.133; .-1._¢ 3
$§‘_'_-§..

3-2..— 4;. 3

 

~.-

3:»

t4:

.-;';

ham.

,, ,

'3’

3 -
‘r' <
ha

 

 

 

 

j

' 33333:“:-

Q‘?‘ 1.; "v‘fi', _

't “$355324: at
him)“ L '

£1

L‘:

1» 1
L‘

k‘fli’u‘flw ,

‘L “-1,“. u

l

.3 us:w

“We?

M537».

2,
' 1 1'}:
$14,)“ .71 '1‘1 ‘ . v? R What!“ -
3

v. 'N
h 1:3
l I '
"\ . > I
V. ‘13:“: 3 “‘5' 7595“ 3‘35 -

, 3335me
"931??“ .1 an-n
H b

. 1w." L.
' an. 3'51

_\ . 3 fl“.
“£96.13, \ .3 , - JWHEESQ.
, 1 115W; “ ' ‘ 3 .‘F‘V‘."‘
«:Efi‘u‘ik‘fi a "I ,
I ,2»? 4-3,“: ,3. ~

~ ’x‘ L J» ~ 2544917351963fi. arm “I
q“ “' *3 3333333333133, .33
3237‘“ 3““ “ V" 31‘“ ‘ 3 1.4 ‘m
“”13”“ .4 a: ““1 mm“ x

' 1347.13 N" WWW” “’ a i'uu: a“.
"‘ 3 1 ‘mfix. www-‘m ‘1.
- 33% 3%‘3352‘33 = - 333333
. M 15,33 mn‘ 11,. '51'137-L

L

o .g _‘
w ' 33$}
“ 3“ “F 33,3

" F1338! kw a??? 1:
WM? V 8‘ kg!
3% u’wfim?

‘3‘ ”9‘ ' i “:3 «t
v}: 3.1
I: m.
. 2m“!

I V. “11‘ FI‘PT I
:33 ”"figfrWW :
3333"":313hfl

‘1?’ .' 3 ‘

‘1'!

4

‘ a}:
-5

4'.
~ i.
,3 ‘_ fizz?
figr; J
x‘
:1
t

6

I
l

(2:

”($35

“3% w r
(4‘5“. 3 ' 3‘ J .,

A E"
«:33

3% ~

11-53 ‘ .
I". .
4

1-1“!

A r
£1 .
..

J?‘
931‘?
3 .3 r
;. é:
.30”.

‘J '13

it‘
. a“???

‘ a?“ V
Effiéfi'

33%

J

h
‘3 154:2
.H 1.‘
13.

1L“? 1’4. I i" Ht FELELk;
\ 4 "3:. 3 . ‘3‘; . .
“YES! :wwtrfigfi M vii“; ‘ I“:

T-
3 ‘q
‘ ‘3
'Nfik 1
h ", p013... :4
m. MW! ~ ‘3“
#8 £3.
a
3 mama,
‘1 3'1"?“
3‘5 ’51» .131:-

A 4. "‘2. ‘71

. .33,
3-3;: 3::

J
' 3.! 'u. 1,“
, ' ' .u‘ .1
' 3'7 Ni: ‘21“
“ ‘- . 1..
Axuifii 51 ME“ .5
SF" 37":
"1'01

J
:J’Ql
~r.‘..:I-x:§"7 3- a .
«13:4 25%?
?°’ r? .
5'1 A.
.L’

,__
3‘4...

’31.

Ad»;

.1 fl 3.
«$313 {’53.
a I 1.3.) .
' 31:33

«1 ' 11'
.4113}:
33-3133 .3

r A ". ‘
l ‘ n
7:3 £133 L‘g'fé V
, a

{r-

' n , ‘

4" pt? 5. 'l q.- "3 .

"5) Tight 3.1,. ' ~ .

35%;}: Wisfififii ' '3'; -:‘

"’1‘ mtwfif‘. ’3‘. 3;;
315%“ Mm"
x,’ 3

3
3
”IE" _

n ’ ‘5”;

tang; -5
i «3.27:: .
.
i" a
.3i:

. hurt. .
.r .
1. z. A

'1 “3‘ 2:1

fig 235 ,

H-‘a ? 43:7
uh. w
"3329:1131?

:..
‘ l
.; m‘
. "z.
..... r.
~ .:.x;., 1
.2; ‘1 ‘
V '3;
a -.‘ A

"5.2.21 .

r

“:32:

I'm ,3
, I. O . ..,
‘3; 0:1? M
S'JLWI‘.
3 "W3“

J‘I::_.

_ 2.4.2353

~-\. ~ .3.
.. w
‘VN‘fig‘th:
n , 1'

\" E1316 1
‘
.

. n .
1‘}:ng 113:1

M‘.’ {3:5 “3:1.
mv§S¢E"~ :5. 7 fig
‘ 7 ‘3 s. ‘33‘ h
1' 1"}:

:3:

It; ' 3 1‘» , , 311.3 ”w
543%” {ff‘fiu‘zt’n‘fiéfi ”‘3‘
331,33

‘ J’ "”3, n,“ 3 : ‘x -
3:- 'L" 3‘,é!).~"é “109); 3&3. .._ - J ‘ggizflagh

‘tE‘ia‘xgy‘
A' 3: “3...; 7
~

"- I
.3}:
hr 4:31- .

I44

4—

a:
at:
£35

3:"

0'

TE: gl'
‘7'-
3113-15,

71

n. if _
, ‘3‘ . «I «4).
“‘3‘! f. _ A 9“, frgfi-x

@‘ifi'a‘ayulfi éfifiim‘gg

{.4 (n gt s 73935;? ‘1‘ng h

w, 3" ‘Wé “ 5

N I" [W .., > ‘-
WW‘Tfihfii 3&3? ‘3
.33.- “3’3 3 m

w

‘13
:-

‘ 3
“1:3

‘43: 3
‘3;

n
'v’q“.i,
~ML‘IE‘?‘3=.&,§ ._
3 33': 1.1311331.” -, . .1
11-31513 Hymfil‘i‘z‘f‘; t
LP"k.:-¢ .3": ‘1.

4,
r.

5‘“
s.
.

f1.
.c '

. 3-513

x ' ‘3 11
fiv‘ifihfk‘ ~ - 3U “y‘tggvnfi
a , 1" ’“($959.4
muk‘fi?

. ,fi: W‘s-9.25%
53333332333323

$J'n“! I
u‘ ‘ .

3“

v

«a.
.g-
4 :3:

t' .-

"< 3‘
:r‘

 

. ill 33333133333333333

3 129 06 57

 

 

 

 

flESls ,
This is to certify that the
thesis entitled
Proper Names: Rigid Designation
And The Causal Theory
presented by
Robert M. Steinman
has been accepted towards fulfillment
of the requirements for
Ph.D. dpgflmin Philosophy
m I
by) u") $0
t
,9
3.“ ,4.» ~
, w ,1» M“ W 0, £44
(A Major professor
Date June 1, 1981 Lu“, 4r
ng~
. W
07539 \\ \ “if,“ if} aw
/W ,\ a / l—" p
5 L
at O ..
f p (X Maw

 

 

 

 

 

 

 

 

 

 

 

RETURNING MATERIALS:
bVIESI_J Place in booE drop to
LIBRAR] remove this checkout from
£5,.,,.:Ei_ your record. FINES will
be charged if book is
returned after the date
stamped below.
’7 [I ’49,
-.é.‘!"';" ’W
K? .
t/‘zz @7
3 WK
1 J
We“ 4%.? I‘M-r; 0. 3;
we {[3 pit-Wm
hym‘w W1 {Ll/t
yéyed f5
‘I s“)
av» .
3‘

 

 

PROPER NAMES:

RIGID DESIGNATION AND THE CAUSAL THEORY

By

Robert M. Steinman

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Philosophy

1981

 

Ramon kw

 

 

 

ABSTRACT

PROPER NAMES:
RIGID DESIGNATION AND THE CAUSAL THEORY

By

Robert M. Steinman

In this essay an attempt is made to clarify and
resolve some of the issues concerning proper names that
have arisen as a result of Saul Kripke's now famous work,

Naming and Necessity. More specifically, the main focus of

 

attention will be the issues that arise in connection with
Kripke's thesis that proper names are rigid designators and
his thesis that proper names refer to their bearers in
virtue of some appropriate causal connection.

In chapter one I give a formally precise and rigor—
ous account of Kripke's notion of a rigid designator by
adopting the following definition. Where a is any name or
description, to say that a is rigid is to say that the

following condition holds:

(A) E]{Elo¢:3(3x)[x=on & D(E!oc2>x=a)]}

(A) is then used to formally express Kripke's claim that
proper names, unlike definite descriptions, are rigid des-
ignators, that is that proper names designate the same
thing in every possible world in which they designate at

all. Finally, in light of my account of a rigid designator

Robert M. Steinman
I attempt to explicate and defend what I take to be
Kripke's modal argument against the description theory of
proper names. In particular I show that the argument has
been misunderstood by various philosophers, and that the
attempt to dodge it by viewing names as definite descrip-
tions that have widest possible scope in modal contexts
fails.

Chapter two is devoted to critically examining a
recent attempt by Michael Dummett to show that at least
some proper names are on a par with definite descriptions
in modal contexts, and hence are not rigid designators.

In chapter three I attempt to show that the thesis
that proper names are rigid designators is not coextensive
with the thesis that proper names refer to their bearers in
virtue of some appropriate causal connection. To accom-
plish this I construct a fairly clear and intuitive case of
reference involving a proper name where there is no causal
connection between the referent and the speaker's utter-
ance of the name. Hence, if successful, I show that a
causal theory of proper names cannot provide a necessary

condition for name reference.

© Copyright by
ROBERT M. STEINMAN

1981

For my parents: Morris and Theresa Steinman

 

ACKNOWLEDGMENTS

I would like to thank Professor Richard J. Hall
and Professor Herbert E. Hendry for accepting the task of
codirecting this thesis, and for their invaluable assis-
tance and support throughout its preparation. I also owe
a large debt of gratitude to Anne Scott without whose
patience, forbearance, encouragement, and love this work

could not have been done.

iv

 

TABLE OF CONTENTS

INTRODUCTION .......................................
CHAPTER ONE. KRIPKE'S MODAL ARGUMENT .............
CHAPTER TWO. CRITICISMS OF THE MODAL ARGUMENT

CHAPTER THREE. THE CAUSAL THEORY ...................

BIBLIOGRAPHY .......................................

 

INTRODUCTION

Saul Kripke in his now famous work, Naming and

 

Necessity, introduced a very new and radically different
theory of proper names, namely, the "causal theory" of
proper names. All it takes is a rough statement of the
theory for it to be easily seen that it breaks radically
with any of the more traditional views of proper names.
Consider, for example, the following account of the theory

given by Michael Devitt.

The central idea of the causal theory of proper names
is that our present uses of a name, say 'Aristotle'
designate the famous Greek philosopher Aristotle, 22E
in virtue of the various things we (rightly) believe
true of him, but in virtue of a causal network stretch-
ing back from our uses to the first uses of the name to
designate Aristotle. Our present uses of a name borrow
their reference from earlier uses. It is this social
mechanism that enables us all to designate the same
thing by a name.

 

 

 

 

According to a view of proper names that is Fregean in
spirit the primary linguistic function of a name is to be

a bearer of sense, whereas according to the causal view the
primary linguistic function of a name is to refer to its

bearer. The fact that names might have reference is only a

 

 

 

1Michael Devitt, ”Singular Terms," The Journal of
Philosophy, LXXI (April 18, 1974), p. 184.
l

.MII

secondary consideration for the Fregean, just as the fact
that names might have sense is only a secondary consider-
ation for the causal theorist.1 When we ask the question
”How do proper names refer?”, Frege and Russell tell us
that proper names refer if and only if there is an object
which satisfies their sense. According to the causal the-
ory, however, a proper name refers if and only if there is

an appropriate causal connection between uses of it and

 

some object.

What we have here is two radically different ways
of explaining how proper names refer to objects. The first
says: via some definite description associated with the
name that is its sense, and the other says: via causal
connections with the object. It seems clear that the two
views are not extensionally equivalent. Given some form
of the sense or description theory of names it is possible
that a proper name such as 'Santa Claus' could refer to
some object via some associated definite description, even
though the name bears no causal relation to the object re-
ferred to. Imagine that some jolly creature was discovered
at the North Pole and that he fit perfectly all of the de-
scriptions commonly associated with Santa Claus. According

to a causal theorist like Donnellan the name 'Santa Claus'

 

llt should be noted that some philosophers, Leonard
Linsky for example, interpret Kripke as holding the View
that names lack sense altogether. See, Names and Descrip-
tions (Chicago: The University of Chicago Press, 1977),
pp. 52-65.

 

 

would fail to refer to this individual since we know that
Santa Claus is only a fictitious entity, and thus the appro-
priate historical or causal connection would not be avail—
able. According to the sense theorist, however, it is
entirely possible that we might come to doubt that Santa
Claus was fictitious, in which case the name would refer to
the individual residing at the North Pole even though an
appropriate causal connection was found to be missing. It
is also possible, according to the causal theory, to have a
case of successful name reference where a name is causally
connected with some object in an appropriate manner, yet
none of the descriptions associated with the name fit the
object in question. Kripke's Godel—Schmidt example is just
such a case. Imagine that someone other than Godel, say
Schmidt, discovered the incompleteness of arithmetic, and
further suppose that the only definite description that we
associate with the name 'Godel' is 'the man who discovered
the incompleteness of arithmetic'. According to Kripke we
refer to Godel and not Schmidt when we use the name 'Godel'.
If the sense theory of proper names was correct, however,
we would be referring to Schmidt and not Godel when we used
the name.

It also seems clear that any plausible View of
proper names that is Fregean or Russellian in spirit would
present a challenge for the causal theory of names. Devitt

shows his awareness of this problem when he states that:

 

—————r i ., .._‘

...a-....

h
’a
1'

.

The main problem in giving the semantics of proper
names is that of explaining the nature of the link
between name and object in virtue of which the former
designates the latter. From Frege and Russell through
to Strawson and Searle, the solution has been sought
in the descriptions of the object that users of the
name associate with the name. Saul Kripke has shown
that all such ”sense-theories" of names are mistaken.
They are mistaken not merely in details but in funda-
mentals.

One reason that Devitt and other supporters of the

causal theory are soconvinced that all such ”sense—theories"

of proper names are incorrect is that they implicitly ac-
cept Kripke's thesis that proper names, unlike definite
descriptions, are rigid designators. If this thesis is
correct, then Kripke has shown that proper names cannot be
disguised or truncated definite descriptions. Roughly,
what it means to say that a proper name, say 'George
Washington', is rigid is that it denotes the same thing in
every possible world where it designates anything. A defi-
nite description such as 'the first president of the United
States', on the other hand, is not rigid since in some pos-
sible world the object denoted by it is other than George
Washington. One argument attributed to Kripke to show that
while names are rigid, definite descriptions are not, goes

as follows.

Consider any nonrigid singular term, say 'the author of
the Iliad'. We may distinguish between the de dicgg

 

1Devitt, ”Singular Terms,” p. 183.

 

 

and de re readings of any modal sentence in which such

an expression occurs in subject position. In partic-

ular we may distinguish two readings of the sentence I
'The author of the Iliad might have existed and not '
been the author of the Iliad'. On the de dicto read—
ing the sentence is false; on the de re reading it is
true. But if we replace the description 'the author of
the Iliad' in both of its occurrences by the proper

name 'Homer' to obtain the sentence 'Homer might have
existed and not have been Homer', we do not find this
sort of ambiguity. To say that the proposition that
Homer exists but is not Homer is possibly true (g3
dicto) is tantamount to saying that Homer is such that
it would be possible for him to exist while not being
Homer (de re). The reason for this is precisely that
the name 'Homer' is rigid, i.e., it denotes the same
thing with respect to every possible world in which

that thing exists. Thus it makes no difference whether
the first occurrence of the name is taken as lying
within (de dicto) or without (de re) the scope of the 1
modal adverb 'might'. Its denotation remains the same.

 

 

 

 

 

|\‘_£

 

In chapter one I will attempt to give a precise and rigor-
ous account of Kripke's notion of a 'rigid designator',
then in light of that account explicate and defend what I
take to be his modal argument against the description view
of proper names. Chapter two will be devoted to criti-
cally examining a recent attempt by Michael Dummett to
show that at least some proper names are on a par with def—
inite descriptions in modal contexts, and hence are not
rigid designators.

Traditionally, the description theory of names has
had a slight advantage over any theory of names that has

been Millian in character. According to Mill, the sole

 

lNathan Ucuzoglu Salmon, review of Names and De-
scriptions, by Leonard Linsky, in The Journal of Philosophy,
LXXVI (August 1979), p. 437.

 

“b-

 

 

 

linguistic function of a proper name is to refer to its

bearer. If, however, we were to ask a proponent of a , _7
strictly Millian view of names ”How do names refer?” or ”In

virtue of what do names denote their bearers?”, no answer

would be forthcoming. The description theorist, on the

other hand, has an intuitively plausible answer to these

questions. Proper names have associated with them some

definite description which is their sense, and it is in

virtue of this sense or meaning that they refer to their

bearers. Looked at from a slightly more epistemological '3?
rather than an ontological perspective, Kripke makes this

point in the following manner.

The basic problem for any view such as Mill's is how E
we can determine what the referent of a name, as used ‘ .
by a given speaker, is. According to the description

View, the answer is clear. If 'Joe Doakes' is just l
short for 'the man who corrupted Hadleyburg', then -
whoever corrupted Hadleyburg uniquely is the referent

of the name 'Joe Doakes'. However, if there is not

such a descriptive content to the name, then how do

people ever use names to refer to things at all? Well,

they may be in a position to point to some things and

thus determine the references of certain names osten-

sively. This was Russell's doctrine of acquaintance,

which he thought the socalled genuine or proper names

satisfied. But of course ordinary proper names refer

to all sorts of people, like Walter Scott, to whom we

can't possibly point. And our reference here seems to

be determined by our knowledge of them. Whatever we

know about them determines the referent of the name as

the unique thing satisfying those properties. For

example, if I use the name 'Napoleon', and someone

asks, ”To whom are you referring?”, I will answer some—

thing like, ”Napoleon was emperor of the French in the

early part of the nineteenth century; he was eventu—

ally defeated at Waterloo.”, thus giving a uniquely

identifying description to determine the referent of

 

 

the name. Frege and Russell, then, appear to give the
natural account of how reference is determined here;
Mill appears to give none.

Now the causal theory of names, which is essentially
Millian in nature, may be viewed in part as an attempt to
answer this question left unanswered by Mill's theory. It
is important here to notice that if Kripke is correct in
holding that names, unlike definite descriptions, are rigid
designators and consequently they cannot be disguised de—
scriptions, then the Frege—Russell answer to the question
”How do names refer?” is no longer viable. The thesis that
proper names are rigid designators simply eliminates any
possibility of a Frege—Russell solution to the problem of
explaining proper name reference. Thus the causal theory
of names might be viewed as an essential adjunct to the
thesis that proper names are rigid designators if one is
looking for a systematic and comprehensive theory of naming
and reference. In chapter three, however, I will attempt to
show that the thesis that proper names are rigid designators
is not coextensive with the thesis that names refer to their
bearers in virtue of some appropriate causal connection.
To accomplish this I construct a fairly clear and intuitive
case of reference involving a proper name (which may be

intuitively seen to be a rigid designator in Kripke's sense

 

 

lSaul Kripke, Naming and Necessity (Cambridge:
Harvard University Press, 1980), pp. 27—28.

"a
I
I
J

 

 

of the term) where there is no causal connection between
the referent and the speaker's utterance of the name.
Hence, I attempt to show that a causal theory of proper
names cannot provide a necessary condition for reference,

and that the question "How do names refer?” has yet to be

answered by Kripke and his supporters.

 

 

CHAPTER ONE

KRIPKE'S MODAL ARGUMENT

Section One

In Naming and Necessity Saul Kripke makes use of a
battery of different arguments to show that a description
view of proper names is false. One sort of argument that is
used repeatedly by him is basically designed to show that
proper names cannot be disguised definite descriptions since
proper names are rigid designators and definite descriptions
are not. The contrast between proper names and definite
descriptions is drawn by Kripke in either of two ways. At
times the distinction is made by noting the truth conditions
of simple sentences with respect to counterfactual situa-
tions. Thus, for example, great emphasis is laid on the
fact that the truth conditions with respect to counter—
factual situations of a simple sentence with a proper name
in the subject position are not the same as those for that
same sentence with some co—designative definite description
substituted for the name. This point is made explicit by

Kripke in his preface to Naming and Necessity (of. especial-

 

ly pp. 6-7).

 

'F- _ _.fi, _ -
315'.- .’.-.':-.-.- - ..._- -.,-- ‘-.

10

At other times Kriple seems to argue for the differ-
ence between proper names and definite descriptions by
noting that they display different behavior in modal con—
texts. Here the emphasis is on an explanation in terms of
scope. It is pointed out that proper names in modal sen-
tences display a feature with regard to their scope that is
not displayed by definite descriptions. This latter manner
of drawing the contrast between proper names and definite
descriptions, however, has not been made explicit by Kripke,
and thus has been the subject of a great deal of misunder-
standing by both his critics and supporters alike. My
primary concern will be with the second of these two meth—
ods, and what I would like to do in this essay is look at
two different sorts of responses that have been made to
Kripke's modal argument based on scope distinctions against
the description view of proper names.

In section 7 of this chapter I will attempt to
justify what may seem to some to be a glaring disparity,
namely, my exclusive concern for Kripke's argument regarding
scope rather than his argument regarding truth conditions
with respect to counterfactual situations of simple sen-

tences. In his preface to Naming and Necessity Kripke

 

places great emphasis on the fact that his thesis that
proper names are rigid designators can be intuitively argued
for by simply considering counterfactual situations, and

need not rely in any way on scope distinctions. His reason

 

ll

for saying that is, I believe, clear enough. His remarks
concerning scope have led many to misinterpret his thesis
that names are rigid designators, and he would like, if
possible, to circumvent the whole issue in order to prevent
any further misunderstandings of his view. Thus Kripke goes
to some length to show that not all of our intuitions con-
cerning rigidity and names can be handled in terms of scope.
He states that ”no hypothesis about scope” can express his
view when only simple, nonmodal sentences are considered.

I will try to show, however, that in his effort to avoid

any further misinterpretations of his view, Kripke over-
states his case. In particular, I will try to show that the
counterfactual truth condition's argument is equivalent to
one about scope. Insofar that this can be accomplished,

and provided that appropriate care is taken not to misinter—
pret his claims about rigidity for names in terms of scope,
I see no reason why his view expressed as a doctrine con-
cerning truth conditions, with respect to counterfactual
situations, requires extensive discussion.

One common misconception shared by all of the crit—
ics of Kripke's modal argument that I will consider in this
essay concerns his doctrine of rigidity for proper names
and the relation it has to scope. According to these
philosophers Kripke‘s thesis that proper names are rigid
designators reduces to the requirement that proper names are

always to be read as having widest possible scope in the

 

12

modal sentences in which they occur. They also maintain
that according to Kripke definite descriptions in modal
contexts are always to be viewed as having narrowest pos-
sible scope. While all of the philosophers that I consider
share this misconception of Kripke's thesis, not all of them
respond in the same way to what they take to be his modal
argument based on this misconception.

Some philosophers, Michael McKinsey and Brian Loar,
for example, have attempted to dodge Kripke's modal argument
with regard to scope by requiring that the definite descrip-
tions that are substituted for proper names be given widest
possible scope in modal contexts. I will show that although
these philosophers are successful in their attempt to cir-
cumvent the modal arguments that they respectively consider,
they succeed only in virtue of the fact that the modal argu—
ment each of them considers is Egg the argument that Kripke
in fact gives. I will also argue that the modal argument

with regard to scope that is used by Kripke in Naming and

 

Necessity can simply not be dodged by the move to wide scope

 

definite descriptions.

Other philosophers, for example Michael Dummett,
also misinterpret Kripke's thesis concerning rigid designa-
tors but nevertheless can be viewed as responding to his
modal argument in what I take to be a more direct manner.
Dummett, for example, seems at times to argue that some

jprOper names, as they occur in modal contexts, are such that

 

 

13

they induce a scope ambiguity that is semantically signifi-
cant. Hence, the sentences in which they occur are seman—
tically ambiguous with regard to their scope. Viewed in
this light Dummett can be seen as trying to show that at

least some proper names display the same behavior in modal 1i]

  

contexts as do definite descriptions. In chapter two I will
try to show that Dummett fails to do what he sets out to
accomplish and that consequently his reply to Kripke is
inadequate.
Section Two
One of the main theses held by Saul Kripke in his

book, Naming and Necessity, is that ordinary proper names

 

are rigid designators, that is, that proper names are such
that they denote the same thing in every possible world in
which they denote at all. What this means in the semantics
of modal logic is that the value of a proper name remains

fixed as we evaluate the sentence in which it is contained
at each possible world. Consider, for example, the follow—

ing sentence:
(1) Carnap might not have been Carnap.
It is often pointed out that sentences such as (l) are (at

least) syntactically ambiguous between a 'de re' or 'large fig

scope' reading and a 'de dicto' or 'small scope' reading.

 

 

14

Classically the dg_£g/de dicto distinction has been drawn
between different modalities, viz., modality de dicto and
modality d§_£g. In a de dicto modality the argument of the
modal operator is taken to be a proposition and in a d§_£g
modality the argument of the modal operator is taken to be

a predicate.1

(1) under a large scope reading would attri—
bute the possibility of not being Carnap to a certain in-
dividual, viz., Carnap, and we could accurately express this

reading of (l) by the following sentence:

(2) Carnap is such that he possibly is not
Carnap.

(1) under a small scope reading would attribute possibility
to a proposition, and we could accurately express this

reading of (l) by the following sentence:

(3) It is possible that: Carnap is not Carnap.

(I) read either de re as in (2) or de dicto as in (3) is
false, since the name 'Carnap' picks out the very same
individual at each possible world, where the sentence
'Carnap is not Carnap' is evaluated. That is, of course,

if the name picks out anything at all. Since the name

 

lSee G. E. Hughes and M. J. Cresswell, An Intro—
duction to Modal Logic (London: Methuen and CO LTD, 1968),
pp. 183-184, for this way of characterizing the de dicto/
de re distinction.

 

_— :i,

 

\J

15

'Carnap' rigidly refers to Carnap it is clear that at each
possible world no one other than Carnap is denoted by the
name in either of its occurrences. There simply is no
possible world where the sentence 'Carnap is not Carnap' is
true, and hence there is no reading under which (I) is true.
This would not be the case if 'Carnap' were not a
rigid designator (i.e., if 'Carnap' picked out different
things in different worlds). In this case while (3) would
remain false, (2) would be true. This can be more readily
seen by substituting some definite description, say 'the

author of Meaning and Necessity', for each occurrence of

 

'Carnap' in (l). The sentence that results,

(4) The author of Meaning and Necessity might not
have been the author of Meaning and Necessity.

 

 

has a de re reading, viz.

(5) The author of Meaning and Necessity is such
that he possibly is not the author of Meaning
and Necessity.

 

 

and a de dicto reading, viz.

 

(6) It is possible that: the author of Meaning
and Necessity is not the author of Meaning
and Necessity.

 

 

16

When we evaluate a modal sentence containing a definite
description in the subject position that has narrow scope
with respect to the modal term we reevaluate the description
at each possible world. We do this because the value of the
description may change from world to world. If, on the
other hand, the definite description has wide scope with
respect to the modal term, its value remains fixed through
all possible worlds in which it has a value. Now, what (5)
says is that that very person who is the author of Meaning

and Necessity in the actual world is such that in some pos-

 

sible world he is not the author of Meaning and Necessity.

 

(5) is true because while the value of the definite descrip-
tion that precedes the modal operator remains fixed from one
world to the next, the value of the definite description
that follows the modal operator may change from world to
world. Thus in some possible world the author of Meaning

and Necessity might be Hegel and it is certainly true that

 

Carnap is not Hegel. What (6) says, on the other hand, is
that there is some possible world where the author of

Meaning and Necessity is not the author of Meaning and

 

 

Necessity. Clearly, there is no such world. The value of

 

the definite description may change from world to world, but
it must have the value it in fact has in each world. Since

'the author of Meaning and Necessity' is not a rigid desig-

 

nator (4) will be false read de dictq as in (6), and it will

be true read de re as in (5).

 

 

 

l7

Kripke, in his article, "Identity and Necessity”,
gives a similar, though more homey, account of rigidity.1
In Naming and Necessity, he characterizes his intuitive test

for rigidity in the following manner.

What's the difference between asking whether it's
necessary that 9 is greater then 7 or whether it's
necessary that the number of planets is greater than K
7? Why does one show anything more about essence than "”1
the other? The answer to this might be intuitively

'Well, look, the number of planets might have been

different from what it in fact is. It doesn't make any M
sense, though, to say that nine might have been diff- J
erent from what it in fact is.‘2 ‘

Thus according to Kripke's test, 'nine' is a rigid designa-

tor and 'the number of planets' is not. In general, then,

it appears as though we can say of any singular term a that

it is rigid if and only if @ might have different than d1 5:
is false. (I have slightly altered Kripke's account because

it is not entirely clear that ”it doesn't make any sense” or

is "meaningless” to say that nine might not have been nine;

it seems more likely that it is just false or perhaps nec-

essarily false.) Given Kripke's intuitive test for rigidity

we can return to the examples given earlier and see that

'Carnap' is rigid by noting that the de re reading of (1)

 

lSaul Kripke, "Identity and Necessity,” in Identity
and Individuation, ed. by Milton Munitz (New York: New York
University Press, 1971), pp. 148-149.

 

2Kripke, Naming and Necessity, p. 48.

 

IF
A

18

(i.e., (2)) is false. We can also see that the 'author of

Meaning and Necessity' is not rigid by simply noting that

 

the gg_£g reading of (4) (i.e., (5)) is true.

Now, what I would like to do is formalize Kripke's
intuitive account of rigidity in order to make his claim
that proper names are rigid designators more perspicuous.
To do this it will be useful to find some syntactic device
that will allow us to make scope distinctions in a system-
atic way that exhibits the distinction that has been drawn
between (2) and (3), and (5) and (6) respectively. Tradi-

tionally the de re/de dicto distinction has been drawn in

 

a quantified modal logic. Consider, for example, the fol-

lowing sentence:

(7) Everything is necessarily extended.

(7), it would be pointed out, is ambiguous since we can
distinguish a modality de re from a modality de dicto. If
we let ”E” abbreviate 'is extended' we can formally illus-

trate the distinction in the following manner:

(8) (X)C]Ex;
(9) [](X)Ex

(8) expresses a de re modality, and it says that every
thing is such that it is necessarily extended. (9) ex-

presses a de dicto modality and it says that necessarily

l9

everything is extended. Notice, that if one is a material-
ist who also believes that spirits are possible (although
not actual) and who holds that being extended is an essen-
tial property of everything that actually exists, then (8)
will be taken to be true, while (9), on the other hand,
will be believed false. Our imagined materialist in this
case may be viewed as objecting to the validity of the
Barcan formula, (x)E]¢x 22> E] (x) (bx, since he assumes that
there are some objects (spirits) that exist in some possible
world and are such that they not only have a different
property (the property of being nonextended) than those
objects that exist in the actual world, but are objects
that do not exist in the actual world at all. It should
also be noted that given Kripke's or any other plausible
semantics for a quantified modal system like SS, the con-
verse of the Barcan formula is also invalid. To see this
let the predicate 'E' used in (8) and (9) stand for exis-
tence. In SS (9) would be provable, while (8) would not be
valid, since as Kripke states, ”although it is necessary
that every thing exists, it does not follow that everything
has the property of necessary existence."1

The de re/de dicto distinction can also be made for

 

modal sentences with singular terms occurring in the subject

position. The sentence 'Carnap is necessarily a man'

 

.ISaul Kripke, "Semantical Considerations on Modal
Logic,‘ in Reference and Modality, ed. by Leonard Linsky
(London: Oxford University Press, 1971), p. 70.

 

20

exhibits the same ambiguity as does (7), and we can again
express the distinction in a modal logic by simply adding
identity. Let "c” abbreviate 'Carnap' and "M" abbreviate

'is a man', then the following sentences result:

(10) (3X) (x=c 8: DMX)
(ll) [3(3X) (x=c 8: MX)

(10) expresses a de re modality, and the occurrence of the
name 'Carnap' is given wide scope relative to the modal
operator. (ll) expresses a de dicto modality, and the
occurrence of the name 'Carnap” is given narrow scope rel-
ative to the modal operator.

Now, in order to formalize sentences like (2) and
(3), and (5) and (6), and clearly indicate the scope of the
occurrences of the singular terms contained in them rel-
ative to a modal operator, a less cumbersome way of making
the distinction illustrated above would be helpful. What
I propose is that both names and definite descriptions be
treated as singular quantifiers.1 Let me introduce the
following definition of a singular quantifier for names.

Where a is any name and Ex is any sentence (open or closed)

 

1I owe this suggestion to Herbert E. Hendry, as
well as the definition of a singular quantifier for names
that follows in the text. Also, for an equivalent but
alternative way of making the de dicto/de re distinction
see David K. Lewis, Counterfactuals (Cambridge: Harvard
University Press, 1973), pp. 36-37.

 

 

21

[a x]Fx = Df (3X) (x=c & FX)

In a sentence of the form [azx]Fx, [dzx] is to be read as:
a is an x such that ——. Let us again return to sentence
(2). Using our singular quantifier for names we can repre-
sent (2) as the de re or large scope reading of (l) by
giving the first occurrence of the name 'Carnap' wide scope
with respect to the modal operator. The formula that re-

sults is as follows:

(12) [c;x1<>x¢c.1

(12) it will be recalled is false, as is the sentence that
results if both occurrences of 'Carnap' in (l) are given
wide scope relative to the modal operator: namely, [sz]
[c:y]€>x%y. If we give both occurrences of 'Carnap' in
(1) narrow scope with respect to the modal operator the

formula that results,

(13) OC7‘C.

is a translation of (3), and it is false as well.
Given Kripke's intuitive test for rigidity men-

tioned earlier one might be tempted to claim that (12)

 

1Where names appear as arguments of predicates they
are to be interpreted as having smallest possible scope.

22

(but not (13)) expresses his claim that 'Carnap' is a rigid
designator. Notice, however, that (12) entails that Carnap
exists in the actual world and such a requirement might

eventually prove to be problematic. I will return to this

matter shortly.

Let me first introduce the following definition of
a singular quantifier for descriptions analogous to the one
introduced for names. Where a is any definite description

of the form (7y) Gy and Fx is any sentence
[a:X]Fx = Df (3x) ((y)(Gy ES y=x) & Fx)1

Using the singular quantifier for descriptions defined above

we can now translate (5) as follows:

(14) [(7X)FX:X] O X7é(’7X)Fx2

(14) it has been noted is true. The formula that results,
however, when both occurrences of the definite description
in (4) are given narrow scope relative to the modal

operator is false, viz.

 

1The idea of treating definite descriptions as
singular quantifiers is credited by W.V.Quine to Richard
Sharvey, See Quineksreply to Grice in Words and Objections,
ed. by D.Davidson and J.Hintikka (Dordrecht-Holland: D.
Reidel Publishing Company, 1969), p. 327.

2A5 with names, definite descriptions where they
appear as arguments of predicates are to be interpreted as
'having smallest possible scope.

23

(15) 0(7X)Fx?‘-(7X)Fx

(15) is a translation of (6) and it is false as is the
sentence that results if both occurrences of 'the author of

Meaning and Necessity' in (4) are given wide scope with

 

respect to the modal operator, viz., [(7x)Fx:y][(7x)Fx:z]
Oyfi-

Again, given Kripke's intuitive test for rigidity
one might be tempted to hold that (14) expresses the fact

that 'the author of Meaning and Necessity' is not a rigid

 

designator. One might even take (12) and (14) to yield the
following rough definition of rigidity. Where a is any
name or description, to say that a is rigid is to say that

the following condition holds:

(A) [oc:X][:7x=oc.

To do so, however, would be, I believe, a mistake. Given
my definitions of the singular quantifiers for names and
descriptions it should be clear that (A) entails that q
denotes something that exists in the actual world. What
this in effect rules out are cases where u is rigid yet the
object denoted by a only exists in some possible world
other than the actual world. Now there may or may not be
such cases and I would like to remain neutral on this issue,

but it seems that we need to allow for such a possibility

24

if we are to retain the spirit, if not the letter, of
Kripke's rough statement of rigidity. Recall, that accord-
ing to Kripke a singular term a is rigid if it designates
the same thing in every possible world where it designates
at all. Hence, this would seem to allow for cases where a
term fails to denote in the actual world, yet denotes the
same object in every possible world where it designates at
all.

In Naming and Necessity, Kripke repeatedly asks the

 

reader to ignore complications that might arise from the
possible nonexistence of the object. For the most part, he
merely assumes that we are dealing with cases where the
term in question denotes an object that exists in the actual
world. Rather than extend his caveat to a more formalized
account of rigidity, I suggest that we seek a definition
that allows for names that fail to denote in the actual
world.

lk1"Identity'and Necessity", Kripke is also aware of
the problem posed by names or definite descriptions that
fail to refer. He notes in that article that someone might
object that the claim that every object is necessarily
self-identical (i.e., (x)C]x=x) is false, since some sin-
gular terms might fail to denote in the actual world or
denote in the actual world but not in all possible worlds.

In an attempt to handle the problem he states:

 

25

Let us interpret necessity here weakly. We can count
statements as necessary if whenever the objects men-
tioned therein exist, the statement would be true. If
we wished to be very careful about this, we would have
to go into the question of existence as a predicate
and ask if the statement can be reformulated in the
form: For every x it is necessary that, if x exists,
then x is self-identical.1

Following Kripke's suggestion in the above quotation we

might propose to replace (A) with the following condition:

(B) D(E!0L:> [oux]Dx=o¢).

Condition (B), as opposed to (A), clearly eliminates the
need to make the assumption that the term in question de-
notes an object which exists in the actual world. Unfortu-
nately, (B) is not a condition that Kripke could live with.
(B) has the unacceptable consequence that if (7x)Fx is rigid
then E!(7x)Fx entails E3E!(7x)Fx. To see this suppose that
(7x)Fx is rigid and that E!(7x)Fx holds in some world w.
Then [(7x)Fx:x]E} x=(7x)Fx holds in w. Since [(7x)Fx:xJC]
x=(7x)Fx holds in w, there is some object designated by
(7x)Fx in w and that object is designated by (7x)Fx in each
world. Thus, (7x)Fx=(7x)Fx holds in each world. But (7x)Fx
=(7x)Fx entails that El(9x)Fx. So E!(7x)Fx holds in every
world. Therefore, le!(7x)Fx holds in world w. Whether or

not a parallel argument for names can be given is a question

 

1Kripke, ”Identity and Necessity,” p. 136.

26

that need not detain us, since it is clear that when a is
a rigid definite description (B) entitles us to hold that
Ela 43 E3E!a.

Given Kripke's views concerning rigidity it seems
clear that he simply could not accept condition (B). In

”Identity and Necessity", for example, he states that:

in talking about the notion of a rigid designator,
I<k>not mean to imply that the object referred to has
to exist in all possible worlds, that is, that it has
to necessarily exist. Some things, perhaps mathematical
entities such as positive integers, if they exist at
all, necessarily exist.1

Now, the result that we have obtained from (B) might not
seem so bad, at least as far as some rigid definite descrip-
tions are concerned. Consider, for example, 'the smallest
prime'. As Kripke points out in the passage quoted above,
perhaps a mathematical entity like the smallest prime nec-
essarily exists. Kripke, however, clearly would not want
to hold that the object referred to by 'Quine' necessarily
exists. Yet this is precisely what (B) would commit him to,
and the name 'Quine' need not be the culprit. The definite
description '(7x)x=Quine' would appear to satisfy (B), and
assuming that the object denoted by it, viz. Quine, exists
in the actual world, (B) has a result that Quine necessar-

ily exists. Hence, condition (B) simply will not do.

 

lIbid., p. 145.

 

27

At first blush one might suppose that replacing (B)
with fij(E!q.:9 [3(Eld :3 [azx][jx=a)? avoids our present
difficulty. It does not. For it is, as may be readily
verified, equivalent to condition (B). What is needed is
a definition of rigidity that does not rule out names that
fail to denote in the actual world, yet at the same time
does not entail that if the object referred to exists, it
necessarily exists. In order to accomplish this I suggest
that we adopt the following definition. Where a is any
name or description, to say that a is rigid is to say that

the following condition holds:
(C) D(Eloc :3 [oc:x]Lj(E!oc:D x=oc)).

Not only does (C) allow for terms that fail to refer in the
actual world, it also does not commit us to holding that
the object referred to by '(0x)x=Quine' necessarily exists.
One possible drawback, however, is that it seems as though
some singular terms will vacuously satisfy (C) by not denot-
ing in any possible world. Thus, for example, terms like
'the largest prime' or 'the round square' or possibly even
names of "fictional entities” like 'Pegasus' will all be
rigid designators according to (C). Notice, however, that
if one were bent on disallowing such terms as 'the largest
prime' and 'the round square' as rigid designators, all

that need be done is to conjoin the following clause to

 

28

(C): Q0Eld7. Since there is no possible world where the
largest prime exists, 'the largest prime' would not be a
rigid designator. Notice also that the addition of this
clause to (C) allows one to remain neutral on the issue as
to whether or not the names of fictional entities are rigid
designators. If one holds a position similar to the one
held by Kripke in his article "Semantical Considerations

on Modal Logic", it would appear as though names of fic—
tional entities could count as rigid designators. In that
article Kripke states that ”Holmes does not exist, but in
other states of affairs, he would have existed." Kripke
thus seems to be allowing for the possibility that 'Holmes'
denotes in some possible world, but not in the actual world.
If, on the other hand, one takes the position that Kripke

seems to hold in the Addenda to Naming and Necessity (cf.

 

pp. 156-158), then names of fictional entities could not
count as rigid designators. In the latter work Kripke
takes back his earlier claim mentioned above and seems to
hold that by definition a fictional entity does not exist
in any world. Thus what at first glance appeared to be a
possible drawback of (C) ultimately turns out to be a vir—
tue. By adding COElaj to (C) we can rule out cases like
'the largest prime', and at the same time remain neutral
with respect to names of fictional entities.

The above definition of rigidity would also seem

to accord well with Kripke's intuitive test. 'Carnap'

 

29

turns out to be a rigid designator as evidenced by (12),

while 'the author of Meaning and Necessity' does not as

 

evidenced by (14). There is, however, yet another way in
which the thesis that proper names are rigid designators
can be made more perspicuous, and that is to note that it
is equivalent to the thesis that a scope operator has no
semantic role to play for a proper name as it occurs in a
modal sentence.l That is, the thesis that proper names are
rigid designators is equivalent to the thesis that proper
names as they occur in modal contexts do not induce seman-
tic scope ambiguity. Strictly speaking, of course, given
my definition of the singular quantifier for names the
above claim is false. (2) and (3), translated as (12) and

(13) respectively, are not equivalent, since (12) entails

 

1To the best of my knowledge Leonard Linsky in his
book, Names and Descriptions, was the first to explicitly
take note of this equivalence (cf. chapter three, pp. 42—
65). Since then the equivalence of de dicto/de re con-
structions involving a rigid designator has been noted by
Christopher Peacocke in ”Proper Names, Reference, and Rigid
Designation,” in Meaning, Reference, and Necessity, ed. by
Simon Blackburn (Cambridge: Cambridge University Press,
1975), pp. 109—115. Kripke also notes the equivalence in
a footnote in the preface to Naming and Necessity (p. 12,
n. 15). In that footnote Kripke asks the reader to ignore
complications that might arise from the possible nonexis—
tence of the object. There will, however, be no need to
extend his caveat to my formulation of the equivalence.

It should also be noted that given Kripke's state-
ment of the equivalence, his thesis that proper names are
rigid designators is ”not the same as the doctrine that
natural language has a convention that only the large scope
reading is allowed.” This is the misconception of Kripke's
thesis that Loar, McKinsey, and Dummett labor under, which
I noted earlier. But as Kripke goes on to point out ”the
sequivalence makes sense only for a language where both
treadings are admissible.”

 

 

30

that Carnap exists, while (13) does not. Hence, strictly

speaking we cannot express Kripke's claim that 'Carnap' is

H“

a rigid designator in the following manner: '[c:x]E]Fx
DFc'. However, while r[d:x][]Fx E UFO-L1 does not suf-
fice to define rigidity, strikingly close conditions do.
Let Dad; be short for C](E!o: :3 ¢) and let 0%) be short for
0(Elo: 8: ab). [30' and (>0: are thus restricted modal quanti-
fiers ranging only over those worlds in which a exists.
Then, rigidity can be defined by either of the equivalent
conditions where a is any name or description and Pa is

atomic:

(DD) DOL([0L:X][:IO'FXE mammary)
(DO) [18([azx100‘r‘x2 OO'[0::X]FX)

Notice, that given our new notation (i.e., E]a¢) we can go
back and rewrite condition (C) as follows: 'Tja[a:x][ja
x=&7. In fact, it can be shown that (C) is equivalent to

(DD) and (Do).1 Also, as was the case with (C), one can
conjoin '13Eld1 to (DD) and (D0) respectively if one wishes
to disallow such terms as 'the largest prime' and 'the

round square' as rigid designators. We can now slightly

 

1Since all modal Operators are restricted to worlds
in which a denotes, the problem of showing that (C) is
equivalent to (DD) and (Do) reduces to the problem of show-
ing that [uzx]Dx=oc is equivalent to [j([o::x]E]Fx E D
[OL:X]FX) and E]([o::x] OFx E ()[d:x]Fx) under the assumption
that a denotes in all worlds. And the reader may easily
verify that the latter equivalences hold.

 

 

31

alter Kripke's claim that the thesis that proper names are
rigid designators is equivalent to the thesis that "if a
modal operator governs a simple sentence containing a name,
the two readings with large and small scope are equivalent"
so that it will be true. The thesis that proper names are
rigid designators is equivalent to the thesis that proper I?
names satisfy conditions (DD) and (Do).

Now, while proper names satisfy condition (D), def-
inite descriptions, in general, do not. That is, in gen—
eral definite descriptions are not rigid designators and in
general they do induce semantic scope ambiguity in modal
contexts. I make the qualification ”in general” because,
as we have already noted, there are some definite descrip-
tions that do take the same value at each possible world
where the sentences that contain them are evaluated. Linsky,

in Names and Descriptions, cites (7x) (3<x<5) as just such

 

a description. A better example might be 'the least prime'
since names are not involved in the description itself.
Other examples might be 'the darkest color' or 'the smallest
set'. It is interesting to note that what all of these
descriptions seem to have in common is that they designate
abstract objects which necessarily stand in certain rela-
tionships to other members of their respective kinds; and

it is in virtue of necessarily standing in these relation-
ships that the objects necessarily have the properties that

they have. A modal sentence involving one of these definite

32

descriptions seems to have the same truth value regardless
of the scope given to the description that occurs in the
sentence. Hence, some definite descriptions like 'the
smallest prime' will satisfy (D).

Typically definite descriptions, however, do not
satisfy condition (D), and hence are not rigid designators.
Recall that (5) and (6), translated as (14) and (15)respec—
tively differ in truth value, hence we should find that'the

author of Meaningyand Necessity' does not satisfy (D). To

 

see this let "F" abbreviate 'is an author of Meaning and

 

Necessity' and let a abbreviate 'the author of Meaning

 

and Necessity'. Now, according to (DD), since Carnap exists

 

(timelessly),
(16) [azx1DO‘Fx
should be equivalent to
(17) DawleFx.

(16) is false, since being an author of Meaning and Neces-
gity is only a contingent property of Carnap's, that is
r[o::x] 00' “’qu is true. (17), on the other hand, is true.
At any world where the object designated by a exists (that
object may or may not be Carnap) it is true that that object
is an author of Meaning and Necessity; that is,r;<>a[a:x]

.1

chx is true. Hence, (16) is not equivalent to (17), and

33

hence 'the author of Meaning and Necessity' does not sat-

 

isfy (D). Definite descriptions are not characteristi-
cally rigid designators. Proper names, on the other hand,
are always rigid if Kripke is correct. Thus proper names
will always satisfy conditions (C) and (D), while definite
descriptions in general will not.

Throughout this section I have attempted to give a
formal account of the notion of rigidity that does not
dodge the crucial issues concerning nonexistence that are

affected by it. In the preface to Naming and Necessity,

 

Kripke shows an acute awareness of these issues when he

states:

Concerning rigidity: In many places, both in this
preface and in the text of this monograph, I deliber-
ately ignore delicate questions arising from the pos-
sible nonexistence of an object. I also ignore the
distinction between 'de jure' rigidity, where the ref-
erence of a designator is stipulated to be a single
object, whether we are speaking of the actual world or
of a counterfactual situation, and mere 'de facto'
rigidity, where a description 'the t such that Fx' hap—
pens to use a predicate 'E' that in each possibIE world
is true of one and the same unique object (e.g., 'the
smallest prime' rigidly designates the number two).
Clearly my thesis about names is that they are rigid
de 'ure, but in the monograph I am content with the
weafier assertion of rigidity. Since names are rigid
de jure ... I say that a proper name rigidly desig-
nates its referent even when we speak of counterfactual
situations where that referent would not have existed.
Thus the issues about nonexistence are affected. Var-
ious people have persuaded me that all these questions
deserve a more careful discussion than I give them in
the monograph, but I must leave them here.

 

 

 

lKripke, Naming and Necessity, p. 21.

 

34

I believe that the account of rigidity that I have given,
in terms of condition (C) and conditions (DD) and (Do), not
only allows for Kripke's distinction between de jure and

de facto rigidity, but does so in such a way that the "del—

 

icate questions arising from the possible nonexistence of
the object” are not ignored. Having accomplished this,
however, I will now ask the reader to extend to me the same
license to ignore the various complications concerning non-
existence that has been extended to Kripke. I do this
merely to facilitate the discussion of various arguments
involving rigidity that occur later in the paper. The
reader should keep in mind, however, the fact that any of
my latter discussion of arguments involving rigidity can
always be reformulated in terms of condition (C) or condi-
tion (D), so that no questions concerning nonexistence are

begged.

Section Three

 

Michael McKinsey in "The Reference of Proper Names'
also notes that in general definite descriptions do not be-
have in modal contexts as do proper names. That is, he
notes that proper names are rigid designators and definite
descriptions, in general, are not. McKinsey further notes
that "if ordinary names are rigid designators, then, as

Kripke points out, they cannot in general, be synonymous

35

with definite descriptions.”1

After stating this, however,
he then curiously goes on to give the following character—
ization of Kripke's modal argument.

Consider the following two sentences:

(1) It is possible that the 1973 winner of the
Triple Crown is not the 1973 winner of the
Triple Crown.

(2) It is possible that Secretariat is not the
1973 winner of the Triple Crown.

McKinsey rightly points out that (l) is false since both
occurrences of the definite description 'the 1973 winner
of the Triple Crown' fall within the scope of the modal
operator so that ”(1) entails that in some possible set of
circumstances, something is not identical with itself.”2
He also correctly points out that (2) is true since 'Sec—
retariat' rigidly designates the same thing in every pos-
sible world, and in some world the horse that was the 1973
winner of the Triple Crown may not have been Secrteariat.

Thus, according to McKinsey ”since (1) and (2) have differ— 'W>

ent truth-values, it follows that 'Secretariat' and 'the
“HE

1973 winner of the Triple Crown' are not synonymous in

English."3 All that remains to be done to complete Kripke's

 

lMichael McKinsey, "The Reference of Proper Names”
(unpublished Ph.D. dissertation, Indiana University, 1976),
p. 32.

2 3

Ibid. Ibid., p.33.

36

argument, according to McKinsey, is to generalize the re-

sult as follows:

for any name and any contingent description... if
a name a rigidly designates and a description (the F7
is contingent, rit possible that a # the Flyill be
true apd ’it is possible that the F # the F will be
false.

McKinsey's response to the argument as stated is simply to
require that the definite description that is substituted
for a proper name be given widest possible scope in modal

contexts. He states that

Given this supplement, an ordinary name like 'Secretar-
iat' is typically used as a rigid designator, since it
typically has largest possible scope in the sentences
in which it occurs. But this is consistent with sup-
posing that whenever a speaker uses 'Secretariat' on a
particular occasion, he is using it as short for a def—
inite description having large scope. For instance a
speaker of (2) might be using 'Secretariat' as a rigid
designator while the thought in his mind is explicitly
expressible by

(3) The individual 5 which won the Triple Crown
in 1973 is such that it is possible that g is
not the 1973 winner of the Triple Crown.
Since (2) and (3) have the same truth value, it is con-

sistent to suppose that in uttering (2) a speaker is
expressing a thought which is also expressible by (3).2

It should be clear from the above quotation that McKinsey

feels that he has been able to dodge Kripke's argument by

 

l 2

Ibid. Ibid., p. 34.

 

37

showing that (2) can have the same truth value as a sen-
tence which has a coreferential definite description sub—
stituted for the name, as long as that description has

latgest possible scope.

 

I am willing to grant that McKinsey's response to
to the argument he considers is adequate. What I am not
willing to grant, however, is that the argument he responds
to is in fact the modal argument that Kripke gives. I will
also attempt to show, later in this essay, that, in general,
any attempt to dodge Kripke's modal argument by a move to
wide scope definite descriptions is doomed to failure. Be-
fore]ch>so, however, I would like to consider yet another

argument that runs along the same lines as McKinsey's.

Section Four

 

Brian Loar has also interpreted Kripke's modal ar-
gument against the View that ordinary proper names are
disguised definite descriptions in such a way that the
argument can be met by requiring that the definite descrip-
tion that is substituted for a proper name be given widest
possible scope in modal contexts. According to Loar one
popular form of the argument, which derives from Kripke's

Naming and Necessity goes as follows:

 

(1) If n were used to mean the F, then 'n might
not have been the F' would be false.

3.8

(2) But 'n might not have been the F' is true.

(3) Therefore, n is not used to mean the F.1

Loar's claim is that the argument is defective.
What he argues is that premise (2) is true because the
n-position in the sentence mentioned in the premise is
given widest possible scope.2 But, Loar goes on to argue,
if we treat the n-position in the sentence mentioned in
the consequent of premise (l) as also having widest pos-
sible scope, then premise (l) is false and the argument is
unsound. We can easily see that premise (l) is false
according to Loar by simply substituting 'the F' for 'n'.

In order to evaluate Loar's argument and in order
to make his claim concerning the scope of the n-position
more perspicuous let me switch his statement of Kripke's

3

argument from the formal mode to the material mode. Put

 

1Brian Loar, ”The Semantics of Singular Terms,”
Philosophical Studies, 30 (1976). p. 373.

 

2Of course, if Kripke is correct that proper names
are rigid designators, then scope ambiguity would not
affect the second premise. That is, the sentence men—
tioned in premise (2) would be true on both a wide scope
and a narrow scope reading of the n-position. To simply
assume at this point, however, that proper names are rigid
designators would be to beg the question against Loar,
since it is not an assumption that he seems to hold.

3The change from the formal mode to the material
mode is justified insofar as it does not appear to alter
the essential nature of the argument. Also, it should be
noted that such a change could always be justified by argu—
ing that the following equivalence holds. Where '8' is
any sentence: S is necessary if and only if 'EJS' is true.

39

into the material mode the modal argument attributed to
Kripke is as follows: Let'Thqfl'abbreviate 'x means y'.
(1') Mn(7x)Fx + [n:x]EJFx1
(2') ~[n:x] DFx

(3') foean(7x)Fx

To see how Loar's argument works let us consider a
particular case. For example, let us take the name 'George
Washington' and the definite description 'the first pres—
ident of the United States'. According to Loar "names are
normally read as having wider scope than modal operators”.2
Thus premise (2') with the name given widest possible scope
is clearly true. What (2') says is that it is not the case

that George Washington is such that necessarily he is a

first president of the United States, and surely we can

 

1It should be noted that strictly speaking my
translation of the argument is incorrect since I have re-
placed the definite description in the sentence mentioned
in premise (l) and in premise (2) with an indefinite de-
scription. The reader can easily verify, however, that
this does not in any way affect the basic point that Loar
is trying to make, nor does it significantly alter his
formulation of Kripke's argument. The change to an indef—
inite description is simply to facilitate dicussion later
in the essay of an attempt by Hudson and Tye to revise the
premises of the argument in such a way that an appeal to
scope distinctions becomes irrelevant.

Also, it should be noted that I am assuming, as
does Loar, that 'n' denotes in the actual world.

2Loar, "The Semantics of Singular Terms,” p. 373.

 

40

agree that being a first president of the United States is
only a contingent property of George Washington. He might
have died in the Revolutionary War and never become a

president at all.

Loar contends, however, that the scope consideration

which makes premise (2') true is the same feature that makes

premise (1') false. His key claim is that (l') with the
name in the consequent given widest possible scope is
false. What (1') says is that if 'George Washington' means
the first president of the United States, then George Wash-
ington is such that necessarily he is a first president of
the United States. To see that (1') is false Loar claims

that all we have to do is substitute 'the F' for n'. Thus

we have the following:

(1”) M(7x)Fx(7x)Fx + [(7x)Fx:x][]Fx

What (1”) says is that if 'the first president of the
United States' means the first president of the United
States, then the first president of the United States is
such that necessarily he is a first president of the United
States. But surely that very person who was in fact the
first president of the United States need never have be—
come a president at all. Hence, the consequent of (1”) is
false which makes (1") false, and Kripke's argument is

unsound.

41

In essence what Loar has done is save the descrip-
tion View of proper names from Kripke's modal argument by
requiring that proper names be identified only with those
definite descriptions that take widest scope in modal con-
texts. Again, however, the crucial question that needs to
be raised is not whether Loar's move to wide scope definite
descriptions is an adequate response to the stated form of
Kripke's argument. Rather, the crucial question is wheth-
er or not Loar's formulation of Kripke's modal argument is

the argument that Kripke in fact gives in Naming and Neces—

 

ity, and if it isn't will his response remain adequate?
Before I answer this question I would like to examine a
recent attempt to salvage Kripke's argument, as stated by
Loar, by modifying it in such a way that an appeal to scope

distinctions becomes irrevelant.

Section Five

 

In a recent article in Analysis James Hudson and

 

Michael Tye attempt to revise the premises of Kripke's ar-
gument in such a way as to thwart the move to wide scope
descriptions. They feel that Loar's response is an ade-
quate rejoinder to Kripke's argument as stated, but that
their modification of it decisively shows that "the view
that proper names are synonymous with definite descriptions

cannot be saved by giving the descriptions widest possible

42

scope".l Their restatement of the argument is as follows:

(4) If n were to mean the F, then 'n is an F'
would be necessary on the assumption that
there is a unique F.

(5) 'n is an F' is not necessary on the assump—
tion that there is a unique F.

(6) Therefore, n does not mean the F.2

Hudson and Tye maintain that the primary advantage
of their revised argument is that (4) is immune to a
possible scope ambiguity in a way in which (1) of section
four is not. This is so, they claim, because ”while in (l)
we have a remark about a sentence which contains a modal
operator, in (4) we have rather a modal remark about a non-
modal sentence”.3 What I would like to show, however, is
that Hudson and Tye are simply wrong. That is, I will
attempt to show that (4) properly understood is not only
syntactically ambiguous with regard to scope but that it is
semantically ambiguous in this regard as well. To accom-
plish this end itwd11.again be useful to switch the argu—
ment from the formal mode to the material mode. Here it
might be objected that keeping the argument in the formal

mode is an essential feature of the argument. I would

 

1J.Hudson and M.Tye, ”Proper Names and Definite
Descriptions With Widest Possible Scope,” Analysis,
40 (January 1980) p. 64.

21bid., p.63. 31bid., p.64.

43

again, however, suggest that one could always argue that
the change is justified based on the equivalence referred
to in footnote 3 on page 38.

Also it could be pointed out that Hudson and Tye
rely rather heavily on the fact that they are dealing with
a nonmodal sentence. Because of this and because of the
fact that they are working in the formal mode they seem con—
vinced that a nonmodal sentence like 'n is an F' will not be
ambiguous with regard to scope when a definite description
is substituted for 'n'. It is not entirely clear, however,
why it is that they are so confident about this. Certainly
there is no general principle to the effect that no nonmodal
sentence with a definite description in the subject position
will be ambiguous with regard to scope that they can appeal
to. Clearly there are at least some nonmodal sentences that
are ambiguous with regard to scope regardless of whether
they are treated in the formal mode or the material mode.

Consider the sentence,

(A) The present king of France is not bald.

According to Russell it is ambiguous with regard to scope
since the definite description 'the present king of France'
may have either a primary occurrence in (A) or a secondary
occurrence in (A). If the description has wide scope or

primary occurrence then the proposition expressed by (A)

44

entails that the present king of France exists. In this

case (A) would read as follows according to Russell.

(B) (3X)(Kx & (y)(Ky :: x=y) & ~Bx)

Since 'the present king of France' is an improper descrip-
tion (i.e., lacks a denotation) the sentence is false. If
on the other hand, the description has narrow scope or sec-
ondary occurrence, then the proposition expressed by (A)
does not entail that the present king of France exists. In
this case the sentence would be true and according to

Russell (A) would read as:

(C) ~(EX)(Kx 8. (y)(Ky Dx=y) & Bx)

It should also be clear from the above discussion
that the claim that the sentence 'the present king of France
is not bald' is false is ambiguous as well. It could mean
that the sentence 'there is a present king of France who is
not bald' is false, in which case the claim is correct, or
it could mean that the sentence 'There is no present king
who is bald' is false, which would be incorrect.

It might, of course, be objected that even though
Hudson and Tye do not explicitly emphasize the fact that it
is atomic, nonmodal sentences that they are talking about,

it clearly was their intent. Hence, trotting out Russell's

45

well worn example is merely a red herring, since it is not
an atomic sentence. It might also be pointed out that while
(A) is ambiguous with regard to the scope of the definite
description relative to the tilde, it need not be ambiguous
with regard to the scope of the definite description rela—
tive to a modal operator.

I think that both of these points are sound; never-
theless I find Russell's example instructive in this context
since its limitations indicate what would be required to
show that (4) is false. What we need is an atomic nonmodal
sentence with a definite description in the subject position
that is ambiguous with regard to the scope of the descrip-
tion, and that is also ambiguous with regard to the scope of
the description relative to a modal operator. Consider the

following sentence:
(D) The murderer of Smith is a murderer.

Now according to Hudson and Tye if we say of (D) that what
it expresses is necessary, our claim will not be ambiguous
with regard to scope. I would like to suggest, however, a
plausible argument to the contrary. The following argument
was first suggested to me by Richard Hall. First let us
consider (D) itself. If one holds that Donnellan's refer-
ential/attributive distinction for definite descriptions is

a semantically relevent distinction, then it could be argued

46

that (D) is ambiguous depending on whether or not 'the
murderer of Smith' is used referentially or attributively.
If the definite description in (D) is used referentially,
then what (D) expresses is that some particular individual
is a murderer, regardless of whether or not he happens to
be the murderer of Smith. Now suppose (D) is uttered in a
context where 'the murderer of Smith' picks out Jones and
Jones is not a murderer. In this case (D) will be false.
If we treat referential descriptions as simply descriptions
that have wide scope we can express this reading of (D) as

follows:
(E) [(7y)Fy:x}Fx

If, on the other hand, the definite description in (D) is
used attributively, then what (D) expresses is that the
murderer of Smith whomever he might be is a murderer. In
the same case as described above, (D) could be true on this
second interpretation, and if we similarily treat attribu-
tive definite descriptions simply as descriptions that have

narrow scope we can express this reading of (D) as follows:

(F) F(7y)Fy

So far then we have shown that an atomic, nonmodal

sentence can be ambiguous with regard to the scope afforded

47

to the definite description, as evidenced by (E) and (F),
which have been shown to be false and true respectively.

Now notice that if we say of (E) that what it expresses is
necessary, then what we say is false. (E) will only be true
at those worlds where Jones is in fact a murderer, but we
know that there is at least one possible world, namely the
actual world, where Jones is not a murderer. Hence, our
claim that what (E) expresses is necessary is false. To
express this in the material mode we need only give the def-
inite description in (E) wide scope relative to the neces-
sity operator.

If, on the other hand, we say of (F) that what it
expresses is necessary (barring the nonexistence of Smith's
murderer in some world w), then what we say is true. There
will be no possible world in which the murderer of Smith in
that world is not a murderer. Thus the proposition ex—
pressed by (F) is necessary while the proposition expressed
by (E) is not. Again, to express in the material mode our
claim that what (E) expresses is necessary we need only give
the definite description in (E) narrow scope relative to the
necessity operator.

We have been able to show that there is a scope
ambiguity in an atomic, nonmodal sentence (i.e., (D) read
as (E) and (F)) and that it affects the necessity of that
sentence (i.e., the claim that what (E) expresses is neces-

sary, is false, while the claim that what (F) expresses is

48

necessary, is true). Thus we have shown that premise (4)
is false.

Those, who like myself, find questionable the claim
that Donnellan's referential/attributive distinction for
definite descriptions is a semantically relevant distinc-
tion will not find the above argument to be very convincing.
Those philosophers who agree with Kripke, for example, that
Donnellan's distinction is only a pragmatic one would never
let the argument get off the ground.1 Aside from this dif-
ficulty with the argument there are a number of other prob-
lems associated with it that I will merely suggest here,
but not elaborate on. For one thing the above argument
treats Russell's notion of scope as a simple dichotomy to be
associated with attributive and referential definite de-
scriptions. Also, it might be plausibly argued that
Russell's notion of scope doesn't even apply to our (D),
since in an atomic sentence like 'The murderer of Smith is
a murderer' there does not appear to be any room for scope

ambiguity, at least at the level of surface structure".
Further, this last point might be used by some to buttress
their claim that (E) and (F) are actually equivalent, and

hence do not express different propositions.

One last difficulty that might be raised goes as

 

1See Saul Kripke, "Speaker's Reference and Semantic
Reference,” Midwest Studies in Philosophy, Vol. II: Studies
in the Philosophy of Language (February 1977), pp. 255-276,
for Kripke's account of the distinction.

 

..

49

follows. It is natural to assume that if our (E) and (F)
are not equivalent, it is because the definite description
in (E) is rigid and the definite description in (F) is non-
rigid. Now one thing it might mean to call a definite de-
scription, say (bx)¢x, rigid is that "it denotes, with re-
spect to all possible worlds the unique object that (actu-
ally) ¢'sJJ' If this is the case, then we cannot associate
the rigid definite description in (E) with Donnellan's re-
ferential definite description. This is so because if a
rigid definite description is defined as above, it "deter-
mines its referent via its unique satisfaction of the asso—
ciated property"2 Donnell's referential definite descrip-
tion, on the other hand, might pick out some individual who
does not satisfy the associated property of the description,
as was the case in our example with Jones.

For these reasons then, as well as for some that I
am sure that I failed to consider, I find it necessary to
continue in my attempt to show that Hudson and Tye are
wrong by switching their argument to the material mode.

Hudson and Tye start out by considering the sen—

tence 'The F is an F'. Concerning that sentence theynote it

. is not itself a necessary truth, but it is entailed
by the sentence 'There is exactly one F' (given a fixed

 

l 2

Ibid., p. 259. Ibid., p. 260.

50

value of 'F').... it is necessary on the assumption
that there is a unique F, meaning by this simply that

it is a necessary consequence of 'There is exactly one
F'.... if n were to mean thg F, then n would be substi-
tutable salva veritate for the F in '"The F is an F" is
necessary on the assumption—fhat there is a unique F'.1

Now this leaves us with five possible translations for the

consequent of (4). They are as follows:

(7) E!(’}x)Fx :> DFn

(8) E!(7x)Fx :9 [nleL'JFx

(9) D(El(?x)Fx:D Fn)

(10) D([n:x](E!(7x)Fx ‘3 Fx))
(ll) [n:x]D(E!(’7x)Fx:9 Fx)

Notice that (7) and (8) are, nearly enough for present pur-
poses of the form P :DEJQ, while (9) and (10) and (11) are
again nearly enough for present purposes of the form

[J (P D Q). Now, clearly a translation of the form [J (P ‘3 Q)
would be a more charitable translation of the consequent of
(4), although a translation of the form P 3 DO would seem
to be a better rendition of the English, as given. To see
this consider the argument with (7) as the translation of

the consequent of (4).

 

1Hudson and Tye, "Proper Names and Definite De-
scriptions With Widest Possible Scope,” p. 64.

 

51

(4') Mn(7x)Fx + (E!(7x)FxD EJFn)
(5') El(7x)Fx:D~DFn

(6') .3 ~Mn(7x)Fx

The argument as it stands is not even valid. What premises
(4') and (5') entitle us to conclude is not (6') but
'E!(7x)Fx:D'VMn(7x)Fx'. Of course, the argument can be
made valid by either adding 'E!(7x)Fx' as a premise or by
translating (5) as '~(E!(9x)Fx D DFn)’. Essentially the
same point can be made if we consider the argument with (8)
as the translation of the consequent of (4). This, however,
is not the chief difficulty with treating a translation of
the form P ZDEJQ as the correct translation of the conse-
quent of (4). When we substitute 'the F' for 'n‘ in (7)

and (8) we can get the following two readings of (4):

(12) Mn(7x)Fx -> (E!(7x)Fx D EJF(7x)Fx)
(l3) Mn(7x)Fx -> (E!(7x)Fx D [(7x)Fx:x]EJFx)

The first thing to note about (12) and (13) is that they
are syntactically distinct. Thus (4) at least displays a
syntactic ambiguity with regard to scope. If, however, we
let ”n” abbreviate 'Benjamin Franklin', ”(7x)Fx” abbreviate
'the inventor of bifocals', and ”F” abbreviate 'is an in-
ventor of bifocals', we will see that (12) and (13) have

the same truth value. This might lead one to hold that

52

Hudson and Tye have been successful in their attempt to
make Kripke's argument immune to possible scope ambiguity,
at least as far as semantic scope ambiguity is concerned.
This, however, would be a mistake, for while it is true
that the consequents of (12) and (13) both have the same
truth value under the interpretation given above, the truth
value they have is not the one that allows the argument to
succeed. Both (12) and (13) are false. What the conse—
quent of (12) says is that if there is a unique inventor of
bifocals, then there is a unique inventor of bifocals in
every possible world. But surely, assuming that there is a
unique inventor of bifocals in the actual world, say
Benjamin Franklin, it is still possible that in some world
other than the actual that bifocals were never invented or
that two people invented bifocals.l What the consequent of
(13) says is that if there is a unique inventor of bifocals,
then that very person is such that in every possible world
he is an inventor of bifocals. But, clearly being an in-
ventor of bifocals is only a contingent property of
Franklin. In some possible world he might have died at
birth and never invented anything at all. Only the vacuous

case would make the consequent of (12) and (13) true.

 

1It has been pointed out to me by Richard Hall that
the situation isn't saved by stipulating some standard
entity a, say the null set, to be the denotation, in a
world, of definite descriptions which ”intuitively" fail to
denote anything in that world. For such an entity won't be
an inventor of bifocals, and hence F a will be false there

(at those worlds); hence E]F(7x)Fx is false

53

Excluding that possibility the consequent of both (12) and
(13) are false, hence (12) and (13) are false (assuming, of
course, as we have all along that their antecedents are
true). Therefore, the more charitable reading of the con-
sequent of (4) is clearly a translation of the form

[HP 3 Q), and not one of the form P 3 DO.

Now, this still leaves us with (9), (10) and (11)
as possible translations of the consequent of (4). What
(9), (10) and (11) show us, however, is that not only is
(4) syntactically ambiguous with regard to scope, but that
it is semantically ambiguous in this regard as well. When
we substitute 'the F' for 'n' in (9), (10) and (11) we get

the following three readings of the consequent of (4):

(9') D(E!(’)X)Fx9 F(7x)Fx)
(10') D([(?x)Fx:x](El(9x)FxD Fx))
(ll') [(7X)FX:X]D(E!(7X)FXDFX)

Both (9') and (10') give the definite description substi-
tuted for 'n' narrow scope relative to the modal operator,
and both are true under the interpretation previously given,
as the reader can easily verify. Hence, neither (9') nor
(10') will make (4) false. Now, according to Loar premise
(5) is true because the name in it is normally read as
having wider scope than the modal operator. However, he

goes on to argue that if we also give the definite

54

description substituted for the name in the consequent of
(4) widest possible scope, as we have done with our (11'),
we will see that (4) is false. To see that it is false let
”(4x)Gx” abbreviate 'the man who discovered electricity',
and suppose that '(9X)Fx=(7x)Gx' holds. If we substitute
(7x)Gx for the first occurrence of (7x)Fx in (11'), the

resulting sentence

(l4) [(7x)Gx:x] DUE! (7x)Fx) 9 FX)

is clearly false. Since (11') results in a reading of the
consequent of (4) that makes (4) false Loar's response to
Kripke's argument as stated remains intact, contrary to
the claim made by Hudson and Tye,

Now, it might be objected that my (11') is an ”il-
legal" or "incorrect” translation of the consequent of (4).
This is so because a modal remark about a nonmodal sentence
can only be translated in such a way that the nonmodal sen-
tence has to occur, in tact, within the scope of the modal
operator. Another way of making this point would be to
argue that if we say of a given sentence, call it ¢, that
what ¢ expresses is necessary, then our remark can only be
taken in the de dicto sense and can only be translated as
E]¢. Giving the definite description substituted for 'n'

widest possible scope, as I do in my (11'), violates this

condition.

55

Let me make three points concerning the above objec-
tion in particular, and one final point concerning my tran-
sition of the argument from the formal mode to the material
mode in general. I'll begin by making the more general
point first.

To the best of my knowledge, Kripke's ”modal argu-

ments” in Naming and Necessity are stated by him in the

 

material mode. Hence, it looks like the argument attributed
to him by Loar and revised by Hudson and Tye is taken from
the material mode and put into the formal mode. In his
article, ”A Puzzle About Belief”, however, Kripke states
that "given the arguments of Church and others, I do not
believe that the formal mode of speech is synonymous with
other formulations."l Thus one might well argue that any
objection that would be telling against my transition from
the formal mode to the material mode would be equally tell-
ing against their transition from the material mode to the
formal mode.

My second point is that in the present context what
the above objection amounts to is that a sentence like 'n
is an F' is necessary simpliciter, and can only occur within
the scope of the necessity operator. I would like to sug-
gest, however, that sentences are only necessary or possi-

ble relative to a given reading. When we say of the

 

lSaul Kripke, ”A Puzzle About Belief,” in Meaning
and Use, ed. by A. Margalit (Dordrecht-Holland: D. Reidel
Publishing Company, 1979), p. 272.

56

sentence 'The number of planets is greater than 5.’ that

it is necessary we are claiming that it is true at every
possible world. But to determine its truth value at each
possible world we need to know whether or not the values

of the singular terms in the sentence remain fixed or
change from world to world. In particular we need to know
whether or not the value of 'the number of planets' remains
fixed at nine or if in some world it changes,i.e., we need
to know whether it is '[the number of planets:x][]x>5' or
'E](the number of planets>5)' that is being asserted when
one asserts that 'the number of planets is greater than 5'
is necessary. Basically what I am arguing here is that be-
fore a sentence can be determined to be necessary or possi-
ble a prior question needs to be asked. Namely, 'What are
the values of the singular terms in the sentence under con-
sideration?’ This prior question can only be answered by
giving the sentence a reading with regard to scope. In so
far that the above account is correct I suggest that my
(11') is a legal translation of the consequent of (4).

The third point that I would like to make concern-
ing the above objection is that it is not an objection that
could be raised by Hudson and Tye. This is because to
argue that the consequent of (4) can only be given a_dg
dicto or small scope reading is merely to concede the point
that Kripke's argument cannot be made independent of scope

considerations. But, this is precisely what Hudson and Tye

57

claim is not the case. That is, they argue that their re-
vision of Loar's first premise "makes an appeal to scope
distinctions irrelevantU. Hence, scope considerations
should no longer be an issue. But, it seems to me, to
raise the above objection is tantamount to holding that
scope is a relevant issue.

My final point is that matters are not improved
much if one argues that Hudson and Tye could easily resist
the temptation to raise the above objection, but argue in-
stead that the consequent of (4) simply cannot be translated
into the material mode at all. That is, they could argue
that not only is (11) an illegal translation of the conse-
quent of (4), but so are (10) and (9) as well. It should be
pointed out, however, that to argue in this way would be to
part company with Kripke. Kripke goes on in the footnote
referred to above to say that the formal mode ”can be used
as a rough way to convey the idea of scope”.1 Hence, it
would appear that Kripke would allow (9) or (10) as a legit—
imate translation of the consequent of (4). Again, notice
that this would be tantamount to granting that scope is a
relevant concern, and this is something that Hudson and Tye
take pains to deny. It might still be the case, however,
that Hudson and Tye do not mind parting company with Kripke
at this juncture. If this is so, then it seems to me that

the burden of proof is now on them to show that the

 

lIbid.

Fl

58

following does not hold: 'n is an F' is necessary if and
only if'EJn is an F'is true. It is one thing to agree with
Church that the formal mode of speech is not synonymous with
the material mode, it is quite another, however, to argue as
they would have to,that there appears to be no or verylittle
connection between the modes at all. Thus it appears as
though Kripke's "modal argument” can be dodged by simply re-
quiring that the definite descriptions that are substituted
for proper names in modal contexts be given widest possible
scope. We still, however, need to address what I feel is
the crucial issue concerning this type of response to
Kripke's modal argument. Namely, is the argument that Loar
and McKinsey criticize and the argument that Hudson and Tye
attempt to salvage by revision, the argument that Kripke in

fact gives in Naming and Necessity. I think that it is not,

 

and it is to this issue that I will next turn.

Section Six

 

McKinsey and Loar, and by implication Hudson and
Tye as well, all seem to feel that the modal argument that

Kripke gives in Naming and Necessity can be met by the move

 

to wide scope definite descriptions. The time has now come

to make good on my claim made earlier that the argument they
consider is simply not Kripke's argument. Before doing so,

however, let me first try to indicate why I think it is

that Loar and McKinsey are led astray.

59

As I pointed out in my introduction both seem to
mistakenly believe that the crucial feature to note con-
cerning proper names is that they are normally read as hav-
ing wider scope than the modal operator. They believe this
because they misunderstand Kripke's thesis that proper names
are rigid designators. What Loar and McKinsey fail to real-
ize is that if proper names are rigid designators, then
scope distinctions are simply irrelevant for the modal sen-
tences in which they are contained. That is, both fail to
realize that proper names do not induce semantic scope ambi-
guity in modal contexts.

It might be objected that if we take Loar and
McKinsey to be claiming that natural language has a conven-
tion such that proper names are always to be read as having
widest possible scope in modal contexts, then it is also
true that proper names do not induce scope ambiguity.l
Thus, it would be argued, that they, as well as Kripke, can
account for the fact that proper names are not ambiguous in
modal contexts with regard to scope, and consequently they
have not radically misunderstood his thesis that proper
names are rigid designators. The problem with this sort of

objection, however, as I pointed out in an earlier footnote

 

1Notice that on this account the claim is not that
particular wide scope tokens of proper names are rigid
designators, hence they do not induce scope ambiguity;
rather the claim is that proper names are not rigid desig-
nators, but as they occur in English we are to read them
as always having widest possible scope, hence we never get
two readings.

60

(n.1, p. 29) is that it obscures a crucial feature of
Kripke's account of the rigidity of proper names. On
Kripke's account the claim that proper names do not induce
scope ambiguity in modal contexts only makes sense for a
language where both a wide scope reading of the name and a
narrow scope reading of the name are admissible. Once this
is made clear we can then see that proper names display a
certain feature in modal contexts (i.e., they are not ambi-
guous with regard to scope) that is not displayed by defi-
nite descriptions.

It should also be noted that Kripke would reject the
thesis that proper names are nonrigid designators that are
to be read as having wide scope in modal contexts, since he
does not feel that all of ”our" intuitions regarding proper
names can be accounted for along these lines. What Kripke
has in mind here is what he refers to as his "intuitive
test for rigidity" in terms of counterfactual situations.

As I pointed out in my introduction, Kripke sometimes estab-
lishes his thesis that proper names are rigid designators
by considering the truth conditions, with respect to cer-
tain counterfactual situations, of simple, nonmodal sen-
tences with proper names in the subject position. In these
cases, according to Kripke, the sentences considered will be

such that "there is no room for any scope distinctions".1

 

lKripke, Naming and Necessity, p. 11.

 

61

Hence, our intuitions concerning the rigidity of the names
involved cannot be accounted for along the lines of the
hypothesis considered above in terms of viewing proper names
as wide scope nonrigid designators. More will be said about
this way of looking at rigid designators later in the sec—
tion.

Given then this misunderstanding of Kripke's thesis
regarding proper names as rigid designators, Loar and
McKinsey go on to interpret his argument in the following

manner. Consider the sentence

(1) It is possible that Aristotle was not a
philosopher.

Both Loar and McKinsey note that a sentence like (1) is true
because the name 'Aristotle' is normally read as having
wider scope than the modal operator. Now suppose'Aristotle'
really was a disguised definite description, say 'the
greatest philosopher of antiquity'. In that case it should
be possible to substitute this definite description for the
name in (l) and have (I) retain its same truth value. But

when the substitution is made the sentence becomes false.

What we get is

(2) It is possible that the greatest philosopher
of antiquity was not a philosopher.

62

Since (1) and (2) are not materially equivalent Kripke has
shown, according to Loar and McKinsey, that 'Aristotle'
cannot be the disguised definite description 'the greatest
philosopher of antiquity'.

Now to dodge the argument both note that the reason
why (2) is false is that the definite description in it is
read as having smaller scope than the modal operator. But,
they point out, if we give the definite description substi-
tuted for the name wider scope than the modal operator, as
was implicitly done with the proper name in (1), then the
sentence that results will have the same truth value as (l).

The sentence that now results

(3) The greatest philosopher of antiquity is such
that it is possible that he is not a
philosopher.

is materially equivalent with (l), and hence according to
Loar and McKinsey, Kripke has not been able to show that
proper names are not disguised descriptions.

In an effort to clear up some of the misunder-
standings concerning his views Kripke in the preface to

Naming and Necessity, has the following to say about the

 

above example:

'It might have been the case that Aristotle was not a
philosopher' expresses a truth, though 'It might have
been the case that the greatest philosopher of

 

~ s—u. rc— .— _‘v n'“ - n . . - ...
.- ....Aah. _..~-._....- '___._- _ ‘

63

antiguity was not a philosopher' does not, contrary to
Russell's theory. Now the last quoted sentence would
express a truth if the description used were read, con-
trary to my intent, with wide scope. So perhaps it
might be supposed that the problem simply arises from
an (unaccountable!) tendency to give 'Aristotle' a wide
scope reading while the descriptions are given a small
scope reading; sentences with both names and descrip-
tions, however, would be subject in principle to both
readings. My point, however, was that the contrast
would hold if all the sentences involved are explicitly
construed with small scopes.l

It should be obvious from the above quotation that
the move by Loar and McKinsey to a wide scope definite
description is simply ”wide” of the mark. As Kripke points
out the contrast between (1) and (2) remains even if, con-
trary to Loar and McKinsey, we give the name in (1) narrow
scope relative to the modal operator.

It is important, however, to notice that there is
nothing in Kripke's argument stated in the passage above to
prevent Loar and McKinsey from maintaining that the contrast
will not hold as long as it is only a wide scope definite
description that is substituted for the name in (1). All
that is lost in this case is the symmetry in scope.
Further, they could point out that a particular token of a
definite description which has been given wide scope is
such that it picks out the same thing in every possible
world where it picks out anything at all, just as is the

case for proper names. In this case we could view McKinsey

 

lIbid., p. 13.

64

and Loar as agreeing with Kripke that proper_names are
rigid designators, but also arguing that the particular
token of a wide scope definite description that has been
substituted for the name is a rigid designator as well.
Hence, it looks like proper names can be disguised descrip-
tions after all, since the claim that names have a property
not shared by definite descriptions cannot be upheld.

I am, of course, merely assuming for the sake of
argument that it does in fact make sense to view a partic-
ular wide scope token of a definite description as a rigid
designator. It is not at all clear, however, that a token
is the sort of thing that can be substituted, nor is it
clear that a token is the sort of thing that can be rigid.
In regard to the latter point, if my account of Kripke's
notion of rigidity in section two is correct, then it would
appear that rigidity is a concept that applies to singular
terms (i.e., names and definite descriptions) and not
tokens of singular terms. Gerald Vision in a recent arti-

cle in the Australasian Journal of Philosophy has attempted

 

to make the same point in connection with proper names. In

that article he states:

N is not a name-token, nor is it sim l a name-type: it
is a instance of a (type) name (henceforth, a name-
instance): viz., a type—name as applied to one individ-
ual in our world. Thus, the type-name 'Francis Bacon'
has in our history at least the following two

65

name-instanceszone the name of a philosopher, the
other the name of an artist.1

Unfortunately, Vision does not go on to give us a proper
account of what a name-instance is, but it is possible to
get at least a rough idea of the distinction that he is
groping for. A name-type properly understood is simply a
sign—type. Hence, it is an abstract entity defined, for
the most part, in purely syntactic terms. (The qualif-
ication, for the most part, is meant simply to allow for
the possibility of counting words like 'honor' and 'honour'
as two tokens of the same word-type, or for the possibility
of counting words like 'taxi' and 'taxicab' as two tokens
of the same word-type.) Name-types are recognized by their
form alone; hence it is possible that two tokens of a given
name-type denote different objects. Consider, for example,
the sentence, ”London, England is bigger than London,
Ontario." Here we have two tokens of the same name-type
referring to different cities. Clearly, then the name-type
'London' is not the sort of thing that can be rigid.

Now, this might lead one to hold that it is a par-
ticular token of the name—type 'London' that is rigid, but
this would be a mistake as well. A particular token of a

given name-type need not be a token of a singular term.

 

1Gerald Vision, ”Linsky on Rigid Designation and
Sense,” Australasian Journal of Philosophy, Vol. 58
(September 1980) p. 291.

 

66

Thus, for example, the token of the name-type 'London' that
occurs in the following sentence, ”There is many a London
in the world” is a token of a general term, since it pur-
ports to pick out more than one thing. Clearly the token
of the name-type 'London' that occurs in the sentence men-
tioned above is not the sort of thing that can be rigid
either. Hence, it is not the name-type 'London' that can
be said to be rigid, nor is it the name-token of the type,
rather it is simply the name (i.e., the singular term)
'London' that can be said to be rigid.

But if a name is not to be identified with a type,
and it is not to be identified with a token either, then
with what is it to be identified? More is needed than
Vision's notion of a name-instance. Perhaps a name might
be identified with the ordered pair <N,O> where N is some
name-type and O is some object associated with it in some
way; and each such ordered pair identifies a singular term.
Clearly, more has to be said and some obvious problems
(i.e., problems posed by names that lack bearers) have to
be ironed out before such a suggestion could be accepted,
but I believe that it is enough to give some substance to
Vision's claim.

We need not, however, rule out the possibility of
tokens being rigid in order to uphold Kripke's claim that
proper names have a property (i.e., rigidity) not shared by

definite descriptions, and hence that names cannot be

67

disguised descriptions. That is, even if we assume that it
makes sense to call a token of a definite description rigid,
the contrast between proper names and definite descriptions
can still be drawn in terms of sc0pe. Vision in a footnote
to the passage quoted above makes this point in connection
with his notion of an expression-instance. He states in the

footnote that

Perhaps it is because being a rigid designator is a
property of expression-instances, and not their uses or
tokens, that Kripke seems uneasy about definite descrip-
tions as rigid designators. For although uses of such
descriptions may take wide scope, no description-
instanpe is such that all its (possible) uses take wide
scope.

 

The basic point that Vision makes in the passage quoted
above need not, however, be made in terms of his notion of
an expression—instance. It simply has to be pointed out
that while it may be true that a particular token of a def-
inite description that has wide scope is rigid, the definite
description of which it is a token is not. Thus as we

have already noted some tokens of 'the greatest philosopher
of antiquity' will occur in sentences like (2) and some will
occur in sentences like (3). Thus the definite description
will be such that it is ambiguous in modal contexts, and
hence not be a rigid designator. Proper names, on the other

hand, simply do not display this ambiguity with regard to

 

11bid.

 

68

scope in modal contexts. There simply is no sense in which
Aristotle might not have been Aristotle. There is, however,
a sense in which the greatest philosopher of antiquity might
not have been the greatest philosopher of antiquity.

In a footnote in Naming and Necessity Kripke states

 

the following:

The facts that 'the teacher of Alexander' is capable

of scope distinctions in modal contexts and that it is
not a rigid designator are both illustrated when one
observes that the teacher of Alexander might not have
taught Alexander (and, in such circumstances, would

not have been the teacher of Alexander). On the other
hand, it is not true that Aristotle might not have been
Aristotle

One clear implication of this quotation is that proper names
are rigid designators and definite descriptions are not,

and that while definite descriptions are capable of scope
distinctions in modal contexts, proper names are not.

Notice also that while a particular wide scope occurrence

w)

of 'the teacher of Alexander' may be rigid, this does not
alter the fact that the description which it is an occur—
rence of is nonrigid, and hence capable of scope distinc-
tions in modal contexts.

The astute reader will, of course, observe that
Kripke in the above quotation speaks of twp facts. One fact

being that 'the teacher of Alexander' is capable of scope

 

lKripke, Naming and Necessity, p. 62.

 

69

distinctions and the other being that it is not a rigid
designator. I believe that the reason for this is that
Kripke in the above quotation wants to emphasize his intu-
itive test for rigidity that I referred to in section two.
'The teacher of Alexander' is nonrigid since ”it does make
sense” to say that the teacher of Alexander might have been
someone other than the teacher of Alexander. We can explain
this by noting that 'the teacher of Alexander' is capable
of scope distinctions in modal contexts in the sense that
it is semantically ambiguous with regard to scope, while
Kripke is apparently speaking of syntactical scope, so that
no such syntactical scope difference comes in here between
'the teacher of Alexander might not have been the teacher
of Alexander', and 'Aristotle might not have been
Aristotle'.

The proper name 'Aristotle', on the other hand, is
a rigid designator. It doesn't make any sense to say that
Aristotle might have been someone other than Aristotle. X
This, however, can not be explained by an appeal to scope.
'Aristotle' is not capable of scope distinctions in modal
contexts in the sense that it is semantically ambiguous
with regard to scope. Perhaps the following passage from

Naming and Necessity where the above quoted footnote occurs

 

will help to clarify the issue.

Most of the things commonly attributed to Aristotle
are things that Aristotle might not have done at all.

70

In a situation in which he didn't do them, we would
describe that as a situation in which Aristotle didn't
do them. This is not a distinction of scope, as hap-
pens sometimes in the case of descriptions, where some-
one might say that the man who taught Alexander might
not have taught Alexander; though it could not have
been true that: the man who taught Alexander didn't
teach Alexander. This is Russell's distinction of
scope. It seems to me clear that this is not the case
here. Not only is it true pf the man Aristotle that

he might not have gone into pedagogy; it is also true
that we use the term 'Aristotle' in such a way that,

in thinking of a counterfactual sitation in which
Aristotle didn't go into any of the fields and do any
of the achievements we commonly attribute to him, still
we would say that was a situation in which Aristotle
did not do these things.1

 

Again, Kripke's point here seems to be that we cannot
account for the fact that it is possible that Aristotle
might not have been a philosopher by an appeal to scope
distinctions. 'Aristotle' is simply not capable of scope
distinctions that are semantically relevant. We don't
have a wide scope reading that makes the claim true and a
narrow scope reading of the claim that makes it false; as
would be the case if we were dealing with a definite de-
scription instead of a proper name. We use the term
'Aristotle' in such a way that '[a:x][]x=a' is true, and
what this means is that when we think of Aristotle in some

counterfactual situation it is Aristotle that we are con—

 

sidering.

 

lIbid., pp. 61-62.

71

Section Seven

Now one issue that I have carefully avoided through-
out the entire chapter is whether or not there are some in-
tuitions regarding rigidity and names that can be explained
by Kripke's intuitive account of rigidity that cannot be
explained by the claim that his thesis regarding names is
equivalent to the claim that ”if a modal operator governs
a simple sentence containing a name, the two readings with

”1

large and small scope are equivalent. In the preface to

Naming and Necessity, it is clear that Kripke wants to argue

II II

that at least some of our linguistic intuitions regarding
rigidity cannot be handled by some hypothesis about scope
conventions to the effect that proper names are simply non-
rigid designators that are read as having widest possible
scope. He starts by giving the following account of his

idea of rigid designation.

Consider:
(l) Aristotle was fond of dogs.

A proper understanding of this statement involves an
understanding both of the (extensionally correct) con-
ditions under which it is in fact true, ppd of the
conditions under which a counterfactual course of
history, resembling the actual course in some respects
but not in others, would be correctly (partially) de—
described by (l). Presumably everyone agrees that
there is a certain man—~the philosopher we call
'Aristotle'——such that, as a matter of fact, (1) is
true if and only if pp was fond of dogs. The thesis

 

lIbid., p. 12.

72

of rigid designation is simply——subtle points aside——
that the same paradigm applies to the truth conditions
of (l) as it describes counterfactual situations. That
is, (l) truly describes a counterfactual situation if
and only if the same aforementioned man would have been
fond of dogs, had that situation obtained. By contrast,
Ruisell thinks that (1) should be analyzed as something
1 e:

 

(2) The last great philosopher of antiquity was
fond of dogs,

and that this in turn should be analyzed as

(3) Exactly one person was last among the great
philosophers of antiquity, and any such
person was fond of dogs.

The actual truth conditions of (3) agree extensionally
with those mentioned above for (1), assuming that
Aristotle was the last great philosopher of antiquity.
But counterfactually, Russell's conditions can vary
wildly from those supposed by the rigidity thesis.

With respect to a counterfactual situation where some-
one other than Aristotle would have been the last great
philosopher of antiquity, Russell's criterion would
make that other person's fondness for dogs the relevant
issue for the correctness of (l)!

 

Given the intuition about proper names that underlies the

above account, Kripke wants to make it clear that his doc—

trine of rigidity cannot be explained in terms of scope.

That is, he wants to make it clear that those philosophers,

like Loar and McKinsey, who misinterpret his thesis of rigid

designation by viewing it merely as the doctrine that nat-

ural language has a convention to the effect that names are

to be read as having widest possible scope in modal contexts

cannot account for our intuition concerning the rigidity of

 

lIbid., pp. 6—7.

73

names in the above quoted example. In his effort, however,
to show that some of the linguistic intuitions that he
adduces on behalf of rigidity cannot be explained by any
hypothesis involving scope conventions it appears as though

Kripke simply overstates his case. He states:

(1) and (2) are 'simple' sentences. Neither contains
modal or other operators, so there is no room for any
scope distinctions. No scope convention about more
complex sentences affects the interpretation of these
sentences. Yet the issue of rigidity makes sense as
applied to both. My view is that 'Aristotle' in (l) is
rigid, but 'the last great philosopher of antiquity' in
(2) is not. No hypothesis about sc0pe conventions for
modal contexts expresses this view; it is a doctrine
about the truth conditions, with respect to counter-
factual situations, of (the propositions expressed by)
El; sentences, including simple sentences.

Contrary to Kripke's claim made above it does seem
as though the view that natural language has a convention
that names are read only as having wide scope in modal con-
texts could explain why 'Aristotle' in (l) is rigid (i.e.,

why it is Aristotle that is relevant to the truth condi-

 

tions, with respect to counterfactual situations, regarding
(1)). All that is necessary is that the view be expressed
as a conditional claim. That is, we could explain why
'Aristotle' in (l) is rigid by noting that'tf a modal Oper-
ator preceded (1) then the name 'Aristotle' would be given

widest possible scope, and that is why Aristotle is the

 

llbid., pp. 11-12.

74

value of the name in this world, as well as all possible

worlds. It could then be pointed out that no such scope

convention exists for definite descriptions and that is why

'the last great philosopher of antiquity' in (2) is non—

rigid. It is important to notice, however, that this sort

of explanation as to why 'Aristotle' in (l) is rigid and

'the last great philosopher of antiquity' in (2) is not,

will be of no value to those who want to hold that proper

names are really disguised definite descriptions. The

reason for this is simply that it seems extremely implau- J M1
sible to suppose that names could be abbreviated or dis-
guised descriptions if in natural language there exists a
scope convention regarding the one but not the other. If,
on the other hand, it is claimed that the same scope con—
vention that exists for names, exists for definite de-
scriptions as well, then it seems as though we could no
longer claim, as Kripke does, that, 'the last great phi-
losopher of antiguity' in (2) is nonrigid. This is so
because if the description in (2) is given only wide scope
when (2) is preceded by a modal operator, then the value
of 'the last great philosopher of antiquity' would be
Aristotle in all possible worlds; assuming, of course, that
Aristotle is the value of the description in this world.
Thus it seems that Kripke's view can be expressed as an
hypothesis about scope conventions, but expressing it in

this way will be of no help to the description theorist.

 

75

I would like to again stress the fact, however,
that any view which places a restriction on the scope of
names in natural language will not be the same as Kripke's
view that a name is a rigid designator if and only if the
wide and narrow scope readings of the sentence that con-
tains the name are equivalent. Kripke's View just would
not make any sense for a language where both readings were
not allowed. It is also interesting to note that in a foot-
note concerning the above quoted passage, Kripke also
expresses his thesis that names are rigid designators in
terms of a conditional. That is, he states that his view
is equivalent to the thesis that ”if a modal operator gov-
erns a simple sentence containing a name, the two readings

1 Thus insofar

with large and small scope are equivalent."
as his View, expressed in this way, counts as an "hypoth-
esis about scope” we also can use it to explain why
'Aristotle' in (l) is rigid and 'the last great philosopher
of antiquity' in (2) is not. 'Aristotle' is rigid because
if (1) were preceded by a modal operator, the wide and nar—
row scope readings of the sentence would be equivalent.
'The last great philosopher of antiquity', however, is not

rigid since, if (2) were preceded by a modal operator, the

two readings of the sentence regarding scope would not

 

1Ibid., p. 12. Emphasis added.

76

be equivalent.1

It should also be noted that it is possible to ar—
gue, in a much more direct manner, that Kripke cannot make
good on his claim that all of our intuitions regarding ri-
gidity can be explained without getting involved with the
question of scope. That is, up to now I have only argued,
contrary to what Kripke has claimed, that an "hypothesis
about scope" 23p explain all of the intuitions that he has
adduced on behalf of rigidity. One might also argue, how-
ever, that the account of rigidity that Kripke gives in
terms of the truth conditions of (l) and (2) as they de-
scribe counterfactual situations actually commits him to an
explanation in terms of scope. Thus, one might argue as
follows. Let C be a ”counterfactual condition", and let
'[P” represent the counterfactual conditional. Now,
whether or not (1) is true in C is simply the question of

whether or not

(1') C D (1)

is true in this world, and whether or not (2) is true in C

is simply the question of whether or not

 

1A wide and narrow scope reading of (2) preceded
by a modal operator will, of course, be materially equiv—
alent. The two readings, however, will not be logically
equivalent since the value of the description within the
scope of the modal operator may change from world to
world.

77

(2') C D» (2)

is true in this world. But (1') and (2') are potentially
ambiguous with regard to scope. To see this let us consider
(2'). It has a wide scope reading, viz., 'the greatest
philosopher of antiquity is such that if it were true that
C, then he would have been fond of dogs', and a narrow scope
reading, viz., 'if it were true that C, then the greatest
philosopher of antiquity would have been fond of dogs'. The
same point can be made regarding (1'), only in this case the
syntactical scope ambiguity displayed will not be semanti-
cally relevant as is the case for (2').

Thus it appears that not only can all of the intu—
itions that Kripke adduces on behalf of rigidity be

expressed by some form of hypothesis in terms of scope,

 

namely, his own; but that also his own view may actually
commit him to an explanation in terms of scope. It is for
this reason that I have neglected to consider at length his
"intuitive test" for rigidity as a separate and possibly

more fundamental account of his doctrine.

Section Eight

 

It should be clear from what has been covered so far
that the move to avoid Kripke's argument by employing wide
scope definite descriptions has no chance of success. One

can stipulate that the definite description that is

 

78

substituted for a proper name be given widest possible sc0pe
in modal contexts, but as long as that definite description
is ordinarily thought to be capable of scope distinctions,
the fact remains that it will not have a property that
proper names do. Namely, it will not be a rigid designator.
That is, it will not be such that scope distinctions are
irrelevant, apart, of course, from ad hoc stipulation.
Kripke's modal argument would, of course, be met if
the definite descriptions that one substituted for proper
names were such that they did not induce scope ambiguity in
modal contexts. We have already noted what such a definite
description would be like. Definite descriptions like 'the
smallest prime number' and 'the darkest color' are rigid
designators, and as such would fit the bill if, of course,
they were the sort of descriptions that could be substi-
tuted for prOper names. One feature that these descriptions
all seemed to have in common, however, was that they took
as their values abstract objects. We also noted that they
seemed to be rigid designators in virtue of the fact that
certain sorts of relationships held between their values
and other members of the respective kinds to which they
belonged. Thus, while there does seem to be some definite
descriptions in natural language that are rigid designators,
they do not seem to be the sort of descriptions that will
be of much value for those who wish to save a description

view of proper names and avoid Kripke's argument.

79

This is not to say, however, that such a definite
description could not be invented. Linsky notes in Names

and Descriptions (p.54) that Quine's proposal for elimi-

 

nating names in favor of predicates provides us with just
such a description. According to Quine any proper name can
be eliminated in favor of a predicate in a regimented lan-
guage. Thus, for example, the proper name 'Socrates' could
be replaced by the predicate 'socratizes'. The definite
description that would result is (7x)(x Socratizes), and it
is such that it is not ambiguous in modal contexts. Con—

sider the following sentences:

(1) «0(7x)(x Socratizes) ¢ (7x)(x Socratizes)

(2) [(7y)(y Socratizes):x]€>x # (7z)(z Socratizes)

Both (1) and (2) are false. This is so because as Linsky
correctly points out ”if 'Socrates', as a name, is a rigid
designator so is the predicate 'socratizes' which has with-
in its extension in each possible world (in which it has
anything within its extension) just that individual who is

”1 Thus Kripke's modal argument can be met by

Socrates.
requiring that the definite descriptions that are substi-
tuted for proper names be of this special Quinean sort.

It is interesting to note that in a footnote in

Naming and Necessity (p. 29, n.5) Kripke, himself, suggests

 

 

1Linsky, Names and Descriptions, p. 54.

 

80

the basis for this strategem. In that same footnote, how-
ever, he claims that all of the problems he poses for the
Frege-Russell description view of proper names (which I take
to include his modal argument as well) will apply mutatis
mutandis to a view that makes use of the special sort of
Quinean definite description we have just considered.
Kripke's claim here is simply too strong. Mppt of the prob-
lems he raises for the Frege-Russell view will apply to a
reformed language view as well, but not his modal argument.

There is another sort of argument employed by Kripke
to show that a description view of names is false that also
would seem to be blunted by the adoption of the special sort
of Quinean definite description that we have been consid-

ering. In Namingyand Necessity Kripke gives some cases that

 

are designed to show that a description view of names cannot
provide us with necessary conditions for determining the
referent of a name. One such case is his Godel-Schmidt ex-
ample. Assuming that the only description we associate with
'Godel' is 'the man who discovered the incompleteness of

arithmetic' we are to imagine the following scenario:

Suppose that Godel was not in fact the author of this
theorem. A man named 'Schmidt', whose body was found
in Vienna under mysterious circumstances many years ago,
actually did the work in question. His friend Godel
somehow got hold of the manuscript and it was there-
after attributed to G6del.1

 

lKripke, Naming and Necessity, pp. 83-84.

 

81

Given the situation above, Kripke goes on to conclude,

On the view in question (the description view) then,
when our ordinary man uses the name 'G6del', he really
means to refer to Schmidt, because Schmidt is the unique
person satisfying the description, 'the man who discov-
ered the incompleteness of arithmetic'. ...So, since
the man who discovered the incompleteness of arith-,
metic is in fact Schmidt, we, when we talk about Godel',
are in fact always referring to Schmidt. But it seems
to me that we are not. We simply are not.1

Kripke in the above example is, of course, assuming that
'the man who discovered the incompleteness of arithmetic' is
a nonrigid definite description. The important point to
notice, however, is that the situation described above is
not improved one iota by claiming that it is a rigid defi—
nite description that is used to determine the referent of
the name 'Godel'. Even if this were so the referent of the
description in question would still be Schmidt, since a
rigid definite description still determines its referent via
its unique satisfaction of the associated property. This
would not be the case, however, if the rigid description
used to determine the referent of the name were of the spe-
cial Quinean sort that we have been considering. '(7x)(x
G6delizes)‘ would correctly pick out Godel in this world as
well as all possible worlds, and the intuition that Kripke
appeals to in his case is preserved.

Overall, however, I do not think that an appeal to

 

llbid., p. 84.

82

the sort of Quinean descriptions we have considered will
prove to be satisfactory for the description theorist. For
one thing, it seems totxivializetheir position and for an-
other as Kripke notes ”the question, 'How is the reference
of 'Socrates' determined?‘ yields to the question, 'How is
the extension of 'x Socratizes' determined'”1
Before I leave this section I would like to make two
last points concerning rigidity and the possibility of de-
feating Kripke's modal argument by use of rigid definite
descriptions. First, if descriptions like 'the greatest
philosopher of antiquity' are ambiguous in natural language
between a rigid and nonrigid sense, and it is the rigid
sense of the definite description that we are dealing with,
then Kripke will simply claim that his principle thesis is
not affected. This is so because, as he states, his prin-
cipal thesis "contrasts names with nonrigid descriptions,
as advocated by Russell"2
Second, it should also be pointed out that even if
definite descriptions are ambiguous in natural language be-
tween a rigid and nonrigid sense, this ambiguity could not
be used to replace Russell's idea of scope. The reason for
this is simply that definite descriptions are such that they

may occur in modal sentences where they have neither widest

possible scope nor narrowest possible scope. That is,

 

lIbid., p. 29.
2Ibid., p. 6.

83

definite descriptions may have intermediate scopes in some
modal sentences in which they occur. Kripke has made this
point in a number of places (see, for example, "Identity

and Necessity", n.10, p. 149, and "Speaker's Reference and
Semantic Reference,” p. 259) by use of the following exam-

ple. Consider the sentence,

(C) The number of planets might have been
necessarily even.

The only sense in which (C) could be true is when the defi-
nite description in it is given intermediate scope. If 'the
number of planets' is given widest possible scope, (C) is
false, and if it is given narrowest possible scope, (C) is

false as well. Consider the following translations of (C):

(D) O[(7x)Nx:x]DEx
(E) OUE(7X)NX
(F) [(7X)NX:XJOE1EX

(D) is the reading that gives the definite description
intermediate scope and it is the reading that may very well
'make (C) true. This is because the value of the descrip-
tion might be eight at some possible world and it is nec-
essary that eight is even. (E) gives the definite descrip-
tion narrowest possible scope and it is false since the

'value of the description might be three at some possible

84

world in which case it is not possible that it be necessary
that the number of planets be even. (F) gives the definite
description widest possible scope and it is false as well.
In this case the value of the description is nine in all
possible worlds and it is such that it could not possibly
be necessarily even.

The important point to notice here is that if we try
to replace Russell's notion of scope with a simple rigid
(i.e., wide scope reading as in (F)), nonrigid (i.e., narrow
scope reading as in (E)) dichotomy, then we cannot use the
distinction to get the reading which makes (C) true (i.e.,

(D)). In his article in the Midwest Studies in Philosophy

 

Kripke states that ”no twofold distinction can do the job"
of replacing Russell's idea of scope distinctions. He
clearly intends that this claim hold for Donnellan's
referential-attributive as well, but the example he uses to
make his case is far less convincing than the one given
above. I won't argue for this here, but merely leave it to
the reader to form his own opinion. The point I would like
to make, however, is that insofar that it is possible to
view Donnellan's distinction as capable of replacing
Russell's notion of scope, we have good reason for not in-
dentifying referential descriptions with rigid descriptions
and attributive description with nonrigid descriptions.
This is because it is clear that the rigid-nonrigid dichot-

omy cannot replace Russell's idea of scope.

CHAPTER TWO

CRITICISMS OF THE MODAL ARGUMENT

Section One

 

In his book, Frege: Philosophy of Language, Michael
Dummett makes the following claim concerning the question of
possible scope ambiguity for proper names and definite de-

scriptions in modal contexts.

the question may be resolved for all contexts by
adopting some uniform convention determining the scope;
thus Frege may be represented as having adopted the
convention that the scope of a proper name or a defi-
nite description is always to be taken as the widest
possible, i.e., the whole sentence.1

I believe that it has been this remark or others similar to
it that have led some philosophers, like Hudson and Tye, for
example, to hold that Dummett's approach to Kripke's argu-
ment is essentially the same as that taken by Loar and
McKinsey. That is, that the way to avoid the argument is

to move to wide scope definite descriptions. The quotation

from Dummett referred to above not withstanding, I think

 

1Michael Dummett, Frege: Philosophy of Languagp
(New York: Harper and Row, 1973), p. 115.

85

86

that to view his argument against Kripke along these lines
would be a mistake. It would be a mistake not because that
interpretation is insufficiently supported by the text.
Unfortunately, Dummett's lack of clarity in discussing the
issue seems to allow for such an interpretation. Rather,
as I hope will become clear in the pages to follow, it
would be a mistake because Dummett can also be seen as
attacking Kripke head on. That is, Dummett is best viewed
as denying the thesis that proper names are rigid desig-
nators, since he attempts to show that at least some proper
names are ambiguous with regard to scope in modal contexts.
Taken in this way Dummett can be seen as denying Kripke's
claim that proper names and definite descriptions are ppt
on a par in this respect.

One of the issues that Dummett focuses his atten- ‘”r\

tion on in his appendix to chapter 5 of Frege: Philosophy

 

of Language is Kripke's distinction between using a definite

 

description to ”fix the reference" of a proper name verses
using one to "give its meaning” (section 2, pp. 111-135).
According to Kripke whenever a definite description 'I9x)¢x1
is used to introduce a name a and '77x)¢x' is used to fix
the reference of a and not give its meaning, a so introduced
will be a rigid designator. Now, supposedly one consequence
of introducing a in this way will be that we can know 3
priori that rba=(')x)¢x1 (this is so because the reference of

a is fixed by r-(Oxmx1 even though rb>a¢(7x)¢x1 is true

87

(this is so because r_(’)x)cbx1 does not ”fix the meaning" of
a; while the value of a remains fixed from world to world,
the value of F17x)¢il may change at each world). Since a
is a rigid designator and '—(7x)¢>x1 is not F =(7x)¢x7 would
not be a necessary truth.

Possibly one of Kripke's clearest examples of having
a name introduced by a reference fixing definite description

is given in a footnote in Naming and Necessity. There he

 

states:

An even better case of determining the reference of a
name by description, as opposed to ostension, is the
discovery of the planet Neptune. Neptune was hypoth-
esized as the planet which caused such and such dis-
crepancies in the orbits of certain other planets.

If Leverrier indeed gave the name ”Neptune” to the
planet before it was ever seen, then he fixed the ref—
erence of ”Neptune” by means of the description men—
tioned. At the time he was unable to see the planet
even through a telescope. At this state, an a priori
material equivalence held between the statements
”Neptune exists” and ”Some one planet perturbing the
orbit of such and such other planets exists in such
and such a position," and also such statements as ”if
such and such perturbations are caused by a planet,
they are caused by Neptune” had the status of a priori
truths. Nevertheless they were not necessary truths,
since ”Neptune” was introduced as a name rigidly des—
ignating a certain planet. Leverrier could well have
believed that if Neptune had been knocked off of its
course one million years earlier, it would cause no
such perturbations and even that some other object
might have caused the perturbations in its place.1

Some philosophers appear to take issue with Kripke,

concerning the passage just quoted, over whether or not

 

lKripke, Naming and Necessity, n. 33, p. 79.

 

'8“)

88

'Neptune' was in fact introduced by Leverrier as a rigid
designator as opposed to a mere abbreviation for some def-
inite description. In this case it would be argued that
the definite description 'the planet that caused the per-
turbations in the orbit of Uranus', as uSed by Leverrier

to introduce the name 'Neptune', fixed its meaning. Keith
Donnellan in his article 'The Contingent A Priori and Rigid
Designators' attributes such a position to Dummett, refer-
ring to the section in Dummett's book previously noted. In
that article Donnellan rather halfheartedly attempts to
justify the concern over the factual question as to how
'Neptune' was introduced, which he attributes to Dummett,
by noting that Kripke, in the last two sentences in the
above quotation, "may seem to be giving an argument" to the
effect that Neptune was in fact introduced as a rigid des-
ignator and not a mere abbreviation. According to

Donnellan the argument goes as follows:

Take the modal sentence,

(A) Neptune might have existed and not been the
cause of the perturbations in the orbit of
Uranus.

Following Kripke, it seems that having just introduced
the name yip the description contained in (A),
Leverrier might nevertheless believe without any incon-
sistency what (A) expresses. But that seems to show
that the following sentence expresses a contingent
truth:

(B) If Neptune exists, Neptune is the cause of the
perturbations in the orbit of Uranus.

89

But if "Neptune” were a mere abbreviation for the
description in question then (B) would be equivalent
by substitution of the description for the name, to a
mere tautology. Thus, or so it might seem, we can
show that Neptune was not introduced as an abbrevia-
tion.

Now, Donnellan takes the unstated conclusion of this
argument to be that since 'Neptune' was not introduced as a

mere abbreviation, it was introduced as a rigid designator.

 

But, why should we take this to be the conclusion of
Kripke's argument? Especially since as it stands it is
fallacious. The argument foisted on Kripke assumes that
only two alternatives are possible. Either a proper name,
so introduced, is introduced as a rigid designator or it is
introduced as a mere abbreviation, but not both. But why,
without further argument to the contrary, can't we suppose
that the name was introduced as neither or that some as of
yet unstated third possibility exists?

If, however, the argument isn't intended by Kripke
as an argument to show that 'Neptune' was introduced as a
rigid designator, then what is it an argument for? Notice
that one could take Donnellan's reading of Kripke to be an
argument to show that 'Neptune' doesn't mean 'the planet
that caused the perturbations in the orbit of Uranus'. In

this case we would simply have a particular instance of the

 

lKeith Donnellan, "The Contingent A Priori and
Rigid Designators,” Midwest Studies in Philosophy, Vol. II:
Studies in the Philosophy of Language (February l977),p. l4.

 

90

general form of argument that Loar attributed to Kripke.

The argument would be as follows:

Mn(9x)Fx -> 1:] (EanD Fn)
~D(Eln '3 Fn)
f.e’Mn(7x)Fx

We have already seen, however, that there is good reason
for not taking this to be Kripke's argument. What I sug-
gest is that we take Kripke's remarks in the section noted
by Donnellan, at their face value. That is, that we simply
take them as an argument to show that sentences like 'If
Neptune exists, then Neptune is the planet that caused the
perturbations in the orbit of Uranus' are not necessary
truths. The reason for this is that 'Neptune' is a rigid
designator and the definite description is not. Leverrier
might well have believed that Neptune might not have been
the cause of the perturbations in the orbit of Uranus, since
that belief could have been true. The belief could have
been true because 'Neptune' rigidly designates the very
same planet, while the nonrigid definite description might
have an object other than Neptune as its value.

The confusion that arises over Kripke's Neptune
example results from the fact that Dummett, and to some ex-
tent Donnellan, take Kripke to be giving us a case where a

name is made rigid by introducing it via a reference fixing

91

definite description. Rather, what Kripke is actually
doing is showing us, that a proper name which is a rigid
designator because all proper names are rigid designators,
can be introduced into the language one of two ways. Either
by ostension or by fixing the referent with the definite
description. Kripke is not arguing that 'Neptune' was in-
troduced by Leverrier as a rigid designator. Rather he is
arguing that a rigid designator, 'Neptune', can have its
referent determined by a reference fixing description.

It is instructive to consider Donnellan's reply to
the argument, as originally set out by him, since it is a
response that he attributes to Dummett. According to
Donnellan the argument can be met simply by pointing to
scope differences in (A). That (A) is ambiguous with re—
gard to scope can be readily seen, it is argued, when we
consider the sentences that result when we substitute for
'Neptune' the definite description that is claimed to be
definitionally equivalent to it. The narrow scope reading

that results,

(D) It might have been the case that [the cause
of the perturbations in the orbit of Uranus
did not cause the perturbations in the orbit
of Uranus].1

is plainly false, and hence not a belief that we could

 

lIbid.

92

charitably attribute to Leverrier. The wide scope reading

that results,

(E) The cause of the perturbations in the orbit
of Uranus might have been such that it did
not cause the perturbations in the orbit of
Uranus.1

however, is true and might well have been believed by

Leverrier. Given (D) and (E) Donnellan then goes on to

point out,

That Leverrier might well have consistently believed
what is expressed by (A) does not then show that
”Neptune” was not introduced by him as an abbreviation
for the description nor that (B) is contingently true.
For on the hypothesis that it was introduced as an
abbreviation (E) expresses one of the propositions (A)
could express and (E) is something Leverrier might well
have believed.2

It should be obvious by now that the above is simply
another instance of the basic move by McKinsey and Loar.
The sense in which (A) is true is the sense in which the
definite description substituted for the name in (A) is
given widest possible scope. I hope that it is equally
clear, at this point, that this sort of rejoinder to Kripke
does not refute his claim that proper names are not dis-

guised descriptions; even when the description is used to

 

 

lIbid.

21bid., p. 15.

93

introduce the name. Kripke makes this abundantly clear in

a footnote concerning Donnellan's article. He states:

. Donnellan asks whether I think proper names (in
natural language) are alwa s rigid: obviously, he
thinks, proper names couId he introduced to abbreviate
nonrigid definite descriptions. My view is, that
proper names p33 always rigid. In particular this
applies to "Neptune”. It would be logically possible
to have single words that abbreviated nonrigid defi-
nite descriptions, but these would not be names. The
point is not merely terminological: I mean that such
abbreviated nonrigid definite descriptions would
differ in an important semantical feature from (what we
call) typical proper names in our ordinary speech.1

I believe that Kripke's claim that nonrigid definite
descriptions ”differ in an important semantical feature"
from proper names is not only consistent with the position
that I have attributed to him, but supports my claim that
it is in fact his position as well. I take his point here
simply to be that proper names do not induce scope ambigu-
ity in modal contexts and nonrigid definite descriptions
do. And that it is this aforementioned "semantical feature”
that would prevent 'Neptune' from being a proper name if it
were introduced to abbreviate some nonrigid definite de-
scription. Kripke also makes this same point in his pref-

ace to Naming and Necessity. He states:

 

 

1Kripke, ”Speaker's Reference and Semantic Refer—
ence," n. 9, p. 272.

94

It also became clear that a symbol of any actual or
hypothetical language that is npt a rigid designator
is so unlike the names of ordinary language that it
ought not to be called a 'name'. In particular, this
would apply to a hypothetical abbreviation of a non-
rigid definite description.1

The basic premise that Donnellan wants to establish
is that there are no theoretical reasons that would block
the introduction of a proper name as a rigid designator via
a definite description as long as it was explicitly stipu-
lated that the name was being introduced as such. That is,
he is primarily concerned to show that the concept of intro-
ducing a name as a rigid designator via a reference fixing
definite description is a coherent one. The only lesson to
be drawn from what he takes to be Dummett's position is
that ”in the absence of stipulation that the name shall be
one or the other, it would be indeterminate whether a name
introduced by means of a description is a rigid designator
or an abbreviation, so long as the name continues to be

2 What Dummett supposedly

pegged to the description.”
shows us, concerning Kripke's 'Neptune' example, is that an
"argument" from modal beliefs to show that 'Neptune' was
introduced by Leverrier as a rigid designator can always be
.

evaded, and it is equally plausible to suppose that'Neptune

was introduced as a mere abbreviation. Thus on the latter

 

lKripke, Naming and Necessity, n. 5, p. 5.

2Donnellan, ”The Contingent A Priori and Rigid
Designators,” p. 16.

 

95

assumption, according to Donnellan, (A) will have two non-
equivalent readings as evidenced by (D) and (E), but on the
assumption that 'Neptune' was introduced as a rigid desig-
nator no such ambiguity of scope results. If 'Neptune' was

in fact introduced as a rigid designator, then

(1) On7‘(9x)Px

is equivalent to

(2) [n:X] ()Xr(7x)Px.

Now, Donnellan wrongly takes the heart of Dummett's attack
on Kripke to be no more than what has been set out above.
What he fails to realize, however, is that in the very same
section of Dummett's book that he refers to, Dummett offers
a much more direct attack of Kripke's position. In partic-
ular, Dummett may also be viewed as arguing that even if we
assume, a la Kripke, that the description 'the planet that
caused the perturbations in the orbit of Uranus' only fixes
the referent of 'Neptune' and not its meaning, it is still
the case that while (2) is true, (1) is false.

Donnellan goes wrong because he assumes that
Dummett is merely arguing that names like 'Neptune' won't
be rigid designators if the definite descriptions that

introduce them are abbreviations of the names and fix their

96

meanings. I grant that viewing Dummett's argument in this
manner gives us the same result (i.e.,"Neptune' will not
be a rigid designator), but it also requires some argument
to the effect that Kripke cannot produce a case where the
introducing description is only a reference fixer and not a
meaning fixer as well. Nowhere in his book, however, do we
find Dummett offering such an argument. Rather, what we
find is Dummett meeting Kripke head on. He would accept
Kripke's claim regarding the introduction of 'Neptune'
(i.e., that the introducing description only fixes its ref-
erence and not its meaning), but still argue that the name
is semantically ambiguous with regard to scope in modal

contexts, and hence is not a rigid designator.

Section Two

 

Dummett begins his attack on Kripke by laying out ,//
what he takes to be Kripke's argument to show that a proper
name cannot be a disguised definite description when the
description in question fixes the referent of the name. He
feels that among personal proper names 'St. Anne' would be
a good candidate to bear out Kripke's claim. This is be-
cause assuming that all we really know about her is that
she was the mother of Mary it is natural to take the ref-
erence of 'St. Anne' as fixed by the description 'the
mother of Mary'. Thus according to Dummett, Kripke's ar—

gument is as follows:

‘4‘ ...-na- ‘s‘l ‘- ..

97

It is evidently true, for example, to say, 'The

mother of Mary was necessarily a parent', at least
where this is understood as meaning, 'It is necessarily
true that, if there was such a person as Mary, and
there was one and only one woman who was her mother,
then that woman was a parent'. But it is not so evi-
dent that it would be true to say, 'St. Anne was neces—
arily a parent', meaning thereby, 'It is necessarily
true that, if there was such a woman as St. Anne, then

she was a parent'. For surely we can truly say, 'St.
Anne might have died in infancy' or 'St. Anne might
have remained a virgin all her life'. It appears to

follow that 'St. Anne' and 'the mother of Mary‘ cannot
be synonymous.1

Now, if Donnellan's characterization of Dummett's
position was essentially correct, we would expect to find
Dummett pointing out that the sentence, 'The mother of Mary
was necessarily a parent' is false when the description in
it is read as having widest possible scope. This, however,
is not what we find. (Nor do we find Dummett explicitly

arguing that 'the mother of Mary' fixes the meaning of 'St.

Anne'.) Rather what Dummett goes on to argue is that there

ip p sense in which the sentence, 'St. Anne was necessarily
a parent' is true. Dummett acknowledges the fact that 'the
mother of Mary' is ambiguous with regard to scope in modal
contexts, but holds that the same is true of 'St. Anne'.

What he wants to show with his St. Anne example is that

no difference between proper names and definite
descriptions appears at all. In both cases, there

 

1Dummett, Frege: Philosophy of Language, p. 112.

 

s...

.—- 9%....

98

seems to be an ambiguity in modal sentences containing

them: the very same ambiguity in both cases, which ac-

cordingly cannot be used to differentiate the two types
of expression.1

In order to make his case Dummett admits that there
is a sense in which it is correct to say 'St. Anne might
never have become a parent'. In this case 'St. Anne' would
be given wide scope relative to the modal operator as

follows:

[a:x]O"Px.

But he also holds that there is ”an equally clear sense in
which we may rightly say, 'St. Anne cannot but have been a
parent' provided always that this is understood as meaning
that if there was such a woman as St. Anne, then she can

"2

only have been a parent. Hence, Dummett's chief concern

is to show that the following sentence is true:

(3) If there was such a woman as St. Anne, then
she can only have been a parent.

Unfortunately, (3) is ambiguous. Dummett could be claiming

either that

 

1Ibid., p. 113.
Ibid.

99

(4) Ela D DPa

is true or that

(5) E] (Ela ‘3 Pa)

is true. Clearly, however, the principle of charity dic-
tates that we take (5) and not (4) as the intended reading
of (3). The idiom, 'If ... then it must be (or can only

have been) the case that -—— is ambiguous in English, and

very often what is intended by a speaker to be an uncon-
tested truth can be interpreted in such a way that what he

says is false. Consider the following sentence:

(6) If the Tigers win every game, then they must
have a perfect season.

(6) could be interpreted as either

(7) P DL'JQ

or

(8) [3(P 9Q).

(7) says that if the Tigers win every game in the actual

world, then there is no possible world where they do not

 

 

 

have a perfect season. Clearly, (7) is false since we could

assume that the antecedent is true and still imagine a
possible world where the Tigers lose one or even all of
their games. Similarly, (4) is clearly false. It says that
if St. Anne exists in the actual world, then there is no
possible world where she is not a parent. But certainly we
could assume that the antecedent of (4) is true and still
hold that in some possible world St. Anne remained child—
less all her life. (8), on the other hand, is the reading
of (6) that makes it true. (8) says that there is no world
where the Tigers win every game and do not have a perfect
season. Clearly (8) is the more charitable reading of (6),
and similarly (5) would seem to be the more charitable read-
ing of (3). But while it is clear that (6) read as (8) is
true, it is not at all clear that (3) read as (5) is true.
It simply is not at all obvious that there is no world where
St. Anne exists and she is not a parent. (3) read as (5)

is doubtful at best, unless we have some account to help
guide our intuitions regarding its truth. Dummett realizes
this and suggests that Kripke, himself, acknowledges a sense
in which (3) interpreted as (5) is true; only not in connec-
tion with proper names. Concerning Kripke's metre rod

example Dummett states:

He [Kripke] comments on Wittgenstein's example of the
standard metre rod in Paris, and insists, as against
Wittgenstein, that it is perfectly proper to ascribe

 

 

 

 

101

to that rod the property of being 1 metre long, on the
ground that we can truly say of it that it might not
have been 1 metre long. But, in arguing this, he also
grants that, in another way, it is a priori true that
the standard metre isl metre long. Of c0urse, this
case is not one to do with what is ordinarily called a
proper name: but Kripke wishes to apply, his distinc-
tion between 'fixing the reference' and 'giving the
meaning' to this case also, holding that taking the
metre rod as the standard is a way of fixing the refer-
ence, but not of giving the meaning, of the word 'metre.
Hence the concession must be taken to apply to proper
names in the more usual sense, at least whenever there
is something specific which may be taken as fixing the
reference.1

What Dummett wants to establish here is that Kripke merely
replaces necessity de dicto with a priori truth. Quine in a
his review of "Identity and Necessity" also reads Kripke as
making the same move. He states, "Kripke sums up matters in

other words: genuine names he calls rigid designators, ne—

 

cessity de re he calls metpphysical necessity, and necessity
2

 

de dicto he calls a priori truths.” Now, what Dummett

 

apparently wants to argue is that insofar as it is a priori
true that St. Anne is a parent, and given that a priori

truth is just de dicto necessity, (3) read as (5) is true.

Thus 'St. Anne' is ambiguous with regard to scope in modal
contexts, and hence there is no reason in principle why 'St. ,5

Anne' cannot be the disguised definite description 'the

 

lIbid.

2Willard Van Orman Quine, review of Identity and
Individuation, ed. by M.Munitz (New York: New York Univer-
sity Press, 1971) in The Journal of PhiloSophy, LXIX (1972),
p.493.

 

 

 

 

 

102

mother of Mary'. Dummett sums up his position as follows:

.. we have one and the same phenomenon occurring for
both proper names and definite descriptions. There
cannot, therefore, be any argument from this fact alone
to the conclusion that a proper name can never be equiv—
alent to a definite description. In the case of def-
inite descriptions, Kripke explains the phenomenon in
terms of the notion of scope. For proper names, on the
other hand, he considers the notion of scope inappli-
cable, and therefore invokes a distinction between two
kinds of possibility. The argument for saying, in this
case, that there are two kinds of possibility seems no
stronger than it would be in the case of definite de-
'scriptions. When we say that the mother of Mary can
only have been a parent, in the sense in which it is
true to say this are we not expressing a priori knowl-
edge, based solely on our understanding of the words,
precisely similar to that expressed by saying that the
standard metre rod can only be 1 metre long? When, on
the other hand, we say that the mother of Mary might
not have been a parent, are we not concerned with the
very same kind of metaphysical necessity involved in
saying that St. Anne might not have been a parent? To
explain the ambiguity, in the definite description case,
in terms of uncertainity of scope, however, requires
that the modal operator be taken as unambiguous: if its
sense shifted, should we not need also to suppose that
its scope altered, as we pass from one interpretation
of the sentence to the other. Quite plainly, these
considerations, as far from providing grounds against
the assimilation of proper names to definite descrip—
tions, supply substantial evidence in its favor.

Dummett, however, fails to substantially support the two
crucial claims that he needs to make his case. Namely,
that it is a priori true that St. Anne is a parent, and
that Kripke merely trades ambiguity of scope regarding def-
inite descriptions for ambiguity of modality (i.e., episte-

mic vs. metaphysical necessity) regarding proper names.

 

1Dummett, Frege: Philosophy of Language, p. 116.

 

103

It seems to me that both claims are false. Notice, though,
that if it can be shown that it is not an a priOri truth
that St. Anne is a parent, that alone will be sufficient to
show that (5) is false.

At first glance one might suppose that Dummett is
arguing that we know a priori that St. Anne is a parent
simply because 'St. Anne' mgppp 'the mother of Mary'. In
the passage quoted above he does seem to indicate that this
is his position when he says that our a priori knowledge is
”based solely on our understanding of the words". For
Dummett, however, the phrase "our understanding of the

words” refers to their use as reference fixers and not the

 

meanings thay have. This is made clear when he states con-
cerning his St. Anne case and Kripke's metre rod example

that:

our knowledge is genuinely a priori knowledge,

given in advance of any particular observations or ex-
perience relating to the subject-matter of the sen-
tence. It is knowledge derived solely from a grasp of
the way in which the words are used, i.e., from the
fact that 'the length of the standard metre rod' is
used to fix the reference of '1 metre' and 'the mother
of Mary' to fix the reference of 'St. Anne'.1

 

 

 

 

Section Three

 

Before I go on to give my criticism of Dummett's

claim that (3) read as (5) is true I would like to briefly

 

1Ibid. Emphasis added.

 

104

consider a slightly different account of and response to

his St. Anne case. Leonard Linsky in his book, Names and

 

DescriptiOns, also views Dummett as meeting Kripke's modal

 

thesis head on (i.e., as arguing that names like 'St. Anne
are not rigid designators). .But Linsky, like Donnellan,
takes Dummett to be giving a much different argument than
the one I have attributed to him. According to Linsky,
Dummett has produced a case where the definite description

in question fixes the meaning of the name, since it is a

 

case where the description exhausts all we know about the
bearer of the name. Thus for Linsky, 'the mother of Mary'
is "as good a candidate as we can find for a meaning fixer
of the name”.1
Linsky, uses the name 'Homer' to construct a case
parallel to Dummett's St. Anne case. If suCh a person as
Homer existed at all, then we are to suppose that all we
know of him is that he was the author of the liipd and the
Odyssey. In this case the definite description 'the
author of the liin and the Odyssey' is to be taken as

fixing the meaning of 'Homer'. But if this is done, then

(9) DAh

is true, while

 

1Linsky, Names and Descriptions, p. 57.

 

 

.m—u-~ . . 1. ~ ..=.~.q...__*_’=

105

(10) [hzx][_'_'[Ax

 

is false. (10) is certainly incompatible with
[th]O "AX

which is clearly true. Kripke, of course, would probably
deny the assumption that a ppmp could have its meaning fixed
by some definite description. He would, however, grant that
the pogo 'Homer', in the case described above, could have ,)
its meaning fixed by some definite description. But he
would then argue, as we have already seen, (cf. p. 93) that
the word 'Homer' is simply an abbreviated nonrigid definite
description and not a proper name. Linsky seems to totally
disregard this as a possibility and merely assumes that
'Homer' and 'St. Anne' would count as proper names for
Kripke even if their meaning was fixed by 'the author of
the llipd and the Odyssey' and 'the mother of Mary'
respectively.
Now, Linsky correctly maintains that if we count
sentences such as (5) and (9) as true, then Kripke's claim '/4
that all proper names are rigid designators is false. (For '
purposes of explication I suggest that we do not contest
Linsky's implicit assumption that the singular terms occur-

H

ring in (5) and (9) are what we would normally call names”,

even when they are taken as abbreviated nonrigid definite

106

descriptions.) He does not, however, want to give up
Kripke's thesis that proper names are rigid designators.
Yet, he also seems to hold that his Homer case and Dummett's 9)
St. Anne case pose a genuine threat. That is, he also seems
to hold that without any argument to the contrary, our in-
tuitions are such that we would count (5) and (9) as true.
In order to find a solution to this quandry Linsky turns to
Kripke's metre rod example.

According to Linsky, given Kripke's discussion of
the metre rod example, one might plausibly suppose that it ,3
is de dicto necessary that the length of the standard metre
rod in Paris is 1 metre long. But in this case Linsky main-
tains that the purported de dicto necessity can be chal—
lenged by Kripke's claim that it is only a priori true
that the standard metre rod in Paris is 1 metre long and it

is not in any sense necessary (cf. Names and Descriptions,

 

pp. 62 and 66). Hence, Kripke can handle what might seem

to be a de dicto/de re ambiguity regarding the standard

 

metre rod case by being able to show that the ambiguity in F
question is really between a priOri truth and metaphysical
necessity. The sentence, 'The standard,metre rod in Paris

is 1 metre long' is both contingent and a priori true.

This is because the phrase '1 metre' was introduced into the 1
language by stipulating that its reference be fixed by the
description 'the length of the standard metre rod in Paris'.

Kripke supposedly can explain how contingent a priori truth

 

‘Lp" *"nv—"P ‘ a‘m _ "1 v , _ ' . r I

107

is possible in this case simply by pointing out that a
"stipulative definition" was used to fix the reference of
the term introduced into the language. The phrase '1 metre'
so introduced rigidly designates a certain length in all
possible worlds, while the value of the description 'the
length of the standard metre bar in Paris' may vary from
world to world. Hence, we are dealing with a contingent
truth since 1 metre might not be the length of the standard
metre bar in Paris in some possible world. Yet, we know p
priori what length the standard metre bar has, viz., 1
metre, since it was stipulated that '1 metre' would have '}
its reference fixed by the description used to introduce the
term.

The trick now, according to Linsky, is for Kripke
tolxaable to explain, in the same manner, how it is possible
for it to be a priori true that St. Anne is a parent and
Homer is an author. For, if a similar account can be given
of these cases, then Kripke can explain why (5) and (9)
appear to be true, and thus protect his thesis that proper ‘}
names are rigid designators against the purported counter-
examples. Linsky, however, places great stress on the ante-

cedent of the preceding conditional. He states:

I doubt that it can be met. The difficulty lies in the
fact that we really do not have any clear account of p
prioricity. In what way are the St. Anne and Homer
cases like that of the standard metre bar? I do not
see how the cases can be made similar at all. Can it

 

108

be claimed that St. Anne is a mother "by definition”?
Perhaps it can be said that, in some sense, it is true
that St. Anne is a mother by virtue of the definition
(0r meaning, or sense) of 'St. Anne', certainly not of
'mother'. But then what happens to the claim that names
do not have senses? Indeed, it can appear that the
appeal to a difference in extensions of the concepts of
the a priori and the necessary, the a poSteriori and

the contingent, is only an ad hoc device for avoiding
counterexamples.1

 

As a result of the St. Anne and Homer cases Linsky feels
that ”Kripke must abandon either the distinction between

the a priori and the necessary, or the thesis that names are

‘n‘}

rigid designators (do not induce de dicto/de re ambiguities),

 

or the thesis that they lack sense."2 Linsky's answer is

to take the latter course of action. That is, he wants to
hold that it will be a priori true that St. Anne is a parent
and Homer is an author only if we allow the proper names of
these individuals to have senses. Thus, "we can safely

explain the apparent de dicto/de re ambiguities induced by

 

these names which threaten their rigid designator status as
really differences of modality, epistemic and metaphysi-
cal."3

I am sure that by now it has become painfully obvi-
ous to the reader that either something has been left out

of my account of Linsky's argument or that there is

 

llbid., p. 63.

21bid., p. 66.

31bid., p. 83.

109

something amiss with it. Why, we might wonder, does Linsky
go through all of these contortions to show us that a name
such as 'St. Anne' requires a sense, when by hypothesis we
start out with a case where the definite description used

to introduce the name is taken to fix its meaning? If, as

 

Linsky states, 'the mother of Mary' fixes the meaning of
'St. Anne', then 'St. Anne' already has a sense. What
further argument is required? Also, since it would be ana-
lytically true that St. Anne is the mother of Mary, cer-
tainly no further argument is needed to show that we know 3)
a priori that St. Anne is a parent.

Actually, something has been left out of my account
of Linsky's argument. Linsky does need to argue that proper
names require a sense because he also wants to hold that the
sense of a name like 'St. Anne' will not be the same as the
sense of the definite description that is used to introduce
the name and fix its meaning. He is also clearly aware of
the fact that it simply won't do for it to be analytically
true that St. Anne is the mother of Mary. In this latter

regard he states:

I am not claiming that in cases in which we acquire the
sense of a name with a single definite description that
the sense acquired is the same as that of the definite
description. In fact that cannot be the case, for it
is de dicto necessary that the mother of Mary is a
mother, and the author of the Iliad and the Odyssey an
author. Thus if the sense of the name 'St. Anne is
identical with the sense of 'the mother of Mary' it
would follow that it is also necessary de dieto that

110

 

St. Anne is a mother, and this I have been at some
ppins to deny.1 -

 

Linsky's answer to the problem raised in the passage quoted

above is to suggest that:

the name gains only part of its sense from the
description which introduces it to us. Another constit-
uent of its sense accrues to it solely by virtue of the
fact that it is a proper name. It is by virtue of this
constituent of its sense that any proper name is a
rigid designator.2

Hence, it would appear as though Linsky wants to argue that
we know a priori that St. Anne is a parent not because it
is analytically true that St. Anne is the mother of Mary
(in which case (5) would be held by many to be true), but
because being a mother (i.e., a parent) is only part of the
sense of St. Anne. It is hard to see, however, how this
sort of explanation could help Linsky avoid his problem.
Being a male is only part of the meaning of 'bachelor' yet
clearly it is analytically true that a bachelor is a male
and as a consequence it is de dicto necessary as well.
Similarly, if 'the mother of Mary' is part of the sense of
'St. Anne', then it is analytically true that St. Anne is
the mother of Mary and it follows that (5) is true. But

this is precisely what Linsky has taken great pains to deny

 

lIbid., pp. 83-84. Emphasis added.

2Ibid.

111

since he wants to maintain Kripke's thesis that proper
names are rigid designators.

Perhaps Linsky would argue that since being a rigid
designator is part of the sense of a proper name it is not
analytically true that St. Anne is the mother of Mary.
Gerald Vision has pointed out, however, that any attempt
that Linsky might make in that direction to reconcile us to
the claim that while 'the mother of Mary' is part of the
sense of 'St. Anne' it is not analytically true that St.
Anne is a mother would seem to undermine his claim that

proper names have senses. According to Vision,

since, on ordinary conceptions, 'X is Y' is analytic
(and hence de dicto necessary) if 'Y' forms only part
of the meaning of 'X', we must suppose that the re-
mainder of the sense cancels this usual expectation.
However, all Linsky discloses about the remainder of the
sense is that it consists in the item's being a proper
name and thus a rigid designator. How could this cancel
the analyticity of 'St. Anne is a mother' without being
tantamount to the explanation that, despite an intimate
relation of the terms, proper names don't take sense?1

Thus it would appear that Linsky faces the following problem.

If he tries to explain how it is possible that we know p
priori that St. Anne is a parent in terms of the sense of
the name 'St. Anne', then he can no longer hold that all
proper names are rigid designators. This is because if 'the

mother of Mary' fixes the meaning of 'St. Anne' (and gives

 

lVision, "Linsky on Rigid Designation and Sense,”
pp. 294-295.

 

 

112

us part or all of its sense), then it is analytically true

that St. Anne is the mother of Mary and it follows that (5)

is true. If, on the other hand, he tries to argue that it

is not analytically true that St. Anne is the mother of

Mary and hence, it does not follow that (5) is true because

part of the sense of any proper name is that it is a rigid

designator, then, as Vision points out, it would appear that

names do not take sense. But, according to Linsky, it is.

only in virtue of the sense of 'St. Anne' that we can know

a priori that St. Anne is a parent. 4‘
Linsky could, of course, avoid the problem alto-

gether if he could show that a definite description's being

part of the sense of a name is sufficient for a prioricity

 

but not analyticity. Unfortunately, he nowhere gives us
any indication as to how this could be possible. Notice
that in those cases where the introducing definite descrip-
tion merely fixes the referent of the name and not its
meaning we can a la Kripke produce the desired explanation.
Thus, for example, according to Kripke we know a priori that
Neptune is the cause of the perturbations in the orbit of
Uranus, and it is not de dicto necessary (and hence, not
analytic) since the definite description used to introduce
the name 'Neptune' merely fixes the reference of the name.
For Linsky, however, the introducing definite description
fixes the meaning of the name and gives us, at the very

least, part of its sense, in which case it would seem to be

113

much more difficult to be able to drive a wedge between 3

prioricity and analyticity.

Section Four

 

It is possible that I have not done justice to
Linsky's argument. It is also possible that he is simply
confused or that he never intended to claim that 'the
mother of Mary' fixes the meaning of 'St. Anne'. Whatever
the case may be it does not affect the crucial feature I

wish to note concerning his treatment of the St. Anne case.

at}

Linsky feels that in order to view (5) as false we have to
be able to show that it is a priOri true that St. Anne is a
mother, and this he feels can only be accomplished by re-
quiring that the name 'St. Anne' have a sense attached to
it. Dummett, on the other hand, wants to argue that (5) is
true only because we do know a priori that St. Anne is a
parent. His position is that all a priori truths are also
de dicto necessary, but not all de dicto necessary truths
are a priori true. (Dummett, pp. 117 and 121)

While I find Linsky's account of both Dummett and
Kripke to be seriously flawed, I do find it particularly
instructive in one respect. Recall that for Dummett, 'the
mother of Mary' is only taken to fix the reference of 'St.
Anne' and not its meaning; and this fact alone is supposed
to show us why it is a priori true that St. Anne is a

parent, in a manner precisely analogous to Kripke's metre

114

rod example. But Linsky has, I believe, argued convinc-
ingly that the two cases cannot be made similar at all. In
the St. Anne case, as opposed to the metre rod example, we
have no plausible way of accounting for the purported g

prioricity, since no stipulative definition is involved.

 

(We could, of course, do as Linsky suggests and attach a
sense to 'St. Anne', but there is no reason for Kripke to
be burdened with this unnecessary tactic. Kripke's claim
that names are rigid designators can be protected in a
manner that is much more direct and that is consistent with
his overall view concerning proper names.) Dummett's claim,
however, that (3) read as (5) is true is based on the as-
sumption that it is a priori true that St. Anne is a parent.
Without this assumption he cannot show that 'St. Anne' is
not a rigid designator.

Linsky, unlike Dummett, seems to hold that one might
take (5) to be true without appealing to any further
support. Thus in order to protect Kripke's thesis that
names are rigid designators he feels that we have to account
for any intuitions we might have regarding the truth of (5)

in terms of a prioricity (i.e., he feels that Kripke has to

 

show that what appears to result in a de dicto/de re ambig-

 

uity is really an ambiguity between two kinds of necessity,
metaphysical and epistemic). It is only because he starts
out by assuming that Dummett has produced a plausible coun-

terexample (i.e., (5) might be taken as true) that he is

115

able to then go on and argue that proper names require
senses. Linsky simply misunderstands the structure of
Dummett's argument and as a result he takes (5) too seri-
ously. That is, he fails to realize that (5) is false, and
that given Dummett's view all one has to do to show this is
make it clear that it is not a priori true that St. Anne is
a parent.

Dummett, on the other hand, argues in precisely the
opposite direction from Linsky. He feels that our intu- S

itions regarding the truth of (5) hinge on its being p

“x"!

priori true that St. Anne is a parent. According to
Dummett all a priori truths are de dicto necessary. Hence,
if it is a priori true that St. Anne is a parent, then (5)
is true. Notice also that given the structure of Dummett's
argument he cannot argue, as does Linsky, that 'St. Anne'
has a sense attached to it and this accounts for our know-
ing a priori that St. Anne is a parent. This would be
simply to beg the question on Kripke, since he need only
maintain that there is no a priori truth to be accounted
for in this case.

One might object at this point that Dummett's only
mistake is his choice of example. His St. Anne case fails j
simply because it is not a case where the name was intro-
duced into the language via a stipulative definition. But
surely this can be easily corrected. All we need do is

consider a name like 'Neptune'. If we assume, as does

116

Kripke, that Leverrier introduced 'Neptune' into the lan-
guage by stipulating that its referent be fixed by the def-
inite description 'the planet that cauSed perturbations in
the orbit of Uranus', then we have the required example.
The account of how it is possible to know a priori that
Neptune caused the perturbations in the orbit of Uranus is
thus readily at hand. Dummett can now argue that insofar
as we have a case of a priori truth, we also have a case of
de dicto necessity.

For the sake of argument I will merely assume that

3...!“

Dummett's somewhat eclectic argument, in this regard, has w
some force behind it. Hence, rather than try to evaluate
it, I would merely like to point out what I find to be an
extremely counter-intuitive consequence of accepting it.

If Dummett is correct, then 'Neptune' is not a rigid desig-
nator, and this follows without having to assume that it is
definitionally equivalent to the introducing definite de-
scription. But if 'Neptune' is not a rigid designator,
then there is some sense in which Neptune might not have
been Neptune, viz., the reading on which the first occur—
rence of 'Neptune' has wider scope than the modal operator,
and the second occurence narrower scope. Hence, if Dummett

is right the following will be true:

(11) [n:x]0 xsén

117

It is, however, extremely difficult to imagine how this
could be so given that the description associated with
'Neptune‘ by hypothesis only fixes its reference and not
its meaning. In order for it to be possible that Neptune

be other than it is it has to be possible for the value of

'Neptune' in the actual world to be different from its value

in some possible world. But for this to be the case it
would seem that 'Neptune' would have to abbreviate some
definite description which was its meaning or sense. Thus
in order for (11) to be true the second occurrence of
'Neptune' in (11) would have to have as its value something
other than Neptune. Now, if 'Neptune' abbreviated some de-
scription which was its meaning, it would be plausible to
suppose that its value changed from world to world. But by
hypothesis the description used to introduce 'Neptune' only
fixes its reference, and is not abbreviated by the name.

Thus if Dummett is right and (11) is true, he will have to

explain how it is possible for the value of a name to change

from world to world when the name in question is not a dis-
guised definite description. Nowhere does he offer such an
explanation. We, therefore, cannot accept Dummett's coun-
terexample without paying a price, and for many this price
will surely be too high.

Thus it appears that Kripke's claim that proper
names, unlike definite descriptions, are rigid designators

can be upheld, and that his argument that names cannot be

4“ ’t

118

disguised definite descriptions remains intact. As I noted
in my introduction, however, this would seem to rule out
any possibility of a Frege-Russell answer to the question
”How do proper names refer?" It is at this point that the
causal theory of names becomes an essential adjunct to the
thesis that proper names are rigid designators, and it is

the causal theory that I will turn to next.

 
 

nn 1.! .m l.

‘lflfl 3 ‘il l|l

CHAPTER THREE

THE CAUSAL THEORY

Section One

 

When Saul Kripke first introduced the causal theory
of proper names he made it perfectly clear that he was not
attempting to give a set of necessary and sufficient con-
ditions for reference. He realized that his view concern-
ing prOper names required further refinement and elaboration
before this could be accomplished, and he took himself to
be giving a theory of reference only in the sense that it ,
afforded a ”better picture” of the way reference involving
proper names actually works. “I want to present a better
picture without giving a set of necessary and sufficient
conditions for reference. Such conditions would be very
complicated, but what is true is that it's in virtue of our
connection with other speakers in the community, going back
to the referent himself, that we refer to a certain man.”1

Others, such as Michael Devitt2 have been less

cautious, and have attempted to give a causal theory of

 

lKripke, Naming and Necessity, p. 94.

 

2Devitt, ”Singular terms," pp. 183-205.

119

120

proper names in the stricter sense of the term 'theory'.

Devitt may be viewed as making some of the refinements that

are required if a causal theory of proper names is to give

us a set of necessary and sufficient conditions for ref-

erence. It seems to me that such work is crucial if the

causal theory of proper names is to be worthy of wide accep-

tance. One might be inclined to agree with Kripke that the

causal View gives us a "better picture” of the way reference

involving prOper names actually works as long as the possi—

bility exists that it ”might be refined so as to give us 4

”1 Such

more exact conditions for reference to take place.
a refinement (i.e., an analysis yielding a set of necessary
and sufficient conditions for such reference) would be most
welcome by the philosophical community in general, and the
defenders of the causal theory in particular. On the other
hand, if it can be shown that a causal theory of proper
names cannot provide us with such conditions, then, although
we may still be inclined to agree with Kripke that it gives
us a "better picture" (than do other traditional accounts),
we must nevertheless continue our search for an even more
adequate analysis.

In this chapter I attempt to give a fairly clear
and intuitive case of reference involving a proper name

where there is no causal connection between the referent

and the speaker's utterance of the name. Hence, I attempt

1Kripke, Naming and Necessity, p. 94.

 

121

to show that a causal theory of proper names cannot yield

a necessary condition for reference. Also, since the name
in question will intuitively satisfy condition (C) of
chapter one I will in affect be showing that the thesis that
proper names are rigid designators is not coextensive with
the thesis that names refer to their bearers in virtue of
some appropriate causal connection. I construct my case by
first considering what I take to be a parallel case for the
causal theory of knowledge. Finally I argue that a familiar
move made to salvage the causal theory of knowledge in

light of this case won't work for the causal theory of ‘

proper names .

Section Two

 

One approach to the analysis of knowledge inspired
by the Gettier1 counterexamples was an attempt to spell out
the characteristic causal relations that hold between one's
belief and what is believed. This approach was taken by
Alvin Goldman.2 Goldman attempted to replace the epistemol-
ogical concept of justification with causal concepts that
are naturalistic and given in purely descriptive terms. He
viewed his analysis as yielding necessary and sufficient

conditions for knowledge; we, however, need only be

 

1E.Gettier, ”Is Justified True Belief Knowledge?,"
W. XXIII (June. 1963), pp. 121-123.

2Alvin Goldman, ”A Causal Theory of Knowing,” The
Journal of Philosophy, LXXIV (June, 1967), pp. 357-372.

 

122

concerned with the weaker claim that a causal connection

is a necessary condition for knowledge:

Thesis 1. S knows that p only if an appropriate
causal connection exists between S's
belief that p and the fact that p.

”Appropriate,” knowledge—producing causal processes include

the following for Goldman:

(l) perception
(2) memory 3
(3) a causal chain, exemplifying either Pattern l

or 2,1 which is correctly reconstructed by

inferences, each of which is warranted, (Back-

ground propositions help warrant an inference

only if they are true)

(4) any combination of (l), (2) and (3).2

Brian Skyrms3 designed the following case to show
that Thesis 1 is false. Suppose that while Harry is walk-
ing home from work one evening he stumbles across some poor
soul lying on the sidewalk with his head severed from his
body. Harry, being a sharp-witted fellow and a keen

student of anatomy realizes that the man must be dead.

 

lIbid., pp. 369-370.

2Ibid.

3Brian Skyrms, ”The Explication of 'x knows that p'fl'
The Journal of Philosophy, LXIV (June 22, 1967), pp. 373-
389.

 

123

Sure enough, Harry is right. The man is dead. Harry could
certainly be said to know that the man is dead in this case,

and the causal reconstruction would be as follows. Let:

'B' stand for belief.

'h' stand for Harry.

Solid arrows represent causal connections.
Dotted arrows represent inferences.

p = the fact that the man is dead.

the fact that his head is severed from his body. ,

9

..P

a "background" proposition, describing the
effect that losing one's head has on one's life.

r

This is what Goldman calls a ”Pattern 2" case, and it

reveals that (q) is a cause both of p and of h's belief

(p)
(94)

\

\
\
~;,Bh(p)

of (p).

I

I
Bh(rY

Let us suppose, however, that before Harry stumbled across
the body the following events transpired. The man lying

on the sidewalk had actually died of a heart attack.

124

After he had lain there for some time a young medical
student happens along and neatly amputates the man's head
with his shiny new scalpel. Harry would still be said to
know that the man is dead in this case, but notice that (q)

is not causally responsible for the man's death.

(p)

(q)———>Bh(q)(

\

Bh(p)

I
/

Bh(rY

\3 LI
ft

It seems quite apparent from the above case that
Goldman's causal analysis cannot be taken as specifying
necessary conditions for knowledge. The point that I would
like to stress about Skyrms' case, however, is that we
count it as a legitimate counterexample because there are
sufficient evidential considerations to warrant knowledge
independent of any appropriate causal connection. That is,
we are forced to grant that Harry knows the man is dead
solely on the basis of epistemological considerations. In
the next section of this paper I will present a similar
sort of case for reference involving a proper name. Hope—
fully, it will be a case where there are sufficient eviden-
tial considerations to grant successful reference, yet it

will be a case where there is no causal connection between

125

the referent and the speaker's utterance of the name.

Section Three

 

The causal theory of knowledge attempts to spell
out the characteristic relation that has to hold between a
person's belief in a certain proposition p and the fact p
for knowledge of p to be obtained. Similarly, the causal
theory of proper names attempts to spell out the relation
that has to hold between a speaker's utterance of a name

and the thing named for reference to take place. In both

5"

cases the relation is a causal one. The thesis, then, that

I attempt to show is false is as follows.

Thesis 2. In uttering the name n, S refers to x
only if an appropriate causal connection
exists between S's utterance of the name
and x.

Unfortunately, the proponents of the causal theory of
proper names are not quite as clear as Goldman is when it
comes to spelling out the "appropriate" causal processes.
Kripke, for example gives us the following rough account

of his theory.

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 ref-
erence as the man from whom he heard it. If I hear the

126

name 'Napoleon' and decide it would be a nice name for
my pet aardvark, I do not satisfy this condition.1

Devitt characterizes the causal theory of proper names in

the following manner.

The central idea of the causal theory of proper names
is that our present uses of a name, say 'Aristotle'
designate the famous Greek philosopher Aristotle, not
in virtue of the various things we (rightly) believe
true of him, but in virtue of a causal network stretch-
ing back from our uses to the first uses of the name

to designate Aristotle. —Our present uses of— names
borrow their reference from earlier uses. It is this
social mechanism that enables us all to designate the
same thing by a name.2

 

 

 

 

 

And, Donnellan gives us yet another account of the theory.

In general, our use of proper names for persons in
history (and also those we are not personally acquainted
with) is parasitic on uses of the names by other
people-—— in conversation, written records, ect. In-
sofar as we possess a set of identifying descriptions
in this case they come from things said about the pre-
sumed referent by other people. My answer to the
question, 'Who was Thales?’ would probably derive from
what I learned from my teachers or from histories of
philosophy. Frequently, as in this example, one's
identifying descriptions trace back through many levels
of parasitic derivation. Descriptions of Thales we
might give go back to what was said, using that name,

 

1Kripke, Naming and Necessity, p. 95.

 

2Devitt, "Singular Terms,” p. 184.

 

[Pl \ ‘. 1T1: \ 1‘-‘i: \ nnnnn ‘4

127

by Aristotle and Herodotus. And,if Thales existed, the
trail would not end there.1

The general picture that emerges from the preceding accounts

is that of a name being hooked up with some object via an

appropriate causal process such as an initial baptism and

then being passed along to others in virtue of their par-

ticipation in the causal network that is generated from the

original naming ceremony. Jaegwon Kim has pointed out in

his article "Perception and Reference Without Causality"

that viewing things in this way enables one to think of the r
causal chain that connects a name with an object in two

parts:

(1) the part extending from a given use of a name to
the baptismal event (this causal chain transmits the
reference initially established), and (2) the part
corresponding to the connection between the baptismal
act and the object baptized (this part establishes the
initial referential connection between the name and the
object named).2

Hence, if we want we can focus our attention on the link of
the causal chain that exists between a speaker's utterance

of a name and its first uses or we can focus our attention

 

1Keith Donnellan, "Proper Names and Identifying
Descriptions," in Semantics of Natural Language, ed. by D.
Davidson and G.Harman (Dordrecht-Holland: D.Reidel
Publishing Company, 1972), p. 373.

 

2Jaegwon Kim, ”Perception and Reference Without
Causality," The Journal of Philosophy, LXXIV (Oct., 1977),
pp. 606-621.

 

 

128

on the causal link between the first uses and the object.
The question that can now be asked is how does this way of
looking at things affect Thesis 2. According to Thesis 2
the causal connection that is necessary for reference is
one that obtains between an utterance of a EEEE and an
object. No mention is made of ”first uses", ”naming cere—
monies”, or "initial baptisms". Thus Thesis 2 might be

reformulated as follows.

Thesis 2a. In uttering the name n, 8 refers to x
only if (1) an appropriate causal 1
connection exists between S's utterance
of the name and some initial naming
ceremony, and (2) an appropriate causal
connection exists between that naming
ceremony and x.

Any legitimate case of reference involving a proper name
where the use of the name is not causally hooked up with
the object would falsify both Thesis 2 and Thesis 2a.
Notice, however, that in the case of Thesis 2a this might
be done in such a way that the causal link between the use

of the name and the naming ceremony is left intact.1 This

 

1See John Canfield, "Names and Causes," Philosophi-
cal Studies, 35 (1979), pp. 71-80, for an example of some-
one who attempts to show that the causal account does not
provide us with necessary conditions for a name's naming
its referent, and blurs the distinction between the link of
the causal chain that exists between a speaker's utterance
of a name and its first uses and the link that exists
between the first uses and the object. As a result,
Canfield's cases might be viewed as falsifying clause (2)
of Thesis 2a, but not clause (1).

Regarding Canfield's cases it should also be noted

 

 

129

might lead one to hold that falsifying Theses 2 and 2a poses
no real threat to a causal theory of proper names. In fact,
Kripke and Donnellan at times seem to suggest that the
crucial connection is the one that links present uses of a
name to first uses. Thus it seems as though we need to

consider yet another formulation of Thesis 2.

Thesis 2b. In uttering the name n, S refers to x
only if an appropriate causal connec-
tion exists between S's utterance of
the name and an initial naming ceremony.

I strongly suspect that Thesis 2a is the favored formulation
of the proponents of the causal theory of proper names;
nevertheless, insofar that Thesis 2 and 2b remain options,

I will show that all three are false. To do so, I first
construct a case that falsifies Thesis 2a and 2b, but leaves
Thesis 2 unscathed. Then by making a slight alteration in

the case I show that all three are false.

 

that their success would appear to depend on the speaker's
descriptions fitting a certain individual when using the
name in question (cf. especially pp. 75 and 77). Unfor-
tunately, these are precisely the sorts of cases that
someone like Donnellan appeals to in order to show that a
causal/historical connection is necessary for a name to
name its referent (see Keith Donnellan, ”Speaking of
Nothing," The Philosophical Review, LXXXIII (January, 1974),
p. 18.) The cases that I will go on to present, however,
will attempt to circumvent this entire issue by relying
for their success on certain evidential considerations
which are quite independent ofithe correctness of the
speaker's descriptions.

 

 

130

Section Four

 

Suppose that while at a party one evening I am in-
formed by my friend Dave's wife that Dave has the peculiar
habit of naming all of his pets 'Sal'. I am told that as
a lad Dave named his very first pet 'Sal', and that all of
his pets ever since have been given the same appellation.
Thinking that things might be a bit confusing around the
house, if they ever have more than one pet, I asked how
they are able to determine which pet is being referred to
when the name 'Sal' is used. As it turns out, however,
this was never a problem. Dave, she informed me has always
been a one pet man. When his first pet died he got a snake,
when it died a dog, and so on over the years. Always one
pet is replaced by another and they all received the same
name.

”At least," I said, when she had finished ”this
saves you the problem of thinking up new names for your
pets since Dave's procedure seems to be pretty automatic.”
This I was quickly informed was not quite correct. Dave it
seems never refers to his new pet as Sal until he has con-
sidered a long list of names. Other names, besides 'Sal',
are suggested and bantered about, but the end result is
always the same. He looks at the list, then looks at his
pet, then in the end he looks at his wife and says "Let's
call it Sal" (i.e., a naming ceremony takes place). Since

Dave's wife is known to be an extremely reliable and

 

 

131

truthful person and not the sort to invent a story simply
for the sake of being an entertaining party guest I readily
accepted what she had related as an interesting fact about
Dave.

Unbeknownst to me, shortly after this conversation
took place Dave's latest pet died and he replaced it with
a goldfish. As usual when he brought the fish home he con-
sidered at length a list of possible names. And just as in
the previous cases he rejected all of them in favor of 'Sal'.

Suppose now, that shortly after this I am over at
Dave's house and while pointing to Sal Dave asks me what I
think of his new fish. Suppose further that previous to
this I have not come into contact with Dave, his wife or
anyone else who knew Sal after the naming ceremony took
place. The question is, if I reply to Dave by saying "Gee,
Sal sure is a fine specimen," has my utterance of the name
'Sal' referred to the object swimming around in the bowl?
It seems clear to me that the name 'Sal' as it occurred in
my utterance did refer to the goldfish. Moreover, it seems
clear to me that reference was achieved in this case inde—
pendently of there being any causal connection between my
utterance of the name and the initial naming ceremony.
Notice that as the case was described I was at no time part
of the causal network that gets generated from the initial
naming ceremony. Yet, according to Thesis 2a and 2b this

is just the condition that must be met for reference to

 

 

132

occur.

Why do I think it is clear that the name 'Sal' as
used by me referred to the goldfish? Recall that in the
Skyrms' case we grant that Harry knows the man is dead
because there are sufficient evidential considerations to
warrant knowledge. Similarly, it seems to me that there
are sufficient evidential considerations in the case now
under consideration to grant that reference was achieved.
Certainly, it wasn't just a lucky guess on my part that
'Sal' was the fish's name. I had been informed that Dave
had named all of his pets 'Sal' in the past, and my use of
the name was based on good inductive evidence.

What about Thesis 2? Was it shown to be false as
well? Was there a causal connection between my utterance
of the name and the goldfish? According to Thesis 2 my use
of the name 'Sal' refers to the goldfish only if there is

I

an "appropriate' causal connection between my utterance of
the name and the goldfish. Since I was perceiving the
goldfish at the time of my utterance, and since to perceive
something is to be causally affected by it, it might seem
as though Thesis 2 has not been shown to be false. I seri-
ously doubt, however, that this is the sort of causal con-
nection a proponent of the causal theory of proper names

would call appropriate”. While it is true that the name
'Sal' as it occurred in my statement was causally connected

with the goldfish (i.e., I was seeing Sal when I uttered

133

the name and was prompted to utter the name by seeing Sal
in Dave's house) it doesn't seem as though my use of the
name was causally linked to the fish in a way that the
causal theorists of proper names deem appropriate for ref-
erence. Some may feel that the case I have presented does
in fact falsify Thesis 2, as well as Thesis 2a and 2b.
Others may feel that the causal connection that obtained
when I uttered the name while looking at the goldfish was
”appropriate". I don't see how this issue can be settled
until the causal theorist spells out in detail exactly what
is to count as an "appropriate" causal connection. In order
to eliminate debate on this question and bypass this issue
altogether I make a slight alteration in my case in the
next section of the paper. Before doing so, however, I
would like to note what I take to be a much more compelling
reason for altering the original case.

So far I have merely assumed that the expression
'Sal', as it occurred in my statement, functioned as a
proper name. One might argue, however, that the reason why
'Sal' refers to the goldfish and consequently the reason
why Thesis 2, and for that matter Thesis 2a and 2b, have
not been clearly shown to be false is that the expression
'Sal' in this context can be paraphrased away by or is a
surrogate for an indexical sign or perhaps a Russellian
logically proper name. That is, 'Sal' as used in my utter—

anceluuithe extension it did because of the context in

 

134

which it was used, and not because it is a proper name that
refers. The expression 'Sal' on this account picked out
the goldfish in just the way the words 'this' or 'that'
would if either of them had been used instead, and what
'Sal' denoted was relative to me just as what a demonstra-
tive denotes is relative to a speaker. Thus, it could be
argued that it is not evidential considerations that lead
us to hold that 'Sal' refers to the goldfish, but rather it
is the realization that the expression in the context of
my utterance stood as a surrogate for a demonstrative. If
this account of why my use of 'Sal' refers to the goldfish
is correct, then it could be argued that I have not really
shown Theses 2, 2a, or 2b to be false, since I have not
given a case where a proper name, qua proper name, refers
to an object.

But is this account correct? Suppose that I had
uttered the name 'Saul' instead of 'Sal' when I was comment-
ing on Dave's goldfish. (Imagine that Dave's wife pro-
nounces all of her a's as soft a's so that instead of pro-
nouncing Sal\\Sal\ she pronounces it\\Sol\, and that this
led me to believe that all of Dave's pets had been given
the name 'Saul'.) My own intuitions are that in this case
the expression 'Saul' would not refer to the goldfish, and
that this would be sufficient to show that the expression
'Sal' in the context of my original utterance was a name

and not a substitute for a demonstrative. Others, however,

 

135

might feel that I would still refer to the goldfish, espe-
cially those who think that something like Donnellan's1
referential/attributive distinction for definite descrip-
tions holds for proper names. If it does, then the index-
ical account would seem to gain plausibility since 'Saul'
could be viewed as being used referentially, no matter what
the ”true" name of the fish is. On this account I could
have used any name and still referred to the fish since the
name is incidental for reference. Intuitions, then con—
cerning this case may conflict. Thus in order to avoid
begging any crucial questions it is clear that I need to

alter the original case if I want to avoid this problem.

Section Five

 

As matters now stand I have presented a case which
may at best falsify Theses 2a and 2b but leave open the
possibility that Thesis 2 is correct, and which at worst is
not a counterexample to any of the three theses. In this
section of the essay I alter the original case so that both
of these problems may be avoided, and I show that all three
theses are false.

Let everything in the original case remain the same,
except instead of going over to Dave's house suppose that I

just happen to bump into him on the street. He informs me

 

1Keith Donnellan, ”Reference and Definite Descrip-
tions," The Philosophical Review, LXXV (July, 1966), pp.
281-304.

 

 

136

that his pet aardvark has just died and that he has replaced

it with a goldfish. Suppose now that upon hearing this I

say to Dave, "I have a lot of extra food for Sal if you

need it." As in the previous case I would like to suggest

that there are sufficient evidential considerations for ref-

erence to be granted. We concede that my utterance of the

name 'Sal' refers to the goldfish, not because it is

causally connected with Sal (which in this case it clearly

is not), but rather because certain epistemological con-

ditions obtain. Dave's wife has, to use Devitt's phrase, (

"given me the ability to designate 331 by 'Sal'.”1 This

ability was gained, however, independently of Dave's wife
being causally affected by Sal, and independently of her
being causally linked to the original naming ceremony.
Notice, also, that there is less temptation in this case to
view 'Sal' as standing as a surrogate for a demonstrative.
In the previous case it might have seemed that we could
paraphrase away the expression 'Sal' by substituting the
word'tfds' for'Sal'. In this case no such paraphrase is
possible, since no element of spatial selectivity is in-
volved. There is no direct spatiotemporal connection
between the expression 'Sal' as it occurs in my utterance
and the goldfish.

It might be pointed out, however, that not all

 

lDevitt, "Singular Terms," p. 185.

137

indexicals indicate their objects directly.1 For example,
in the case of pronouns we sometimes refer indirectly to an
object that has been previously named or described in dis—
course. (Some linguists refer to such coreferential uses

of pronouns as anaphoric.) Thus, it may be argued that
while we can't paraphrase away the occurrence of 'Sal' with
a demonstrative in this case, we may adequately paraphrase
it away with a pronoun. Consider the sentence "I have a

lot of extra food for it if you need it" uttered by me after
Dave informs me that he has acquired a new pet goldfish.
Here reference to the goldfish is achieved because the pro—
nown 'it' in some way "refers back to" Dave's words, and in
virtue of this back reference the fish. On this account
then the reason why the expression 'Sal' refers to the gold-
fish is not because it is a name, but because it is a surro-
gate for the pronoun 'it'.

This time, however, I think it can be shown that the
alternate account of reference is not a plausible one.
Recall that one observation that made the indexical account
of the previous case plausible was that the expression 'Sal'
might be viewed as being used referentially. That is, the
name would be inessential for successfully referring to the

goldfish since any name might have been used to do this.

 

1Richard Gale makes this point in his article "In-
dexical Signs, Egocentric Particulars, and Token-Reflexive
Words," The Encyclopedia of Philosophy, ed. P.Edwards (New
York: MacMillan, 1967), Vol. 4, p. 151.

 

 

138

In this case, however, it is far less plausible to suppose
that I would have successfully referred to goldfish had I
used another name. If 'Sal' was merely a surrogate for the
pronoun 'it', then it seems as though I should have been
able to refer to the fish no matter what name I used. It
seems clear to me, however, that if I had said "I have a
lot of extra food for George if you need it” after Dave
informed me that he has a new pet goldfish, my use of
'George' would not have referred to the fish. Thus we
realize that the name 'Sal' is essential for reference in
this case, and there is no temptation to suppose that it
could be adequately paraphrased away.

Just in case there are any lingering doubts as to
the legitimacy of my counterexample that need to be erad-
icated, however, let me alter my case one last time.
Suppose that while at the party Dave's wife informed me
that Dave's latest pet had died, and that he was going to
replace it the next day with.a goldfish. That is, suppose
I acquired sufficient evidence to know that Dave would
shortly have a new pet goldfish. Now again suppose that I
bump into Dave on the street a few days after the party,
only this time before Dave has a chance to mention his new
pet I say "I have a lot of extra food for Sal if you need
it.” In this case there is no causal link of any sort be-
tween my use of the name 'Sal' and the goldfish (Thesis 2

is false) nor is thereaicausal connection between my use of

139

the name and some initial naming ceremony (Theses 2a and
2b are false). More important, however, is the fact that
in this case there isn't the slightest possibility of
viewing 'Sal' as a mere surrogate for a demonstrative or a

pronoun.

Section Six

 

One obvious observation that could be made concern-
ing the Skyrms' case is that while it is true that the man's
head being cut off was not causally responsible for his
death it was causally sufficient for his death. Unfortu-
nately, this piece of information is of no avail to Goldman,
but it is useful to someone like Marshall Swain1 who
utilizes causal conditions in a ”defeasibility” approach to
knowledge. On this view it is noted that what prevents
knowledge in the Gettier type case is some kind of defect
in one's justification. Swain states that ”one way of
characterizing this defect is to say that, even though the
justification involved is sufficiently strong to render the
proposition evident, the justification is nevertheless de-

2

feated by some special counterevidence.” His analysis of

knowledge is as follows:

 

1Marshall Swain, "Knowledge, Causality, and Justi-
fication," The Journal of Philosophy, LXIX (May 25, 1972),
pp. 291-300.

2

 

Ibid., p. 293.

140

S has nonbasic knowledge that p if and only if

(i) p is true;

(ii) S believes that p;

(iii) S's justification renders p evident for S;

(iv) The causal chain leading to 8'3 belief in e
either (1) contains the event or state of
affairs referred to by p, or (2) contains some
other event or state of affairs that is, in the
context of the evidence possessed by S, either
causally or logically sufficient for the occur-
rence of the state of affairs referred to by p.

(v) There is no true statement q such that q in
conjunction with S's evidence E fails to render
p evident for S and such that q is true because
of events in the causal chains referred to in
(iv).

The condition in this analysis that defeats Skyrms' counter—
example is (iv). This condition rules out the problem
raised by Skyrms because the second half of the disjunction
allows Harry to know the man is dead on causal grounds.
Might not a similar move be made to salvage the causal
theory of proper names in light of the case I present? I
think not, at least not in any way that would ultimately
prove acceptable to a proponent of the causal theory of
proper names.

Swain's move relies on the observation that having
one's head severed from one's body is causally sufficient
for death. In light of this observation it could be
pointed out that Harry makes use of the following "causal"

law

 

Ibid., p. 294.

 

 

141

(A) (x)((x is a person & x has his head out off) +
x is dead)

to learn that the man is dead. Similarly one might plau—
sibly argue that in the Sal case I make useof the

following ”causal" law

(B) (x)(x is a pet of Dave's + x is named 'Sal')

to learn that the goldfish is named 'Sal'. It might then
be argued that (B) seems to be an empirical (i.e., causal) 3
law about Dave and his habits just like (A) is an empirical
(i.e., causal) law about people. It might even be argued
further that both sustain counterfactual and subjunctive
conditional statements. (A), for example, clearly supports
statements such as "If Bob were a person and had his head
been cut off, then he would be dead." Similarly (B) seems
to support such statements as "If the lion at the Potter
Park Zoo had been a pet of Dave's, then it would be named
'Sal'.” Thus it might finally be argued that (B) is
causally sufficient for me to learn that the goldfish is
named 'Sal'.

It seems to me that all of this might be granted to
the proponent of the causal theory of proper names who opts
to make a move similar to Swain's in order to avoid my
counterexample. I do not think, however, that this would

leave him with very much in the way of a theory that makes

142

use of Thesis 2, 2a, or 2b. For one thing, while it might
be plausible to suppose that (B) is causally sufficient £95
m§_£9_1e§£n that the goldfish is named 'Sal', I do not see
how it is plausible to suppose that (B) is causally suffi-
cient for the name 'Sal' as it occurs in my utterance to
refer to the goldfish. But it is precisely this latter
point that is at issue. Swain is able to avoid the problem
raised by Skyrms because he is able plausibly to argue that
the evidence Harry bases his belief on (i.e., that the man's
head is severed from his body) is causally sufficient for
death. In the Sal case, however, the evidential consider-
ations that lead us to grant that reference was achieved
are simply not amenable to a causal reconstruction of the
sort required by the causal theorist of proper names. The
important point to notice is that once the causal theorist
appeals to (B) to avoid the Sal counterexample it seems as
though the causal connection between the present utterance
of a name and some initial naming ceremony drops out of the
picture altogether as a requirement for reference. It is
also equally important to notice that once an appeal to

(B) is made it seems as though there no longer need be a
causal connection between an utterance of a name and the
object named for reference to take place. This is so
because none of the instances that confirm (B) involve the
goldfish 'Sal'. Thus the proponent of the causal theory

of proper names who makes use of (B) in order to circumvent

 

 

143

my counterexample gives up all that originally seemed to

be important and interesting concerning Theses 2, 2a, and

2b.

BIBLIOGRAPHY

BIBLIOGRAPHY

Canfield, John. "Names and Causes.” PhilosophiCal Studies,
35 (1979), 71-80.

 

Devitt, Michael. ”Singular Terms.” The Journal of Philos—
ophy, LXXI (April 18, 1974), 183-205.

 

Donnellan, Keith. "The Contingent A Priori and Rigid
Designators.” Midwest Studies in Philosophy, Vol.
II: Studies in the Philosophy of Language (February
1977), 12-27.

 

"Proper Names and Identifying Descriptions."
Semantics of Natural Language. Edited by D.Davidson
and G.Harman. Dordrect-Holland: D.Reidel Publishing
Company, 1972.

 

 

. ”Speaking of Nothing." The Philosophical Review,
LXXXIII (January 1974), 3-31.

 

 

. "Reference and Definite Descriptions." The
Philosophical Review, LXXV (July 1966), 281-304.

 

 

Dummett, Michael. Frege: Philosophy of Language. New York:
Harper and Row, 1973.

 

Gale, Richard. ”Indexical Signs, Egocentric Particulars,
and Token-Reflexive Words.” The Encyclopedia of
Philosophy. Edited by P.Edwards. New York:

Mac Millan, 1967.

 

 

Gettier, E. ”Is Justified True Belief Knowledge?” Analysis,
XXIII (June 1963), 121-123.

 

Goldman, Alvin. "A Causal Theory of Knowing.” The Journal
of Philosophy, LXXIV (June 1967), 357-372.

 

 

Hudson, J. and Tye, M. ”Proper Names and Definite Descrip-
tions With Widest Possible Scope." Analysis, XL
(January 1980), 63-64.

 

Hughes, G.E., and Cresswell, M.J. An Introduction to Modal
Logic. London: Methuen and CO LTD, 1968.

144

 

11..
V . . .M in. m ,1... .. ....
11 1 l . s-.|. ‘||.|||

 

145

Kim, Jaegwon. "Perception and Reference Without Causality."
The Journal of Philosophy, LXXIV (October 1977),
606-621.

Kripke, Saul. Naming and Necessity. Cambridge: Harvard
University Press, 1980.

 

"Identity and Necessity.” Identity and Individ-
uation. Edited by Milton Munitz. New York: New
York University Press, 1971.

 

. "Semantical Considerations on Modal Logic."
Reference and Modality. Edited by Leonard Linsky.
London: Oxford University Press, 1971.

 

. "Speaker's Reference and Semantic Reference."
Midwest Studies in Philosophy, Vol. 11: Studies in
the Philosophy of Language (February 1977), 255-276.

 

. ”A Puzzle About Belief." Meaning and Use. Edited
by A.Margalit. Dordrecht-Holland: D.Reidel
Publishing Company, 1979.

 

Lewis, David K. Counterfactuals. Cambridge: Harvard Univer-
sity Press, 1973.

 

Linsky, Leonard. Names and Descriptions. Chicago: The Uni-
versity of Chicago Press, 1977.

 

Loar, Brian. "The Semantics of Singular Terms." Philosoph-
ical Studies, 30 (1976), 353-377.

 

 

McKinsey, Michael. ”The Reference of Proper Names.” Unpub-
lished Ph.D. dissertation, Indiana University, 1976.

Peacocke, Christopher. ”Proper Names, Reference, and Rigid
Designation." Meaning, Reference, and Necessity.
Edited by Simon Blackburn. Cambridge: Cambridge
University Press, 1975.

 

Quine, W. V. ”Reply to Grice.” Words and Objections. Edited
by D.Davidson and J.Hintika. Dordrecht-Holland: D.
Reidel Publishing Company, 1969.

 

. Review of Identity and Individuation. Edited by
M.Munitz. The Journal of Philosophy, LXIX (1972),
491-495.

 

 

Salmon, Nathan U. Review of Names and Descriptions, by
Leonard Linsky. The Journal of Philosophy, LXXVI
(August 1979), 436-452.

 

 

 

 

146

 

Skyrms, Brian. "The Explication of 'x knows that p'." The
Journal of Philosophy, LXIV (June 22, 1967), 373-
389.

Swain, Marshall. ”Knowledge, Causality, and Justification.‘
The Journal of Philosophy, LXIX (May 25, 1972),
291-300.

Vision, Gerald. "Linsky on Rigid Designation and Sense."
Australasian Journal of Philosophy, Vol. 58
(September 1980), 291-297.

 

 

 

 

 

 

 

HICHIGQN STQTE

V.

UNI LIBRARIES
1| 1 IIHHHIHH HI‘HIIHIHIHIIHHI
0571

3129300706