

tradition/corpus linked to Conan Doyle. If another author, Jones, had coincidentally written a

novel about a brilliant detective named ‘Sherlock Holmes’ living on Baker Street in London, and

speaker A, familiar with the Jones character but ignorant of the Conan Doyle literary corpus used

the name ‘Sherlock Holmes’ in the sentence ‘I wish I had the crime-solving abilities of Sherlock

Holmes,’ intuitively he would not express the same proposition as another speaker, B, who

uttered the same sentence but who was familiar only with the Conan Doyle literary corpus. The

speakers would express different propositions even if the descriptive beliefs they associated with

the name ‘Sherlock Holmes’ were identical.

Kripke’s causal-historical picture of reference, or at least a modified version of it, would

have an advantage over Descriptivism in explaining some of the phenomena cited in the above

paragraph. We can adapt the causal-historical picture of reference to explain why A and speaker

B refer to different fictional characters when they utter the name ‘Sherlock Holmes.’ A and B

were in causal contact with different literary corpuses when the name ‘Sherlock Holmes’ entered

their respective lexicons—corpuses that can be causally traced back causal-historically to the

different writers who created the different fictional characters, and therefore, to the source of the

fictional characters themselves. In the case of Santa Claus, one might argue that the name ‘Santa

Claus’ refers to Santa Claus not because of a causal connection between a dubber and the actual

152
Santa Claus (there is no such flesh and blood person 87) but because the person uttering the name

‘Santa Claus’ is linked by a causal chain to the person who originated/created the fictional

work/myth and his act of creation and transmitting the myth to others either orally or in writing.

The initial baptism by the author/creator of the legend would proceed by a reference-fixing

definite description, which would refer to a fictional entity concretely instantiated in the token

brain state of its creator. Thus, the reference of Santa Claus would be causally grounded in a

particular brain state constituting the act of creation of the character. This is of course only a

sketch of a view, and it does nothing to resolve the question of the ontological status of fictional

characters.

In agreement with a growing body of literature on empty and fictional names (see Kroon

2011), I take the thesis that these so-called empty names fail to refer at all to be highly

implausible. Salmon (1998) and Soames (2002, 89-95) have argued forcefully against the notion

that (most) empty names lack referents. The intuitive and correct view, I believe, is that

fictional/mythical names refer to mythical or fictional characters/objects. ‘Santa Claus does not

exist’ means that Santa Claus is not real—that he is fictional/mythical. But Santa Claus is

something—he is a mythical/fictional character. Nevertheless, the metaphysical status of

fictional/mythical characters or objects is mysterious. Are they abstract objects, non-existent

concrete particulars, or some other sort of entity? If fictional characters are abstract objects, how

do our physical brains interact with them, and how can we know about them? This is a difficult

and thorny metaphysical issue that can be bracketed. I shall not delve further into this issue here.

Whatever the correct and complete theory of reference turns out to be, that theory can be

incorporated into the notion of dossier subjecthood, the precise nature of which we can leave

87
Here, ignore the fact that St. Nicholas was a real person vaguely connected to the fictional
character Santa Claus in the story told to children.
153
open.

4.5 Rigidified Descriptivism

Some Descriptivists (e.g., Braddon-Mitchell and Jackson, 1996, 70-74) have suggested

defusing Kripke’s Modal Argument by proposing that the meanings of proper names are

equivalent to the meanings of rigidified definite descriptions (or clusters thereof). They contend

that this modified version of Descriptivism addresses Kripke’s modal argument. Consider the

rigidified description ‘the actual superhero that protects Metropolis.’ This rigidified description,

as used in the actual world, designates Kent-Super in every possible world in which he exists,

even in those in which he is not the superhero protector of Metropolis. Hence, some

philosophers have remained unconvinced that Kripke’s modal argument has undermined

Descriptivism and maintain that the meaning of a proper name is equivalent to a rigidified

reference-fixing definite description.

However, rigidified descriptivism is still vulnerable to Kripke’s Epistemic and Semantic

arguments. To see why, consider sentence (a) and (b). (b) differs from (a) in that we have

substituted into (b) a rigidified definite description picking out Superman in place of the second

occurrence of ‘Superman’ in (a).

(a) If Superman exists, then Superman is Superman

(b) If Superman exists, then Superman is the actual superhero that protects
Metropolis.

Here, we see that (a) and (b) differ in epistemological status, with (a) expressing an a priori

proposition and (b) expressing an (intuitively) a posteriori proposition. The fact that (a) and (b)

intuitively differ in epistemological status shows by reductio that (a) and (b) cannot mean the

154
same thing. Thus, the rigidification of the definite description does not blunt the force of

Kripke’s epistemic argument against descriptivism.

Nor does it blunt the force Kripke’s semantic argument against Descriptivism, for rather

obvious reasons. Whether the descriptions are non-rigid or rigid, either way there will be cases in

which agents are ignorant of the descriptions constituting the meaning of a name, or associate

erroneous descriptions with a name, and yet, intuitively, still manage to refer when uttering that

name.

4.6 Causal Descriptivism

According to Causal Descriptivism, the meaning of an utterance of proper name ‘NN’ is

equivalent to the description ‘the individual/object that bears relation R to this utterance of

‘NN’’, where R represents the causal relation that fixes the reference of utterances of ‘NN’. 88

Causal Descriptivism is designed to be immune to the critique that Kripke directed at

Descriptivism via his Modal, Epistemic, and Semantic arguments, as described above. David

Lewis (1984), Frederick Kroon (1987), and Frank Jackson (1998), have proposed preliminary

sketches of Causal Descriptivist theories, though none of them has worked out the details

precisely to elaborate a full-fledged theory.

On Causal Descriptivism, the meaning of an utterance of ‘Clark Kent’ would be the

individual/object bearing relation R to this very utterance of ‘Clark Kent,’ and the meaning of

‘Superman’ would be the individual/object bearing relation R to this very utterance of

‘Superman.’ Causal Descriptivism seems to solve Frege’s puzzle about identity sentences

because (1) and (2) would express different propositions. Sentence (1):

88
This way of stating it is due to Lewis (1984).
155
 (1) Clark Kent is Clark Kent

would express the uninformative, tautological, and trivial proposition that the individual/object

bearing relation R to this very utterance ‘Clark Kent’ is the individual/object bearing relation R

this very utterance ‘Clark Kent.’ Sentence (2):

(2) Clark Kent is Superman

would express the informative and non-tautological proposition that the individual/object bearing

relation R to this very utterance ‘Clark Kent’ is the individual/object bearing relation R this very

utterance ‘Superman.’ Since the same individual, Kent-Super, is the individual bearing the

relation R to the speakers’ utterances of both names (since Kent-Super is causally connected to

both utterances in the right sort of way [although via different causal chains], as sketched under a

causal theory of reference in accord with Kripke’s causal-historical picture), on Causal

Descriptivism sentences (1) and (2) would express distinct propositions. Yet they would both be

true because in both (1) and (2) Kent-Super bears relation R to both utterances, and Kent-Super

is Kent-Super.

However, Causal Descriptivism lacks the resources to solve the puzzle about

propositional attitude ascriptions. Consider sentence ¬ (4) as uttered by Jimmy Olson to describe

Lois’ ignorance of the fact that Clark Kent is Superman.

¬ (4) Lois Lane disbelieves that Clark Kent is Superman

On Causal Descriptivism, (4) would mean the same as ¬ (4)CD:

¬ (4)CD Lois disbelieves that the individual/object bearing relation R to this very utterance

156
 ‘Clark Kent’ is the individual/object bearing relation R to this very utterance
‘Superman’

Problematically, ¬ (4)CD does not capture what Lois fails to realize about the identity of Clark

Kent and Superman. Suppose that Jimmy Olson uttered ¬ (4) outside of Lois’ presence. She

could therefore have no opinion whatsoever about the causes of Olson’s utterances of ‘Clark

Kent’ and ‘Superman.’ How could she? She did not hear Olson’s utterance, so how can she

have any beliefs regarding whether the utterance took place, what caused his utterances of the

names, or whether these utterances of the names bear the same relation R to their bearers.

Nevertheless, clearly ¬ (4) is true. We might sum this up problem for causal descriptivism by

saying that they render propositional attitude ascriptions ‘too autobiographical’: the ascriber’s

utterance of the ascription sentence, ostensibly about Lois’ beliefs, are really (at least in part)

about the ascriber himself and the causal relationship of his own utterances to the bearer of the

name. Causal Descriptivism appears especially implausible when the ascribee does not know that

the ascriber exists. For example, suppose Ralph, an ordinary citizen of the US unknown to

President Obama, utters “Obama believes ISIS is dangerous.” On Causal Descriptivism Ralph

has said that Obama believes that the cause of his (Ralph’s) utterance ‘ISIS’ is dangerous. The

ascription sentence is too autobiographical—it is about Ralph and the causal relationship

between Ralph’s utterance of the name ‘ISIS’ and ISIS. Obama has no belief about which

objects or individuals cause Ralph’s utterances nor what properties those objects or individuals

have. Obama has no idea who Ralph is or even that he exists. Yet, the ascription sentence Ralph

utters, “Obama believes that ISIS is dangerous,” is true nonetheless. This shows, by reductio,

that Causal Descriptivism is false.

Furthermore, Causal Descriptivism is too metalinguistic to be an adequate theory of

157
proper names. Consider Jennifer Saul’s (1998) critique of metalinguistic theories of proper

names involving a story in which a woman named Nicole never learns either the names ‘Clark

Kent’ or ‘Superman’ yet forms the opinion, after meeting Kent-Superman in both personas

(without realizing she had met the same individual), that Clark Kent is boring and drab, but

Superman is witty and urbane. On Causal Descriptivism, the true ascriptions sentence, as uttered

by Olson, “Lois believes that Superman is witty and urbane but Clark Kent is not,” would mean

the same as “Lois Lane believes that the cause of Olson’s utterance ‘Clark Kent’ is boring, but

that the cause of Olson’s utterance ‘Superman’ is witty and urbane.” Nicole cannot have any

beliefs about the causes of Olson’s utterances of ‘Superman’ and ‘Clark Kent’ because she does

not even recognize the names ‘Clark Kent’ or ‘Superman.’ She has never heard these names and

they mean nothing to her.

Devitt (1996) proposes a superficially similar theory (although his theory is not a species

of Descriptivism) on which the meaning of a proper name ‘NN’ that refers to object o is the

property of referring to o via a d-chain d, where d is a causal chain involving tokens of ‘NN’

causal-historically traceable back to the baptism of o with the name ‘NN.’ ‘Clark Kent’ and

‘Superman’ would have different meanings because they are causally connected to Kent-Super

by different d-chains. Devitt claims that speakers need not have beliefs (even tacit beliefs) about

these d-chains, so speakers need not (and in most cases will not) know the meanings of the

names they understand and use competently (and this is the prime reason why Devitt’s theory

must be classified as non-Descriptivist). Devitt thinks that the fierce resistance to the claim that

names have might have meanings unknown to the speakers who use them and are competent

with them is attributable to a “ubiquitous Cartesianism about meaning.” Devitt explains (2012):

“…. clearly, the name’s reference must be determined somehow, presumably causally in
something like the way that Kripke sketched. So the name must have some sort of causal

158
 mode of reference. So perhaps that mode is the meaning. This surely should be a
candidate for being the meaning, perhaps not the right candidate, but still a candidate. Yet
the idea that this causal mode is a meaning is clearly alien to the semantic tradition; it is,
as I have said in the title of a paper, “A Shocking Idea about Meaning” (2001). Why is it
so shocking? I think the main cause is a ubiquitous Cartesianism about meaning. It is a
truism that competent speakers of a language “know the language”. The Cartesian
assumption is that this involves (tacitly) knowing facts about meanings: if an expression
has a certain meaning in the language then speakers know that it does. Then, since the
typical speaker knows nothing about causal modes of reference, those modes cannot be
meanings. Yet this popular Cartesianism is almost entirely unsupported and is, I have
argued, undermined by the revolution. We should embrace the much more modest view
that linguistic competence is an ability or skill, a piece of knowledge-how not
knowledge-that.”

Devitt’s point with respect to Cartesianism about meaning is well taken. Speakers do not

need to know all the meaning facts to be competent users and understanders of language.

However, Devitt seems to elide over the real worry for his proposal. The difficulty is that if

speakers lack beliefs or knowledge about d-chains, his theory cannot account for the difference

in cognitive significance of ‘Clark Kent’ and ‘Superman.’ If ‘Clark Kent’ and ‘Superman’

meant different things but speakers were unaware aware of this difference, they would not take

the names to differ in meaning. Perhaps Devitt could claim that speakers tacitly know the

meanings, and this tacit knowledge explains the different cognitive significance of ‘Clark Kent’

and ‘Superman’. (However, I believe Devitt claims speakers have typically have no beliefs

whatsoever, tacit or non-tacit, about these meanings—they are simply not in the speaker’s head

at all). Another possible explanation of the difference in cognitive value between ‘Clark Kent’

and ‘Superman’ that Devitt might endorse (and I speculate) would be the speaker’s realization

that these names are syntactically different, which is evident even if the speaker has no

knowledge of or beliefs about d-chains. However, it is not clear if this proposal would work for

Paderewski-style cases, where two uses of ‘Paderewski’ are connected to the same man via

different d-chains but there are no syntactic differences. The details of Devitt’s account would

159
need to be fleshed out to better explain cognitive value differences.

4.7 Metalinguistic Descriptivism

Metalinguistic Descriptivism, in very rough sketch, claims that a proper name ‘NN’

means the bearer of ‘NN’. There are various formulations of the theory (e.g., Bach 2002, Katz

1997, Geurts 1994, et al.) that differ somewhat in the details. It is claimed that the theory can

explain the cognitive value difference between ‘Clark Kent is Clark Kent’ and ‘Clark Kent is

Superman’. The former sentence would say that the bearer of ‘Clark Kent’ is the bearer of

‘Clark Kent’—which is obvious and trivial, while the latter sentence would say that the bearer of

‘Clark Kent’ is the bearer of ‘Superman’, which is informative. In fact, Lois is ignorant of this

latter fact—she would deny that the bearer of ‘Clark Kent’ is the bearer of ‘Superman.’ So, it

seems, at first blush, that Metalinguistic Descriptivism is plausible and can solve Frege’s identity

sentence puzzle.

However, this theory is inadequate. Suppose that Lucy is a golden retriever who bases

her recognition of people on smell alone. Lucy realizes that Clark Kent is Superman, for he

smells the same to Lucy no matter how he is dresses or acts. So, the sentence ‘Lucy realizes that

Clark Kent is Superman’ is true. However, if Metalinguistic Descriptivism were true, this

sentence would claim that Lucy realizes that the bearer of ‘Clark Kent’ is the bearer of

‘Superman.’ This cannot be true. Lucy has no beliefs about names or who bears them. She is a

dog and does not use names nor understand names (save perhaps for reacting to her own name).

Her belief that Clark Kent is Superman is therefore not a belief about language. Lucy is

incapable of having such beliefs, but she can realize that Clark Kent is Superman. Hence, by

reductio, the names ‘Superman’ and ‘Clark Kent’ cannot mean the same as the bearer of

160
Superman and the bearer of Clark Kent, respectively. 89

We can also see that the metalinguistic theory is insufficient because of Jennifer Saul’s

argument (1998), cited in the preceding section. Saul asks us to imagine a woman named Nicole

who never learns either the names ‘Clark Kent’ or ‘Superman’ yet forms the opinion, after

meeting Kent-Superman in both personas (without realizing she had met the same individual),

that Clark Kent is boring and drab, but Superman is witty and urbane. On Metalinguistic

Descriptivism, the true ascriptions sentence ‘Lois believes that Superman is witty and urbane but

Clark Kent is not’, would mean the same as ‘Lois Lane believes that the bearer of “Clark Kent”

is boring, but that the bearer of “Superman” is witty and urbane’. However, Nicole cannot does

not have any beliefs about the names this individual bears because she never learned his names.

Yet, the ascription sentence is true, showing by reductio that Metalinguistic Descriptivism is

inadequate.

Finally, consider the fact that the sentence ‘The Maya discovered that Hesperus was

Phosphorus independently of the Babylonians and the Greeks’ is intuitively true. However, on

Metalinguistic Descriptivism, the sentence expresses the proposition that the ancient Maya

discovered that the bearer of ‘Hesperus’ was the bearer of ‘Phosphorus’ independently of the

Babylonians and the Greeks. This is clearly false. The ancient Maya knew no Greek. Similarly,

Attila, a monolingual speaker of Hungarian, may utter ‘Magyarország Európában’ and I may

report him, in English, as saying that Hungary is in Europe, even though Attila would not

89
This should call to mind Frege’s early metalinguistic view of proper names in der
Begriffsschrift, where he had claimed that ‘Hesperus is Phosphorus’ expresses a proposition
about the co-reference of these names, ‘Hesperus’ and ‘Phosphorus’, which both refer to the
planet Venus. In Über die Grundlagen der Arithmetik and On Sense and Reference, Frege
rejected this earlier view, arguing that the discovery that Hesperus was Phosphorus was an
astronomical discovery, not a linguistic discovery (i.e., it was not a discovery about the co-
reference of names).
161
understand the name ‘Hungary’ or the sentence ‘Hungary is in Europe.’ The proposition Attila

expressed was not that the bearer of ‘Hungary’ is in Europe.

In addition to Metalinguistic Description theories, there are two notable metalinguistic

versions of Millianism, the theories of Larson and Ludlow, and that of Mark Richard. According

to Braun (2002: 371), “these theories … claim that the content of a name consists partly of its

bearer and partly of the name itself.… Both theories say that agents bear attitudes towards

linguistically enhanced propositions, which are (roughly) amalgams of Russellian propositions

with words.” See Soames (2002) for a critique of these theories as being overly metalinguistic.

162
 CHAPTER 5

MILLIANISM

5.1 Introduction. The Millian Strategy and Some Worries

Millianism is the view that a proper name’s sole contribution to the proposition expressed

by a sentence in which it occurs is its referent. On most forms of Millianism, names are directly

referential, i.e., they are devoid of any sort of linguistic meaning whatsoever (such as, e.g.,

character); they are merely “tags” attached to their referents, as Ruth Barcan Marcus famously

put it (1961, 310). The bearer fully exhausts the meaning of a name. (Pelczar and Rainsbury’s

indexical view, discussed supra in Chapter 3, is a different sort of Millian view on which names

have character meaning and therefore the bearer does not fully exhaust the meaning of a name;

this section will deal exclusively with the sorts of Millianism on which the bearer exhausts the

meaning of a name). I will focus my discussion of Millianism in this chapter on the two most

influential accounts—that of Nathan Salmon (1986) and Scott Soames (2002).

Millians recognize that Millianism has counterintuitive implications with respect to

Frege’s puzzle (discussed at section 1.1., supra) and design their theories to explain away these

intuitions. They usually appeal to the semantics/pragmatics distinction to do so. In rough sketch,

most Millians, including Salmon and Soames, claim that ‘Clark Kent is Clark Kent’ and ‘Clark

Kent is Superman’ semantically express the same proposition, the trivial proposition that Kent-

Super is Kent-Super, but pragmatically communicate different propositions. They claim that the

propositional attitude ascription ‘Lois Lane does not believe that Clark Kent is Superman’ is

false (since the sentence just expresses the proposition that Lois does not believe that Kent-Super

is Kent-Super, and clearly, she realizes that), but it pragmatically communicates a true

163
proposition. Ordinary speakers do not normally distinguish between what a sentence

semantically expresses and what it pragmatically communicates, so that ordinary speakers

incorrectly judge that ‘Lois Lane does not believe that Clark Kent is Superman’ is true. They

base this judgment on the fact that the proposition pragmatically communicated is true. When

ordinary speakers intuit that ‘Clark Kent is Clark Kent’ expresses a different proposition from

‘Clark Kent is Superman’, they base their judgment on the fact that these sentences

pragmatically communicate different propositions, the former uninformative and the latter

informative.

With respect to the Problem of Inconsistent Rational Belief, most Millians appeal to

‘propositional guises’ to solve the problem (Both Salmon and Soames appeal to guises; although

Soames does not mention guises in his 2002 book, in response to critics he stated subsequently in

his 2006 that he thinks guises are an essential part of Millianism). A propositional guise is

analogous to a mask covering a face. You might not recognize a person you know if he or she is

wearing a mask. It would be incorrect to say that you do not recognize that person in general.

Rather, we would say that you recognize the person when he or she appears to you in certain

ways, but fail to recognize the person when they appear to you in other ways. Likewise, a

proposition wears a sort of mask, a guise (from “disguise”). A proposition may present itself in

certain ways (via certain sentences or ways of characterizing the proposition), but in other ways

that very proposition might be presented under a different guise such that you would not

recognize it as the very same proposition. You might rationally have different attitudes with

respect to that proposition depending on how it is presented to you. When presented under one

guise, you might believe it, while when it is presented under a different guise, you might

disbelieve it or suspend belief with respect to it, not realizing that you believe and disbelieve (or

164
suspend belief towards) the same proposition. The guise prevents you from realizing that your

propositional attitudes are inconsistent, because, without additional evidence, you cannot tell that

the one and the same proposition is hidden behind the different guises. Your inconsistent beliefs

are rational provided that you believe the inconsistent propositions under different propositional

guises that you do not realize are guises of the same proposition.

Both Salmon and Soames, who are proponents of the most influential Millian theories,

ascribe to some version of both claims. First, the claim that speakers confuse semantics and

pragmatics (to explain away our Fregean intuitions with respect to Frege’s puzzle). Second,

propositional guises to solve the Problem of Rational Inconsistent Belief (although Soames says

little about guises). I will not attempt to offer a knockdown refutation either of these theories,

but limit myself to pointing out my four principal worries about Millianism (which do not effect

Salmon and Soames’ account in the same way or to the same extent), listed (a)-(d) as follows:

Four worries about Millianism

(a) The Guise Definition Problem. This is the failure to specify: (a) what propositional

guises are, and (b) the exact nature of their relation to singular propositions they disguise. I

consider what propositional guises could be in section 5.3.

(b) The Pragmatic Mechanism Problem. This is the failure to specify the pragmatic

mechanisms by which speakers semantically expressing singular propositions are supposed to

convey further propositions to their audience, which propositions are supposed to explain our

Fregean intuitions. This affects Salmon’s theory more than Soames’.

(c) The Ignorance of Identities Problem. This is the failure to state with specificity
165
what proposition Lois Lane is ignorant with respect to the identity of Clark Kent and Superman.

We want our theory of proper names to identify the proposition of which Lois lane is ignorant of

with respect to Kent’s identity with Superman. What proposition does she fail to realize?

Ideally, we would want a theory according to which the proposition Lois fails to realize is the

one literally expressed ‘Clark Kent is Superman.’ After all, when we say that ‘Lois Lane does

not realize that Clark Kent is Superman,’ that identity sentence is embedded in the ‘that’-clause

of this ascription. But even if a theory claimed that the proposition that Lois fails to realize was

not the one semantically expressed by that identity sentence, we should insist that the proposition

be articulable in the theory. However, Millians have difficulty stating what Lois Lane fails to

realize about the identity, as Thomas McKay and the Michael Nelson, themselves Millians,

recognize in their entry ‘Ignorance of Identities’ in the Stanford Encyclopedia of Philosophy.

https://plato.stanford.edu/entries/prop-attitude-reports/ignorance.html. Millians would not say

that she fails to realize the singular proposition that Kent-Super is Kent-Super. Clearly, she

realizes that. But they also cannot plausibly say that she fails to realize a purely descriptive

proposition, such the proposition that the mild-mannered reporter from Smallville working for

the Daily Planet is the superhero that protects Metropolis. The proposition Lois fails to realize is

object-dependent, a proposition that is essentially about Kent-Super, so a purely descriptive

proposition is insufficient to capture what she is ignorant of. 90 By contrast with Millianism, the

90
That a purely descriptive proposition cannot be the proposition of which Lois’ ignorant is
made evident by a thought experiment, a variation of the Superman story, in which a pair of
indistinguishable identical twins from Krypton, Kal-El and Xal-El, both of whom work at the
Daily Planet in shifts as reporters going by the name ‘Clark Kent’. However, only Kal-El acts as
the superhero known as ‘Superman’. Xal-El prefers to keep a low profile. Suppose Lois has only
interacted with Kal-El while working at the Daily Planet. Her belief that Clark Kent and
Superman are distinct people is a belief about Kal-El, and not about Xal-El. It is not a belief that
Kal-El and/or Xal-El is not identical to Superman. Yet her descriptive beliefs may be insufficient
to distinguish between Kal-El and Xal-El. Her beliefs are causally grounded in the individual
166
TIUT offers an answer to this question, as set out in section 2.4-2.5 supra. The TIUT’s answer

combines Lois’ descriptive conception with a causal grounding of her belief in the person her

belief is about, via the subjecthood of her ‘Clark Kent’ and ‘Superman’ dossiers.

(d) The No Direct Expressibility Problem. Millians differentiate the semantic

proposition expressed from the pragmatic proposition communicated, both with respect to

identity sentences and propositional attitude ascriptions. However, they do not provide that there

is any idiomatic linguistic convention we may resort to in order to semantically express directly

the pragmatically communicated proposition, and this is rather suspect.

There is no doubt that we can and often do express one proposition that pragmatically

conveys another. You can utter ‘I just ate’ to implicate the proposition I’m not hungry, utter ‘I

enjoy a cigarette after dinner’ to communicate the proposition I enjoy smoking a cigarette after

dinner, or utter ‘You’re not going to die’ to communicate the proposition You’re not going to die

from that cut. In these cases, it is always possible to utter a sentence that semantically expresses

the pragmatically communicated proposition directly if one so chose. The use of these pragmatic

mechanisms is optional. Instead of saying ‘I just ate’ to communicate the proposition I’m not

hungry, you could utter “I am not hungry” to express directly the latter proposition, instead of

implicating it. Instead of uttering, ‘I enjoy a cigarette after dinner’ to communicate I enjoy

smoking a cigarette after dinner, you could utter “I enjoy smoking a cigarette after dinner.”

Instead of uttering ‘You’re not going to die’ to communicate the proposition You’re not going to

die from that cut, you could utter “You’re not going to die from that cut.” But this is not so in

the Frege’s puzzle cases: with propositional attitude reports such as ‘Lois Lane does not believe

about whom she has them, Kal-El, and a purely descriptive belief lacks that grounding.
167
that Superman can fly,’ there is no idiomatic way to semantically express directly the proposition

that Millians claim is pragmatically communicated. Perhaps there is a complex roundabout way

to express these propositions in the sophisticated language of Millians (it’s not clear what it

would be, since Millians have trouble saying what Lois Lane fails to realize about the identity,

but perhaps Millians could come up with something), but the point is that there is no ordinary

idiomatic way to express this proposition in ordinary, non-theoretical English. Normally, when a

speaker relies on some pragmatic mechanisms to express a proposition other than the one

literally expressed, the speaker always has the option of stating more directly what he means

without relying on that pragmatic mechanism. This is not the case in the Puzzle cases, and this

strongly suggests that in the puzzle cases the sentences already semantically express the

propositions we intuit they do. There is no more direct way of stating what these sentences say;

using these sentences is essential and unavoidable.

5.2 Salmon’s theory

As a Millian, Salmon claims that the content of a proper name is simply its bearer and

that a name contributes its bearer to the proposition expressed by the sentence in which it occurs.

Hence, in the sentence ‘Superman flies’, as well as in the sentence ‘Clark Kent flies’,

‘Superman’ and ‘Clark Kent’ both contribute Kent-Super, the man himself, to the proposition

expressed by the sentences. The singular proposition expressed by these sentences has two

constituents: Kent-Super, and the property of flying. This proposition can be schematized as

< Superman, ability to fly >

Believing a proposition, such as, e.g., that Superman flies, is to have a certain psychological

168
attitude toward it. According to Salmon, an agent can have different attitudes towards one and

the same proposition depending on how one “grasps” or “takes” it. Lois Lane, for example, has

different attitudes towards < Superman, ability to fly > depending on how she grasps it. When

she grasps it via “Clark Kent can fly”, she does not accept the sentence and does not inwardly

assent to the proposition, but when she grasps it via “Superman can fly”, she accepts the sentence

and internally nods assent to the proposition < Superman, ability to fly >. Grasping a

proposition “under a guise” is to grasp a proposition when it is presented in a certain way. Lois

Lane grasps < Superman, ability to fly > under two different guises, one guise associated with

the sentence ‘Superman can fly’ (in which case she accepts the sentence and inwardly nods

assent at the proposition < Superman, ability to fly >), and the other associated with the sentence

“Clark Kent can fly” (in which case she does not accept the sentence and does not inwardly

assent to the proposition < Superman, ability to fly >).

For Salmon, belief is a binary relation between an agent and a singular proposition.

There is also a ternary relation that Salmon calls “BEL”. It is the relation between an agent, a

singular proposition, and a guise (i.e., a way of taking/grasping that proposition). An agent

BELs a proposition p under guise g just in case she inwards assents to p when she grasps p under

g. Belief ascriptions of the form α believes that p are to be construed as existential

generalizations on a ternary BEL relation:

BEL RELATION ∃g [α grasps that p under g, & BEL(α, that p, g)]

where a represents the agent, g is a variable ranging over propositional guises, and p the singular

proposition believed by a. In plain English, this means, again, that an agent a believing a

proposition p involves a grasping p under guise g and inwardly assenting to p under g. Since

169
Lois inwardly assents to <Superman, ability to fly> under the guise associated with “Superman

flies,” Lois believes that proposition. The fact that she withholds inward assent (or dissents)

when that proposition is presented under the guise associated with “Clark Kent flies” does not

negate the fact that she believes the proposition. For her to believe it, it is sufficient that there be

at least one guise under which she inwardly assents to the proposition. Lois fails to realize that

the same proposition, <Kent-Super, ability to fly> is presented twice over under different guises

via sentences ‘Clark Kent flies’ and ‘Superman flies.’ In Salmon’s terminology (1989, 261),

Lois Lane suffers from “propositional recognition failure’ because she fails to recognize a

proposition when presented under one propositional guise (the one associated with ‘Clark Kent

flies’) as a proposition she already believes but under a different propositional guise (the one

associated with ‘Superman flies’). According to Salmon, Lois is not irrational as long as her

differing behavior with respect to the same proposition, sometimes nodding assent and sometime

withholding assent, occurs when she takes the proposition under different guises such that she

does not realize that these are guises of the same proposition. Voilà, we have a solution to the

Problem of Rational Inconsistent Belief.

Thus far I have discussed how Salmon solves the problem of rational inconsistent belief,

dealing with the case of the unenlightened Lois—how and why she can believe inconsistent

proposition vis-à-vis Kent-Super. Now let’s turn to Salmon’s proposal to solve Frege’s puzzle

about propositional attitude ascriptions. Why, according to Salmon, are we as enlightened

speakers subject to (what Salmon takes to be) an illusion of a difference in truth-value between

(5) and (6)?

(5) Lois Lane believes that Clark Kent flies

(6) Lois Lane believes that Superman flies

170
Here is what Salmon says about this. (5) and (6) are both true, since Lois inwardly nods assent

to the proposition expressed by ‘Superman flies’ when that proposition presented in a way like

‘Superman flies’, so she believes the proposition and (6) is therefore true. Since (5) expresses

the same proposition as (6), (5) is true as well. So why do we strongly intuit that (5) is false?

Because, according to Salmon, in addition to the proposition (5) and (6) semantically express, a

speaker would typically utter (5) and (6) to pragmatically “communicate” distinct propositions.

These pragmatically communicated propositions would really differ in truth value, with (5) being

false and (6) true. Ordinary speakers do not normally distinguish between the propositions

semantically (or literally) expressed by the sentences they utter and the propositions

pragmatically communicated. Thus, these speakers judge, in error, that the semantic truth

conditions of the proposition(s) expressed by the (5) and (6) differ because the truth conditions

of the communicated propositions differ. What is, according to Salmon, the pragmatic

mechanism by which these further propositions are communicated? Salmon suggested in

Frege’s Puzzle (1986) that the mechanism could be Gricean implicature, and subsequent authors,

e.g., Recanati, have referred to Salmon’s theory as the “Implicature Theory.” Recanati (1993)

and other authors have claimed that implicature is an implausible candidate for the pragmatic

mechanism. (I examine the prospect of implicature as a possible mechanism for Salmon’s theory

in section 5.4 and I show it to be implausible because the alleged implicatures are neither

cancellable nor detachable). In subsequent years Salmon has stated that he is not committed to

the pragmatic mechanism being implicature and has left the matter open. Not taking a stand on

which pragmatic mechanism is involved detracts from the value of Salmon’s theory, given how

central the issue is the theory.

171
 Although Salmon tends to rarely address the identity sentences version of Frege’s puzzle,

we may presume that guises explain the difference in cognitive value between ‘Clark Kent is

Clark Kent’ and ‘Clark Kent is Superman’ because an agent such as Lois Lane will take the

proposition expressed under different guises when presented by these sentences. These sentences

present the proposition Kent-Super is Kent-Super under different guises. Salmon does not

clearly say what propositional guises are. He suggests that a propositional guise is to a

proposition as a mask is to a face. When a person wears a mask, we may fail to recognize a

person we would ordinarily recognize without the mask. Likewise, we when a proposition wears

a certain guise, we may fail to recognize it as a proposition we believe under some other guise.

In the case of a person wearing a mask covering their face, the relation between the mask and the

face is clear — it is the “wearing” relation. The mask covers the face. But what exactly is the

relation between a proposition guise and the proposition it masks? Salmon says we take

proposition “under” guises, but what exactly is it to take a proposition under a guise? I do not

want to claim that there is no hope for making this highly metaphorical picture more precise,

only pointing out that the picture is very metaphorical.

In section 5.3 I explore the issue of propositional guises. What are they? And what the

relation between them and the singular propositions they disguise? In section 5.4 I explore the

possibility that pragmatic implicature could be the pragmatic mechanism at work (by which the

propositions semantically express pragmatically convey further propositions), and I show that

this is implausible. If Salmon’s theory is to work, some other pragmatic mechanism other than

implicature should be proposed. In section 5.5, I state how I believe Salmon’s theory fares on

the four problems (a-d), which I identify as my principal worries for Millianism.

172
5.3 What are Propositional Guises?

Salmon states in his 1986 that it is difficult to say what propositional guises are. He has

tended to be vague about the nature of propositional guises since that time. In his 1986, he

suggested (126) that propositional guises may be Fregean senses, ‘ways of taking’ a proposition,

a ‘mode of presentation’, or a ‘mental file.’ Salmon writes in his 1989b that he has little by way

of specifics to say about propositional guises except to note that

“[t]he important thing is that, by definition, they are such that if a fully rational believer
adopts conflicting attitudes (such as belief and disbelief, or belief and suspension of
judgment) toward propositions p and q, then the believer must take p and q in different
ways, by means of different guises, in harboring the conflicting attitudes toward them-
even if p and q are in fact the same proposition.” (246)

According to Heimir Geirrson (2012, 58), Salmon’s view regarding guises is best

captured by saying that Salmon hold that they are “sentence-like entities.” On Salmon’s view,

sentence (5)

(5) Lois Lane believes that Clark Kent flies

would say that Lois Lane believes the singular proposition that Kent-Super flies under a

propositional guise like (in some specific way to be eventually more precisely specified) the

sentence ‘Clark Kent flies.’ Sentence (6),

(6) Lois Lane believes that Superman flies

would say that Lois Lane believes the singular proposition that Kent-Super flies under a

propositional guise like (in some specific way to be eventually more precisely specified) the

sentence ‘Superman flies.’ The theory owes us an account of what precisely propositional guises

173
are and exactly how they resemble these sentences. In what way does the guise under which the

‘that’-clause of (5) is presented more closely resemble the sentence ‘Clark Kent flies’ rather than

‘Superman flies’?

5.3.1 Guises as Descriptive Propositions

The simplest proposal is that guises are descriptive propositions. Salmon suggests in

passing in his 1986 that guises might be like Fregean senses or modes of presentation. If

Fregean senses were construed as descriptive modes of presentation, definite descriptions

associated with proper names, or what I call ‘conceptions,’ then propositional guises would be

descriptive propositions of the sort expressed by definite descriptions. Presumably, these

descriptive propositions would be a compositional function of the descriptive modes of

presentation attached to all the expressions within a sentence. 91 For example, the propositional

guise of the singular proposition, PROP-1, when expressed by sentence (2):

(2) Clark Kent is Superman

would be composed of the descriptive modes of presentation/conceptions associated by the

speaker with the names ‘Clark Kent’ and ‘Superman.’ Hence, the propositional guise of PROP-

1, when presented by sentence (2), would roughly be the descriptive proposition that

Descriptivists would claim is its content. It would be a descriptive proposition such as: the mild-

mannered reporter for the Daily Planet is the caped superhero protecting Metropolis.

On this view, propositional guises are descriptive propositions of the sort expressed by

91
It’s not clear whether Salmon would agree with this. But it’s hard to believe that it could be
otherwise, if propositional guises are descriptive propositions. This seems to follow from the
principle of compositionality.

174
definite descriptions. The speaker would have transparent introspective access to these

propositions—these are the propositions he ‘sees’ before his mind’s eye when he thinks to

himself that Clark Kent is Superman or utters ‘Clark Kent is Superman.’ But the content of the

proposition he entertains and expresses would be the singular proposition PROP-1—that Kent-

Super is Kent-Super. 92 The speaker’s descriptive conceptions might in fact misrepresent the

properties of Kent-Super or fail to denote him uniquely. A speaker such as Lois Lane fails to

realize that (1) and (2) express the same proposition because the contribution/content of the

proper names depends on externalist relational properties (e.g., causal-historical relations) that

are ‘outside the head’ of the speaker, to which s/he has no immediate cognitive access. The

speaker has immediate transparent cognitive access only to the descriptive propositions built

from the descriptive properties, which s/he associates with the names and which s/he takes to

apply to the referents of the names.

On this theory of propositional guises, a speaker might rationally believe the inconsistent

singular propositions that Kent-Super flies and that Kent-Super does not fly because the

descriptive propositions s/he believes—the descriptive propositional guises under which s/he

believes the singular proposition—are consistent. The proposition that the mild-mannered

reporter for the Daily Planet does not fly is consistent with the proposition that the strong caped

superhero protecting Metropolis flies. In this case, the speaker is rational provided that s/he fails

92
To be clear: the content of the proposition expressed by ‘Clark Kent is Superman,’ according
to Millianism, is this singular proposition PROP-1, not because the conceptions associated with
‘Clark Kent’ and ‘Superman’ determine this proposition (not because the definite descriptions
associated with names ‘Clark Kent’ and ‘Superman’ denote Kent-Super twice over, as would be
the case on Descriptivism), but rather because the names both refer to Kent-Super in virtue of the
causal-historical links between the both names and Kent-Super. I do not mean to imply that
Millianism entails acceptance of the causal-historical theory of reference. But I think that it does
entail rejection of a purely satisfactional theory of reference, such as Descriptivism.

175
to realize that the definite descriptions the mild-mannered reporter for the Daily Planet and the

strong caped superhero protecting Metropolis denote the same individual.

5.3.2 Guises as Metalinguistic Propositions

Another proposal is that propositional guises are metalinguistic propositions. On this

proposal, the propositional guise associated with (2) is the metalinguistic proposition: that the

referent of ‘Kent’ is identical to the referent of ‘Superman.’ By contrast, the propositional guise

associated with (1) would be the metalinguistic proposition: that the referent of ‘Clark Kent’ is

identical to the referent of ‘Clark Kent.’ On this theory, Lois Lane assents to (1) because she

realizes the obvious fact that ‘Clark Kent’ and ‘Clark Kent’ co-refer but she dissents from (2)

because she thinks that ‘Clark Kent’ and ‘Superman’ do not co-refer.

This proposal does not work, inter alia, because names can have multiple bearers.

Suppose that Lois Lane knows that Perry White calls his dog by two names: ‘Clark Kent’ and

‘Superman.’ Lois knows the two names of Perry’s dog, so she believes that ‘Clark Kent’ and

‘Superman’ sometimes co-refer. To explain why Lois dissents from certain utterances of (2), we

must say something like: she dissents because she takes the utterer to be tokening ‘Clark Kent’ to

refer to a person named ‘Clark Kent,’ who is her mild-mannered reporter colleague (and not to

refer to a dog) and to be tokening ‘Superman’ to refer to a superhero (and not to a dog).

Furthermore, she believes that the name ‘Clark Kent,’ when used to refer to her mild-mannered

reporter colleague, is not co-referential with the name ‘Superman,’ when used to refer to a

superhero. We must bring in various descriptions that Lois associates with the names to explain

her linguistic behavior when dissenting from a particular utterance of (2). Her meta-linguistic

beliefs alone are insufficient to explain her patterns of linguistic behavior. Thus, the

176
metalinguistic guise theory is inadequate. Lois’ metalinguistic beliefs would have to be

combined with her descriptive beliefs to explain her linguistic behavior and account for her

rationality despite inconsistency.

5.3.3 Guises as Natural Language Sentences

On this view, guises are not propositions, but sentences. Since guises are not

propositions, the psychological relation between agents and the guises is not one of belief, but

rather one of acceptance. An agent believes proposition P via accepting sentence S that

expresses P. A rational agent might accept S1, but not S2, even if S1 and S2 express the same

proposition. A proponent of this view posits that believing a proposition consists in assenting

towards the proposition by accepting a sentence that expresses that proposition. Disbelieving a

proposition consists in dissenting from a sentence expressing a proposition.

This view is idle unless it explains why the agent accepts one sentence, but dissents from

another expressing the same proposition. We need an account of the speaker’s mental state when

entertaining the proposition via various sentences and an explanation of why this causes the

agent to inwardly assent to the proposition when presented via one sentence and to dissent or

suspend judgment from it when presented by a different sentence. How do the natural language

sentences coordinate with internal representations or internal belief states? Why/how do these

internal representations play different causal and functional roles in the agent’s inferential and

behavioral economy? Short of such an account, the view of guises as natural language sentences

is not an explanation of the underlying phenomenon but a mere restatement of the very problem.

Furthermore, as David Braun (1998) and Nathan Salmon (1993b, 87-88) have pointed

out, Kripke’s Paderewski Puzzle (1979) forecloses the possibility that guises could be natural

177
language sentences. For in the Paderewski Puzzle, Peter alternately assents to and dissents from

a proposition when it is presented by one single sentence, ‘Paderewski had musical talent.’ To

explain why Peter sometimes assents and sometimes dissents, we need a theory of guises that

delivers two different guises. Here, we would have only one single guise, the natural-language

sentence ‘Paderewski had musical talent.’

Finally, the view of propositional guises as sentences is insensitive to the fact that

instances of Frege’s Puzzle about identity sentences can be generated under circumstances where

neither sentence (1) nor sentence (2) is uttered. Suppose the enlightened Jimmy Olson points

first at a photo of Clark Kent wearing business attire and glasses and then points to a photo of

Superman in his spandex outfit and cape, and he utters “those are pictures of the same guy.’’

Lois would dissent from Olson’s utterance. If Olson points at a photo of Clark Kent wearing

business attire and glasses and utters “that guy is identical to himself,’’ Lois would assent. Or

suppose that Lois is a deaf-mute who communicates in sign language. Lois would have sign-

language terms corresponding to ‘Clark Kent’ and another sign-language term corresponding to

‘Superman.’ She would dissent from the sign-language translation of the English sentence (2)—

‘Clark Kent is Superman,’ but not from sign language translation of the English sentence (1).

Clearly, she dissents in all three cases for the same reason. The reason for her dissent in these

alternate scenarios is not a function of the specific sentence Olson utters—not all these cases

involve the utterance of either sentence (1) or (2). The key to understanding Lois’ patterns of

assent and dissent in all three cases has little to do with the specific structure of sentence (2) or

the specific words used, but her patterns are rather to be explained by some common feature

between the utterance of (2) and the two alternate scenarios. It is therefore implausible to claim

that propositional guises are specific sentences. It would be far more plausible, in one’s search

178
for the correct account of propositional guises, to identify the property that these three speech

acts—the utterance of (2), the utterance of its sign language counterpart, or the case of pointing

to photos depicting Kent-Super in different attire accompanied by the utterance “those are the

same guy” or “he is he”—have in common with one another, such that the unenlightened Lois

Lane is disposed to dissent from them all.

5.3.4 Propositional Guises as Belief States

The view that propositional guises are ‘belief states’ has its metaphysical roots in the

writings of John Perry and David Kaplan in their study of indexicals. Perry and Kaplan noticed

that a speaker could entertain the same proposition in two different ways and thus come to

believe and disbelieve the same proposition. To adapt Kaplan’s example (1989, 533), an agent

looks into a mirror and sees the reflection of a guy whose pants are on fire. He does not realize

that he is looking into a mirror and does not realize that he is the guy whose pants are on fire. He

might utter “that guy’s pants are on fire. Thank God my pants are not on fire.” Here, according

to Kaplan, the agent has expressed inconsistent singular propositions. The singular proposition

in question is the one containing as constituents: the agent himself, and the property having one’s

pants on fire. The agent has said that this proposition is true (by saying “that’s guy’s pants are

on fire” and pointing to himself without realizing it), and he has also said it is false (by uttering

“my pants are not on fire”). Perry and Kaplan conclude that a rational agent may believe

inconsistent singular propositions provided that s/he is in different belief states when doing so

and does not realize that being in these belief states constitute believing the same proposition.

Analogously, Lois Lane is in different belief states when she entertains sentence (1) and

179
(2). 93 She can believe the proposition that Kent-Super is Kent-Super when presented one way

and can disbelieve it when presented another way. Indeed, she remains rational because she is in

different belief states when she believes it and disbelieves it. She intuits that the two belief states

are different from one another, since the belief states play different inferential, functional, and

causal roles in her mental life and they manifest themselves behaviorally in very different ways.

Due to the very distinct causal and inferential roles that the belief states play in her mental

economy, she erroneously intuits that the propositions she believes when she is each of those

belief states are distinct, when in fact, being in either one of those belief states constitutes

believing PROP-1.

This proposal is lacking in specifics. There is no doubt that Lois is in different belief

states when she considers sentences (1) and (2) and whether they are true or false. Until the

details are filled in, this account of guises is merely a promissory note. One needs an account of

the exact relation between the belief state and the propositions believed—why does being in the

belief state constitute believing the proposition(s)? What are belief states? In addition, how and

why are the belief states correlated with certain sentences? Only once this account is fleshed out

will it be fully explanatory rather than a mere explanatory sketch or picture. (The TIUT fleshes

out this picture in chapter 2, supra, explaining Lois’ different belief states in terms of proper

names as indexicals whose characters refer to distinct dossiers in Lois mental architecture).

93
Belief states must be distinguished from beliefs. Two agents in very different beliefs states
might believe the same singular proposition in virtue of being in these different belief states.
And a rational agent such as Lois Lane might believe and disbelieve the same proposition when
in different belief states when considering that proposition, i.e., she might assent towards that
proposition when in one belief state and dissent from it when in another.
180
5.3.5 Guises as Sentence-like Entities

As stated above, for Salmon belief is a binary relation between an agent and a singular

proposition, but there is another relation, which Salmon terms ‘BEL’, which is a three-part

relation between an agent, a singular proposition, and a propositional guise under which the

agent takes the proposition. A propositional attitude ascription is an existential generalization of

the BEL relation. Hence, the ascription

(5) Lois Lane believes that Clark Kent flies

says that there exists a propositional guise similar or analogous to (in a relevant way to be

specified) the natural language sentence ‘Clark Kent flies,’ under which Lois Lane grasps the

singular proposition that Kent-Super flies, and she inwardly assents to that proposition when

grasping under that guise. The ascription sentence (6)

(6) Lois Lane believes that Superman flies

says that there exists a sentence-like propositional guise similar or analogous to (in a relevant

way to be specified) the natural language sentence ‘Superman flies,’ under which Lois Lane

grasps the singular proposition that Kent-Super flies and under which she inwardly assents to it.

Attitude ascriptions (5) and (6) literally express the proposition that Lois believes the singular

proposition that Kent-Super flies. However, (5) and (6) do not say that Lois BELs the same

thing, for the third relatum in the ternary BEL relation (unlike the binary belief relation, which is

the relation between an agent and a singular proposition) differs as between (5) and (6). The

sentence-like propositional guises linked to sentences (5) and (6) differ.

Salmon maintains that ascription sentences such as (5) and (6) would pragmatically

181
implicate information about the propositional guises under which Lois Lane, the ascribee,

believes the singular proposition that Kent-Super flies. (5) would implicate that Lois Lane

believes that Kent-Super flies under a sentence-like propositional guise like ‘Clark Kent flies.’

(6) would implicate that Lois Lane believes that Kent-Super flies under a sentence-like

propositional guise like ‘Superman flies.’ According to Salmon, ordinary speakers (with non-

Millian intuitions) intuit what the truth-value of the propositions expressed by (5) and (6) is

based on what these sentences say Lois BELs, rather than what they say she believes. They

intuit that (5) is false because Lois does not believe the singular proposition that Kent-Super flies

under a propositional guise like ‘Clark Kent flies.’ So ordinary speakers tend to say that (5)

expresses a false proposition, which in fact, according to Salmon, expresses a true proposition

about the singular proposition Lois believes. This is more of a picture of a theory rather than a

full-fledged theory. Guises cannot be sentences. In what way do they resemble them?

5.3.6 Guises as Theoretical Misconceptions

Salmon has suggested that a misconception or a “misconceived way of taking a

proposition” can serve as a propositional guise (see final paragraph of his 2006). This was

touched on in section 2.8, supra. The enlightened Jimmy Olson is disposed to utter both (3) and

(4n).

(3) Lois Lane believes that Clark Kent is Clark Kent

(4n) Lois Lane does not believe that Clark Kent is Superman

On Millianism, (3) and (4n) express inconsistent propositions—the proposition that Lois believes

PROP-1 and the proposition that Lois fails to believe PROP-1. Millianism gives the verdict that

182
in uttering (3) and (4n), Olson is inconsistent. Generally, proponents of the Guise Millians

Approach such as Salmon want to explain rational inconsistency (why speakers who contradict

themselves can nevertheless be rational) by invoking propositional guises. To explain how

Olson can express/believe inconsistent propositions and be nevertheless rational, a proponent of

the Guise Millian Approach will want to say that Olson takes the propositions expressed by (3)

and (4) under different propositional guises that he does not realize are guises of the same

proposition (and therefore does not realize that (3) and (4n) express inconsistent propositions).

However, unlike the unenlightened Lois Lane, who fails to realize Clark Kent’s identity with

Superman, Olson is not ignorant of the identity. He knows all the facts about who Clark Kent

and Superman are—that they are the same person. Hence, the difference between the

propositional guises under which Olson takes the propositions expressed by (3) and (4) cannot be

explained by, or rooted in, Olson’s failure to realize the identity, as it is for Lois. The

explanation of Olson’s rational inconsistency, and the reason for his taking the (allegedly)

inconsistent propositions under different guises, must proceed along different lines for the

enlightened Olson as it does for the unenlightened Lois Lane. In his 2006, Salmon posits that the

propositional guises under which Olson would take the propositions expressed by (3) and (4) are

rooted in or explained by a “misconception”—to wit, the theoretical misconception that the

proposition expressed by both (3) and (4) is two different propositions (2006, final paragraph).

Olson takes a single proposition under two different propositional guises, and he does not realize

that these are propositional guises of the same proposition, despite knowing all the facts about

who Clark Kent and Superman are and fully understanding the nature of Lois’ confusion. His

failure to realize that the guises are guises of the same proposition arises not from his ignorance

of the facts about identity—about who Clark Kent and Superman are, as they do for Lois Lane

183
when she believes inconsistent propositions about Clark Kent/Superman—but rather his failure

arises from Olson’s erroneous belief that Millianism is false. Olson is in error about facts—but

they are not facts about who or what Lois Lane, Clark Kent, or Superman are—but about

theoretical facts concerning the correct theory of propositional attitude ascriptions and the

content of the propositions semantically expressed by sentences (3) and (4).

The sorts of propositional guises that are proposed to explain why the unenlightened Lois

can believe the inconsistent propositions expressed by (1) and ¬ (2) and (7) and ¬ (8)

(1) Clark Kent is Clark Kent

¬ (2) Clark Kent is not Superman

(8) Superman flies

¬ (7) Clark Kent does not fly

are rooted in her ignorance concerning the properties of Kent/Superman and who he is. But now

Salmon claims that the propositional guises that explain the enlightened Jimmy Olson’s beliefs

about the ascription sentences he utters are of a very different sort. They are rooted in theoretical

confusions. Propositional guises are a heterogeneous class of entities. I think this heterogeneity

renders the notion of guises substantially less plausible, although I would not claim it refutes

Salmon’s theory. 94

94
Although the TIUT appeals to a notion similar to propositional guises (propositional guises
explained via the notion that names are indexicals) to explain the inconsistent beliefs of
unenlightened agents such as Lois Lane (see section 2.8), it does not require any such notion to
explain the utterances of an enlightened ascriber such as Jimmy Olson. On the TIUT ascription
sentences (3) and (4) express different propositions (see section 2.6.1). Hence, Olson’s utterance
of (3) and (4n) is consistent. Olson does not contradict himself, so there is no need to posit
propositional guises to save his rationality.

184
5.4 Could Implicature be the pragmatic mechanism in Salmon’s theory?

For Salmon’s theory to be a full-fledged theory, he must identify the pragmatic principle

that explains how speakers use sentences to communicate/convey propositions other than the

propositions that the sentences semantically express. Salmon has stated that the mechanism

could be implicature. I will focus my criticism of this proposal on the failure of the putative

implicatures to pass Grice’s cancellability and non-detachability tests. These tests apply to both

generalized and particularized conversational implicatures (but not to conventional implicatures),

and therefore we may subject Salmon’s claim to these tests as long as he is not claiming the

implicatures to be conventional. 95 Although the failure to pass these tests does not strictly entail

that the implicatures are absent, as Grice recognized, it does render less plausible the claim that

the implicatures exist. Salmon’s putative implicatures fare poorly on these two tests.

Implicatures, first discussed by Grice (1975), have, inter alia, two salient properties.

They are cancelable:

… a putative conversational implicature that p is explicitly cancelable if, to the form of

words the utterance of which putatively implicates that p, it is admissible to add but not p,
or I do not mean to imply that p, and it is contextually cancelable if one can find
situations in which the utterance of the form of words would simply not carry the
implicature. (Grice 1975: 44.)

And they are non-detachable:

95
Conventional implicatures are, according to Grice, not cancellable. If Salmon is claiming that
the relevant implicatures are conventional, the cancellability tests would be inapplicable to his
claims. Salmon suggests in passing in footnote 11 of his 1989b that the implicatures could be
conventional. However, he does not elaborate on this claim. It is difficult to see how the
implicatures could be conventional, especially since we need to account for the difference in
apparent truth-value of ‘Lois Lane believes Clark Kent is Clark Kent’ and ‘Lois Lane believes
that Clark Kent is Superman’, but conventional implicatures do not affect (or give the
appearance of affecting) the truth or falsity or what is said. Moreover, it is controversial whether
there is any such thing as conventional implicature. See Bach (1999).

185
 … it will not be possible to find another way of saying the same thing, which simply
lacks the implicature in question, except where some special feature of the substituted
version is itself relevant to the determination of an implicature (in virtue of one of the
maxims of Manner). (Grice, 1975: 57) 96

We can test a putative conversational implicature (whether particularized or generalized) by

determining whether it has these properties—cancellability and non-detachability. Note again

that Grice did not consider these tests definitive or absolute. Nevertheless, the implicatures

alleged to exist by the Implicature theory fare poorly on the cancellability and detachability tests,

suggesting that the Implicature theory is implausible (without thereby refuting it). I'll subject

sentences (4), (4n), and ¬ (4), below, to the cancellability and detachability tests and show that

they do poorly on them.

(3) Lois believes that Clark Kent is Clark Kent

(4) Lois believes that Clark Kent is Superman

(4n) Lois does not believe that Clark Kent is Superman

¬ (4) Lois believes that Clark Kent is not Superman 97

However, before I test the above sentences, I will state the propositions that the above

sentences semantically express by Millian lights (or at least according to a Guise Millian view

like Salmon’s). (3) and (4) express the same proposition. They express the proposition that Lois

believes the singular proposition PROP-1—that Kent-Super is Kent-Super. This proposition is

true if and only if Lois believes it under at least one propositional guise. ¬(4) says that she

believes the singular proposition ¬PROP-1—that Kent-Super is not identical to Kent-Super.

96
Grice also noted metalinguistic implicatures, where reference is made to the specific words
used by a speaker, are detachable.
97
Sentence (4) is logically equivalent to the sentence ‘Lois disbelieve that Clark Kent is identical to Superman’.
Disbelieving that it is raining is the same thing as believing that it is not raining.
186
¬(4) expresses a true proposition if there is at least one propositional guise under which she takes

the proposition that Kent-Super is not identical to Kent-Super and assents to it. Importantly, the

truth of the proposition expressed by ¬(4) is consistent with Lois also believing, under some

other propositional guise, the singular proposition that Kent-Super is Kent-Super (e.g., via her

acceptance of the sentence ‘Clark Kent is Clark Kent’ or ‘Superman is Superman’). Therefore,

the proposition expressed by ¬ (4), according to the Millians, is consistent with the proposition

expressed by (3)-(4). (4n) says something quite different from ¬ (4). (4n) says that Lois fails to

believe (which encompasses withholding belief from) the singular proposition PROP-1—that

Kent-Super is Kent-Super. (Withholding belief or failing to believe a proposition is not the same

thing as disbelieving it or believing its negation. I may fail to believe that quantum mechanics is

inconsistent with general relativity; this does not entail that I believe quantum mechanics is

inconsistent with general relativity; I may have never considered the issue, or I may have no

opinion on the matter.) (4n) entails that Lois fails to believe that Kent-Super is Kent-Super under

any propositional guise whatsoever (even when the proposition is presented via the sentence

‘Clark Kent is Clark Kent’ or ‘Superman is Superman’).

5.4.1 The Cancellability Test Applied to the Alleged Implicatures of (4), (4n), and ¬ (4)

5.4.1.1 The cancellability Test Applied to the Alleged Implicature of (4)

According to Millians, (3) and (4) semantically express the same proposition. Ordinary

speakers judge incorrectly, according to the Implicature theorist, that (4):

(4) Lois believes that Clark Kent is Superman

187
expresses a false proposition, because they intuit (correctly) that (4) implicates a false

proposition. For example, (4) would implicate (on a version of Millianism on which

propositional guises are descriptive propositions) the proposition that Lois is aware that her mild-

mannered reporter colleague and the strong flying superhero, with whom she in love, are one and

the same person. 98 On Salmon’s theory on which propositional guises are sentence-like entities,

(4) implicates that Lois Lane believes that Clark Kent is Superman under a sentence-like

propositional guise like ‘Clark Kent is Superman.’ This proposition is false according to the

story—Lois believes no such thing. She believes that Clark Kent is Superman, but not under that

sentence-like guise. She believes it only under a sentence-like guise like ‘Clark Kent is Clark

Kent’ or like ‘Superman is Superman.’

How would a speaker uttering (4) cancel the putative implicature to signal to his audience

that he means to convey just the singular proposition semantically expressed—the true

proposition that Lois believes (under some propositional guise, but not under any particular

guise) the proposition that Kent-Super is Kent-Super? By the speaker cancelling the implicature,

the hearer/audience should realize that the speaker intended only to convey the semantically

expressed proposition and should intuit that the speaker has said something true (according to the

Millians, this is true because Lois believes PROP-1. Indeed, according to the Superman story,

there is at least one propositional guise under which she believes that Kent-Super is Kent-Super).

Let us first consider sentence (4-i) below as potential means of cancelling the implicature. This

way of capturing the cancellation is based on Salmon’s theory of propositional guises as

sentence-like entities:

98
It is hard to say exactly what proposition it would implicate on the Implicature theory, since
the Implicature theory has not sufficiently well developed to allow such granularity of analysis.
188
 (4-i) Lois Lane believes that Clark Kent is Superman, but not under a sentence-like
propositional guise like ‘Clark Kent is Superman.’

(4-i) is a poor prospect for cancelling (4)’s putative implicature. Few ordinary speakers (and

likely no ordinary speakers unfamiliar with Salmon’s theory of propositional guises) would

understand (4-i). Furthermore, it is unclear what a sentence-like propositional guise is. One

must be familiar with the notion of propositional guises and Salmon’s notion of guises as

sentence-like entities (which stands in need of explication) to stand any chance of understanding

(4-i).

A simpler version of (4-i), one more understandable to an ordinary speaker untutored in

the philosophy of language, would be (4-ii):

(4-ii) Lois Lane believes that Clark Kent is Superman, but she would not put it that way

But (4-ii) is insufficient to cancel (4)'s putative implicature. There are many ways for a hearer of

(4-ii) to interpret the speaker's communicative intent such that the hearer would be unlikely to

infer that the speaker meant to communicate merely that Lois Lane believes that Kent-Super is

Kent-Super. The hearer might suppose that the speaker believes, erroneously, that Lois knows

about Superman's secret identity, but she cannot put her knowledge into language—she “would

not put it that way”—because the speaker thinks that Lois suffers brain damage in the story,

injuring her language center and rendering her incapable of recalling proper names. Or perhaps

Lois is deaf and communicates only through sign language. She realizes that the reporter

colleague is the strong superhero but she does not use the names ‘Clark Kent’ or ‘Superman’ to

refer to him in either of his personas. I deem it quite unlikely that any ordinary hearer would

interpret a speaker uttering (4-ii) as saying that Lois Lane realizes the trivial proposition PROP-

189
1—that Kent-Super is Kent-Super. Of course, one could stipulate in the story that Lois suffers

no brain damage and that she is fully competent with ordinary English. Nevertheless, I doubt

that any amount of stage-setting would result in a cancellation of the implicature. Speakers

would be inclined to seek some sort of explanation of Lois’ disinclination to “put it that way”

relating to her competence or dispositions to speak in certain ways rather than grasping the

proposition that Millians claim is semantically expressed by the sentence ‘Lois Lane believes

that Clark Kent is Superman.’

Consider whether a metalinguistic cancellation, such as (4-iii) would work.

(4-iii) Lois Lane believes that Clark Kent is Superman, but she does not realize that
‘Clark Kent’ and ‘Superman’ co-refer

This has the same problem as (4-ii). Most hearers of (4-iii) would interpret the speaker as saying

that Lois is, for some reason or other, ignorant of the fact that ‘Clark Kent’ and ‘Superman’ co-

refer—either due to incompetence with proper names due to brain damage or because she is deaf

and communicates only through sign language. The truth of (4-iii) is consistent with Lois

realizing that the reporter colleague is the superhero but she does not use the names ‘Clark Kent’

or ‘Superman’ to refer to him in either of his personas.

Would a definite description such as in (4-iv) cancel the putative implicature?

(4-iv) Lois believes that Clark Kent is Superman, but this is not to say that she
realizes that the mild-mannered reporter colleague is the caped Superhero

(4-iv) does not seem to me to cancel the putative implicature. First, I believe that ordinary

speakers would say that (4-iv) is inconsistent, saying in response to hearing (4-iv) uttered, some

version of: “if Lois Lane fails to realize that her mild-mannered colleague is a caped superhero,

190
she simply does not realize that Clark Kent is Superman.” Furthermore, (4-iv) could be

plausibly interpreted as saying that there are some descriptive modes of presentation under which

Lois recognizes Clark Kent and Superman as the same individual, although she fails to do so

when Clark Kent is presented to her as a bespectacled mild-mannered reporter and Superman

presented as a caped superhero. For example, Kent-Super might have a flirtatious phone

relationship with Lois, where Lois has no idea that the caller is the Superhero she admires or the

reporter with whom she works. Sometimes Kent-Super calls her from his Fortress of Solitude

when in his Superman persona (he's rather forward and aggressive) and sometimes he calls her

when in his Clark Kent persona from his Office at the Daily Planet (he's rather timid and passive

and Lois has to do most of the talking). Lois realizes, based on the similarity of his voice on the

telephone regardless of the persona he is in, that the guy she is flirting with is the same guy

whether he is being aggressive or shy/passive. In this scenario, there is a pair of “Clark Kent-y”

and “Superman-y” conceptions under which she realizes that Clark Kent is Superman, even

though she does not realize that the bespectacled mild-mannered reporter is the caped Superhero.

(4-iv) could be uttered intending to convey this, rather than conveying the information that Lois

believes the trivial singular identity proposition PROP-1.

I believe the only certain way to cancel the implicature of (4) would be with (4-v):

(4-v) Lois Lane believes that Clark Kent is Superman, but only in the sense that
Superman and Clark Kent are the same individual and Lois obviously
realizes that Clark Kent is Clark Kent, so she also realizes that Clark Kent
is Superman

This cancellation of the implicature works only if ordinary speakers can understand it. I doubt

an unsophisticated competent speaker of English who understands the Superman story would

understand (4-v). (It would be interesting to test this empirically). An ordinary speaker would

191
likely respond to an utterance of (4-v) by uttering: “But Lois does not realize that Clark Kent and

Superman are the same person.” Furthermore, it is hard to imagine anyone uttering (4-v) in a

real-life conversation. Only a convinced Millian philosopher would say (4-v). Hence, I

conclude that the implicature of (4), were it to exist, would be difficult to cancel.

5.4.1.2 The Cancellability Test applied to the Alleged Implicature of (4n)

The Implicature theorist maintains that (4n) expresses a literally false proposition, but it

implicates a true proposition.

(4n) Lois does not believe that Clark Kent is Superman.

The Implicature theorist maintains that (4n) literally says that Lois fails to believe the trivial

singular identity proposition that Kent-Super is Kent-Super under any propositional guise

whatsoever. (This is false according to the Superman story, because Lois does not fail to believe

this proposition—she believes it because she assents to ‘Clark Kent is Clark Kent’ and

‘Superman is Superman.’) Suppose that, in uttering (4n), the speaker meant to convey just the

semantically expressed (false) proposition. The speaker means to convey the proposition that

there is no guise under which Lois believes PROP-1—that Kent-Super is Kent-Super. She does

not believe this proposition no matter how that proposition is presented (even if presented by the

obviously true (1)). There are three possible reasons Lois might fail to believe this semantically

expressed proposition:

(a) because she is unfamiliar with (not en rapport with) Kent-Super and therefore cannot
entertain propositions of which he is a constituent,

192
 (b) there is an individual with which Lois is familiar, Kent-Super, such that that she fails
to recognize him as an individual with which she is already familiar even if he is
presented under the same mode of presentation on different occasions, or

(c) Lois believes that Kent-Super is not self-identical.

(a) is foreclosed as a possible interpretation because it directly contradicts the facts of the

Superman story. According to the story, Lois is en rapport with Kent-Super. She has met Kent-

Super many times, both in his Clark Kent and Superman personas. (b) is foreclosed as a possible

interpretation because the truth of (b) would entail that Lois had severe dementia—how can a

mentally spry agent such as Lois Lane fail to recognize an individual from one moment to the

next when that individual is presented in the same way to her at both points in time? (c) is

foreclosed as a possible interpretation because it is irreconcilable with Lois’ rationality. What

rational agent entertains doubts about the self-identity of an object?

Would (4n-i) cancel the alleged implicature?

(4n-i) Lois does not believe that Kent is Superman, in the sense that there is no
propositional guise under which the trivial singular proposition PROP-1—that
Kent-Super is Kent-Super—can be presented to her so that she would assent to the
proposition (and not in way that she just fails to recognize him as the same person
when he is presented to her under the conceptions associated with the names
‘Clark Kent’ and ‘Superman’).

I doubt that this would cancel the alleged implicature. To understand (4n-i) one must be familiar

with the notion of propositions being presented under propositional guises. A competent speaker

of English, untutored in Salmon’s theory, would not understand (4n-i). The implicature would

be difficult to cancel.

5.4.1.3 The Cancellability Test Applied to the Alleged Implicature of ¬ (4)

193
 The implicature theorist maintains that ¬ (4) expresses a true proposition.

¬ (4) Lois believes that Clark Kent is not identical to Superman

The proposition expressed is true iff there is at least one guise under which Lois disbelieves that

Kent-Super is Kent-Super. This is a true proposition, for she disbelieves that Kent-Super is Kent-

Super when that proposition is presented by the sentence ‘Clark Kent is Superman.’ This

follows from the weak disquotation principle: Lois is disposed to sincerely, reflectively, and

competently utter ‘Clark Kent is not identical to Superman’ (e.g., in response to a suggestion by

Jimmy Olson that Clark Kent and Superman are the same person), so, in this case, she believes

the proposition expressed by the sentence. (Of course, according to the Millians Lois also

believes the proposition that Kent-Super is Kent-Super, for she assents to both ‘Clark Kent is

Clark Kent’ and ‘Superman is Superman’.) Ordinary speakers agree with the Millians that ¬ (4)

expresses a true proposition, but for different reasons. According to the Implicature theorist,

ordinary speakers say that ¬ (4) is true because they confuse the proposition literally expressed

with the implicated proposition. For in addition to expressing a true proposition, Millians claim

that ¬ (4) also implicates a true proposition (e.g., the proposition that Lois disbelieves a mild-

mannered reporter is a strong superhero), and it is for this reason that ordinary speakers intuit

that ¬ (4) is true. Now suppose that one wanted to cancel the alleged implicature to

communicate merely the literally expressed proposition. How would this implicature be

cancelled? We could try

¬ (4-i):

¬ (4-i) Lois disbelieves that Clark Kent is Superman, for she dissents when this is put to
her under certain guises, e.g., via the sentence or a propositional guise similar or
analogous to the sentence ‘Clark Kent is Superman,’ but that is not to say that she

194
 doesn’t also believe it when presented by a different guise, e.g., the sentence
‘Clark Kent is Clark Kent’ or a guise similar/analogous to this sentence.

This proposal is highly technical and requires a speaker to understand the notion of propositional

guises. Furthermore, it requires an ordinary speaker to understand how a rational speaker could

believe and disbelieve the same proposition. That is beyond the intellectual capacities of many

ordinary competent speakers, even highly sophisticated speakers. Understanding this

cancellation presupposes familiarity with (and acceptance of the truth of) the Millian theory, and

it presupposes that the notion of sentence-like propositional guises can be fleshed out.

Would ¬ (4-ii) cancel the implicature?

¬ (4-ii) Lois disbelieves that Clark Kent is Superman, for she dissents from ‘Clark Kent is
Superman,’ but she also believes that Clark Kent is Superman, for she assents to
‘Clark Kent is Clark Kent.’

To any speaker unfamiliar with Millian theory, ¬ (4-ii) appears false on its face. Ordinary

speakers would likely respond to an utterance of ¬ (4-ii) by stating: “the fact that she dissents

from ‘Clark Kent is Superman’ entails that she disbelieves that Clark Kent is Superman, but the

fact that she assents to ‘Clark Kent is Clark Kent’ does not entail that she believes Clark Kent is

Superman. It merely entails that she recognizes that Clark Kent is identical to himself, not that

she realizes that he is Superman.”

5.4.2 The Non-Detachability Test Applied to the Alleged Implicatures of (4), (4n), & ¬ (4)

Sentences (4), (4n), and ¬ (4) fare poorly on the non-detachability test. A conversational

implicature (except when based on the maxim of Manner) is non-detachable when, after the

195
replacement of what is said with another expression with the same literal meaning, the same

conversational implicature remains.

5.4.2.1 The Non-Detachability Test Applied to the Alleged Implicature of (4)

For (4) to be non-detachable,

(4) Lois Lane believes that Clark Kent is Superman

it would have to be possible to replace (4) with another sentence with the same literal meaning

that would carry the same implicature. Suppose we substituted (4¶) for (4)

(4¶) Lois Lane believes that Superman is Superman

According to the Millians, (4¶) means the same thing as (4), for the only difference between

them involves substituting ‘Superman’ for ‘Clark Kent’—and these are synonymous on the

Millian view. (4¶) would manifestly carry a different implicature than (4). On Salmon’s theory,

(4) implicates that Lois Lane believes the singular proposition that Kent-Super is Kent-Super

when she takes the proposition under a sentence-like entity like ‘Clark Kent is Superman.’ This

implicated proposition would be, according to Salmon, false. (4¶) would implicate that Lois

Lane believes the singular proposition that Kent-Super is Kent-Super when she takes the

proposition under a sentence-like entity like ‘Superman is Superman.’ This implicated

proposition would be, according to Salmon, true.

Sentence (4) therefore fails the non-detachability test. The (alleged) implicature is

detachable because after the replacement of ‘Clark Kent’ with another expression with the same

literal semantic meaning—here ‘Superman,’ the conversational implicature does not remain.

196
5.4.2.2 The Non-Detachability Test Applied to the Alleged Implicature of (4n)

Sentence (4n) also fails the non-detachability test for the same reason as (4) failed.

(4n) Lois Lane does not believe that Clark Kent is Superman

Suppose we substituted ‘Clark Kent’ for ‘Superman’ (which, according to Millianism, are

synonymous) at its first occurrence in (4n). This yields sentence (4n¶):

(4n¶) Lois Lane does not believe that Superman is Superman

(4n¶)’s implicature would be very different from (4n)’s. (4n) would implicate the proposition

that Lois Lane dissents from the proposition that Kent-Super is Kent-Super when she takes that

proposition under a sentence-like entity like ‘Clark Kent is Superman.’ This is true. (4n¶) would

implicate the proposition that Lois Lane dissents from the proposition that Kent-Super is Kent-

Super when she takes that proposition under a sentence-like entity like ‘Superman is Superman.’

This would be false.

(4n) fails the non-detachability test. The (alleged) implicature is detachable because after

the replacement of ‘Clark Kent’ with another expression with the same literal meaning—here

‘Superman’, the conversational implicature does not remain.

5.4.2.3 The Non-Detachability Test Applied to the Alleged Implicature of ¬ (4)

Sentence ¬ (4) also fails the test for the same reason as (4n) failed the test.

¬ (4) Lois Lane believes that Clark Kent is not identical to Superman

197
Suppose we substituted into ¬ (4) ‘Superman’ in place of ‘Clark Kent’ ‘Superman’ (which,

according to Millianism, are synonymous). This would yield ¬ (4¶).

¬ (4¶) Lois Lane believes that Superman is not identical to Superman

¬(4¶)’s implicature would be very different from ¬(4)’s. ¬(4) would implicate the proposition

that Lois Lane assents towards the proposition that Kent-Super is not identical to Kent-Super

when she takes that proposition under a sentence-like entity like ‘Clark Kent is not identical to

Superman.’ This would be true. ¬ (4¶) would implicate the proposition that Lois Lane assents

towards the proposition that Kent-Super is not identical to Kent-Super when she takes that

proposition under a sentence-like entity like ‘Superman is Superman.’ This would be false.

¬ (4) fails the non-detachability test. The (alleged) implicature is detachable because

after the replacement of ‘Clark Kent’ with another expression with the same literal meaning,

‘Superman’, the conversational implicature does not remain.

5.4.3 Conclusion: The Alleged Implicatures are not Detachable and are not Easily
Cancellable

The alleged Implicatures are difficult to cancel. The implicatures can be cancelled

perhaps, but only for an audience which is already familiar with (and committed to) Millian

theory (provided that the Millian notion of propositional guises could be ultimately explicated).

The cancellation would only work for an audience of speakers able to carry out relatively

complex philosophical and/or linguistic reasoning. Without such an understanding of Millianism

or a sophisticated reasoning ability, ordinary speakers are not able to recognize via the attempts

at cancellation the proposition Millians claim is the one literally expressed. The alleged

198
implicatures are non-detachable. This is a serious worry for the proposal that implicature could

be the pragmatic mechanism Salmon requires.

5.5 How Salmon’s theory fares on the four principal worries for Millianism

In section 5.1, supra, I set out four worries (listed a-d) that I have with respect to Millian

theories in general: the Guise Definition Problem, the Pragmatic Mechanism Problem, the

Ignorance of Identities Problem, and the No Direct Expressibility Problem. I’ll now state

how I think Salmon’s theory fares with respect to these worries.

5.5.1 The Guise Definition Problem

Salmon does not clearly define guises. The most plausible claim is that they are

descriptive ways of taking. So the guise of ‘Clark Kent is Superman’ is the descriptive

proposition that the mild-mannered reporter from Smallville is the Superhero that protects

Metropolis. What is the relationship between the guise and the singular proposition, that Kent-

Super is Kent-Super, that is semantically expressed? Salmon does not say specifically, which is

a gap in the theory. 99

99
The account I would give is one that would be very much in the spirit of the TIUT. I would say
that the descriptive proposition that constitutes the guise is made up of the descriptive
conceptions in Lois’ two dossiers, her ‘Clark Kent’ and her ‘Superman’ dossiers. Lois believes
that the subject of her one dossier is not the same as the subject of the other dossier. But she is
wrong, for each of the dossiers still refer to the same individual, have the same subjecthood. This
way of explaining guises using the language of the TIUT connects the singular proposition that
Kent-Super is Kent-Super to the descriptive guise under which Lois takes it. The connection is
that the dossiers have the descriptions conceptions associated with the names ‘Clark Kent’ and
‘Superman’ in them, and these dossiers are both about Kent-Super because they are both causally
linked to him in the right sort of way. Perhaps Salmon could give such an account of guises
based along the lines of the TIUT’s dossier model. But it is not so clear that this can be done
without thereby implicitly accepting the TIUT indexical approach to the Problem of Rational
Inconsistent Belief.
199
5.5.2 The Pragmatic Mechanism Problem

This problem is, to recap, to explain the pragmatic mechanism by which the semantic

propositions expressed by informative identities sentence such as ‘Clark Kent is Superman’ and

propositional attitude reports such as ‘Lois Lane believes that Clark Kent is Superman’

pragmatically communicate further propositions. Salmons’ theory fares poorly here because he

has not specified a specific pragmatic mechanism. Furthermore, implicature is an implausible

candidate for the mechanism, as I have explored supra in section 5.4.

5.5.3 The Ignorance of Identities Problem

Salmon does not clearly say what Lois Lane is ignorant of. It cannot be a purely

descriptive proposition, such as the proposition that the mild-mannered reporter from Smallville

working for the Daily Planet is the superhero that protects Metropolis. The proposition of which

Lois is ignorant is object-dependent, causally tied to Kent-Super. This purely descriptive

proposition is not object-dependent. Her belief is essentially about Kent-Super, not someone who

fits his description.

Perhaps she is ignorant of the fact that the singular proposition Kent-Super is Kent-Super,

when taken under a guise like ‘Clark Kent is Superman’, is the same proposition as Kent-Super

is Kent-Super when taken under a guise like ‘Clark Kent is Clark Kent’ or a guise like

‘Superman is Superman’. Thus, the following sentence could express her ignorance:

The singular proposition that Kent-Super is Kent-Super, when taken under a guise like
‘Clark Kent is Superman’, is the same proposition as the singular proposition that Kent-
Super is Kent-Super when taken under a guise like ‘Clark Kent is Clark Kent’ or a guise
like ‘Superman is Superman’

200
 This sentence captures what Lois fails to realize because, according to Salmon, she

suffers from is “propositional recognition failure.” Her ignorance consists of an inability to

recognize a proposition when it is presented in one particular way as one she recognizes when it

is presented a different way. This answer is consistent with Salmon’s theory, but it is not very

satisfying because it neither gives us insight into the reasons for the propositional recognition

failure, nor any idea as to what Lois needs to find out for her propositional recognition failure to

be cured.

5.5.4 The No Direct Expressibility Problem

As discussed in the above section, it is not clear what, on Salmon’s theory, is the

proposition Lois fails to realize about Kent’s identity with Superman. But let us suppose, as we

did in the above section, that what she fails to realize is captured by the following sentence:

The singular proposition that Kent-Super is Kent-Super, when taken under a guise like
‘Clark Kent is Superman’, is the same proposition as the singular proposition that Kent-
Super is Kent-Super when taken under a guise like ‘Clark Kent is Clark Kent’ or a guise
like ‘Superman is Superman’

The problem is that this sentence, while grammatically well-formed, does not belong to

ordinary, idiomatic English. Ordinary speakers do not understand it, and they certainly would

never use it. Not even a Millian philosopher would use it as a regular convention to express

informative identities. It is a sentence belonging to Millian philosophical theory. There seems to

be no ordinary idiomatic way to express the proposition about Kent’s identity with Superman in

plain English that is as good as ‘that Clark Kent is Superman’. If the semantically expressed

proposition and the pragmatically conveyed proposition were truly distinct, it should be possible

to find an idiomatic sentence that would express that pragmatically conveyed proposition

201
directly. There is none, and that militates in favor of the view that informative identity sentences

like ‘Clark Kent is Superman’ are the conventional (and hence semantic) direct way to express

informative identities.

5.6 Soames’ Descriptive Enrichment Account

Like Salmon, Soames (2002) proposes a theory that attempts to show that ordinary

speakers’ intuitions about cognitive and truth-value in Frege’s puzzle cases do not in fact falsify

Millianism. Like Salmon, Soames claims that these erroneous intuitions are rooted in the failure

of ordinary speakers to distinguish semantics from pragmatics, i.e., the failure of ordinary

speaker to distinguish between the propositions that sentences semantically express—singular

propositions, and the descriptively enriched singular propositions that they pragmatically convey.

Unlike Salmon, who is non-committal with respect to the pragmatic mechanisms involved,

Soames specifically identifies descriptive enrichment as the pragmatic mechanism. Another

salient difference is that Soames, in his 2002, never mentioned propositional guises and some

readers took his failure to mention propositional guises to imply that his account dispensed with

them. However, in response to criticism from Salmon and Braun after publication of his 2002, in

his Précis of Beyond Rigidity (2006) Soames clarified that he never intended his account to leave

out guises and he accepted propositional guises is an essential part of Millianism. Soames wrote

in his 2006:

“Although this is the conclusion of part 1, I did not argue in the book, and I do not
believe, that descriptive enrichment is a universal solvent for dissolving all anti-Millian
intuitions. In particular, I intended it to augment, not replace, the Millian idea --
championed by Nathan Salmon -- of different ways of entertaining and believing the
same proposition.” (5)

With respect to our Clark Kent/Superman case, Soames would claim that the proposition

202
that speakers would always semantically express with their utterances of identity sentences such

as (1) and (2):

(1) Clark Kent is Clark Kent

(2) Clark Kent is identical Superman

would be the trivial, uninformative, and a priori singular proposition PROP-1, the proposition

that Kent-Super is Kent-Super. But what speakers would pragmatically convey in uttering (1)

and (2), in certain conversational contexts, would be different descriptively enriched singular

propositions differing in cognitive value (both in informativeness and epistemological status).

The descriptively enriched proposition the speaker pragmatically conveys by uttering (1) is

supposed to be uninformative and a priori. By contrast, the descriptively enriched singular

proposition the speaker pragmatically conveys by uttering (2) is supposed to be informative and

a posteriori. In uttering (1) and (2), a speaker may pragmatically convey the descriptively

enriched singular propositions that would be semantically expressed, e.g., by utterances of

sentences (1 DESP) and (2 DESP):

(1 DESP) Clark Kent the mild-mannered reporter is Clark Kent the mild-mannered
reporter

(2 DESP) Clark Kent the mild-mannered Reporter is Superman the strong

superhero 100

The propositions that (1 DESP) and (2 DESP) express could be schematized as:

100
Soames thinks that numerous propositions may be asserted by a single utterance.
Furthermore, the exact nature of the propositions asserted may be a partially indeterminate
matter.

203
 DESP-1 <<<Kent-Super, the mild-mannered reporter >, <Kent-Super, the mild-
mannered reporter >> identity>

DESP-2 <<<Kent-Super, the mild-mannered reporter >, <Kent-Super, the strong

superhero >> identity >

On Soames’ theory, (1) and (2) would typically be uttered by speakers to convey propositions

such as DESP-1 and DESP-2 (as well as a range of other descriptively enriched propositions).

DESP-1 and DESP-2 are singular propositions because the proper names occurring in (1 DESP)

and (2 DESP) directly refer (via the directly referential names ‘Clark Kent’ and ‘Superman’) to

Kent-Super. They are “descriptively enriched”, meaning that the descriptive material

(underlined in (1 DESP) and (2 DESP)) supplements or enriches the singular propositions with

descriptive information salient to or presupposed by the participants in the conversational

setting. 101 Descriptively enriched propositions are hybrid propositions—part singular and part

descriptive.

According to Soames, speakers do not need to actually utter (1 DESP) and (2 DESP)

(although perhaps they could if they wanted to) because, in the context of the particular

conversational setting in which they utter (1) and (2), there is mutual knowledge between the

speaker and the audience hearing the utterance in the conversational setting and this shared

knowledge enables the audience to ascertain the speaker’s assertive intent—that he intends to

convey the descriptively enriched singular propositions, DESP-1 and DESP-2 (which could also

be directly semantically expressed by utterances of (1 DESP) and (2 DESP) if the speaker chose

to do so).

The difference in cognitive value between the descriptively enriched propositions DESP-

101
The names in (1 DESP) and (2 DESP), as in (1) and (2), are Millian and directly referential.

204
1 and DESP-2, which are pragmatically conveyed by the speaker in uttering (1) and (2), is

supposed to explain our Fregean intuition that (1) and (2) express distinct propositions. Ordinary

speakers fail to distinguish between the proposition that (1) and (2) semantically express—which

is the same trivial a priori singular proposition PROP-1, that Kent-Super is Kent-Super—and the

descriptively enriched propositions, DESP-1 and DESP-2, that speakers convey pragmatically.

So they tend to judge, in error, that (1) and (2) express different propositions.

Likewise, in the propositional attitudes puzzle, Soames maintains that we have the

erroneous intuition that (3) differs from (4), and (5) from (6), in truth-value.

(3) Lois Lane believes that Clark Kent is Clark Kent

(4) Lois Lane believes that Clark Kent is Superman

(5) Lois Lane believes that Clark Kent flies

(6) Lois Lane believes that Superman flies

According to Soames, (3) semantically expresses the same proposition as (4), and (5) the same as

(6), since the simple sentences embedded in the ‘that’-clauses of (3) and (4), and (5) and (6),

have the same semantic content. Our erroneous judgment that these attitude ascriptions differ in

truth-value is based on what they pragmatically convey about the descriptively enriched

propositions Lois believes. For example, ascription (5) pragmatically conveys that Lois Lane

believes that Clark Kent the mild-mannered reporter flies (which is false), and (6) pragmatically

conveys that Lois Lane believes that Superman the strong superhero flies (which is true). So we

judge, in error, that (5) is false and (6) is true.

205
5.7 Descriptive Enrichment in the Frege’s Puzzle Cases: Genuine Cases of
Descriptive Enrichment?

Soames claims that the pragmatic mechanism by which the propositions semantically

expressed pragmatically convey descriptively enriched singular propositions is descriptive

enrichment. This pragmatic process is often referred to as ‘free enrichment’ (to distinguish it

from the sort of enrichment that arises from the determination of the semantic values of

indexicals). In this section I will point out that descriptive enrichment that Soames claims occurs

in the Frege cases is quite different in several respects from the descriptive enrichment that

occurs in the two examples he gives in his book to illustrate descriptive enrichment, anecdotes

which he calls Coffee, please and Smoking and Drinking (Soames 2002: 78). It is not clear to me

if these differences are benign, or whether they signal a significant worry for Soames’ theory.

Perhaps Soames is running two sorts of pragmatic mechanisms together.

Coffee, please is an anecdote about a man who goes into a coffee shop, sits at the counter,

and says to the waitress “I would like some coffee, please.” Although he literally stated just that

he wanted coffee—not specifying whether he wanted brewed coffee, coffee beans, coffee

grounds, a truckload of coffee, etc.—the speaker descriptively enriches the word ‘coffee’ with a

description such as the brewed drinkable kind. Hence, the enriched proposition communicated,

what is said, is the proposition I would like coffee, the brewed drinkable kind, please. The

descriptive enrichments are not spoken aloud, but they are an essential part of the proposition the

speaker conveyed with his utterance. In Soames’ second anecdote, Smoking and Drinking,

Jeeves the butler utters: “I enjoy a cigarette after breakfast in the morning and a brandy before

retiring in the evening.” Jeeves does not expressly state how he enjoys the cigarette or the

brandy—does he enjoy consuming them, looking at them, or putting them in his ear? In the

conversational setting (which Soames presumes is a normal conversational context), Jeeves

206
means he enjoys drinking the brandy and smoking the cigarette. Jeeves descriptively enriches the

words ‘cigarette’ and ‘brandy’ so that what is said is I enjoy smoking a cigarette after breakfast

in the morning and drinking a brandy before retiring in the evening. Again, these descriptive

enrichments are not spoken aloud, but they are an essential part of the proposition the speaker

conveyed with his utterance.

The descriptive enrichments Soames claims occur in the Frege’s puzzle cases differ from

those in the Coffee, Please and Smoking and Drinking cases in several ways:

First, the descriptive enrichments change the reference of the expressions in the Coffee,

Please and the Smoking and Drinking cases, but they do not do so in the Frege’s puzzle cases.

Consider the Coffee, Please case. The word ‘coffee’ can refer to different things in different

conversational settings, depending on how it is descriptively enriched. If enriched by the

brewed drinkable kind at the counter in the coffee shop, it would mean brewed coffee, and if

enriched by the whole bean kind in another setting (for example, if the word ‘coffee’ were

uttered in a store exclusively selling whole bean coffee), it would mean whole bean coffee. The

expressions ‘brewed coffee’ and ‘whole bean coffee’ do not co-refer. Brewed coffee is not the

same thing as whole bean coffee. Hence, the following descriptively enriched identity sentences

(with the enrichments made explicit), (15) and (16), express false propositions:

(15) Coffee, the brewed drinkable kind, is coffee, the whole bean kind

(16) Brewed coffee is whole bean coffee

Brewed coffee and whole bean coffee are not the same thing (although they are, in part, made

out of the same thing). By contrast, Clark Kent and Superman are the same person. That is

precisely what a speaker would state in uttering the true informative identity sentence (2):

207
 (2) Clark Kent is Superman

Sentence (2) is true, full stop. No matter what any name referring to Kent-Super might be

descriptively enriched with, according to Millianism, it rigidly refers to Kent-Super. The

descriptive enrichment does not change the name’s reference. Hence, according to Soames the

following (synonymous) descriptively enriched identity sentences are true:

(1 DESP) Clark Kent the mild-mannered reporter is Clark Kent the mild-mannered
reporter

(2 DESP) Clark Kent the mild-mannered Reporter is Superman the strong superhero

Clark Kent the mild-mannered reporter is identical to Superman the strong superhero—they are

the same person, whereas brewed coffee is not identical to whole bean coffee. In the Frege

cases, unlike in the examples of enrichment Soames gives, there is co-reference of the

expressions flanking ‘=’ no matter what sort of descriptive enrichment there may be. The

descriptively enriched identity sentences are true, just as the unenriched identity sentences are.

In the Coffee, Please case, the descriptively enriched identity sentences, (15) and (16), are false

because of the different enrichments. Hence, it is not clear to me that the same pragmatic

process is present in the Frege’s puzzle cases—at least not the same sort of descriptive

enrichment that Soames’ presents through the Coffee, Please case. 102 There is a disanalogy.

Secondly, in the Coffee, Please and Smoking and Drinking cases, the descriptive

enrichment is unconscious and automatic. Neither the speaker nor the hearer would typically be

consciously aware that certain salient facts in the conversational setting are furnishing a

102
The same change of reference happens in the Smoking and Drinking case. For example,
smoking a cigarette ≠ eating a cigarette.
208
significant element of the meaning of the utterance. However, in the Frege cases, by contrast, a

speaker who uttered ‘Lois believes that Superman can fly but that Clark Kent cannot’ would be

clearly consciously aware that he uses ‘Clark Kent’ and ‘Superman’ to communicate different

information. The speaker chooses syntactically distinct terms, reflecting conscious design on his

part. If Soames is right that the speaker is descriptively enriching ‘Clark Kent’ and ‘Superman,’

he is doing it by conscious design, not automatically and infra-consciously as in the Coffee,

Please case.

Thirdly, in the examples of enrichment Soames gives us, the sentences explicitly spelling

out the enrichments are true only if the sentences with the enrichments left implicit are true. ‘I

want coffee, the brewed drinkable kind’ (explicit), uttered by a patron to a waitress at the counter

of a coffee shop, expresses a true proposition only if ‘I want coffee’ (implicit), expresses a true

proposition in that situation. ‘I enjoy smoking a cigarette after breakfast’ (explicit) expresses a

true proposition only if ‘I enjoy a cigarette after breakfast’ (implicit) express a true proposition in

that context. However, in Soames’ cases of descriptive enrichment vis-à-vis Frege’s puzzle, the

sentences in which the enrichment is made explicit are true but the sentences with the

enrichments left implicit are false (according to Millianism). For example, the sentence ‘Lois

Lane does not believe that Clark Kent the mild-mannered reporter is Superman the strong

superhero,’ where the enrichments are made explicit, would, according to Soames, semantically

express a true proposition. Lois does not believe that a mild-mannered reporter is a superhero.

However, the sentence ‘Lois Lane does not believe that Clark Kent is Superman,’ where the

enrichments are left implicit, would semantically express a false proposition (according to

Millianism the latter sentence is false because Lois Lane does believe the singular proposition

PROP-1, the singular proposition that Kent-Super is Kent-Super), and that sentence semantically

209
expresses the proposition that Lois fails to believe PROP-1, the singular proposition that Kent-

Super is Kent-Super).

I do not know what import these differences between Coffee, Please and Smoking and

Drinking and the Frege cases has. The differences do give me some pause. Perhaps there are two

sorts of descriptive enrichment going on here.

5.8 How Soames’s theory fares on the four worries set out in 5.1

In section 5.1 supra I set out four worries (list a-d) that I have with respect to Millian

theories in general. I’ll now state how I think Soames’s theory fares with respect to these

worries: the Guise Definition Problem, the ‘Pragmatic Mechanism Problem, the Ignorance

of Identities Problem, and the No Direct Expressibility Problem.

5.8.1 The Guise Definition Problem

Soames stated (2006), in response to criticism by Salmon and Braun, that propositional

guises are an essential part of any Millian account. In his 2002, Soames had appeared to attempt

a guise-free purely pragmatic solution to the puzzles. Given that the notion of propositional

guises is not clear, and Soames does not attempt on his own to define the notion of propositional

guises, Soames fares about as well as Salmon in this point. We are left with an intuitive and

metaphorical understanding of guises.

5.8.2 The Pragmatic Mechanism Problem

Soames’ theory Fares better here than Salmon’s theory because Soames clearly identifies

a pragmatic mechanism, descriptive enrichment, by which semantically expressed propositions

210
pragmatically convey further, richer propositions. I raised a worry about Soames’ use of

descriptive enrichment – there seem to be some differences between the sort of descriptive

enrichment in the Frege cases and in the Coffee, please and Smoking and Drinking cases he uses

to illustrated descriptive enrichment. I do not know what significance, if any, these differences

have, or what they reveal about Soames’ theory.

5.8.3 The Ignorance of Identities Problem

Here, Soames’ theory runs into a problem. Soames seems to be saying that Lois Lane is

ignorant of the descriptively enriched descriptive proposition expressed by (2 DESP):

(2 DESP) Clark Kent the mild-mannered reporter is Superman the strong superhero

As I mentioned above, that proposition may be schematized as 2-DESP.

DESP-2 <<<Kent-Super, the mild-mannered reporter >, <Kent-Super, the strong

superhero >> identity >

But what is this descriptively enriched singular proposition? I believe it is a hybrid

proposition, which can be analyzed as a proposition composed of the following parts:

(a) < < Kent-Super, Kent-Super >, identity >

(b) < Kent-Super, the mild-mannered reporter >

(c) < Kent-Super, the strong superhero >

(d) < Kent-Super, the mild-mannered reporter and the strong superhero >

(e) < < the mild-mannered reporter and the strong superhero >, identity >

211
 I’ll now show that Lois either already knows all the above propositions, or would be able

to infer them from what she knows. Hence, Lois does not in fact fail to realize DESP-2, and it

therefore cannot be the proposition of which she is ignorant.

Lois Already believes proposition (a), i.e., the a priori singular proposition that Kent-

Super is self-identical. According to Millianism (and the facts of the Superman story) Lois Lane

already believes proposition (b), i.e., the singular proposition that Kent-Super is a mild-

mannered reporter, given her disposition to assent to the sentence ‘Clark Kent is a mild-

mannered reporter’. According to Millianism (and the facts of the Superman story) she already

believes proposition (c), or the singular proposition that Kent-Super is a strong superhero, given

her disposition to assent to the sentence ‘Superman is a strong superhero.’ (d) is also a

proposition that Lois already believes, or at least one she could come to believe merely using

logical inference. She already believes that Kent-Super is a reporter via believing that Clark Kent

is, and she already believes that Kent-Super is a superhero via believing that Superman is. She

should be able to infer effortlessly, via the a priori rule of logical inference, conjunction

(“CONJ”),

(CONJ) if x is f and x is g, x is both f and g

that Kent-Super is both a mild-mannered reporter and a strong superhero. Finally, if Lois

believes (d), then it follows that she also believes (e), since the truth of the descriptively enriched

singular proposition (d) is sufficient for the truth of the descriptive proposition (e).

DESP-2 therefore cannot be the proposition that Lois fails to realize about Kent’s identity

212
with Superman, for she already realizes DESP-2. 103 What then is the proposition of which Lois

is ignorant? Soames does not tell us. Problematically, if someone were to say, ‘Lois Lane does

not realize that Clark Kent the mild-mannered reporter is Superman the strong superhero’, they

would be asserting a falsehood. Lois does in fact believe that, according to Soames, or at least

she could come to believe this proposition merely by reflecting on her beliefs and using logical

inference. This is quite worrisome for Soames’ theory.

5.8.4 The No Direct Expressibility Problem

It is unclear how the state the proposition of which Lois Lane is ignorant with respect to

the identity of Clark Kent and Superman in Soames theory, as discussed above. But even if we

presumed that there were some way to express this proposition directly in Soames’ theory, I

suspect it would be some complex theory-laden sentence, rather than an ordinary, idiomatic

English sentence.

5.9 Conclusion

Although I present no knockdown argument against Millianism, I take Millianism is to be

implausible. I would base this judgment on the fact alone that it runs counter to our Fregean

103
Speaks (2011) makes a somewhat related point. He claims that on Soames’ theory the
following sentence would pragmatically convey an a priori descriptively enriched singular
proposition:

‘If Clark Kent exists and Superman exists, then Clark Kent is Superman’.

The problem, Speaks maintains, is that the descriptively enriched singular proposition
pragmatically conveyed would need to be a posteriori to account for our (erroneous) Fregean
intuition that the sentence semantically expresses an a posteriori proposition. But the
pragmatically conveyed descriptively enriched singular proposition is a priori. So we cannot
account for the Fregean intuition.
213
intuitions with respect to Frege’s puzzle. In addition, both the theories of Salmon and Soames

face the four worries I raised above, and neither theory satisfactory addresses all of them.

214
 CHAPTER 6

THE HIDDEN INDEXICAL THEORY

The Hidden Indexical Theory (“HIT”) is a variant of Millianism specifically designed to

solve Frege’s puzzle about propositional attitudes. Stephen Schiffer first proposed the HIT in his

1977. On HIT, (3) and (4) express different propositions and differ in truth-value, consistent with

the intuition of ordinary speakers. This is contrasted with (directly referential) Millianism

discussed above in chapter 4, which take a counterintuitive stance on truth-value, claiming that

(3) and (4) express the same proposition and have the same truth-value.

(1) Lois Lane believes that Clark Kent is Clark Kent

(2) Lois Lane believes that Clark Kent is Superman

HIT holds that propositional attitude ascriptions semantically express a ternary relation

between the ascribee, the singular propositions expressed by the ‘that’-clauses of (3) and (4), and

a propositional mode of presentation or propositional guise. The HIT theorist proposes that the

‘that’-clause of every propositional attitude ascriptions contains a hidden indexical—an indexical

that has no syntactic representation. This hidden indexical picks out the propositional guise

under which the ascriber (or maybe the ascribee, or both) takes the proposition that ‘that’-clause

would ordinarily express were it not embedded in a propositional attitude ascription, and the

indexical contributes that propositional guise semantically to content. Hence, despite the

identical contributions of the proper names in (3) and (4)—they both contribute Kent-Super, the

‘that’-clauses do not co-refer—they refer to different entities, which I shall call ‘singular-

propositions-under-propositional-guises.’ Because the ‘that’-clauses of (3) and (4) do not co-

215
refer—they refer to different singular-propositions-under-propositional-guises—(3) and (4) do

not express the same proposition and may differ in truth-value. (3) and (4) state that Lois stands

in the belief relation to different singular-propositions-under-propositional-guises.

This account might sound similar to the Guise Millian Approach (e.g., that of Salmon),

except that, according to the Guise Millian Approach, propositional guises are not semantically

contributed to the propositions expressed by (3) and (4). On a Millian view like Salmon’s, the

truth-value of (3) and (4) are a function only of the singular proposition to which the ‘that’-

clauses refer. Since the ‘that’-clauses refer to the same singular proposition, (3) and (4) express

the same proposition and have the same truth-value. Although Salmon recognizes the existence

of the BEL relation, which is ternary, unlike HIT theorists Salmon maintains that the belief

relation reported by attitude ascriptions is binary: it is the psychological relation between the

singular proposition semantically expressed by the ‘that’-clause and the ascribee.

My principal criticisms of HIT are:

(a) HIT is ad hoc because it only solves the Propositional Attitudes puzzle and does

not solve the Identity Sentences puzzle, which is the more fundamental puzzle,

(b) HIT is otiose/superfluous because any solution to the Identity Sentences puzzle

(whether a pragmatic or semantic solution) should automatically entail a solution

to the Propositional Attitudes puzzle; we should start with the identity sentences

puzzle and then solve the propositional attitudes puzzle based on that solution;

and

(c) On the HIT, all propositional attitude ascriptions contain a hidden indexical

referring to a propositional mode of presentation, even those ascriptions in which

216
 the speaker has no intention of indicating any propositional mode of presentation

under which he, the ascribee, or the audience take the proposition. This is a worry

for the HIT because in many cases, speakers express what I call ‘Millian

ascriptions.’ They have no intent to communicate the conceptions or guises under

which they take the singular proposition referred to by the ‘that’-clause.

With the Identity Sentences puzzle, we seek to explain why (1) and (2) differ in cognitive

value. Guise Millians will explain the difference in cognitive value when Lois sees them written

or hears them uttered in the following way: Lois takes (1) and (2) to express different

propositions, because, when presented by these sentences, she takes these propositions expressed

by (1) and (2) under different propositional guises and she fails to realize these are guises of the

same proposition. That is why they differ in informativeness and cognitive value vis-à-vis her.

But an enlightened speaker, by contrast, knows that Clark Kent is Superman. How then can an

enlightened speaker take (1) and (2) under different propositional guises? Well, there is a story

that an adherent of Millian Pragmatist Approach could tell: the enlightened speaker confuses

semantics and pragmatics. That confusion leads the enlightened speaker to say, in error, that (1)

and (2) express different propositions, when what he really should say is that they pragmatically

suggest/convey different propositions. An HIT theorist might adopt the following hybrid

strategy by claiming both that the appearance to enlightened speakers of cognitive value

differences between (1) and (2) is an illusion rooted in the fact that (1) and (2) pragmatically

convey different information and enlightened non-Millian speakers fail to distinguish semantics

from pragmatics. Here, ordinary speakers with non-Millian intuitions are laboring under an

illusion. On the other hand, the apparent difference in truth-value between propositional attitude

217
ascriptions (3) and (4) to enlightened speakers is not an illusion, according the HIT theorist,

because there is a genuine difference in truth-value (due to the different contributions made by

the hidden identical to the semantic content of the ‘that’-clauses of propositional attitude

ascriptions). So in one puzzle, the identity sentences puzzle, there is a conflation of semantics

with pragmatics, whereas in the attitude ascriptions puzzle, there is no such conflation. This is

an ad hoc strategy. We have one strategy to solve one puzzle—claiming there is a cognitive

value illusion in the Identity Sentences puzzle (rooted in a conflation of semantics and

pragmatics by enlightened speakers), and a completely unrelated and dissimilar strategy to solve

another puzzle—claiming there is no truth-value illusion in the Propositional Attitudes puzzle.

This strategy would not be ad hoc if the Identity Sentences puzzle and the Propositional

Attitudes puzzle were distinct puzzles. I take that to be implausible just because I take the

puzzles to be two versions of the same puzzle. Take (1) and (2), with their different cognitive

values (even for enlightened speakers). As soon as you embed them in the ‘that’-clause of a

propositional attitude ascription, you have an apparent difference in truth-value. Given that the

only thing happening in this case is the embedding of (1) and (2), the apparent difference in

cognitive value and the apparent difference in truth-value are intimately related. They seem to

both be explained by the same underlying phenomenon. Ergo, whichever phenomenon explains

the cognitive value difference in the Identity Sentences puzzle must be responsible for the

difference in truth-value (or an appearance of a difference) in the Propositional Attitudes puzzle.

The problem with the HIT is that it does not respect this and unwarrantedly treats the puzzles as

distinct.

We should start out with solving the identity sentences puzzle and see if that leads to a

solution to the propositional attitudes puzzle. It seems to me that whatever solution one arrives

218
at for the identity sentences puzzle —be it a semantic solution claiming that (1) and (2) express

different propositions, or a pragmatic one that claims they express the same proposition but

pragmatically suggest different propositions)— that solution should naturally entail a solution to

the propositional attitudes puzzle. Hence, seeking a solution to the propositional attitudes puzzle

should begin with seeking a solution to the identity sentences puzzle.

Let us suppose the solution to the Identity Sentences puzzle turned out to be pragmatic in

nature, such as proposed by some Millians. We explain the difference in apparent cognitive

value for enlightened speakers between (1) and (2) as an illusion generated by the different

pragmatic implications the sentences have and by the failure of ordinary speakers to distinguish

between semantics and pragmatics. This solution entails the following solution to the

Propositional Attitudes puzzle: the apparent difference in truth-value between (3) and (4) would

be explained as an illusion generated by the further illusion that the propositions expressed by (1)

and (2), which are identical to the propositions referred to by the ‘that’-clauses of (3) and (4), are

different propositions. In other words, one could explain the appearance of a truth-value

difference between (3) and (4) as an illusion directly inherited from the illusion that (1) and (2)

express different propositions. This would obviate the need to posit a hidden indexical. For

example, we would explain Jimmy Olson’s (allegedly) erroneous belief that (3) and (4n) are

consistent as follows: Jimmy thinks that the propositions to which the ‘that’-clauses of (3) and

(4) refer, which are also the propositions that would be expressed by un-embedded utterances of

(1) and (2) respectively, are different propositions. So Olson thinks that ascriptions (3) and (4)

report Lois Lane as standing in the belief relation to different propositions.

Suppose, on the other hand, that the correct solution to the Identity Sentences puzzle were

semantic (as, e.g., on the TIUT), explaining the difference in apparent cognitive value between

219
(1) and (2) via a real and non-illusory difference in the propositions expressed by those

sentences. The Propositional Attitudes Ascriptions puzzle would automatically be solved.

Ascriptions (3) and (4) would in fact express different propositions: they would say that Lois

stands in the belief relation to two different propositions. The appearance of a difference in

truth-value between (3) and (4) would be easily explained as a genuine and non-illusory

difference in truth-value following from the fact that the ‘that’-clauses in (3) and (4) really would

refer to different propositions.

Another problem for the HIT is its failure to distinguish what I call Millian ascriptions

from Conception-indicating ascriptions. They are treated in the same way on the HIT. For the

HIT proposes that all propositional attitude ascriptions contain a hidden indexical that makes

reference to the propositional mode of presentation, and propositional attitude ascriptions express

the relation between ascribees and propositions under the propositional modes of presentation

under which the ascribees take them. But with many, if not most, propositional attitude

ascriptions, the intent of the speaker is not to communicate any information about propositional

modes of presentation. When I utter ‘John believes that Bill Clinton is tall,’ I may very well have

no idea how John conceives Bill Clinton. Therefore, it is highly implausible to claim, as the HIT

does, that in uttering the ascription sentence I say anything about the propositional mode of

presentation under which John takes the proposition referred to by the ‘that’-clause of the

ascription (‘that Clinton is tall’). My intent in uttering the ascription sentence may be just to

inform my audience that John believes of Clinton that he is tall, without taking any position

whatsoever on how John conceives Clinton. By contrast with the HIT, the TIUT recognizes a

distinction between the sorts of propositional attitude ascriptions in which the ascriber does

intend to communication information about conceptions (Conception-indicating Ascriptions),

220
and those in which the ascriber does not (Millian ascriptions). That there are these two sorts of

propositional attitude ascriptions seems to be a datum that any theory of proper names must take

into consideration. The HIT’s failure to do so is a mark against the theory.

221
 CHAPTER 7

FORBES’ THEORY OF PROPER NAMES

On Forbes’ theory of proper names (1990), as on the TIUT, a dossier is locus in the mind

where an agent stores conceptions that s/he takes to be about the subject of the dossier. Forbes

holds that dossiers are labeled with proper names. But there are significant differences in the

semantics of belief reports according to Forbes’ theory and the TIUT. On Forbes’ theory, the

semantics of propositional attitude ascriptions (5) and (6) are schematized as follows (the

semantics of sentences (5) and (6) are schematized by 5-F and 6-F, respectively):

(5) Lois Lane believes that Clark Kent flies

5-F Clark Kent is such that, for Lois’ so-labeled way of thinking of him, α, Believes
(Lois, <α, flies>)

(6) Lois Lane believes that Superman flies

6-F Superman is such that, for Lois’ so-labeled way of thinking of him, β, Believes
(Lois, <β, flies>)

In the case of her ‘Clark Kent’ labeled dossier, Lois has a conception of Kent-Super (the subject

of the dossier), α, that represents him as being incapable of flight. In the case of her ‘Superman’-

labeled dossier, Lois has some conception of Kent-Super (the subject of the dossier), β, that

represents him as being capable of flight. Forbes’ theory seems to work well in the case of

ascription sentences (5) and (6) because it makes plain why (5) is false and (6) is true. Lois has

two dossiers that represent Kent-Super in different ways and as having different properties. Lois

does not realize that these dossiers have the same subject, nor does she realize that the

conceptions in the dossiers are about the same individual, Kent-Super.

222
 Forbes’ theory is known as a “logophoric” analysis of proper names because the term ‘so-

labeled’ is a logophor. A logophor is an expression that refers back to a previously mentioned

expression (contrast this with an anaphor, which is an expression that refers back to a previously

mentioned object/individual.) 104 With 5-F and 6-F (which schematize the underlying semantic

structure of (5) and (6) respectively), the expression ‘so-called’ is logophoric because it refers

back not to Clark Kent or Superman (i.e., Kent-Super, the flesh-and-blood man himself), but

rather to the names ‘Clark Kent’ and ‘Superman’ qua labels for dossiers containing different

conceptions of Kent-Super.

Unlike the TUIT, Forbes’ account claims that no reference is made to the ascriber’s

‘Clark Kent’-labeled or ‘Superman’-labeled dossier(s). Nevertheless, on Forbes’ account it is

implicit that the ascriber has ‘Clark Kent’-labeled or ‘Superman’-labeled dossiers (or perhaps a

single, doubly labeled dossier) by virtue of the fact that the ascriber utters these names. He must

draw these names from a dossier or dossiers that he, the ascriber, labels with these names.

On the TIUT, the ascriber aims to entertain dossiers that match Lois’ in terms of subject

and conception so that his mental architecture approximates hers. Attitude ascriptions are

possible, according to the TIUT, because the ascriber can see things from the perspective of the

ascribee; or at least, he or she attempts to model his or her mental architecture so that it matches

the ascribee’s as much as possible. Then, the ascriber can indexically gesture at his or her own

dossiers and refer to them. These dossiers are of the same type as Lois’, matching hers in both

subject and conceptions, so it is as if they were Lois’. Forbes (1990, 548) call this an “a believer

oriented” theory of belief ascriptions, and he makes it clear that his theory is not of this sort.

104
E.g., in the sentence “Lois thinks Clark Kent is weak, but in reality, he is not,” the expression
“he” refers back to Clark Kent, the man himself, and not to the expression “Clark Kent.” Here,
“he” is an anaphoric expression).
223
According to Forbes’ theory, the ascriber does not aim to discern conceptions that the ascribee

associates with the name ‘Clark Kent’ or ‘Superman’ or the sort of conception that the ascribee

has in his or her respective ‘Clark Kent’-labeled and ‘Superman’-labeled dossiers.

Forbes’ theory comes up against two very serious obstacles. The first consists in

explaining the possibility of attitude ascriptions in cases where the ascribee has dossiers without

proper names labeling them. Consider Jennifer Saul’s case, mentioned above, where a woman,

Nicole, meets Kent-Super in both Clark Kent and Superman personas, and she thinks that Clark

is boring and drab while Superman is witty and urbane. However, she does not learn either the

name ‘Clark Kent’ or ‘Superman.’ Lois would presumably have two dossiers: one for the Clark

Kent persona, and another for the Superman persona. Neither of these dossiers would be labeled

with these names. Perhaps they are labeled with definite descriptions, or with images (i.e. what

the subject of the dossier looks like), or perhaps they would have no labels at all. Despite the

lack of labels on Lois’ dossiers, the ascriptions sentence ‘Nicole believes that Clark Kent is dull

and boring and Superman is witty and urbane’ is manifestly true. But, on Forbes’ account, the

sentence would be false. For Lois does not have ‘Clark Kent’- and ‘Superman’-labeled dossiers.

The second problem comes from Kripke’s Paderewski puzzle. Peter has two

‘Paderewski’-labeled dossiers. The sentence ‘Peter believes that Paderewski had musical talent’

says, on Forbes’ account, that Paderewski’s having musical talent is consistent with the

conception in Peter’s ‘Paderewski’-labeled dossier. The sentence ‘Peter believes that

Paderewski did not have musical talent’ says, on Forbes’ account, that Paderewski’s having

musical talent is inconsistent with the conception in Peter’s ‘Paderewski’-labeled dossier. How,

on Forbes’ account, are we to distinguish between Peter’s two ‘Paderewski’-labeled dossiers, and

the different conceptions in them, if the only information we know about the dossier is the

224
name(s) they are labeled with? These labels are identical, so we have no way to distinguish

between Peter’s two dossiers, one with a musician conception and the other with a politician

conception. On Forbes’ account, the ascriptions sentences contradict one another. The ascriber

is guilty of contradicting himself.

The TIUT is not vulnerable to either of the problems mentioned above. The conception-

indicating character provides that Nicole has a dossier token, which is of the same dossier type as

the ascriber’s ‘Clark Kent’ dossier, and that Nicole has another dossier token, which is of the

same dossier type as the ascriber’s ‘Superman’ dossier token. Nicole does not need to have

dossiers that represent their subjects as bearing any particular names. That Nicole’ dossiers

represent their subjects as bearing different names from the corresponding dossiers of the

ascriber (or even the possibility that Nicole’ dossiers do not contain any representations at all

with respect to the names the subjects of her dossiers bear) does not prevent her dossiers from

being of the same type as the ascriber’s. Her dossiers can be of the same type even if the

conception they contain fail to match the exact conception contained in the ascribers’ dossiers.

So, even if Nicole’ two dossiers lack labels and even if the representation contained in her ‘Clark

Kent’ and ‘Superman’ dossiers fail to match the ascriber’s, her dossiers can be of the same type

as those of the ascriber.

The TIUT also can solve Kripke’s puzzle, where Forbes falters. As argued in section

2.12, the fact that the ascriber has only one name for Paderewski is no obstacle to his coining

partially descriptive names to refer to his two dossiers on Paderewski: one that contains a

musician conception and one that contains a politician conception. He can describe Peter’s

beliefs by uttering the Conception-indicating ascription ‘Peter believes that Paderewski the

musician has musical talent, but not that Paderewski the Politician did.’

225
 CHAPTER 8

THE SAUL CASES

The Saul cases, first discussed by Jennifer Saul in her 1997, presents an additional puzzle

(not presented by either version of Frege’s puzzle) that any theory of proper names must explain.

Namely, the substitution of one proper name for another can (seem to) change the truth-value of

a sentence, even for enlightened agents, and even outside of the context of propositional attitude

ascriptions (i.e., even when the substitution occurs outside of the ‘that’-clause). In Frege’s

puzzle concerning propositional attitude ascriptions, apparent changes in truth-value were

occasioned only by substitutions inside of the ‘that’-clause. Substitutions outside of ‘that’-

clauses in ‘simple’ sentences (i.e., sentences that are not propositional attitude ascriptions) only

changed the cognitive value of the sentence, never its truth-value. For example, enlightened

speakers perceive cognitive value differences between (1) and (2), but they do not perceive any

truth-value differences. An enlightened speaker may rationally think that (1) and (2) express

different propositions, but he or she may not rationally think that they differ in truth-value.

The Saul cases considered here are (17)-(18) and (19)-(20)

(17) Superman is more successful with women than Superman

(18) Superman is more successful with women than Clark Kent

(19) Clark Kent went into the phone booth and Clark Kent came out

(20) Clark Kent went into the phone booth and Superman came out

The intuition of enlightened ordinary speakers is that (17) is necessarily false, (18) is true,

and that (19) and (20) express inconsistent propositions, so (19) and (20) must necessarily have

226
different truth-values.

Saul, a Millian, claims that both (17) and (18) are literally false. (17) is false because a

person cannot be more successful with women than himself. (18) is false because, according to

Millianism, it expresses the same false proposition as (17). Saul claims, contrary to the intuition

of ordinary speakers, that (19) and (20) cannot differ in truth-value, because Clark Kent going

into the booth and Clark Kent coming out is the same thing as Clark Kent going in and Superman

coming out, since Clark Kent is Superman. For the Millian, both (19) and (20) express the same

proposition, the proposition that Kent-Super went into the phone booth and Kent-Super came

out. Ordinary enlightened speakers’ intuitions line up with Millian truth conditions only in the

case of (17) (where the intuition, shared by Millians and ordinary speakers, is that (17) is false).

I contend, in agreement with Moore (1999) and Pitt (2001), that the cases Saul puts

forward do not involve substitution of co-referential names at all, at least, as far as enlightened

speakers are concerned. In agreement with Moore and Pitt, I claim that ‘Superman’ and ‘Kent’

as they appear in (18) and (20) above are not co-referential when uttered by enlightened

speakers. These cases therefore do not constitute cases of substitution of co-referential names.

The appearance of a difference in truth-value in those cases is explained by the TIUT (and

Moore and Pitt) as a genuine difference in truth-value. Indeed, it is explained by ‘Clark Kent’

and ‘Superman’ referring to different things.

With Moore and Pitt, I hold that these names in (18) and (20) (when uttered by

enlightened speakers) refer to something like different aspects of the same individual. Or maybe

they refer to individuals as they outwardly appear differently. ‘Kent’ means something like

Kent-Super-wearing-a-suit-and-glasses, and ‘Superman’ means something like Kent-Super-

wearing-a-superhero-costume. Kent-Super-wearing-a-suit-and-glasses is not identical to Kent-

227
Super-wearing-a-superhero-costume (even if the person who underlies those entities is one single

person). So ‘Clark Kent’ and ‘Superman’ are not referring to something identical in these

sentences. The names fail to co-refer in this context.

Enlightened speakers will utter (18) to express the proposition that Kent-Super-wearing-

a-superhero-costume is more successful with women than is Kent-Super-wearing-a-suit-and-

glasses. This is true. The enlightened speaker will utter (20) to express the proposition that

Kent-Super-wearing-a-suit-and-glasses went into the phone booth and Kent-Super-wearing-a-

superhero-costume came out. This is obviously distinct from the proposition that Kent-Super-

wearing-a-suit-and-glasses went into the booth and Kent-Super-wearing-a-suit-and-glasses came

out. When enlightened speakers are trying to draw a contrast, they use the names to refer to

these different aspects of Kent-Super.

Suppose that a policeman arrives on the scene of a crime and wants to know what was

observed. Two enlightened speakers, bystanders 1 and 2, are asked. Bystander 1 tells the police

office that he observed Clark Kent go into the phone booth and Superman come out. But that is

not what happened according to bystander 2. He observed Clark Kent dressed in a business suit

go in and then Clark Kent came out, still dressed in a business suit. In this case, to contradict the

bystander 1, bystander 2 utters (19) using the names to refer to aspects or appearances. His

second use of ‘Kent’ is contrasted with bystander 1’s use of ‘Superman’.

We can see that the enlightened speakers in the example, bystander 1 and bystander 2, do

not use the names ‘Clark Kent’ and ‘Superman’ co-referentially. Here we will consider what

they would say in case the police officer on the scene were a Millian philosopher, who is

enlightened as to the identity of Clark Kent and Superman:

Bystander 1: I observed Clark Kent going into the phone booth and Superman
came out.

228
 Bystander 2: Bystander 1 does not know what he is talking about. I observed
Clark Kent go into the phone booth but Clark Kent, not Superman,
came out.

Millian Policeman: You guys are not disagreeing with one another. You observed the
same things. You are both saying the same thing: that Kent-Super
went in, and Kent-Super came out. For after all, Clark Kent and
Superman are the same person—Kent-Super.

Bystander 1: No, we genuinely disagree about what happened. I realize, of

course, that ‘Clark Kent’ and ‘Superman’ are names of the same
person. But when I uttered ‘Clark Kent’, I was referring to Kent-
Super qua reporter, dressed a particular way. I was saying that
Kent-Super qua reporter went into the phone booth and Kent-Super
qua Superhero came out. The other guy is saying that Kent-Super
qua reporter went into the phone booth and Kent-Super qua
reporter came out.

The phenomenon of ordinarily co-referential names being used in non-co-referential

ways is omnipresent in language. Consider the German saying: Was Hänschen nicht lernt, lernt

Hans nimmermehr. This means literally: What Hänschen (little Hans) does not learn, Hans (as

an adult) will never learn. Roughly, this is equivalent to the English expression: you can’t teach

an old dog new tricks. Here, Hans and Hänschen are names of the same (hypothetical) person,

but refer to him at different life stages or aspects. Or one may sensibly say, without

contradiction: ‘I’ve never been to Leningrad, but I have been to St. Petersburg.’ If the names as

used here were co-referential, these sentences would make no sense; the speaker would have said

something nonsensical or contradicted himself. Surely, the names must refer here to the city in

different historical periods, i.e., different temporal and historical stages of the city.

The foregoing claim that names such as ‘Clark Kent’ and ‘Superman’ refer to different

aspects of Kent-Super applies only to enlightened speakers in the Saul cases. It is crucial that we

distinguish between enlightened and unenlightened speakers. Suppose an unenlightened speaker,

such as Lois Lane, utters (18). She does not realize that realize that Clark Kent and Superman

229
are the same person, so she would not use the names to refer to different aspects of him. She

uses the name to refer to what she takes to be two different individuals. She uses ‘Clark Kent’

and ‘Superman’ as Millian names. As uttered by Lois, (18) expresses the proposition that Kent-

Super is more successful with women than Kent-Super. That is necessarily false. As uttered by

Lois, (20) expresses the proposition that Kent-Super went into the phone booth and Kent-Super

came out, and this is exactly what (19) would mean if she said it. So (19) and (20) do not differ

in truth-value when she, an unenlightened speaker, utters them. This might seem worrisome, for

suppose that Lois is also on the scene and utters (20) to the police officer to express her

agreement with bystander 1 about what occurred. If Lois just expressed the proposition that

Kent-Super went in and Kent-Super came out, then her statement is no more in agreement with

bystander 1 than with bystander 2. But clearly, Lois’ utterance of (20) and her dissent from (19)

would strongly support bystander 1’s version of what occurred at the scene. How do we explain

this? I am going to bite the bullet and say that in fact an unenlightened speaker such as Lois Lane

would not express the same propositions that an enlightened speaker would express in uttering

the same sentences. Enlightened and unenlightened speakers have radically different knowledge

of the facts, so it should not be surprising that this difference in knowledge leads to a difference

in the propositions they express. I am claiming that when Lois utters (18)

(18) Superman is more successful with women than Clark Kent

she semantically expresses the false proposition that Kent-Super is more successful with women

than Kent-Super, which is obviously false. This is not as counterintuitive as it might seem at

first sight. Although (18), as uttered by Lois, literally expresses a false proposition, nevertheless,

230
enlightened speakers who hear Lois utter (18), knowing that she is unenlightened, can agree with

Lois’ statement by ‘translating’ her statement into other terms: the enlightened speaker will agree

that Kent-Super as Superman is more successful than Kent-Super as Clark Kent.

Even though the proposition expressed when Lois utters (20) does not support bystander

1’s version of the events any more than bystander 2’s does, which is somewhat counterintuitive,

the counter-intuitiveness is blunted once we realize that we would consider not just the

proposition Lois semantically expresses in order to assess what she observed. We would also

ask: who, out of bystander 1 or bystander 2, would Lois have agreed with had she known about

the identity of Clark Kent and Superman? If she says that she saw Superman come out of the

phone booth, given that she is unenlightened, we can infer that she saw Kent-Super dressed as

Superman coming out, and this means also that she would have used the name ‘Superman’ to

refer to his Superman aspect had she been enlightened when she made the statement. That is, we

can use our knowledge about what unenlightened agents know and do not know, and the

descriptive properties that they associate with names, to ‘translate,’ in a manner of speaking,

their statements that are not about aspects into talk about aspects. When we carry out this

translation, we see that Lois’ statement supports bystander 1’s version of the facts.

Moreover, were Lois to become enlightened about the identity after the incident, she

would be disposed to utter sentence (20), both before and after the enlightenment, although the

propositions she would express would be different vis-à-vis her utterances pre- and post-

enlightenment.

Importantly, I deny that names used to refer to aspects (or individuals outwardly

appearing differently) are the same thing as Conception-indicating names. ‘Clark Kent’ and

‘Superman,’ when used as Conception-indicating names, are co-referential—they both refer to

231
the same person—Kent-Super. Names used to refer to different aspects of an individual are, by

contrast, not co-referential.

Hence, we could not solve either of Frege’s puzzles by claiming that ‘Clark Kent’ and

‘Superman’ refer to different aspects of Kent-Super. If the names ‘Clark Kent’ and ‘Superman’

referred to aspects of Kent-Super in sentence (2) (‘Clark Kent is Superman’), then (2) would turn

out to be false. Clark Kent is identical to Superman. But different aspects of Kent-Super, his

Clark Kent aspect and his Superman aspect, are not identical to one another. (Two different

aspects of an individual are not identical—that is what makes them two aspects). If the names in

(1)-(4) referred to different aspects of Kent-Super, Lois Lane’ belief that Kent is not Superman

would be true, and Olson’s belief that Clark Kent is Superman would be false.

The Frege cases involve co-referential names used rigidly to refer to individuals, and the

Saul cases do not. In the Frege cases, ‘Clark Kent’ and ‘Superman’ both refer to Kent-Super. In

the Saul cases, ‘Clark Kent’ refers to one aspect of Kent-Super and ‘Superman’ to a different

aspect of him. The Saul cases show that there is in fact a third sort use to which a proper name

may be put—to refer to an aspect of an individual, in addition to using it in a Millian or

Conception-indicating way. The TIUT should therefore not be understood as a comprehensive

description of all the ways that proper names can be used. Rather, the TIUT is a theory of proper

names as used rigidly to refer to individuals (rather than aspects of them).

232
 CHAPTER 9

ANTI-FUNCTIONALIST THEORIES

In the preceding sections of this dissertation, I have discussed various theories of proper

names—Descriptivism, Millianism, and Indexical theories (including the TIUT). The common

thread running through these theories is the assumption that proper names are primarily devices

of reference. That is, a proper name is an expression semantically used by a speaker to denote a

particular object. In all of these theories, a name determines a function from worlds to individual

objects (or entities of some sort). On Indexical theories, this function varies from one context to

another; on Descriptivism, this function picks out different individuals across worlds; and on

Millianism the function is a constant function, picking out the same individual in all worlds.

By contrast with the above theories, which we may call “referentialist” or

“functionalist theories” (because the semantic value of a name is always some function), in the

past decade “anti-functionalist” or “non-functionalist” theories have become popular. These

theories posit that names are not primarily devices of reference. Instead they have in the first

instance other sorts of meanings that explain why they can be used to refer to objects or entities.

In other words, according to these theories, the primary semantic meaning of a proper name is

non-referential, and the referential use to which they can be (and regularly are) put is explicable

as a consequence of this primary semantic meaning. To clarify, Anti-functionalists do not deny

that names have an important referential use, but merely that this use is best explained in terms

of names’ non-referential meaning. Anti-functionalists think their views are better poised than

functionalism to explain some puzzling data about names.

The two principal versions of anti-functionalism are Variablism and Predicativism.

Variablism (Cumming 2008) takes as it’s starting point the apt observation (probably first made
233
by Dever (1998)) and developed by Cumming (2008) that in some syntactic positions names can

be bound to an indefinite antecedent, much in the way variables can be. Cumming uses this

evidence, as well as further evidence he adduces with respect to names, to claim that names are

also used as bound variables when they occur in other contexts where they look like they are

used primarily as devices of singular reference. The second sort of Anti-functionalism is

Predicativism, which maintains that proper names are predicates. The clearest statement of

Predicativism at present is found in the paper “Names are Predicates” (2015) by the recently

regretted Delia Fara. According to Fara, names such as ‘John’ means the same as the definite

description “the bearer of ‘John’”, where the ‘the’ is an unpronounced or suppressed constituent

of the predicate.

9.1 Variablism

I shall examine Cumming’s version of Variablism, which is articulated in his 2008.

Cumming’s starting point is his observation on page 536 (in section 2.3 of his paper) that in his

sentence (14) (reproduced below as my (21)), the second occurrence of ‘Ernest’ in the second

sentence appears to be bound by the first (which is mentioned but not used).

(21) There is a gentleman in Hertfordshire by the name of

‘Ernest’. Ernest is engaged to two women.

Moreover, this second occurrence does not have singular reference, at least a reading of (21) on

which the speaker has no particular person named ‘Ernest’ in mind, but rather is making the

general claim that some gentleman in that county has that name and is engaged to two women.

One such a reading of (21), which is a plausible reading even if not perhaps the most intuitive

234
one, its truth conditions would be given by:

(21*) ∃xyz (gentleman x ∧ in-Hertfordshire x ∧ named-Ernest x ∧ woman y ∧ woman z

∧ y = z ∧ engaged xy ∧ engaged xz)

The second occurrence of ‘Ernest’ functions as an existentially bound variable, rather than a

referential expression with singular truth conditions. This data alone would not likely be strong

support for Variablism—the view that names are variables, rather than the more modest claim

that there are occasional occurrences of names where they can be used as variables. After all, a

referentialist might admit that names could be put to various uses, but still maintain that they are

principally devises of singular reference that admit of derivative uses that deviate from that

primary semantic use, which it to refer. However, Cumming also expends a great deal of effort in

his 2008 arguing for the thesis that names are variables tout court, and he wants to demonstrate

this by showing how considering names to be variables can explain interesting and puzzling data

about proper names. This would suggest that names are variables in the first instance and that

referential uses are derivative of their use as variables. In these other contexts, he argues that

considering names to be unbound variables (where they appear to be used referentially). Thus,

he claims that names are always variables, sometimes bound (as in (21)) and sometimes unbound

elsewhere.

The first piece of evidence that names can function as unbound variables begins in

section 2.1 on page 529. Cumming claims to have found a distinction he claims is akin to the de

re/de dicto distinction but with proper names rather than with definite descriptions. I advert the

reader to that section of Cumming’s paper for the details of the discussion, which are complex,

and I shall not reproduce his argument here. I shall claim that Cumming has not discovered any

distinction similar to a de re/de dicto distinction, even if he has discovered an interesting

235
linguistic puzzle in need of explanation. Furthermore, I shall argue that positing that names

function as bound variables is not an apt way to solve the puzzle, and I shall gesture at an

alternate way of solving it.

In what follows, I shall here presuppose that the reader is familiar with Cumming’s

discussion in section 2.1 of his 2008. Cumming states the truth-conditions of sentence (KR)

(KR) Biron thinks Katherine (de re) is Rosaline (de dicto)

on page 549 (bottom) and 550:

“... (KR) states that that there is some use that refers to Katherine, but that Biron
believes co-refers with ‘Rosaline’ (or more carefully, with the relevant use of
‘Rosaline’). This is intuitively true in the scenario described in section 2.1—so
long as we allow “uses” to extend to internal deictic symbols—since the thrust of
the example is that Biron has some internal deixis trained on Katherine, which he
inwardly connects with the relevant use of ‘Rosaline.’ [Page 549-550]

As a preliminary matter, I think that the above paragraph mischaracterizes what Biron believes.

Cumming says that for (KR) to be true, Biron must believe that his internal deictic representation

of Katherine is connected to a relevant use of ‘Rosaline.’ But Biron might believe that Katherine

is Rosaline even if he does not know that Rosaline bears the name ‘Rosaline’, for Biron might be

familiar with Rosaline but not know her name. (KR) could still be true under these

circumstances. I think what needs to be said instead is that (KR) is true iff Biron has a deictic

representation of Katherine and some internal representation of Rosaline (which might not be the

name ‘Rosaline’) and he connects these representations (i.e., he believes they co-refer or are

about the same person).

I do not think Cumming has identified a de re/de dicto-like distinction with respect to

names. He wants to draw a parallel with definite descriptions and gives the example of (FL)

236
(page 530).

(FL) Biron thinks the person who entered first (de re) is the person who
entered last (de dicto)

The first definite description in (FL) is read de re and the second de dicto. His regimentation,

which has the first definite description outside the scope of ‘thinks’ and the second inside its

scope, nicely explain this. There is a stark difference between the two definite descriptions in

(FL). The first is used by the speaker merely to refer, to pick out a particular individual on the

basis of a property (having entered first) that the speaker thinks picks out that individual but

which the speaker does not think Biron believes picks out that individual. The second definite

description is used to say that Biron thinks the individual picked out by the first description has

the property attributed by the second definite description.

If the names case involving (KR) is analogous, there will have to be a similar difference

between the use of the two names ‘Katherine’ and ‘Rosaline.’ Now, if ‘Katharine’ is analogous

to the definite description the person who entered first in FL, then the speaker uses ‘Katharine’

merely to refer, to pick out a particular individual on the basis of a property that the speaker

thinks picks out that individual and which the speaker does not think Biron believes picks out

that individual. Now, one candidate for this property is the property being named Katherine. The

speaker does think this property picks out Katherine. But a problem is that, assuming Biron

knows that Katherine bears the name ‘Katherine’, Biron does realize that Katherine, who the

speaker is referring to, bears that name. What Biron fails to realize is rather that the person he

danced with at the masque bears the name ‘Katherine’. So we do not have an analogous use.

Furthermore, if ‘Rosaline’ is analogous to the definite description the person who entered last,

then the speaker uses it to say that Biron thinks the individual picked out by the first name

237
(‘Katharine’) has the property attributed by the second name (‘Rosaline’). Now, let us suppose

that the second name, ‘Rosaline’, is used to say that Biron believes that the property being named

‘Rosaline’ applies to the referent of the first name, ‘Katharine.’ Assuming Biron knows that

Rosaline bears the name ‘Rosaline’, it is true that that Biron thinks ‘Rosaline’ applies to

Katharine. So we do have some analogous use here.

I worry that Cumming’s theory is overly metalinguistic. Let us suppose that Biron does

not know either Rosaline or Katherine’s name. (KR) can be true even if Biron does not know

any names at all for Katherine and Rosaline. He might know both women without knowing their

names. In this case, the speaker of (KR) would not be asserting that Biron believes that Katherine

bears the name ‘Rosaline’. Biron believes no such thing. He believes that Katherine is Rosaline,

not that she bears that name. This shows a great disanalogy with (FL). For in (FL), the speaker

is attributing to Biron a belief about a property putatively borne by an individual. With (KR),

this is not happening. What is being attributed is identity between two individuals, not a

property. The names ‘Katherine’ and ‘Rosaline’ have the same role – to refer to two individuals,

whereas in (FL) the definite descriptions play two very different roles. If the names are being

used just to refer to individuals in (KR), having the same function, the (FL) and (KR) cases

cannot be analogous.

With (FL), the speaker uses the first definite description to refer to an individual and

represents that individual in a way that Biron would not agree with, since he would not agree that

that person came in first. With (KR), what representation about Katherine does the speaker of

(KR) use to identify her but that Biron would not agree with? Again, it is true that Biron would

not agree with her being named ‘Katherine’, but this is also true of the name ‘Rosaline’.

Furthermore, unless we think that the name ‘Katherine’ means something like person named

238
‘Katherine’ It does not seem to me that there is any such representation that Biron would not

agree with. It is not the assertion that she is named ‘Katherine’. That is too metalinguistic.

What Biron fails to realize is that she is Katherine, not that she is named ‘Katherine’.

After all, he might know her without knowing her name. So suppose that the speaker represents

that she is Katherine, but Biron fails to realize this. But this cannot be right. The fact that she is

Katherine is the trivial fact that the person who is Katherine is Katherine. That is just the fact

that Katherine is self-identical. Surely, Biron knows that. He is dancing with Katherine, and he

clearly knows that the person who he is dancing with is self-identical, and that person is

Katherine. He would not express that fact by uttering ‘Katherine is Katherine’, but he clearly

realizes the fact.

As far as I can see, the only thing Cumming achieves through his regimentation is to

explain why the speaker of (KR) asserts that ‘Katherine’ necessarily refers, while the speaker

makes no assertion that ‘Rosaline’ refers. As far as the speaker is concerned, ‘Rosaline’ could be

an empty name, given that it is within the scope of ‘thinks’. But this distinction between a

necessarily non-empty name position (or at least, one taken to be non-empty by the speaker) and

a potentially empty name position is not the de re/ de dicto distinction. Moreover, I am

suspicious of this distinction, since I doubt that empty names exist. Even ‘Santa Claus’ refers to

something— perhaps a fictional character. And I can imagine a true sentence in which the name

in the first position refers to something fictional (and thus “empty”). Thus, I can perfectly well

imagine that the sentence ‘Biron thinks that the Easter Bunny is Santa Claus’ could be true. (In

any case, the emptiness or non-emptiness of the name in the second name position is beside the

point, as in the scenario described in section 2.1, it is stipulated that both the names ‘Katharine’

and ‘Rosaline’ refer.)

239
 In short, I do not see how Cumming’s regimentation accounts for any difference between

‘Katharine’ and ‘Rosaline’ in (KR) to explain the difference between how the names function.

Clearly, there is a difference in need of explaining, but it does not seem to be a de re/de dicto

distinction, or even something like one. I think something else entirely explains the difference.

Contra Cumming, I do not see any reason to doubt the principle of the symmetry of identity in

these sorts of cases. If I believe that A is B, it follows that I believe that B is A. I see sentence

(RK) as true, strictly speaking, since (KR) is true.

(RK) Biron thinks Rosaline is Katherine

Admittedly, (RK) does strike us as intuitively false. However, I think the intuition is erroneous,

and we can explain the apparent falsity of (RK) based on pragmatic considerations.

To see that (RK) is strictly speaking true, consider the following scene taking place the

morning following the masque. Biron approaches Rosaline and says: “You are the person I

danced with at the ball last night”, intending “the person I danced with at the ball” to be a rigid

designator. He says, in effect: You are dthat [the person I danced with at the ball]. Since the

person he went to the ball with was Katharine, he rigidly designates Katherine. Thus, in saying

“You are the person I danced with at the ball” to Rosaline, he is telling her that she is Katherine.

He is not saying she is named Katharine, but he is asserting that she, Rosaline, is Katherine. By

the weak disquotation principle, he believes the proposition that Rosaline is Katherine. (Or

course, he also believes she is not Katherine when he considers the question under different

modes of presentation of Rosaline).

How should we explain the difference between the different occurrence positions of the

240
names in sentences like (KR)? Here’s a salient difference between the names positions in (KR).

Although Biron does technically believe that Rosaline is Katherine (as argued above), the nature

of his confusion is such that he would never think that Rosaline was Katherine in the sense of

calling her ‘Katherine.’ He would never kiss Rosaline by accident thinking she was Katherine, if

it was Katherine he liked and wanted to kiss, for he is not confused about what each of them

looks like. He still thinks of their personalities as distinct, and does not confuse them along this

dimension. His belief that Rosaline is Katherine is limited to a mode of presentation engendering

few interesting potential behavioral consequences. This is an asymmetry between the ways that

he believes that Katherine is Rosaline and Rosaline is Katherine. Much more needs to be said to

explain this asymmetry, but I think it is too quick to judge this asymmetry to being tantamount to

a violation of the principle of the symmetry of identity.

Perhaps there is another difference between the first and second name positions. The first

position tells the audience to focus attention, even perhaps visualize, that individual first. This

focus enables the audience to an utterance of (KR) to best understand the nature and cause of the

ascribee’s confusion. When you contemplate how Biron’s confusion of Katherine with Rosaline

arises, you probably picture Biron approaching Katherine, seeing Rosaline’s favor, then Biron

looking at that favor and calling up a thought about Rosaline. You first consider Katherine, and

then consider Rosaline. When your picture how the confusion arises, an image of Katherine is, at

first, present before your mind’s eye more vividly than Rosaline.

Because I consider this phenomenon to be explained by pragmatic factors or

communication purposes (which I have only gestured at), I do not think it will be necessary for a

theory of semantic content to address it. Cumming’s puzzle is interesting, but I do not see how

the variablist thesis solves it.

241
 The other factor that Cumming adduces in support of Variablism is his claim that it can

solve Frege’s puzzle about propositional attitude ascriptions. Cumming’s discussion of Frege’s

puzzle begins on page 544 in section 3.4. However, I shall now argue that Cumming’s solution

does not work.

Cumming (2008) claims that names can be represented at the level of logical form by

variables, with each name represented by a distinct variable. For example, on a token utterance

of sentence (22)

(22) Hesperus is visible

‘Hesperus’ could be represented by the variable xhes, so that the semantic representation of (1)

would be given by (221)

(221) visible xhes

On a token utterance of sentence (23)

(23) Phosphorus is visible

‘Phosphorus’ could be represented by the variable xpho, so that the semantic representation of

(23) would be given by (231)

(231) visible xpho

The truth-values of (221) and (231) would depend both on variable assignments (to variables

‘xhes’ and ‘xpho’) and world (for the extension of the predicate ‘visible’).

Cummings calls (22) and (23) “open sentences” (544) because the variables occur free.

242
Open sentences denote what Cumming calls “open propositions”, such as (221) and (231), which

are functions from variable assignments to what he calls “closed propositions,” which are open

propositions with assignment made to all variables.

Despite the word “proposition” occurring in the expression “open proposition,” open

propositions lack truth values and are therefore not propositions in the traditional sense,

according to which propositions are the bearers of truth-value and sentences may be said to be

true (or false) only in the derivative sense that they express a true (or false) proposition. This is

important to note, because, as I shall argue below, it will turn out that Cummings’ open

propositions are much closer to being sentences than propositions and this results in Cumming’s

approach to Frege’s puzzle about propositional attitude ascriptions being overly metalinguistic.

Although Cumming contrasts Variablism with Millianism, there is one sense on which he

is a Millian and another on which he is not. Millianism can either be defined as the view that the

meaning of a name is exhausted by its bearer, or as the view that a proper name contributes its

bearer and nothing else to the proposition expressed by the sentence in which it occurs. On the

former definition, Variablism is not classifiable as a species of Millianism because Cumming

posits the pre-assignment value of a proper name as an important semantic entity in addition to

the name’s referent. However, Cumming is a Millian under the second definition, since he holds

that only individual objects, not senses or other intensional entities, may be assigned to the

variables representing proper names, and the semantic value of proper names in closed

propositions is simply the referent and nothing more. At first glance, it appears that Cumming’s

view faces the same challenges that face Millians in accounting for cognitive value differences

(in Frege’s identity sentences puzzle) and apparent truth-value differences in Frege’s

propositional attitudes puzzle), since ‘Hesperus’ and ‘Phosphorus’ have the same closed content.

243
 Nevertheless, Cummings proposes a non-Millian solution to the attitudes puzzle. He

argues that propositional attitude expressions like ‘believes’ or ‘thinks’ operate on the open

propositions denoted by the complement clauses of propositional attitude ascriptions (rather than

the closed propositions, as is usually claimed). In this way, attitude ascription sentences such as

(24) and (25)

(24) Biron thinks that Hesperus visible

(25) Biron thinks that Phosphorus visible

state the relation between Biron and the open propositions denoted to by the complements

clauses, which are the open propositions (221) and (231), respectively. Thus, the content of (24)

and (25) can be schematized by (241) and (251).

(241) Biron thinks visible xhes

(251) Biron thinks visible xpho

Since Biron stands in the ‘thinks’ relation to two different open propositions (containing distinct

variables), (24) and (25) may have different truth values. Voilà, it appears we may have a neat

solution to Frege’s puzzle about propositional attitude ascriptions, showing how (24) and (25)

can differ in truth value despite the co-reference of ‘Hesperus’ and ‘Phosphorus’. This solution

to the puzzle is distinct from the most common Millian solution to the puzzle, which is to deny

that (24) and (25) semantically differ in truth value and claim they differ only in what they

pragmatically suggest.

Cumming explains what it means to stand in the belief relation to an open proposition

beginning in the final paragraph of page 545 and continuing onto page 546. In the final sentence

244
(page 546, top) he says:

“… you believe the aforementioned open proposition (visible xhes) iff you have inwardly
“tagged” that use of ‘Hesperus’ with the property of being visible (or iff you believe that
the referent of that use is visible).”

My worry is that the above definition of the belief relation to an open proposition reveals that

Cumming’s theory is far too metalinguistic to adequately address the propositional attitudes

puzzle. To see this, consider sentence (26) below:

(26) The Babylonians discovered that Hesperus was Phosphorus before the Greeks did.

On Cumming’s theory, an utterance of sentence (26) is true if the Babylonians discovered that

‘Hesperus’ referred to the same celestial body as ‘Phosphorus’ before the Greeks did. But the

problem is that the Babylonians did not use the names ‘Hesperus’ or ‘Phosphorus’ to refer to

Venus. Only the Greeks did. The Babylonians had different names for Venus in its morning and

evening star guises. So, sentence (26), which is intuitively well-formed and true, turns out to be

false on Cumming’s theory. Cumming’s attempted solution is precisely of the sort that Frege

rejected in On Sense and Reference when he stated that the discovery that Hesperus was

Phosphorus was an astronomical discovery, not a linguistic discovery. The worry is that

Cumming’s theory makes the discovery about language.

When used in sentences such as (26), where morning and evening star conceptions of

Venus are an essential and vital part of what is being communicated by (26), we require a theory

on which the names ‘Hesperus’ and ‘Phosphorus’ are used with a character that loads both

Venus and morning star and evening star conceptions into the proposition. The Two Indexicals

Theory of provides for this. On the TIUT, it makes no difference what names the Greeks and

245
Babylonians used for the morning and evening stars—the propositional attitude ascription

attributes the same beliefs/discoveries to the Greeks and Babylonians regardless of what names

they used to refer to Venus.

Another worry for Cumming’s view is pointed out by Bryan Pickel (2015), the essence of

which I think can be captured by the following hypothetical. Suppose that there were two pairs

of heavenly bodies appearing under morning and evening star modes of presentation and that the

name pair Hesperus/Phosphorus was used for both set of heavenly bodies. One heavenly body

was Venus, and the other was a distinct planet, Schvenus. Now suppose someone were to utter

sentence (26). Which heavenly body is being referred to, Venus or Schvenus? If we consider

the complement clause of (26) qua open proposition, there is no answer to the question, for qua

open proposition, no assignment has yet been made to the variables. Until an assignment is

made, there is no fact of the matter as to what (26) is about. Pre-assignment, (26) is about

nothing in particular. It is only the closed proposition that is about anything. Of course, the

open proposition does tell us some important information: we can tell from the open proposition

that there are two names putatively representing the same object, and so two variables in need of

assignments. We can tell from (26) qua open proposition that those variables are to be assigned

the same object, since the speaker is claiming the identity of those assignments. But that is

minimal information, leaving out important information that is communicated by an utterance of

(26): what object is the utterer of (26) referring to, Venus or Schvenus?, and what are the

contrasting conceptions he associates with these names? Open propositions are too skeletal to

constitute the full semantic content of the complement clauses of propositional attitude

ascriptions. A satisfactory solution to Frege’s puzzle about propositional attitude ascriptions

awaits a theory according to which the closed propositions denoted by the complement clauses of

246
(24) and (25) turn out to be distinct. The TIUT is intended to be such a theory. The TIUT

proposes to do this by building on the basic insight that names are indexicals or variables, but

building into the character of names very specific parameters that determine content. In some

cases (‘Millian names’), the content is simply the name’s referent, an object. In others

(‘Conception-indicating names’), the content is an intensional entity, a meaning, that denotes the

name’s referent. With sentences (24) and (25), the result is that the complement clauses have

different closed contents, while at the same time, the names ‘Hesperus’ and ‘Phosphorus’ rigidly

designate the same object, Venus.

9.2 Predicativism

Predicativism is the view that names are predicates on all occurrences, both when they

occur in predicate and in argument positions in a sentence. To say a name is a predicate is to say

the name is a true of its bearer. Thus, ‘Aristotle’ is true of Aristotle iff he is called ‘Aristotle’.

‘Aristotle’ is semantically equivalent to ‘the thing called ‘Aristotle.’’

Delia Fara, whose 2015 “Names are Predicates” is considered the gold-standard of the

Predicativist theory, contrasts Predicativism with “referentialism.” Uniform referentialists hold

that names are never predicates. As Fara correctly points out, this view is implausible, given that

there are examples of sentences in which names are clearly used as predicates, such as in the

sentence “There are many Smiths in the phone book”. “Smiths” clearly means persons called

‘Smith’. And no person or persons named ‘Smith’ are referred to by the use of “Smiths” in this

sentence. Nonuniform referentialists, on the other hand, agree with Fara that names are

sometimes used as predicates, but maintain that when “names occur as bare singulars in

247
argument position, they are referring expressions.” (Fara, 62). The novelty of Fara’s view

consists in the claim that names are predicates in both predicate and argument positions. I shall

argue that non-uniform referentialism is correct.

Here are some sentences in which names appear in the argument position that Fara should

find worrisome. Consider propositional attitude ascription (27).

(27) Jennifer thinks Dave is coming to the party.

Suppose that Jennifer knows Dave only by his middle name, “Scott”, a name Dave sometimes

goes by. If the semantic content of “Dave” in (27), is, as Fara would claim, the thing called

‘Dave’, then (27) would express the proposition that Jennifer believes the thing called ‘Dave’ is

coming to the party. However, she believes no such thing. She does not think that anything

called ‘Dave’ is coming to the party. She thinks rather that a thing called ‘Scott’ is coming to the

party. So according to Predicativism, (27) is false. But (27) is intuitively true; after all, Jennifer

does believe that Dave is coming to the party, even if she would call him ‘Scott’ rather than

‘Dave’.

To disarm this worry, Fara could argue that the name inside the complement clause of a

propositional attitude ascription should (at least sometimes) be read de re. The name ‘Dave’,

which is equivalent to the definite description ‘the thing called ‘Dave’’, could be read de re in

(27), so that it would be true if Jennifer believes of Dave that he is coming to the party. However,

this move would threaten to collapse her view into non-uniform referentialism. For in the vast

majority of propositional attitude ascriptions, we would need a de re reading, i.e., referentialist

reading, of the name. A de re reading would be mandatory wherever the speaker would consider

the sentence true even if it turned out that the ascribee did not know that name or used a different

248
name. This would be the case with most ascriptions, because generally we do not intend the

truth value of a propositional attitude ascription to depend on which name, if any, the ascribee

would use to denote the individual denoted inside the complement clause. The only rare case in

which would want a de dicto reading would be with sentence such as (28), where we are

specifically saying that a person, Oliver, does not realize two names are co-referential.

(28) Oliver does not believe that Tully is Cicero.

Here, there is a reading of (28) where it says merely that Oliver does not know the names ‘Tully’

and ‘Cicero’ co-refer. In this case, we want (28) to express the proposition that Oliver does not

know that the person called ‘Tully’ is identical to the person called ‘Cicero’.

This de dicto reading would not deliver the right truth-value for sentences such as (29).

(29) The Greeks did not originally know that Hesperus was Phosphorus

With (29), we not saying merely that the Greeks were confused about the co-reference of names.

Rather, (29) expresses the much broader proposition that the Greeks did not realize that Venus

under a morning star mode of presentation was Venus under an evening star mode of

presentation.

Since, if Predicativism were true, the reading of the names in complement clauses would

almost always have to be referential, or de re, Predicativism starts to look implausible. For

Predicativism claims that the rare exception case (the case where the name is read de dicto)

should serve as the model for how names are used in general. Why not instead say that names

are largely used referentially, and sometimes admit of other uses, including use as predicates, as

the non-uniform referentialist would say? This would be a simpler explanation.

249
 If Predicativism were true, we would also have to read most occurrences of names in

simple sentences (i.e., outside of propositional attitude ascriptions) as de re. Consider Jennifer’s

utterance of (30).

(30) Scott is coming to the party.

Jennifer’s friend Peter utters (31).

(31) Dave is coming to the party.

Again, ‘Scott’ and ‘Dave’ are names of the same person. Haven’t Jennifer and Peter said the

same thing? Don’t they fully agree on who is coming the party? Don’t (30) and (31) express the

same proposition? If Predicativism were true, it would be difficult to explain why we take

Jennifer and Peter to agree. On a de dicto reading, the semantic content of Jennifer’s utterance

would be: the thing called ‘Scott’ is coming to the party, and the semantic content of Peter’s

utterance would be: the thing called ‘Dave’ is coming to the party. These are different

propositions. It is only on a de re reading that (30) and (31) would express the same proposition,

such that we could claim that Jennifer and Scott made the same claims, which seems intuitively

to be the case. So we need de re readings of names outside of propositional attitude contexts as

well as inside of them.

The most serious worry is that Predicativism delivers the wrong content with sentences

like (26).

(26) The Babylonians discovered that Hesperus is Phosphorus before the Greeks did.

On a de dicto reading of the names, (26) is false. The Babylonians did not discover anything

250
about the names ‘Hesperus’ or ‘Phosphorus’, and even the discovery the Greeks made was not

about the names; it was an astronomical discovery, not a linguistic one. But even on a de re

reading of both names, (26) would not express the right proposition. (26) would say that the

Babylonians discovered that Venus was Venus before the Greeks did. That is not what (26) says.

(26) says they discovered that Venus presented under a morning star mode of presentation was

identical to Venus under an evening star mode of presentation.

I think we should distinguish between the truth conditions of sentences on the one hand,

and truth conditions of the propositions they express on the other. Predicativism does not

distinguish between these. Fara does not want to draw this distinction because, as she says, he

goal is a uniform treatment of names on all occurrences. But there are good reasons to draw this

distinction. Consider sentence (1) again.

(27) Jennifer thinks Dave is coming to the party.

There are numerous conditions that must obtain for sentence (27) to express the proposition it

does. The word “thinks” must refer to the property of thinking in English. “Is coming” must

refer to the property of coming to some place in English. “Party” must refer to parties in

English. These are all things that must be true such that the sentence expresses the proposition it

does, but these are not truth conditions of the proposition expressed. The proposition expressed

is true if there is a certain person of whom Jennifer thinks he is coming the party. What makes

the proposition true are not linguistic facts. Consider that the same proposition expressed by

(27) would also be expressed by (32) in German.

(32) Jennifer glaubt, dass Dave auf die Feier kommt.

251
The linguistic facts that explain why sentence (32) expresses the proposition it does are distinct

from the linguistic facts that explain why sentence (27) expresses the proposition it does. When

we are inquiring into what propositions are expressed by sentences, we seek to understand is

what must be true about the non-linguistic world such that (27) and (32) are both true, and we

thus need inquire into the truth conditions of the common proposition they express, rather than

the distinct truth conditions of the distinct sentences.

Where there are metalinguistic uses of names or other expressions, the fact that a person

or thing is called by a name is part of the truth conditions of the propositions expressed. The

sentence “My parents named me Delia” is such a sentence. This proposition expressed is true if

there is a person, who is Delia Fara, and her parents gave her the name ‘Delia’. That means that

the name itself is what the sentence is about in part—the name is a constituent of the proposition

expressed. But it seems to me this metalinguistic case is not the norm. It is the exception. It

should be considered a distinct use of a name from a referential use. There is no reason to think

that names only have one use, or that there is going to be some grand unified theory reducing all

of their uses to one.

The phenomenon of expressions being used metalinguistically is not limited to proper

names, but exist with many expression types in a language. Thus, consider sentence (33).

(33) Jennifer thinks that Dave is going to a party, but does not realize he is going to a
Feier

Here, “Party” and “Feier” (meaning party in German) are used metalinguistically. Here, ‘Feier’

means the sort of thing referred to by ‘Feier’. (33) is very plainly about Jennifer’s failure to

understand that ‘party’ and ‘Feier’ are synonymous. However, the meaning of ‘party’ is not the

thing in English called a ‘party’, nor is the meaning of ‘Feier’ the thing in German called a

252
‘Feier’. Of course (27) and (32) being synonymous depends on facts about how speakers of

English and German use these words. But the truth-makers of the proposition expressed by both

sentence have to do with facts about what Peter’s believes about what Jennifer believes, not facts

about language.

I would thus propose that Fara’s theory is a plausible theory of the truth conditions of

sentence (27) qua sentence, but not with respect to the truth conditions of the proposition (27)

expresses. With Frege’s puzzle, the relevant question is about the nature of the proposition

expressed by sentences and the truth conditions of these propositions. What are their

constituents? Are content Millian, Fregean, or something else entirely? Fara’s theory does not

address this question. This likely explains why she does not discuss Frege’s puzzle or propose

any solution to it.

Recanati, an indexicalist, also thinks that proper names have a level of meaning similar

to Fara’s. In Direct Reference (1993), he states that the character of a proper name ‘NN’ is the

bearer of ‘NN’, while the content is NN, the referent itself. However, Recanati makes it clear

that this meaning, the bearer of ‘NN’, is confined to the level of character. By contrast, Fara’s

claim, which is that the predicate meaning of a name is its content, requires us to blur this

important distinction between character (i.e., the linguistic meaning of an expression) and

content (what the word contributes to the proposition expressed by the sentence in which it

occurs).

253
 CHAPTER 10

RECANATI’S THEORY OF MENTAL DOSSIERS

Recanati’s central aim in his 2013 book Mental Files is to vindicate Singularism against

Descriptivism. To this end, he attempts to reconcile the theory of direct reference with Frege’s

sense-reference distinction. Recanati notes that Russell’s theory of acquaintance spectacularly

failed to provide a theoretical foundation for singularism. Russell’s theory allowed that one

might be directly acquainted only with sense-data and perhaps with the self. With respect to

names of macroscopic objects such as, e.g., Mont Blanc, Russell was forced to go descriptivist.

Recanati claims that Frege’s sense-reference distinction, on the other hand, could be adapted to

vindicate singularism, despite Russell’s claims to the contrary. The problem, argues Recanati,

was Frege’s construal of sense as descriptive in nature; even though Frege did not expressly

define senses, the examples he gave strongly suggested they were descriptivist. As Recanati puts

it “Russell's claim that a two-level semantics à la Frege is incompatible with Singularism

therefore depends upon an overly narrow, descriptivist construal of ‘sense’, a construal that was

encouraged by Frege himself but which was by no means mandatory.” Descriptivist senses, of

course, cannot be used to ground singularism, as Recanati notes. But non-descriptive senses

could be so used, and this is Recanati’s project. For him, mental files are non-descriptive modes

of presentation or senses. Mental files contain representations with respect to the objects they are

about and refer to those objects in virtue of their being contextually related to the files in the

right sort of way.

What work can Recanati’s non-descriptive senses do? It can solve what Recanati calls

“Frege Cases.” These are the well-known cases in which an agent believes inconsistent singular

254
propositions, e.g., an agent believes that Hesperus is bright but disbelieves that Phosphorus is

bright. One of Frege’s concerns, raised implicitly in his 1892, was how to explain how a rational

agent could entertain these apparently contradictory thoughts. Frege proposed that such an agent

is not irrational because his thoughts are distinct—not in fact inconsistent. For Frege, a Hesperus

is bright thought and a Phosphorus is bright thought are distinct because the sense of ‘Hesperus,’

a constituent of the thought, is distinct from that of ‘Phosphorus.’ Recanati would not want to

claim, with Frege, that ‘Hesperus is bright’ and ‘Phosphorus is bright’ express distinct

thoughts/propositions because as a direct referentialist he wants to say that they express the same

singular proposition, whose constituents are Venus and the property of being bright. But

Recanati does want to use the notion of sense/mode of presentation to solve the Frege cases. He

wants to say that the agent’s rationality is saved because his thoughts involve Venus being

presented under different non-descriptive modes of presentation, which amounts to his thoughts

involving two distinct unlinked/unmerged files that the agent does not realize have the same

referent. In simpler terms, the rationality of an agent who believes a pair of inconsistent singular

propositions is saved by showing that the agent’s mistake is due to ignorance, not irrationality;

the agent is ignorant of the fact that his two files have the same reference. Thus, we can account

for the agent’s rationality in terms of non-descriptive modes of presentation that are based in

acquaintance relations, and thus vindicate a thoroughly singularist picture, but on Frege’s terms

with his sense-reference distinction, not on Russell’s unitary theory lacking that distinction.

The theory I defend in this dissertation, the TIUT, takes a similar approach to solving this

puzzle, which Recanati calls “Frege cases” and I call the “problem of rational inconsistent

belief.” This idea of using mental files to solve this problem is not novel in Recanati—it is

implicit in all file/dossier theories going back to Lockwood and Strawson, and is developed in

255
Forbes (1990). The idea is, again, that an agent suffering from identity confusion has two

clusters of information, files, or dossiers, in his head such that he does not realize they refer to

the same individual or object. Discovering the identity amounts to either linking these files

together, or merging them into one larger file. An agent may rationally entertain inconsistent

singular thoughts about Venus as long those thoughts involve distinct dossiers that the agent does

not realize refer to the same object. An agent is rational as long as he does not knowingly believe

inconsistent propositions, and the dossier/file metaphor elucidates what the agent fails to know—

the co-reference of distinct files.

Despite my agreement with Recanati on this point, the theory I present goes in a very

different direction from Recanati’s theory. Despite addressing the “Frege cases,” Recanati does

not clearly state that he intends to address Frege’s puzzle—either the propositional attitudes

puzzle or the identity sentences puzzle. And it seems to me his theory is not up to solving the

puzzle. To see that why, consider sentence (26).

(26) The Babylonians discovered that Hesperus is Phosphorus before the Greeks did.

This sentence is intuitively true, but for a direct referentialist such as Recanati, it is literally false.

‘Hesperus’ and ‘Phosphorus’ are names of the same object—Venus, so (26) literally expresses

the proposition that the Babylonians discovered the self-identity of Venus before the Greeks.

That is clearly false, and this is clearly not what the utterer of (26) intends to communicate. No

one was ever confused about the self-identity of Venus. Recanati could claim, perhaps, that

‘Hesperus’ and ‘Phosphorus’ corresponded to (or perhaps labeled) different files in the

Babylonian’s and the Greeks’ minds, and this sentence is about the merger or linking of those

files in both groups, saying that the Babylonians merged or linked them first. But this would be

256
insufficient to capture what sentence (26) communicates, for it is clearly about the discovery that

Venus considered under specific morning and evening star descriptive modes of presentation

were identical, not merely that the Babylonians and Greeks merged or linked any old files about

Venus without regard to the descriptive representations in those files. I think we need a theory of

content on which the descriptive representations inside mental files can sometimes become part

of the content of an utterance where a name in the files is tokened. Such a theory should also

explain how those descriptive representations are uploaded into the content of the proposition

expressed (and I claim this happens via the character of the name, which picks it up from the

utterance context, i.e., from the speaker’s dossier). This would account for the names not being

synonymous in some contexts, as in (26). Recanati’s theory does not provide for this.

It is unclear, as a direct referentialist, whether Recanati would endorse the standard

Millian strategy to use pragmatics to explain the phenomena in Frege’s puzzles. He is silent on

this issue. He is in fact silent on the vast majority of literature on Frege’s puzzle. His theory is an

interesting theory when considered as limited to the explanation of how dossiers can be grounded

in acquaintance relations and therefore have externalist reference conditions, as well as

accounting for identity confusion and ignorance without entailing irrationality, but it cannot

constitute a comprehensive theory of content of name containing sentences, and indeed, it does

not seem this is Recanati’s aim.

The TIUT constitutes an attempt to improve in Recanati’s claim that names are indexicals

by describing two sorts of indexical uses—one on which they are not directly referential, and tie

both uses of names into the dossier metaphor. Although Recanati is a proponent of dossiers, he

does not integrate the dossier metaphor into his definition of the character of names, as the TIUT

does. This other type of indexical use, which I call the conception-indicating use, seems to me

257
essential to explain Frege’s puzzle without resorting to the sort of pragmatic explanations

popular with Millians such as Salmon and Soames, which are counterintuitive.

258
 REFERENCES

Bach, Kent. (2002). “Giorgione Was So-Called Because of His Name,” Philosophical
Perspectives 18, pp. 73-103.

Bach, Kent. (1999). “The myth of conventional implicature,” Linguistics and Philosophy, 22:
327–66.

Bach, Kent. (1987). Thought and Reference. Oxford: Clarendon Press.

Barcan Marcus, Ruth. (1961). “Modalities and Intensional Languages.” Synthese 13.4, pp. 303-
322.
Braddon-Mitchell, David and Jackson, Frank. (1996). Philosophy of Mind and Cognition.
Malden, MA: Blackwell Publishing.
Branquinho, João. (2003). “In Defense of Obstinacy.” Nous-Supplement: Philosophical
Perspectives 17: 1-23.

Braun, David. (1993) “Empty Names.” Noûs 27, 449-69

Braun, David. (1998). “Understanding Belief Reports.” Philosophical Review 107, pp. 555-595.

Braun, David. (2002). “Cognitive Significance, Attitude Ascriptions, and Ways of Believing,”
Philosophical Studies 108, pp. 65-81.

Braun, David. (2002b). “Simple Sentences, Substitutions, and Mistaken Evaluations” (with
Jennifer Saul), Philosophical Studies 111, pp. 1-41.

Braun, D. (2003). “Beyond Rigidity: The Unfinished Semantic Agenda of Naming and Necessity
(review).” Linguistics and Philosophy, 26(3), 367-379

Braun, David. (2006). “Illogical, But Rational,” Noûs 40, pp. 376-379.

Braun, David and Sider, Ted. (2006). “Kripke’s Revenge,” Philosophical Studies 128, pp. 669-
682.

Cave, P. (2014). This sentence is false: an introduction to philosophical paradoxes. London:

Bloomsbury Academic.

Chomsky, Noam. (1965). Aspects of the Theory of Syntax. Cambridge, MA: MIT Press.
Crimmins, Mark. (1992). Talk about Beliefs. Cambridge, Mass.: MIT Press/Bradford Books.
Cumming, Samuel (2008), “Variablism.” The Philosophical Review, 117, 525–554.

259
Cumming, Sam, "Names", The Stanford Encyclopedia of Philosophy (Fall 2016 Edition),
Edward N. Zalta (ed.), URL =
<https://plato.stanford.edu/archives/fall2016/entries/names/>.
Devitt, M., 1981. Designation, New York: Columbia University Press.
Devitt, Michael. (1989). “Against Direct Reference,” Midwest Studies in Philosophy 14, pp. 206-
40.

Devitt, M. (2007). Coming to our Senses: A Naturalistic Program for Semantic Localism. New
York: Cambridge University Press.

Devitt, M. (2004). The case for referential descriptions. In A. Bezuidenhout & M. Reimer (Eds.),
Descriptions and beyond (Chap. 7, pp. 280–306). Oxford: Oxford University Press.

Devitt, Michael. (2005). “There is no a priori.” In Contemporary Debates in Epistemology.

Cambridge (MA): Blackwell Publishers (pp. 105-115).

Devitt, Michael. (2007). “Referential Descriptions and Conversational Implicatures,” EUJAP 3,

pp. 7-32

Devitt, Michael. (1990). “Meanings just ain’t in the head,” In George S. Boolos (ed.), Meaning
and Method: Essays in Honor of Hilary Putnam. Cambridge University Press, 79-104.

Devitt, M. (2012). “Still against direct reference.” in Richard Schantz, ed., Prospects for
Meaning. Berlin: Walter de Gruyter (2012): 61-84

Evans, Gareth. (1973). “The Causal Theory of Names,” Proceedings of the Aristotelian Society,
Supp. Vol. 47, pp. 187-208.

Evans, Gareth. (1982). The Varieties of Reference (edited by J. McDowell). Oxford: Clarendon
Press.

Fara, D. G. (2015). “Names Are Predicates,” Philosophical Review 124 (1), pp. 59-117.

Fitch, Greg and Nelson, Michael, "Singular Propositions", The Stanford Encyclopedia of
Philosophy (Fall 2014 Edition), Edward N. Zalta (ed.), URL =
<http://plato.stanford.edu/archives/fall2014/entries/propositions-singular/>.

Fitch, Greg. (1976). “Are There Necessary ‘A Posteriori’ Truths?” Philosophical Studies: An
International Journal for Philosophy in the Analytic Tradition 30 (4), pp. 243-247

Forbes, Graeme. (1989). “Cognitive Architecture and the Semantics of Belief,” Midwest Studies
in Philosophy 14, pp. 84-100.

Forbes. Graeme. (1990). “The Indispensability of Sinn,” Philosophical Review 99 (4), pp. 535-

260
 563.

Frege, Gottlob. (1956). “The Thought: A Logical Inquiry.” Mind 65 (1956), pp. 289-311.
[Originally published in 1918 in German].

Frege, Gottlob. (2007). “On Sense and Reference.” Reprinted in On Sense and Direct Reference,
Davidson, Matthew, ed., McGraw Hill. [Originally published in 1892 in German under
title “Über Sinn und Bedeutung”].

Geurts, B. (1997). “Good News about the Description Theory of Names.” Journal of Semantics
14.4, pp. 319–348.

Geirsson, Heimir. (2012). Philosophy of Language and Webs of Information. New York: Taylor
and Frances.

Grice, Paul. (1969). “Vacuous Names,” in D. Davidson & J. Hintikka (eds.) Words and
Objections, pp.118-45. Dordrecht: Reidel.

Grice, Paul. (1975). “Logic and Conversation,” in P. Cole and J. Morgan, eds., Syntax and
Semantics, vol. 3, New York: Academic Press, pp. 41-58.

Hawthorne, John; Manley, David. (2012). The Reference Book. New York: Oxford University
Press.

Jackson, Frank. (1998). “Reference and description revisited,” Philosophical Perspectives, 12,
pp. 201-218.

Kaplan, David. (1964). “Words,” Proceedings of the Aristotelian Society 64 (1964), pp. 93-119.
Kaplan, David. (1989). “Demonstratives,” in Themes from Kaplan. Joseph Almog, John Perry,
and Howard Wettstein (eds.). New York: Oxford University Press.

Katz, Jerrold J. (1994). “Names without Bearers,” The Philosophical Review, 103.1, pp. 1-39.
Kneale, William. (1962). “Modality de dicto and de re,” in E. Nagel, P. Suppes & A. Tarksi
(eds.), Logic, Methodology, and the Philosophy of Science. Proceedings of the 1960
International Congress, Stanford, CA: Stanford University Press, pp. 622-633.
Kripke, Saul. (1980). Naming and Necessity. Cambridge, Massachusetts: Harvard University
Press.

Kripke, Saul. (1979). “A Puzzle about Belief,” in A. Margalit, ed., Meaning and Use, Dordrecht:
D. Reidel.

Kroon, Frederick. (1987). “Causal descriptivism,” Australasian Journal of Philosophy 65, pp. 1-
17.

261
Kroon, Fred and Voltolini, Alberto, "Fiction", The Stanford Encyclopedia of Philosophy (Fall
2011 Edition), Edward N. Zalta (ed.), URL =
<http://plato.stanford.edu/archives/fall2011/entries/fiction/>.

Lewis, David. (1984). “Putnam’s paradox,” Australasian Journal of Philosophy, 62, pp. 221-36.

Lockwood, Michael. (1971). “Identity and Reference,” In M. Munitz (ed.) Identity and
Individuation, pp. 199-211. New York: New York University Press.
Lycan, William G. Philosophy of Language: a contemporary introduction. New York:
Routledge, 2000.

Marcus, R. Barcan, (1961). “Modalities and Intentional Languages,” Synthese, 13: 303–322.

May, Robert. (2006). “Invariance of Sense,” The Journal of Philosophy 103, pp. 111-144.

McKay, Thomas and Nelson, Michael. (2010). Supplementary Document File “Ignorance of
Identities” in “Propositional Attitude Reports,” The Stanford Encyclopedia of Philosophy
(Fall 2008 Edition), Edward N. Zalta (ed.), URL =
http://plato.stanford.edu/archives/fall2008/entries/prop-attitude-reports/ignorance.html

Moore, Joseph G. (1999). “Saving Substitutivity in Simple Sentences,” Analysis 59, pp. 91–105.

Morris, Michael. (2007). An Introduction to the Philosophy of Language. New York: Cambridge
University Press.

Mount, A. (2008). “The impurity of “pure” indexicals.” Philosophical Studies, 138(2), 193-209.

Neale S. (2004), “This, That, and The Other”, In A. Bezuidenhout & M. Reimer (Eds.),
Descriptions and Beyond (pp. 68–182). Oxford: Oxford University Press.

Ostertag, Gary. (2009). “A Problem for Russellian Theories of Belief,” Philosophical Studies:
An International Journal for Philosophy in the Analytic Tradition, Vol. 146, No. 2, pp.
249-267.

Pelczar, Michael and Rainsbury, Joseph. (1998). “The Indexical Character of Names,” Synthese,
114.2, pp. 293-317

Pelczar, Michael. (2001). “Names as Tokens and Names as Tools,” Synthese, 128.1/2, pp. 133-
155.

Perry, John. (1979). “The Problem of the Essential Indexical,” Noûs 13, pp. 3-21.

Perry, John. (1997). “Reflexivity, Indexicality and Names.” In W. Kunne (ed.), Direct Reference,
Indexicality and Propositional Attitudes. CSLI-Cambridge University Press: Stanford.

Perry, J. (1980). “A Problem about Continued Belief,” Pacific Philosophical Quarterly 61, pp.
262
 317-332.

Pickel, B. (2013). “Variables and Attitudes,” Noûs 49(2), pp. 333-356.

Pitt, David. (2001). “Alter Egos and Their Names,” The Journal of Philosophy, 98.10, pp. 531-
552.

Powell, G. (2016). Language, Thought and Reference. Baskingstoke, U.K.: Palgrave Macmillian.

Recanati, Francois. (1993). Direct Reference, From Language to Thought. Oxford, UK:
Blackwell Publishers Ltd.

Recanati, Francois. (2013). Mental Files. URL =

http://jeannicod.ccsd.cnrs.fr/docs/00/78/40/32/PDF/Mental_Files_RR15.pdf

Reimer, M. (1998). Donnellan’s distinction/Kripke’s test. Analysis, 58(2), 89–100.

Richard, Mark. (1983). “Direct Reference and Ascriptions of Belief,” Journal of Philosophical
Logic 12, pp. 425-452.

Richard, Mark. (1990). Propositional Attitudes: An Essay on Thoughts and How We Ascribe
Them. New York, Cambridge University Press.

Richard, Mark. “Kripke’s Puzzle about Belief.”

https://markrichardphilosophy.files.wordpress.com/2011/07/kripkes-puzzle-about-
belief.pdf [accessed on May 31, 2015].

Russell, Gillian. (2008). Truth in Virtue of Meaning: A Defense of the Analytic/Synthetic

Distinction. New York: Oxford University Press, 2008.

Salmon, Nathan. (1986). Frege’s Puzzle. Cambridge, MA: MIT Press.

Salmon, Nathan. (1986b). “Reflexivity,” in Propositions and Attitudes, eds. Nathan Salmon and
Scott Soames. New York: Oxford University Press, 1988.

Salmon, Nathan. (1989). “How to Become a Millian Heir,” Noûs 23.2, pp. 211-220.

Salmon, Nathan. (1989b). “Illogical Belief,” in Philosophical Perspectives, Vol. 3, Philosophy of

Mind and Action Theory, pp. 243-285. Atascadero, California: Ridgeview Publishing Co.

Salmon, Nathan. (1993). “Analyticity and Apriority,” Philosophical Perspectives, Vol. 7,

Language and Logic, pp. 125-133

Salmon, Nathan. (1993b). “Relative and Absolute Apriority,” Philosophical Studies: An

International Journal for Philosophy in the Analytic Tradition, 69.1, pp. 83-100

Salmon, Nathan. (1998). “Nonexistence.” Nous 32.3, pp. 277.319.

263
Salmon, Nathan. (2006). “The Resilience of Illogical Belief,” Noûs 40.2, pp. 369-375.

Saul, Jennifer. (1997). “Substitution and Simple Sentences,” Analysis, 57.2, pp. 102-108.

Saul, Jennifer. (1998). “The Pragmatics of Attitude Ascription,” Philosophical Studies: An

International Journal for Philosophy in the Analytic Tradition, Vol. 92, No. 3, pp. 363-
389

Saul, Jennifer. (1999). “Substitution, Simple Sentences, and Sex Scandals,” Analysis 59, pp.
106–112.

Saul, Jennifer, and Braun, David. (2002). “Simple Sentences, Substitutions, and Mistaken
Evaluations,” Philosophical Studies 111, pp. 1-41.

Schiffer, S. (1977). “Naming and Knowing.” Midwest Studies in Philosophy, 2, 28-41.

Schiffer, Stephen. (2006a). “A Problem for a Direct-Reference Theory of Belief Reports,” Noûs
40.2, pp. 361-368.

Schiffer, Stephen. (2006b). “Propositional Content”, in E. Lepore and B. Smith (eds.), Oxford
Handbook of Philosophy of Language. New York: Oxford University Press.

Schiffer, Stephen. (1992). “Belief Ascription,” The Journal of Philosophy, Vol. 89, No. 10, pp.
499-521.

Schiffer, Stephen. (1977). “Naming and Knowing,” Midwest Studies in Philosophy, 2, pp. 28–41.

Searle, J. (1958). “Proper Names”, Mind, 67.266, pp. 166–73.

Soames, Scott. (2002). Beyond Rigidity: The Unfinished Semantic Agenda of Naming and
Necessity. New York: Oxford University Press.

Soames, Scott. (2003). Philosophical Analysis in the Twentieth Century. Volume 2: The Age of
Meaning. Princeton: Princeton University Press, 2003.

Soames, Scott. (2006). “Précis of Beyond Rigidity,” Philosophical Studies: An International

Journal for Philosophy in the Analytic Tradition, 128.3, pp. 645-654

Soames, Scott. (2006b). “Reply to Critics,” Philosophical Studies: An International Journal for
Philosophy in the Analytic Tradition, 128.3, pp. 711-738

Speaks, Jeff. (2011). “Frege’s Puzzle and Descriptive Enrichment,” Philosophy &
Phenomenological Research 83.2, pp. 267-282.
Strawson, Peter. (1974). Subject and Predicate in Logic and Grammar. London: Methuen.

264
Strawson, Peter. (1959). Individuals: an Essay on Descriptive Metaphysics, London: Methuen.
Textor, Mark. (2011). Frege on Sense and Reference. New York: Routledge.

Wittgenstein, L. (1953). Philosophical Investigations, translated by G. Anscombe, Oxford: Basil

Blackwell.

265