’ g1 .‘.__.‘“‘-. _ q“‘_-u‘——-.—.u. _. . .~-.—-—~v.. . .. . . w
l

 

   

 

 

 

I
\
.
.
u
A
' .y .
r' .
l . .
, V ‘
, ' ,.
.
I, . ‘ - . . ‘ l 4
, . . ~ . . ,.
» . . ., t
‘ o v A.
. . . . . . ,‘ .
.
~. n' ,
' -> ‘ ' ' 4‘ 'A 2 Hr!“ - “u u:- uw-u. H ‘ .v. 4 4.. .. .., , . “”4”... .

       

 

 

 

 

lllllllllllllllllllllllllllllillllllill‘lllllllllllllllllll

3 1293 00877 0228

This is to certify that the

dissertation entitled

Necessary A Posteriori Truth

presented by

Cynthia Jayne Bolton

has been accepted towards fulfillment
of the requirements for

W degree in —Ph~i—}esephy

é?z/aw/&Oflé€\

Jor professor
Date 4404’ all /7y.2‘
(7 6/

MS U is an Affirmative Action/Equal Opportunity Institution

 

 

l LIBRARY
4 Michigan State
1, Untverstty

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before due due.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

W
ll JV—l

MSU lo An Afflrmetlve ActIoNEquel Opportunlty Institution
cmmut

 

 

 

 

 

 

 

 

 

———

NECESSARY A POSTERIORI TRUTH

BY

Cynthia Jayne Bolton

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Philosophy

ABSTRACT

NECESSARY A POSTERIORI TRUTH
by

Cynthia Jayne Bolton

Traditionally, philosophers have used the analytic/
synthetic, a priori/a posteriori, and necessary/contingent
distinctions to categorize statements. And traditionally,
philosophers have used these distinctions to categorize
necessary statements as a priori statements. Yet certain
statements seem better categorized as necessary and a pos-
teriori. Examples of such statements include: (1) Heat is
the form of energy constituted by the motion of atoms and
molecules in solids; (2) Gold is the element with atomic
number 79; and (3) Water is H20. For while we believe that
these statements express empirical discoveries, we also
believe that they express the essence of their subjects.

I wish to justify the claim that these statements are
both necessary and a posteriori. My justification includes
a synthesis of the causal theory of reference with the net—
work theory of meaning. But once this synthesis is carried
out, we have a rather complicated notion of necessity. We
no longer have one notion of necessity, but four notions.
One notion is the logical positivist's. A statement is

necessary if it expresses a linguistic convention. The

second notion is the network theorist's. A statement is
necessary if it is central to our theoretical network. The
third notion is the causal theorist's. A statement is nec-
essary if it is a singular statement of identity between
two rigid designators. And the fourth notion is the essen-
tialist's. A statement is necessary if the predicate ex-
presses the essence of the subject. But this essence must
be a structural property that plays an explanatory role
within our theoretical network. Of these four notions of
necessity, only the first must be a priori. The other
three are ultimately a posteriori. Statements (1)—(3) fall
under the fourth notion. And thus, they are necessary and

a posteriori.

ACKNOWLEDGEMENTS

I would like to particularly thank my dissertation
director, Professor Richard Hall, for his helpful comments
and criticisms. Special thanks are also due to Betina
Henig, Melanie McLeod, Donna Baker, and Frank Curtis,
all of whom have heard much more about this dissertation

than they ever wished.

iv

TABLE OF CONTENTS

Li st Of Figures 0 O O C O O O C O O O O O O O O O O O .
mapter one C C O O O O O O O O C O O O O C O C O O O 0

Setting the Stage: Leibniz and Hume . . . . . . . . .
Setting the Stage: Kant . . . . . . . . . . . . . . .
The Stage is Set: The Logical Positivists . . . . . .
Problems: . . . . . . . . . . . . . . . . . . . . . .

Cllapter W0 0 O O O O I O O O O I . O O O O C O O C I O

Modifying the Traditional Distinctions . . . . . . . .
Proper Names and Necessary A Posteriori Statements . .
Underlying Assumptions . . . . . . . . . . . . . . . .
De Re or De Dicto . . . . . . . . . . . . . . . . . . .

Chapter Three . . . . . . . . . . . . . . . . . . . . .

The Network Theory . . . . . . . . . . . . . . .
What Happens to the Analytic/Synthetic Distinction? . .
What Happens to the A Priori/A Posteriori Distinction?
What Happens to the Necessary/Contingent Distinction? .

mapter Four . O O C O O O O O O O O O I O O C O O O O

Follesdal: How to Synthesize the Causal Theory and the
Network Theory . . . . . . . . . . . . . . . . . .
Referential Terms . . . . . . . . . . . . . . . .
What Does This Do To The Traditional Distinctions?
Types of Necessity . . . . . . . . . . . . . . . .

Chapter Five . . . . . . . . . . . . . . . . . . . . .
Why We Need An Additional Analysis of Necessity . . . .
What Is An Essence . . . . . . . . . . . . . . . . . .
Essence and Explanation . . . . . . . . . . . . . . . .
What is an Aristotelian Essence? . . . . . . . . . . .
NOtes I O O O O O O I O O O O O O O O O O O I O O O O 0

Bibliography . . . . . . . . . . . . . . . . . . . . .

vi

17
24

30

31
33
42
54

61

61
67
76
84

89

91

95
107
108
123
124
127
135
139
146

153

figure
figure
figure

figure

LIST

OF

vi

FIGURES

43

110

120

Chapter One

The topic of necessary truth centers about two ques-
tions. There is the metaphysical question: Are there such
things as necessary truths? And there is the epistemologi—
cal question: How do we distinguish those truths that are
necessary from those that are not? These two questions are
interrelated since much of our interest in the metaphysical
question would be lost without an answer to the epistemolo-
gical question. Postulating necessary truth would be an
idle exercise if we could not determine which truths are
necessary.

For most of this century, the notion of necessity has
been intertwined with the notion of meaning. According to
this notion, a statement is necessary if and only if its
truth stems solely from the meaning of the words used to
express that statement--if its truth depends upon anything
else, then it is not necessary. This analysis, at the very
least, has the virtue of answering both the metaphysical
and epistemological questions. It tells us that there are
necessary truths--they are just those statements that are
true in virtue of meaning. And it tells us how to distin-
guish necessary truths from other truths--we examine the
role that meaning plays in determining the truth value of
a statement. This examination alone will enable us to
separate those statements that are necessary from those

that are not. This analysis proved quite popular. While

2
it was originally formulated by the logical positivists,
it was soon adopted and advocated by philOSOphers of other
per-suasions as well. And only recently have philosophers
come to admit that this analysis is possibly wrong.

In this chapter, I would like to focus on the logical
positivists' notion of necessary truth. We shall examine
why the logical positivists tied necessity to meaning; and
we shall examine a problem with this analysis. To do this,
we must focus on three distinctions: the analytic/synthe-
tic distinction, the a priori/a posteriori distinction, and
the necessary/contingent distinction. But since the logi—
cal positivists' account of these distinctions owes much to
Leibniz, Hume, and Kant, we should situate their analysis
within some sort of historical perspective because only
then can we understand why the logical positivists tied

necessity to meaning.

Setting the State: Leibniz and Hume

Leibniz and Hume are principally concerned with just
one distinction-~the distinction between knowledge that is
demonstrative and knowledge that is empirical. To make
this distinction, Leibniz contrasts truths of reason with
truths of fact while Hume contrasts relations of ideas with
matters of fact where both truths of reason and relations
of ideas refer to demonstrative knowledge while truths of

fact and matters of fact refer to empirical knowledge.1

3
Since they are concerned with just one distinction, it may
seem rather odd to begin our discussion with Leibniz and
Hume. But the distinctions that concern us-—the analytic/
synthetic, the a priori/a posteriori, and the necessary/
contingent distinctions--originate to a great extent with
Leibniz's and Hume's distinction between knowledge that is
demonstrative and knowledge that is empirical.

While Leibniz and Hume disagree on the origin2 and the
extent3 of demonstrative knowledge, they agree on its foun-
dation. Both believe that demonstrative knowledge is based
upon the principle of contradiction. Leibniz, for example,
argues that

(LD) x is a truth of reason if and only if
its truth is established by the prin—
ciple of contradiction.[4]
while Hume argues that
(HD) x is a relation of ideas if and only
if its truth is established by the
principle of contradiction.[5]
According to Leibniz and Hume, demonstrative knowledge, by
whatever name we choose to call it, owes its truth to the
principle that it is impossible for something to be and not
to be at the same time.

Since demonstrative knowledge is based upon the prin-
ciple of contradiction, it is necessary. By claiming that
demonstrative knowledge is necessary, Leibniz and Hume are

claiming that such knowledge must be true. It is impossi-

ble for such knowledge to be false. To understand the role

that the principle of contradiction plays in guaranteeing
the truth of demonstrative knowledge, consider the follow-
ing statement: 'An equilateral triangle is a triangle'.
Although Leibniz would describe this statement as a truth
of reason while Hume would describe it as a relation of
ideas, they would agree that the statement is necessary.
They would argue that it is impossible for it to be false
since the statement can only be false if there is an equi-
lateral triangle that is not a triangle. Or simply put,
the statement is false only when we have something that is
both a triangle and not a triangle at the same time. But
this would most definitely violate the principle of contra-
diction.

Since demonstrative knowledge is based upon the prin-
ciple of contradiction, it must also be a priori. Accord-
ing to Hume, we discover demonstrative knowledge by means
of reason. He writes

Propositions of this kind [relations of

ideas] are discoverable by the mere op-

eration of thought, without dependence

on what is anywhere existent in the uni-

verse. Though there were never a circle

or a triangle in nature, the truths dem-

onstrated by Euclid would forever re-

tain their certainty and truth. [6]
As this passage shows, demonstrative knowledge is a priori
not only because experience is not needed to determine its
truth but also because its truth does not even depend upon

what exists empirically.

But why would the principle of contradiction lead to

5

a priori knowledge? Traditionally, this principle has been
used to separate the possible from the impossible. What-
ever is in accordance with this principle is possible while
whatever is in violation of this principle is impossible.
Think about a triangle. While it is possible for a tri-
angle to be a triangle, it is impossible for a triangle

not to be a triangle. The former accords with the prin-
ciple of contradiction while the latter violates it. Now
if something is possible, it is thinkable. Similarly, if
something is impossible, it is unthinkable. Whether tri-
angles exist or not, we can certainly think of a triangle
(although we cannot think of a triangle that is not a
triangle). Demonstrative knowledge, which is founded upon
the principle of contradiction, is tied to the notion of
possibility. As such, it is thus tied to the thinkable.
And since it is tied to whatever can be thought, demonstra-
tive knowledge can be discovered by thought alone. In this
explanation, the principle of contradiction acts as a
psychological principle.

But simply because something is possible, this does
not mean that it is actual. While it is entirely possible
for something to be a triangle, this does not mean that
triangles actually exist. Possibility and actuality are
two different things. While demonstrative knowledge con-
cerns the possible, empirical knowledge concerns the ac-

tual. Since the principle of contradiction deals with the

possible and not the actual, empirical knowledge must be
founded on some principle other than that of contradic-
tion. Although empirical knowledge cannot violate the
the principle of contradiction, it cannot be based solely
on it.

When it comes to empirical knowledge, Leibniz and Hume
break company. While they agree that we gain our knowledge
of truths of fact or matters of fact from experience, they
disagree on the basis of this knowledge. Leibniz, on one
hand, argues that

(LE) x is a truth of fact if and only
if it is based upon the principle
of sufficient reason. [7]
Hume, on the other hand, argues that
(HE) x is a matter of fact if and only
if it based on the relation of
cause and effect. [8]
It is at this point that the difference between Leibniz,
the rationalist, and Hume, the empiricist, becomes most
obvious.

Although Hume explicitly states that matters of fact
are based upon the relation of cause and effect, in real-
ity, this comes down to the claim that matters of fact are
based upon experience. Consider the following matter of
fact: 'Caesar is the person who crossed the Rubicon in the
year 49 B.C.'. What is our justification for believing

this? Hume argues that our justification for any matter of

fact always comes back to another fact; and the relation

7

that allows us to connect these facts is that of cause and
effect. But this relation, says Hume, is itself empirical.
He writes that

the knowledge of this relation is not,

in any instance, attained by reason-

ings a priori; but arises entirely

from experience... [14]
Our justification for our belief in the relation of cause
and effect ultimately comes back to experience; in particu-
lar, our experience that one event regularly follows an-
other.

Leibniz concedes to the empiricists that when it comes
to empirical facts, we, as human beings, need experience to
learn these facts. But as a rationalist, he wants to claim
that every truth, truths of fact as well as truths of
reason, can be deduced by means of reason alone. He just-
ifies this claim by arguing that every statement expresses
a relationship between two concepts-—the concept expressed
by the predicate term and the concept expressed by the sub-
ject term. If the statement is true, the concept of the
predicate is contained in the concept of the subject. If
the statement is false, then the concept of the predicate

10 Since

is not contained in the concept of the subject.
the concept of the predicate is contained in the concept
of the subject, we can, in principle, supply an a priori
proof for any true statement, not withstanding its status

as a truth of reason or as a truth of fact.

When it comes to truths of reason, it is fairly easy

8
to see that one concept is contained in the other. Consid-
er the statement 'An equilateral triangle is a triangle'.
It is easy to see that the concept of the predicate term
'triangle' is contained in the concept 'equilateral tri—
angle'. But when it comes to truths of fact, it is not so
easy to see that one concept is contained in the other.
Consider the truth of fact 'Caesar is the person who
crossed the Rubicon in the year 49 B.C.'. The concept
'the person who crossed the Rubicon in the year 49 B.C.'
is not obviously contained in the concept 'Caesar'. But
Leibniz believes that this concept is nonetheless contained
in the concept 'Caesar'. According to Leibniz, "the indi-
vidual notion [concept] of each person includes once for

all everything which will ever happen to him."11

And just
as we can deduce 'triangle' from the concept 'equilateral
triangle' so can we deduce 'the person who crossed the
Rubicon in the year 49 B.C.' from the concept 'Caesar'.
The difference between truths of reason and truths of
fact, according to Leibniz, is not that the former are a
priori while the latter are a posteriori. Rather, the
difference is that truths of reason depend solely upon the
principle of contradiction while truths of fact depend also
upon the principle of sufficient reason. It is this prin-
ciple that enables us to determine what is actual as op-

posed to what is merely possible. Leibniz writes

...we consider that no fact can be
real or existing and no proposition

can be true unless there is a suf-

ficient reason, why it should be

thus and not otherwise... [12]
For anything and everything that is actual, there must be
a reason to explain why this particular possibility, out
of all the various possibilities, became actual. For
anything and everything that has happened to Caesar, for
example, there must be a sufficient reason to explain why
these things happened. If we knew these reasons, we could
deduce everything that happened to Caesar from the mere
concept of 'Caesar'. We could not only deduce that Caesar
crossed the Rubicon in the year 49 B.C., but we could also
deduce the precise point where he crossed the Rubicon and
why he crossed the Rubicon. Since we lack knowledge of
these sufficient reasons, we cannot make these deductions.
We need to rely on experience to learn the truth of 'Cae-

sar is the person who crossed the Rubicon in the year 49

B.C.'. But in principle, we could rely on reason alone.

Setting the Stage: Rant

The logical positivist's primary debt to Kant is that
he is perhaps the first philosopher to recognize that the
analytic/synthetic, a priori/a posteriori, and necessary/
contingent distinctions were distinctions worthy of analy-
sis in their own right. While Leibniz and Hume made use of
the a priori/a posteriori and necessary/contingent distinc-

tions and while Leibniz did everything but identify the

10

analytic/synthetic distinction, their interest in these
distinctions stemmed solely from the standpoint of their
interest in the distinction between demonstrative and
empirical knowledge. It was Kant who gave up one all en—
compassing distinction in favor of three independent dis-
tinctions.

Kant's version of the analytic/synthetic distinction
owes its genesis to Leibniz. With Leibniz, Kant believes
that every statement expresses a relationship between the
concept of its predicate term and that of its subject.
Furthermore, Kant agrees with Leibniz that this relation—
ship takes one of two forms. Either the concept of the
predicate belongs to the concept of the subject or it does
not. But while Leibniz believes that in the case of true
statements, the concept of the predicate always belongs to
the concept of the subject, Kant does not. Even in the
case of true statements, the relationship between subject
and predicate can take one of these two forms. And depend-
ing upon which form it takes, the statement is either
analytic or synthetic. Kant suggests that

(KA) A statement is analytic if and only
if the concept of the predicate is
is contained in the concept of the
subject.

(KS) A statement is synthetic if and only
if the concept of the predicate is
not contained in the concept of the

subject. [12]

Analytic statements, according to Kant, depend solely upon

11

the principle of contradiction. Since the concept of the
predicate is contained in the subject, we cannot deny the
truth of an analytic statements without violating this
principle. Synthetic statements, in contrast, do not
depend solely upon the principle of contradiction. Al-
though synthetic statements may not violate this principle,
they cannot be founded upon it, for as Kant says

...it is evident... that in synthetic

judgments I must have besides the con-

cept of the subject something else (X),

upon which the understanding may rely,

if it is to know that a predicate, not

contained in this concept, nevertheless,

belongs to it. [14]
Up to this point, the discussion may seem vaguely remini-
scent of the discussion in the preceding section. Analytic
statements, like demonstrative knowledge, depend upon the
principle of contradiction while synthetic statements, like
empirical knowledge, depend upon something else.

The difference is that Kant does not stick with just
one distinction but goes on to present two additional
distinctions. Although Leibniz and Hume certainly presup-
pose the existence of the a priori/a posteriori distinc-
tion, Kant is careful to define it. He argues that

(KP) x is a priori if and only if it
meets one of the following con—
ditions:

(a) x can be known independently
of all experience; or
(b) x can be known prior to ex—

perience.

(KT) x is a posteriori if and only
if knowledge of x is dependent

12

on experience. [15]

Kant's notion of the a posteriori should sound familiar
since it is the same notion that we found with Leibniz and
Hume; and since it will be the same notion that we will
find with the logical positivists. His notion of the a
priori, on the other hand, is not so familiar.

Kant's provides us with two conditions for (KP).
These conditions correspond to two different senses for the
phrase 'a priori'. The first condition corresponds to the
epistemological sense. When we use 'a priori' in this
sense, we are asserting that experience is irrelevant. All
that is needed is reason. The second condition corresponds
to the "psychological" sense; where by the term 'psycholo—
gical', we mean to refer to certain sorts of mental pro-
cesses. Experience is not irrelevant to these processes.
But we say that these processes are prior to experience in
that our experience, from the very beginning, is structured
according to these processes.

This "psychological" element becomes even more pro-
nounced when we examine Kant's version of the necessary/

contingent distinction. Early in The Critique of Pure

 

Reason, he presents this distinction as thus:

(KN) x is necessary if and only if x
meets both the following condi-
tions:

(a) x holds with strict univer-
sality; and
(b) x must be the case.

(KC) x is contingent if and only if

13

x meets the following condi—

tions:

(a) x does not hold with strict

universality; or

(b) x may not be the case. [16]
By framing the necessary/contingent distinction in terms of
strict necessity, Kant contrasts universal statements that
must be true with universal statements that merely happen
to be true. Consider the following two statements:

(1) 5+7=12.
(ii) Every president of the United States

during the 19th century was a white

male.
Both (i) and (ii) are universally true; but only (1) holds
with strict universality. While we believe that (i) is
necessary, we believe that (ii) is contingent; and while we
believe that (i) must be true, we believe that (ii) merely
happens to be true. If we examine the class of presidents
of the United States during the 19th century, we will
discover that it is made up entirely of white males; but
even so, we do not think that it had to consist solely of
white males.

But how can we distinguish between those statements
that hold with strict universality from those that simply
hold universally? We can answer this question by examining
the relationship between the necessary/contingent distinc-
tion with the a priori/a posteriori distinction. Kant

argues that

(K1) A statement is necessary if and
only if it is a priori.

14

(K2) A statement is contingent if and
only if it is a posteriori.

Kant equates necessity with the a priori because he agrees
with Hume that "experience can teach us that a thing is so

and so, but not that it cannot be otherwise."17

Experience
may be sufficient to prove that a statement is universally
true; but it is not sufficient to show us that that state—
ment is strictly universal. If we believe that a statement
is strictly universal, we must do so on a priori grounds.18
Reconsider statements (1) and (ii). We accept the truth of
(ii) on a posteriori grounds. Thus, it is contingent. But
in the case of (i), we argue for its truth on a priori
grounds; and that, according to Kant, is the reason why (i)
is necessary.

Kant also argues for the following:

(K3) If a statement is analytic, then
it is a priori.

Since the concept of the predicate term is included within
the concept of the subject term, we need not consider
anything beyond the concepts used within that statement.

In particular, we need not rely on experience.19 From (K1)
and (K3), we can infer

(K4) If a statement is analytic, then
it is necessary.

Given that analytic statements depend upon the principle of
contradiction, and given what we learned about this princi-
ple in the preceding section, (K4) is exactly what we would

expect.

15

Not too surprisingly, Kant also argues for the follow-
ing:

(K5) If a statement is a posteriori, then
it is synthetic.

We have already noted that Kant argues that we need some—
thing, some (X), to unite the concepts of the predicate and
the subject for a synthetic statement. In the case of a
posteriori statements, this (X) is experience.

Although Kant argues that every analytic statement is
both a priori and necessary, he does not argue that every
synthetic statement is both a posteriori and contingent.

In fact, Kant refuses to equate synthetic statements with
a posteriori statement. He argues instead that

(K6) If a statement is synthetic, then it
it is either a priori or a posteriori.

(K7) If a statement is a priori, then it
is either analytic or synthetic.

Kant believes that certain statements are both synthetic
and a priori. And among such statements are the state-
ments of mathematics and various metaphysical principles.
Let us once more consider statement (1). Since this
statement depends upon the principle of contradiction, we
may be inclined to argue that it is a priori and analytic.
But Kant argues that it is synthetic. While the statement
'5+7=12' does not violate the principle of contradiction;
nevertheless, the concept of '12' is not contained in the
concept of the sum '5+7'; and thus, the statement is syn-

thetic. Reasoning in this manner, Kant writes that

16
Arithmetical propositions are therefore
always synthetic. This is still more
evident if we take larger numbers. For
it is then obvious that, however we
might turn and twist our concepts, we
could never, by the mere analysis of
them, and without the aid of intuition,
discover what is the sum. [20]

Kant also believes that certain "metaphysical" prin—
ciples are also synthetic and a priori. For example, the
principle 'Every effect has a cause', according to Kant,
belongs in this category. (This conclusion has not been
widely adopted. Those philosophers who agree with Kant
that the principle is a priori argue, however, that it is
a priori because it is analytic--the concept of 'cause' is
contained in the concept of 'effect'-—while those philoso-
phers who agree with Kant that the principle is synthetic
argue that it is synthetic because it is a posteriori--it
is a principle that is justified by experience.) Kant
concedes that statements that are both synthetic and a
priori are somewhat problematic. Since they are synthetic,
we need something to unite the concepts of the predicate
and subject. But since they are a priori, this something
cannot be experience. But if it is not experience, what
can it be?

It is at this point where Kant begins to rely upon
his second sense of 'a priori'--the psychological sense.
Kant argues that we bring certain a priori intuitions and

concepts to particular statements. In the case of mathema-

tical statements, we bring our a priori intuitions of space

17

and time. In the case of metaphysical principles, we bring
certain a priori concepts--the categories. It is these a
priori intuitions and concepts that enable us to unite the
subject and predicate within mathematical and metaphysical
statements. But Kant does not believe that these intui—
tions and concepts are irrelevant to experience. Rather,
it is these intuitions and concepts that structure our

experience.

The Stage is Set: The Logical Positivists

The logical positivists owe a great deal to the work
of Kant and to the work of Leibniz and Hume. Like Leibniz
and Hume, the logical positivists essentially have one
overall distinction-—the distinction between knowledge that
is demonstrative and knowledge that is empirical. But they
analyze this distinction in terms of Kant's three distinc-
tions. To distinguish demonstrative knowledge from empiri-
cal knowledge, they rely upon the analytic/synthetic dis-
tinction, the a priori/a posteriori distinction, and the
necessary/contingent distinction.

While the logical positivists credit Kant for recog-
nizing the importance of the analytic/synthetic distinc-
tion, they do not adopt his version of the distinction.
Instead, they claim that:

(LA) A statement is analytic if and only

if its truth value is determined com-
pletely by linguistic conventions.

18
(LS) A statement is synthetic if and only

if its truth value is not determined

completely by linguistic conventions.
These linguistic conventions, in turn, are governed by the
principle of contradiction. Hence, analytic statements
are governed entirely by the principle of contradiction
while synthetic statements are not.

Examples of analytic statements include the following:

(a) A bachelor is an unmarried man.

(b) P v -P.

(c) 5+7=12.
According to the logical positivists, the truth value of
each statement is determined by various linguistic conven-
tions. The truth of (a) is determined by the conventions
that govern the use of English. The truth of (b) is deter-
mined by the conventions that govern the use of logical
connectives in an appropriate propositional calculus. The
truth of (c) is governed by the conventions that govern
arithmetic. In each case, we cannot deny the truth of
these statements without also violating the principle
of contradiction.

As we can see, the logical positivists waste little
time breaking with Kant. While Kant believed that the
mathematical statements were synthetic, the logical posi—
tivists believe that they are analytic. Ayer argues that
Kant relied on two different criteria when he formulated

his analytic/synthetic distinction. One criterion was

logical--the principle of contradiction--and it is this

19

criterion that the logical positivists continue to accept.
The other criterion was psychological--whether the concept
of the predicate was thought in the concept of the subject
or not--and it is this criterion that the logical positi-

vists reject.

We can contrast statements (a)-(c) with the following
synthetic statements:

(d) Bachelors are neurotic.

(e) Jones knows that 'P v —P' is true.

(f) This board is seven feet long.
To determine the truth value of these statements, we must
go beyond linguistic conventions. Bachelors may very well
be neurotic; but we cannot discover this fact simply by
examining the conventions that govern the use of the term
'bachelor'. And while 'P v -P' is true, we cannot deter-
mine whether Jones knows that it is true by linguistic
analysis. And the only way we can know that a particular
board is seven feet long is by measuring it. Since lin-
guistic conventions are not sufficient to tell us the
truth values of statements (d)-(f), these statements are
synthetic.

The logical postivists' version of the a priori/a
posteriori distinction also differs from Kant's version.
They claim that:

(LP) A statement is a priori if and only
it can be known independently of ex-

perience.

(LT) A statement is a posteriori if and
only if knowledge of it is dependent

20

on experience. [15]
The logical positivists only worry about Kant's epistemolo-
gical sense of the phrase 'a priori'. Since they reject
Kant's critical project, they have no interest in the
psychological sense. They have no interest in the a pri-
ori conditions that structure experience. Or to be more
accurate, they believe that the conditions that structure
experience are empirical and can be studied in psychology.
When it comes to the necessary/contingent distinction,
the logical positivists again present us with the familiar
version. They argue that
(LN) A statement is necessary if and only if
it must be true--it is impossible for it
to be false.
(LC) A statement is contingent if and only if
it may be false--it is possible for it to
be false.
This is, of course, the same version of necessity that we
first encountered with Leibniz and Hume.
Like Kant, the logical positivists equate necessity

with the a priori. They argue that:

(L1) A statement is necessary if and only
if it is a priori.

(L2) A statement is contingent if and on-
ly if it is a posteriori.

The logical positivists, like Kant before them, are re-
sponding to Hume's scepticism. Ayer, for example, admits
that

...it is impossible on empirical prin-
ciples to account for our knowledge of

21

necessary truths. For, as Hume conclu-

sively showed, no general proposition

whose validity is subject to the test

of actual experience can ever be logi-

cally certain. No matter how often it

is verified in practice, there still re—

mains the possibility that it will be

confuted on some future occasion. [21]
But the logical positivists also believe that necessary
knowledge is a priori because we never allow experience to
falsify such knowledge. Ayer writes that

Whatever instance we care to take, we

shall always find that the situations

in which a logical or mathematical prin-

ciple might appear to be confuted are

accounted for in such a way as to leave

the principle unassailed. ...The prin-

ciples of logic and mathematics are true

universally simply because we never al-

low them to be anything else. [22]
Consider the statement '5+7=12'. Suppose we pushed a pile
of 5 pebbles into a pile of 7 pebbles and discovered that
the resulting pile was something other than 12 pebbles?
Would we then conclude that the principle '5+7=12' has been
falsified. Of course not. We would conclude that we had
miscounted the number of pebbles in one of our original
piles. We do not allow '5+7=12' to be falsified by any
sort of experience. But if we do not allow this principle
to be falsified by experience, then we must not verify it
by experience.

So, how do we verify the truth of a statement such

as '5+7=12‘? The logical positivists argue that:

(L3) A statement is a priori if and only
if it is analytic.

22

To justify this claim, the logical positivists point out
that apart from what they tell us about linguistic conven-
tions, analytic statements have no empirical content what-
soever. And since they lack empirical content, we do not
use experience to determine their truth or falsify. The
statement '5+7=12' owes its truth to the conventions that
govern our use of arithmetic. And this explains why ex-
perience cannot falsify '5+7=12'.

From (L1) and (L3), we can infer that

(L4) A statement is necessary if and only
if it is analytic.

The fact that we can determine the truth or falsity of
analytic statements merely by examining linguistic conven—
tions also explains why these statements are necessary.
Ayer writes

They [analytic statements] simply record

our determination to use words in a cer-

tain fashion. We cannot deny them with-

out infringing the conventions which are

presupposed by our very denial, and so

falling into self-contradiction, and that

is the sole ground of their necessity. [23]
By tying necessity to analyticity, the logical positivists
tie necessity to the principle of contradiction. We vio-
late this principle whenever we violate the conventions
that we have chosen to adopt. And this explains why analy—
tic statements are necessarily true.

Running parallel to these claims are the following:

(L5) A statement is a posteriori if and
only if it is synthetic.

23

(L6) A statement is contingent if and
only if it is synthetic.

Since we cannot determine the truth value of synthetic
statements by linguistic analysis, how can we determine
their truth value? The logical positivists argue that

the truth value of synthetic statements is determined by
experience. Synthetic statements, according to the logical
positivists, make claims about the empirical world; and
thus, it is only fitting that experience verifies or fal—
sifies these statements. But as we have already noted,
experience cannot guarantee necessity. Thus, these state-
ments will all be contingent. This applies even to "meta-
physical" statements, such as 'Every effect has a cause'.
We accept this statement only in so far as it continues

to be corroborated by experience.

Thus, we can see that the logical positivists only
permit two classes of statements: we have those statements
that are analytic, a priori, and necessary and we have
those statements that are synthetic, a posteriori, and
contingent. Demonstrative knowledge consists of those
statements that belong to the former class while empirical
knowledge consists of statements that belong to the latter.
Moreover, we can now understand precisely how necessity has
been tied to meaning. Necessity is equivalent to analyti-

city.

24
Problems

An assumption that underlies the distinctions that we
have examined is that these distinctions are exclusive and
exhaustive. Every statement is either analytic or synthe—
tic, a priori or a posteriori, necessary or contingent.
The purpose that underlies Kant's and the logical positi-
vists' analyses is to explain why a particular statement
falls into one category as opposed to another. But consi-
der the following three statements:

(1) Heat is the form of energy constituted
by the motion of atoms and molecules.
(2) Gold is the element with atomic number

79.
(3) Water is H O.

2
How should we categorize these statements? Are they analy-
tic or synthetic? A priori or a posteriori? Necessary or
contingent? As we shall soon see, the analyses provided by
Kant and the logical positivists give us great leeway in
categorizing these statements.

By using the analysis provided by the logical positiv-
ists, we can argue that these statements are analytic, a
priori, and necessary. If we look in any reasonably decent
dictionary, we will discover that these statements are the
definitions for 'heat', 'gold', and 'water'. Thus, we
could say that statements (1)-(3) are true by definition.
But if they are true by definition, then they express the
linguistic conventions that govern the way we choose to

talk about heat, gold, and water. But this means that

25
statements (1)-(3) are analytic. But if they are analytic,
then they are also a priori and necessary.

But we can also use the logical positivists' analysis
to argue that statements (1)-(3) are synthetic, a poster-
iori, and contingent. We did not originally accept the
truth of statements (1)-(3) by means of linguistic analy—
sis. The fact of the matter is that it was an empirical
discovery that heat is a form of energy associated with
the motion of atoms and molecules. It was an empirical
discovery that gold has an atomic number of 79. It was an
empirical discovery that water is H20. The truth of state—
ments (1)-(3) was settled empirically, not by an examina-
tion of our linguistic conventions. Thus, statements (1)-
(3) are a posteriori. And if they are a posteriori, then
they are also synthetic and contingent.

But while we may be quite willing to concede that
statements (1)-(3) are a posteriori, we should not be quite
so willing to concede that they are contingent. Heat oc-
curs if and only if there is atomic or molecular motion.
Something is gold if and only if it has atomic number 79.
Something is water if and only if it is H20. We do not
believe that statements (1)-(3) express mere correlations.
We tend to believe that if our scientific theories are
correct, then these statements are necessary.

The situation is no better if we use Kant's analysis

rather the logical positivist's. We can use Kant's

26

analysis to argue that statements (1)—(3) are analytic, a
priori, and necessary. Our concept of heat includes the
concept of a form of energy constituted by the motion of
atoms and molecules. Our concept of gold includes the
concept of an element with atomic number 79. Our concept
of water includes the concept of H20. Thus, by Kant's
criterion, statements (1)-(3) are analytic. And since
they are analytic, they are also a priori and necessary.
But once again, we can also use Kant's analysis to
argue that statements (1)—(3) are synthetic, a posteriori,
and contingent. The fact remains that it was an empirical
discovery that heat is a form of energy constituted by the
motion of atoms and molecules, that gold is an element with
atomic number 79, that water is H20. Given the state of
chemistry in his day, Kant undoubtedly did not include the
concept of atomic or molecular motion within his concept of
heat, or the concept of the atomic number 79 within his
concept of gold, or the concept of H20 in his concept of
water. We could argue that Kant has a different conception
of heat, gold, and water than we do. We could say that the
concepts of heat, gold, and water have changed since Kant's
day. But such a suggestion overlooks the fact that these
changes were driven by empirical findings. The fact of the
matter is that the truth of statements (1)-(3) was settled
empirically and not by examining our concepts. Thus, we

can argue that statements (1)-(3) are a posteriori; and

27
since they are a posteriori, they are also synthetic and
contingent.

But once again, we may be unwilling to describe state-
statements (1)—(3) as contingent. Our concept of water,
for example, includes a great many other concepts. It in-
cludes the concept of H20; but it also includes the concept
of a liquid that boils at 100°C and freezes at 0°C; it in—
cludes the concept of a colorless and odorless liquid. Yet
while our concept of water may include concepts other than

H O, we feel that the concept of H20 is somehow more impor—

2
tant than these other concepts.

If we take Kant's and the logical positivists' analy-
ses seriously, then we can categorize statements (1)-(3) as
either analytic, a priori, and necessary or as synthetic, a
posteriori, and contingent. In either case, we go against
certain intuitions that we have about these statements. On
the one hand, we believe that statements (1)—(3) provide
the definitions for 'heat', 'gold', and 'water'. But on
the other hand, we established these definitions empiri-
cally. We believe that these statements are necessary; but
we also believe that they are a posteriori. But every
analysis of necessity that we have examined equates neces-
sity with the a priori.

Why are statements (1)-(3) so difficult to categorize?
The problem is that although we may think that that these

statements are definitions, we do not think that they

28

simply express linguistic conventions. Intuitively, we
feel that these statements express the essence of their
subject; and it is for this reason that we accept these
statements as definitions. Aristotle, for example, argues
that there are two types of definitions. One type gives
the essential nature of its subject while the second type
is only a nominal expression. The first type expresses
the inherent nature of a thing while the second merely

expresses the meaning of a word.24

These two types of
definitions are related. We may use the essential nature
of a thing as the meaning of the word that designates that
thing. For example, the essential nature of water may be
H20. We may then end up using 'HZO' as the definition of
'water'. But we have chosen to define 'water' as 'HZO'
because H20 is the feature that constitutes the inherent
constitution or essence of water. We have reason for
defining 'water' in a particular way; but this definition
is not the result of linguistic convention. Rather, our
linguistic conventions, in the case of 'water', follows
our empirical discoveries.

Now we could bite the bullet, as many logical positi-
vists (and possibly Kant) would recommend, and insist
that statements (1)—(3) are synthetic, a posteriori, and
contingent. The only reason why we think statements (1)-

(3) are necessary is because, intuitively, we feel that

they express the essence of their subject. But perhaps

29

this intuition tells us less about statements (1)—(3) than
it tells us about ourselves. By feeling that statements
(1)-(3) express the essence of their subjects, we have a
certain attitude towards these statements. But an atti-
tude is hardly sufficient to describe these statements as

necessary.

Chapter Two
If we rely on the criteria provided by the logical
positivists, then we have considerable leeway when it
comes to categorizing the following statements:
(1) Heat is the form of energy constituted
by the motion of atoms and molecules
in solids.
(2) Gold is the element with atomic number

79.
(3) Water is H O.

2
On one hand, we can argue, and we can argue persuasively
at that, that these statements are analytic, a priori, and
necessary. But on the other hand, we can argue, and we can
argue just as persuasively, that these statements are syn-
thetic, a posteriori, and contingent. Needless to say, the
logical positivists did not intend their criteria to be
quite so flexible as this.

In this chapter, we shall examine the causal theory of
reference, a theory of meaning that has been proposed by
Kripke and Putnam. Proponents of this theory argue that
statements (1)-(3) are both necessary and a posteriori.
This is a rather convenient claim. It not only takes into
account our intuitive convictions that these statements
express something essential about the nature of their sub-
jects but also the fact that the truth of these statements
was established empirically. But there is a problem. Al-
though the causal theorists may be correct in claiming that
statements (1)-(3) are both necessary and a posteriori,

their justification for this claim leaves something to be

30

31

desired.

Modifying the Traditional Distinctions

Surprisingly enough, the causal theory is reminiscent
of logical positivism. This is because the causal theor-
ists draw the analytic/synthetic, a priori/a posteriori,
and necessary/contingent distinctions in much the same way
as the logical positivists. They accept these distinc-
tions with only a few minor modifications. Yet it is these
modifications, minor as they may appear, that enable the
the causal theorists to reject the conclusion that neces-
sary statements must be a priori.

When it comes to the analytic/synthetic distinction,
for example, the causal theorists agree with the logical
positivists that while analytic statements owe their truth
to linguistic conventions, synthetic statements do not.
But the causal theorists undermine the importance of this
distinction by two moves. First, they deny that this
distinction is exhaustive. Putnam, for instance, argues
that while some statements are analytic and others are
synthetic, many are neither.1 Second, since we decide
whether a statement is analytic or synthetic by examining
the meaning of the signs used within that statement, the
causal theorists argue that this distinction is essentially
a semantic distinction. And since it is a semantic dis-

tinction, the causal theoriests restrict the analytic/

32

synthetic distinction to semantics.2

When it comes to the a priori/a posteriori distinc-
tion, the causal theorists agree with the logical positi-
vists that while a priori statements are independent of
experience, a posteriori statements are not. But just as
they restrict the analytic/synthetic distinction to a
particular area so do they also restrict the a priori/a
posteriori distinction. Since this distinction concerns
the grounds and nature of knowledge, the causal theorists
argue that it is fundamentally an epistemological distinc-
tion. And since it is an epistemological distinction,
the causal theorists restrict the a priori/a posteriori
distinction to epistemology.3

When it comes to the necessary/contingent distinction,
the causal theorists agree with the logical positivists
that a statement is necessary if it is impossible that it
is false and it is contingent otherwise. But once again,
they restrict this distinction. The causal theorists
argue that this distinction is basically a metaphysical
distinction. And since it is a metaphysical distinction,
the causal theorists restrict the necessary/contingent
distinction to metaphysics.4

By restricting these distinctions to particular areas,
the causal theorists have, in effect, divorced the neces-
sary/contingent distinction from both the a priori/a pos-

teriori distinction and the analytic/synthetic distinction.

33
The causal theorists argue that these three distinctions
are independent. And since these distinctions are indepen-
dent, we cannot conclude anything about a statement's sem-
antical or epistemological status on the basis of its
metaphysical status. Simply because a statement is nec-
essary, we cannot conclude that it is also analytic and a
priori. And thus, we cannot rule out the possibility that

certain statements are both necessary and a posteriori.

Proper Names and Necessary A Posteriori Statements
The causal theorists identify two classes of state-
ments that are both necessary and a posteriori. The first
class, as we have already noted, consists of statements
such as
(1) Heat is the form of energy constituted
by the motion of atoms and molecules
in solids.
(2) Gold is the element with atomic number
79.
(3) Water is H20.
The second class consists of statements such as
(4) Currer Bell is Charlotte Bronte.
(5) Jolmo Lungma is Mt. Everest.
(6) The Malvinas Islands are the Falk-
land Islands.
The causal theorists believe that these two classes are
analogous. Statements (1)-(3) are necessary for much the
same reason that statements (4)-(6) are necessary. In

fact, to understand just why statements (1)-(3) are neces-

sary, we need to first understand why statements (4)-(6)

34
are necessary.

Since it is clear that we cannot verify the truth of
statements (4)-(6) by the use of reason alone or by lin-
guistic analysis, there should be little debate over the
claim that these statements are a posteriori. In order to
know that Currer Bell is Charlotte Bronte, we need some
empirical evidence like an admission from Bronte herself
or from some member of her publishing company. To know
that Jolmo Lungma is Mt. Everest, we need to ask both
Tibetans and non-Tibetans the name of a particular moun-
tain. To know that the Malvinas Islands are the Falkland
Islands, we need to ask both the Argentine government and
the British government the name of two small islands in the
Atlantic Ocean. In each case, we need some sort of empiri-
cal evidence.

While it is obvious that statements (4)-(6) are a pos-
teriori, it is less obvious that they are necessary. As a
matter of fact, we may even be tempted to argue that these
statements are contingent. It is certainly contingent that
Charlotte Bronte chose to publish her first novel under the
pseudonym 'Currer Bell'. It is certainly contingent that
the Tibetans chose to name a particular mountain Jolmo
Lungma while the British chose to name it Everest. It is
certainly contingent that while the Argentines chose to
call two islands by one name, the British chose another.

All this seems to suggest that statements (4)-(6) are

35

contingent. But the causal theorists argue that such a
suggestion is mistaken. Statements (4)-(6), contrary to
what we may think, are actually necessary.

To understand why the causal theorists believe that
these statements are necessary, we need to first understand
their analysis of proper names. For much of this century,
the issue of proper names came down to a choice between two
theories. There was Russell's theory and there was Witt—

genstein's. Kripke, in Naming and Necessity, argues that

 

while these theories pose as competitors, they are, for all
intents and purposes, the same theory. They are both de—
scriptive theories of meaning. While Russell argues that a
name is an abbreviation for a definite description, a de-
scription that holds only for the referent of that name,
Wittgenstein argues that a name is an abbreviation for a
cluster of descriptions, some of which may be false but of
which a sufficient number hold for the referent of that
name. They may disagree on the number of required descrip-
tions, but both Russell and Wittgenstein agree that a name
is an abbreviation for some number of descriptions.
Consider the name 'Aristotle'. There are any number
of definite descriptions associated with this name: 'the
last great philosopher of antiguity', 'the most famous
pupil of Plato', 'the teacher of Alexander', 'the author

of the Metaphysics'. According to Russell, the name 'Aris-

 

totle' is simply an abbreviation for some description of

36

this type; and it is the description that gives 'Aristotle'
its meaning. According to Wittgenstein, 'Aristotle' is not
the abbreviation for some single description but for a
cluster of descriptions; and it is this cluster that gives
'Aristotle' its meaning. In either case, the meaning of
the name is due to description.

In Naminggand Necessity, Kripke argues that the prob-

 

lem with such an account is that it fails in counterfactual
situations. Suppose 'Aristotle' is an abbreviation for the
definite description 'the last great philosopher of anti-
quity'. Now suppose that Aristotle had chosen medicine
over philosophy. He certainly could have made such a
choice. But ifAristotle had chosen a career in medicine,
the description 'the last great philosopher of antiquity'
would refer to someone other than Aristotle; so if the name
'Aristotle' is an abbreviation for 'the last great philoso—
pher of antiquity', then the name 'Aristotle' no longer
refers to Aristotle. But our intuitions are very clear
that 'Aristotle' would still refer to Aristotle. When we
say that Aristotle might have gone into medicine, we are
specifically using 'Aristotle' to refer to Aristotle in a
counterfactual situation where Aristotle is not the last
great philosopher of antiquity.

Nor are things much better if we rely upon a cluster
of descriptions rather than a definite description. Not

only can we conceive of a counterfactual situation where

37
Aristotle chose medicine over philosophy but we can also
conceive of situations where he was not the most famous
pupil of Plato, where he was not the tutor of Alexander,

where he did not write the Metaphysics. As Kripke points

 

out, we can conceive of a counterfactual situation for
almost any description that applies to Aristotle. In

such a case, if the cluster of descriptions refer to any-
one, it is not Aristotle. And once again, the name 'Aris-
totle' no longer refers to Aristotle, contrary to our
intuitions.

The major presupposition that underlies the descrip-
tive theory of meaning is that the sense of a name, which
is given by either a definite description or a cluster of
descriptions, constitutes its meaning. It is this presup-
position, according to the causal theorists, that is false.
Kripke argues that names do not obtain their meaning from
their sense. (In fact, Kripke does not even believe that
names have a sense.) He argues that names, instead, obtain
their meaning from their reference. The name, 'Aristotle',
for example, obtains its meaning by a certain causal con-
nection to Aristotle. 'Aristotle' is nothing more than the
tag we use to refer to a particular person. And it is the
person that gives the name its meaning. This is not to say
that description is pointless. Descriptions may be used to
fix a reference. For example, if someone were to ask me

who Aristotle was, I might very well reply that Aristotle

38

was the last great philosopher of antiquity. After all, it
is easier to supply a definition than it is to supply Aris-
totle. But fixing a reference is not the same thing as
fixing a meaning.

This account of meaning has the advantage over the
descriptive theory in that it has no problems with counter-
factual situations. Since names are tags that we use to
refer to particular entities, we can continue to use them
in counterfactual situations. Kripke argues that names are
rigid designators. They designate, or pick out, the same
entities in every counterfactual situation or in every pos-
sible world. The name 'Aristotle' not only allows us to
pick out Aristotle in this world but in every possible
world (or at the very least, in every possible world where
Aristotle exists). Descriptions, on the other hand, are
not rigid designators. The description 'the last great
philosopher of antiquity' may pick out Aristotle in this
world; but it does not pick out Aristotle in every possible
world.

It is the fact that names are rigid designators that
explains why statements (4)-(6) are necessary. Consider
(4). The name 'Charlotte Bronte' picks out the same entity
in every possible world. 'Currer Bell', although a pseudo-
nym, is also a name; and as such, it too picks out the same
entity in every possible world. In this world, 'Charlotte

Bronte' and 'Currer Bell' designate the same object. But

39

if they designate the same object in this world, and since
they are rigid designators, they must designate the same
object in every possible world. Thus, the statement 'Cur-
rer Bell is Charlotte Bronte', if true in this world, is
true in every possible world. This means that it is impos-
sible for statement (4) to be false--it must be true; and
thus, (4) is necessary. Similar arguments can be made for
statements (5) and (6). In general, Kripke argues that if
we have an identity statement 'R1=R2', where R1 and R2

rigid designators, then if that statement is true, it is
5

are

necessarily true.

Kripke argues that statements (1)-(3) are analogous to
statements (4)-(6). These statements, too, are both neces-
sary and a posteriori. There should be little debate over
the claim that (1)—(3) are a posteriori. It was an empiri-
cal discovery, after all, that heat is caused by the motion
of atoms and molecules. It was an empirical discovery that
gold is the element with atomic number 79. It Was an em-
pirical discovery that water is H20.

The causal theorists argue that the same sort of
analysis that proved the necessity of statements (4)-(6)
also proves the necessity of statements (1)-(3). The
causal theorists argue that natural kind terms, terms such
as 'heat', 'gold', and 'water', are analogous to proper

names. In fact, we should think of natural kind terms as

the names for natural kinds. And just as proper names

40

obtain their meaning from their reference so too do the
terms for natural kinds obtain their meaning from their
reference. 'Water', for example, obtains its meaning by

a certain causal connection to water. 'Water' is nothing
more than the tag we use to designate a particular kind of
substance. And it is this substance that provides 'water'
with its meaning.

Since natural kind terms are analogous to proper
names, we can now understand why statements (1)-(3) are
necessary. Natural kind terms, like proper names, are
rigid designators. While a proper name designates the same
entity in every possible world, a natural kind term desig—
nates the same substance (or phenomena or species) in
every possible world. Kripke argues that statements (1)-
(3), as theoretical identities, involve two rigid designa—
tors. For example, consider (3). As a natural kind term,
'water' designates the same substance in every possible
world. While 'HZO' is not exactly a name for a natural
kind, it is a name for a chemical kind; and as such, it,
too designates the same kind in every possible world. In
this world, 'water’ and 'H20' designate the same substance.
Since they designate the same substance in this world and
since they are rigid designators, they designate the same
substance in every possible world. This means that the
statement 'Water is HZO', if it is true in this world,

(which it is), is true in every possible world. Thus, we

41

may conclude that statement (3) is necessary. And accord-
ing to Kripke, similar arguments can be made for (1) and
(2).6

But at this point, we may have a few niggling doubts
about this account. We may agree with Kripke that natural
kind terms obtain their meaning from their reference. We
may agree that natural kind terms are rigid designators.

We may agree that statements (1)-(3) are both necessary and
a posteriori. We may even agree that the statement 'Water
is HZO' is necessary for exactly the reason that Kripke
gives. But we soon run into one little problem. The
phrases 'the form of energy constituted by the motion of
atoms and molecules in solids' and 'the element with atomic
number 79' bear a suspicious resemblance to definite de-
scriptions. In fact, the more cynical among us may even
claim that 'HZO' is actually a definite description rather
than a name of a chemical kind. But this makes it rather
difficult to argue that statements (1)-(3) owe their neces-
sity to the fact that we have two rigid designators that
name the same substance.

This suggests that statements (1)-(3) are not entirely
analogous with statements (4)-(6). This does not imply
that statements (1)-(3) are therefore contingent; but it
does imply that if these statements are necessary, their
necessity hinges on grounds other than those for (4)-(6).

This should not surprise us. If we compare statements

42

(4)-(6) with statements (1)—(3), we intuitively feel that
the latter are, in some sense, necessary while the former
are, in some sense, contingent. This is because we feel
that statements (1)-(3) express something essential about
the nature of their subjects while (4)-(6) do not. However
necessary statements (4)-(6) may be, the fact remains that
they are also trivial. They say nothing of any importance

about the nature of their subjects.

Underlying Assumptions
Although the causal theory provides an account of

natural kind terms, it does not provide an account of
natural kinds. But why is this important? Basically, it
is important because while natural kind terms may be alike,
natural kinds are not. Hull, for example, writes that

From the beginning, a completely satis-

factory explication of the notion of a

natural kind has eluded philosophers.

One explanation for this failure is that

the traditional examples of natural kinds

in the philosophical literature have been

geometric figures, biological species,

and physical elements. By now it should

be clear that all three are very differ-

ent sorts of things. No wonder a gener-

al analysis, applicable equally to all

of them, has eluded us. [7]
But so long as natural kind terms are alike, why should we
care if natural kinds are not? There is no reason why we
cannot agree with Hull that there are different sorts of
natural kinds while we still agree with the causal theor-

ists that natural kind terms obtain their meaning from

43
their reference. Only the issue of meaning is relevant for
the causal theory. Or is it?

Let us assume that Hull is correct. Let us assume
that there are different sorts of natural kinds. And let
us assume further that these different sorts are the sorts
that Hull identified--geometric figures, biological spe—
cies, and physical elements. Now consider the following
statements:

(a) The triangle is a plane figure with
an area enclosed by three straight
lines.
(b) The snow rose is the Hellaborus niger.
(c) Gold is the element with atomic number
(d) Carrer Bell is Charlotte Bronte.
We already noted that the causal theorists believe that (c)
and (d) are necessary for substantially the same reason. We
also noted that this claim is incorrect. But perhaps the
causal theorists merely chose the wrong example. By keep-

ing in mind that natural kinds are not alike, we can devel-

op a slightly different analysis:

 

 

essence non-essence

a priori (a) (b)
The triangle is a plane The snow rose is the
figure with an area en- Hellaborus niger.
closed by three straight
lines. 1

a poster- (c) (d)

iori Gold is the element with Currer Bell is Char-
.atomic number 79. . lotte Bronte.

 

 

 

figure 1

44
The advantage of this chart is that it gives a more accur-
ate picture of which statements are analogous and for what
reasons. While the vertical axis says something about the
epistemic status of our four statements, the horizontal
axis says something about the source of their necessity.
As we can see, while statements (c) and (d) are analogous
in that they are both a posteriori, they are not analgous
in the source of their necessity.

When it comes to the source of their necessity, state-
ments (a) and (c) are analogous. Both (a) and (c) owe
their necessity to the fact that they express the essence
of their subjects. (Granted, we may believe that (a), as
the definition of 'triangle' only expresses a nominal es-
sence while (c) expresses an Aristotelian essence; but for
the time being, let us treat (a) and (c) as though they are
completely analogous in this respect.) Both (a) and (c)
express the essence of their subjects in terms of a defin-
ite description. Since many causal theorists accept essen-

tialism,8

they can certainly accept the necessity of (a)
and (c). But should they? Offhand, it may seem as though
they should not. After all, they have already rejected
description when it comes to the meaning of natural kind
terms. 80, how can they suddenly rely upon description
when it comes to the essence of a natural kind?

The answer to this question is that while the causal

theorists reject any role for description when it comes to

45

the meaning of a natural kind term, they allow description
a role in fixing the reference of that term. Perhaps it is
true that natural kind terms, such as 'gold', obtain their
meaning by a causal connection to gold. But still, how do
we decide what is gold? Natural kind terms may obtain
their meaning from their reference; but what is included
within this reference is often open to question. Compared
to the referent of a proper name, the reference of a natu-
ral kind term is somewhat open-~a fact acknowledged even by
the causal theorists. So, how do we determine the refer-
ence of a natural kind term? Putnam writes that

The use of a word such as 'gold' depends

upon our possessing paradigms, standard

examples that are agreed to be model mem-

bers of the kind. ... What makes some-

thing gold is having the same nature as

the paradigm. [10]
And how do we decide that something has the same nature as
our paradigm? We rely upon description.

Once we have our paradigm, we choose certain proper-
ties to characterize that paradigm. These properties (or
descriptions) then permit us to identity additional members
of that kind. But this set of properties does not consti-
tute any sort of meaning. For one thing, it is subject to
change. Kripke goes so far as to argue that it is the task
of science to produce better sets of properties. He writes
that

...science attempts by investigating

basic structural traits to find the
nature, and thus the essence (in the

46

philosophical sense) of the kind.[11]
As scientists discover the basic structure that members of
a kind share, they reflect this knowledge in their choice
of properties that characterize that kind.

Yet when it comes to certain natural kinds, it seems
as though scientists have actually discovered the essence
of the kind. Sober writes that

A paradigm case [of essentialism] has

been the periodic table of elements,

which seems to tell us that the essence

of each element is its atomic number.

... The atomic number 14 is a character-

istic that all and only Nitrogen atoms

share, and that an atom must have this

atomic number if it is to be Nitrogen.

Essentialism of this sort holds that it

is no accident that the property of be-

ing an atom of Nitrogen and that of hav-

ing a given atomic number go together;

indeed they must covary since the atomic

number is the nature or essence of Nitro-

gen. [12]
Or let's consider gold. Being the element with atomic num-
ber 79 is not some mere property of gold--it is the defin-
ing property. This property tells us what gold is. If
something is gold, it must have atomic number 79; and if
anything has atomic number 79, then it is gold. This is
because the statement 'Gold is the element with atomic num-
ber 79' expresses the essence of gold. Or, according to
Aristotelian terminology, this statement gives the "real"
definition of gold.

Since the causal theorists divorced a statement's

metaphysical status from its epistemic status, it should

ya;

we
all
pa
am
me

Ye

47

not bother us that although (a) and (c) are both necessary,
one is a priori while the other is a posteriori. We can
use the same method to discover the essence of either kind.
And this method has been described by Aristotle. According
to Aristotle, if we wish to find the essence of a kind, we
begin with induction and then follow a method of division.
He writes

We must start by observing a set of sim-

ilar--i.e., specifically identical--in-

dividuals, and consider what element

they have in common. We must then ap—

ply the same process to another set of

individuals which belong to one species

and are generically but not specifically

identical with the former set. When we

have established what the common element

is in all members of this second species,

we should again consider whether the re-

sults established possess any identity

and persevere until we reach a single

formula, since this will be the defini-

tion of the thing. [13]
This process, however, can be either a priori or a poster-
iori. In the case of triangles, we start by examining
various triangles and seeing what features these triangles
have in common. But since triangles are mental entities,
we can carry out this examination by the use of reason
alone. In the case of gold, we start by examining our
paradigmatic examples and seeing what features these ex—
amples have in common. But since gold is a physical ele-
ment, we need experience to carry out this examination.

Yet in either case, the process is similar.

While statements (c) and (d) may not be analogous in

48

the source of their necessity, as the causal theorists
thought, statements (b) and (d) are. If the causal
theorists had compared (b) to (d), their argument would
have been much stronger. Certainly 'snow rose' and 'Hella—
borus niger' are comparable to proper names. Hull argues
that the names for biological species, whether a taxonomic
name or an ordinary name, obtain their meaning from their
connection to the species and not from some description.
Although Hull points out that he is not a causal theorist,
he makes much the same points that they do when he writes

As important as the traits listed in

[zoological] diagnoses and descriptions

may be for a variety of purposes, they

are not definitions. Organisms could

possess these traits and not be includ-

ed in the taxon; conversely, organisms

could lack one or more of these traits

and be clearcut instances of the tax-

on. [14]
Hull continues to borrow features from the causal theory
when he stresses the similarities between the names of
biological species and proper names. While the following
passages are taken from Hull, they could have just as
easily been written by Kripke.

A taxon [species] has the name it has

in virtue of the naming ceremony, not

in virtue of any trait or traits it

might have. If the way in which taxa

are named sounds familiar, it should.

It is the same way in which people are

baptized. [15]

If 'homo sapiens' is or is not a cluster

concept, it will be for the same reason

that 'Moses' is or (more likely) is not.
[16]

49

Hull mentions that if philosophers examined how scientists
designate biological entities like tigers or mallards or
the snow rose, they would have found "rules explicitly for-
mulated in the various codes of nomenclature which were in
perfect accord with Kripke's analysis--but for the wrong
reasons."17 By "wrong reasons", Hull is referring to
Kripke's essentialism. Scientists, according to Hull, do
not even attempt to find essences. Hull accepts Kripke's
account on meaning; but he rejects Kripke's essentialism.

As a matter of fact, we need not presuppose any sort
of essentialism for either (b) or (d). In the preceding
section, we did not need to presuppose essentialism to
argue that (d) was necessary. Nor do we need to presuppose
essentialism to argue that (b) is necessary. Since species
names function like proper names, they are rigid designa-
tors. The name 'snow rose' designates the same kind in
every possible world; and the name 'Hellaborus niger' also
designates the same kind in every possible world. In this
world, 'snow rose' and 'Hellaborus niger' designate the
same kind. Since they are rigid designators, and since
they designate the same kind in this world, they designate
the same kind in every possible world. Thus, statement (b)
is necessary. The very same sort of argument that justi-
fies the claim that 'Currer Bell is Charlotte Bronte' is
necessary also justifies the claim that 'The snow rose is

the Hellaborus niger' is necessary. And neither argument

50
relies in any way on essentialism.
While (a) and (c) owe their necessity to essentialism,
(b) and (d) owe their necessity to some other assumption.
In the arguments we have seen, both in the preceding sec-
tion and this section, the causal theorists appeal to our
intuitions. And many of the assumptions that underlie
these intuitions remain implicit. In "Identity and Neces-
sity", Kripke provides a formal version of this intuitive
argument. And in this formal version, implicit assumptions
become explicit.
Kripke's argument goes like this

(i) (X)(y)[(X=y)+(FX+Fy)].
This premise states that if x is y, then any property of x
is also a property of y.

(ii) (X)D(X=X).
This premise is the principle of self-identity which states
that everything is necessarily identical to itself.

(iii) (X)(Y)(X=Y)+[EKX=X)+EKX=Y)l.
This premise is a substitution instance of (i). From (iii)
and (ii), we can infer the following:

(iv) (X)(Y)[(X=Y)*D(X=Y)]-
This conclusion states that if x is y, then x is necessar-
ily y. Kripke makes it very clear that the substitution
instances for x and y must be rigid designators and not
descriptions. If we substitute rigid designators for x and

y, and these designators refer to the same thing, they do

51
so necessarily. On the other hand, if we substitute de-
scriptions for either x or y, and they refer to the same
object, they only do so contingently. On the basis of
this, we may argue that if the statements 'Currer Bell is
Charlotte Bronte' and "The snow rose is the Hellaborus
niger' are true, they are necessarily true.

This argument makes explicit an assumption that up
until now has only been implicit. This assumption is the
principle of self-identity. Statements such as (b) and (d)
owe their necessity to this principle. Without it, we can-
not argue that these statements are necessary. This fact
is made explicit in the second argument. But the principle
of self-identity is also assumed in the first argument,
albeit implicitly. When we argue that the statement 'Cur-
rer Bell is Charlotte Bronte' is necessary, we are relying
implicitly upon the fact that this statement follows from
the fact that Currer Bell is Charlotte Bronte and the fact
that Currer Bell is necessarily Currer Bell. If it were
not true that everything is necessarily identical to it-
self, then it would not be true that Currer Bell is neces-
sarily identical to Charlotte Bronte where Currer Bell is
Charlotte Bronte.

This second argument is also interesting in that it
makes it obvious that we justify the necessity of state-
ments (b) and (d) via an a priori argument. Whatever their

epistemic status may be, their metaphysical status is de—

52

termined by an a priori argument. And this is just as

true of (d) as it is for (b). In the case of (b), it may

strike us as odd that the metaphysical status of an a
posteriori statement can be determined by an a priori

argument. But when we argue that (b) and (d) are neces-

sary, we are relying upon the following inference pattern

(where once again, the substitution instances for x and y

are rigid designators).

(1) x=y.
(ii) (X=Y)+I:1(X=y).
(iii) D(x=y).

The conclusion, (iii), owes its metaphysical status to

Premise (ii). This premise, (ii), is merely a simpli-

fied version of the conclusion from Kripke's argument
in "Identity and Necessity". The fact that premise (ii)
is necessary explains why our conclusion is also necessary.

The conclusion owes its epistemic status, however, to

Premise (i). In the case of (d), we justify the truth

of premise (i) empirically. Thus, the conclusion will

b
e a- posteriori. In the case of (b), we justify the truth

0f
E’themise (1) without the use of experience. Thus, the

(3011
Q lusion is a priori.

Although statement (b) resembles (d) on metaphysical
grguhds’ since (b) is a priori, the causal theorists cannot
113% :1t to justify their claim that certain statements are
“egg ssary and a posteriori. Or could they? Certainly it

is
1true that philosophers, in the past, believed that (b)

53

was necessary and a priori. Hahn, for example, writes

I talk about a well-known plant... and
I make the stipulation: Let us call
any plant of this kind "snow rose," but
let us also call it "hellaborus niger."
Thereupon I can pronounce with absolute
certainty the universally valid proposi-
tion: "every snow rose is a hellaborus
niger." It is certainly valid, always
and everywhere: it is not refutable by
any sort of observation. ...The state-
ment merely expresses a convention con-
cerning the way we wish to talk about
the plant in question. [18]

But: is Hahn correct?

According to Hahn's account, the statement 'The snow
rose is the Hellaborus niger' is analogous to the state-
Imerlt: 'A bachelor is an unmarried man' in that each express-
es some sort of linguistic convention. We choose to use
'Hellaborus niger' as the taxonomic name for a particular

lcir1<3_ of plant just as we choose to use 'bachelor' as a name
fcxr- Ei.particular class of humans. The problem, though, is
'315‘12- Hahn's account applies just as easily to the statement
”:LIJEVlrer Bell is Charlotte Bronte' as it does to 'The snow
ITDSBTEE is the Hellaborus niger'. After all, didn't Bronte
StiDUlate that she was to be known as Currer Bell in a
eel:N'lr—viain situation? Thus, if the statement 'The snow rose

is
‘t:l]ae Hellaborus niger' is a priori, then so is 'Currer

BE¥I.;11‘

is Charlotte Bronte'.

We can also argue that if 'Currer Bell is Charlotte

Brt:>
Irivte' is a posteriori, then so is 'The snow rose is the

Ik33L_:l‘

aborus niger'. It is not that unusual for botanists,

could
disc:

he

54
particularly botanists who work in the rain forest, to

give a particular species a taxonomic name before they
discover that the plant has a folk name. In such a case,
couldn't we say that the botanists have made an empirical
discovery? Isn't such a discovery just as empirical as
the discovery that Currer Bell is Charlotte Bronte?

The a priori/a posteriori distinction is not as clear-
<nrt for statements such as 'The snow rose is the Hellaborus
niger' and 'Currer Bell is Charlotte Bronte' as we might
lilcee. The problem is that these statements concern names.

Anti naming concerns conventions. We tend to describe con-
\ner11:ions as a priori. But our knowledge of a particular
convention may very well be a posteriori. And our know-
ledge that two conventions coincide may also be a poster-
ic>xer..
De Re or De Dicto?
Towards the beginning of this chapter, we brought up

tmvc) (:lasses of statements. One class consisted of state—

me
hits such as

(1) Heat is the form of energy constituted
by the motion of atoms and molecules
in solids.

(2) Gold is the element with atomic number
79.

(3) Water is H20.

131<Es
second class consists of statements such as
(4) Currer Bell is Charlotte Bronte.
(5) Jolmo Lungma is Mt. Everest.
(6) The Malvinas Islands are the Falkland

”1
tile C

state

is at

and

on

tax

th

is

55

Islands.
The causal theorists claim that both classes consist of
statements that are both necessary and a posteriori. This
is an important point to reiterate. For while the causal
theorists are mistaken in their belief that these two

classes are analogous, they are not mistaken in their
belief that the statements in both classes are necessary
and! a posteriori. But if these statements are necessary
andi a.posteriori, then we need some account of necessity

ottieer than that of logical positivism.
Earlier in this chapter, we noticed that while the

(xaLlssal theorists, accept, more or less, the logical posi-

tivist's version of the necessary/contingent distinction,
tileali’ restrict this distinction to metaphysics. But there

is;

li’eet another difference between the logical positivist's
Vers ion of the necessary/contingent distinction and the

causal theorist's. This difference concerns the nature of
n . . .
ec33s.51ty. Or to be more prec1se, it concerns the nature

0
fr ral«ecessity as a modal property.
Necessity is a modal property. As such, it is either

If it is de dicto, then it is a proper-

de:
‘iinicto or de re.
If it is de re, then

ty

1tZ-Ihat is predicated of statements.
it -

:‘~ 8 a property that is predicated of objects or things.

Cert)
=5=»ider the following two statements:

(1) A Bachelor is an unmarried man.
(ii) Gold is the element with atomic number

79.

pr

51

56

Both statements express necessary truths. But in the case

of statement (i), it is only the statement that is neces—

sary. We certainly do not believe that particular indivi-

duals, who happen to be bachelors, are necessarily bache-

lors and therefore necessarily unmarried. In the case of

(ii), it is the object gold that is necessarily the element
with atomic number 79. Statement (ii) is necessary only
because gold is the sort of thing that is necessarily the

element with atomic number 79. In one case, necessity is

Predicated only of a statement. In the other case, neces-

sity is predicated of an object. Thus, in one case, neces—

Sity is de dicto while in the other case, it is de re.

Why is this distinction between necessity de dicto and
necessity de re so important? Basically, it is because the

logical positivists acknowledge necessity only as a de dic-

to property. This is hardly surprising since they equate

necessity with analyticity; and analyticity is a property

t
hat belongs only to statements. But this also means, as

w
e saw in chapter one, that the logical positivists are

al
SQ forced to acknowledge necessity as an a priori proper-

ty"’ The causal theorists, on the other hand acknowledge
neQQ ssity as both a de dicto and a de re property. In
faq 1'; I statements (1 )-(6) are all necessary de re. Consider
(2 ) s As we already saw, this statement is necessary be-

§Eie gold is necessarily the sort of object that is the

61%
II‘Ient with atomic number 79. Now consider (4). Thanks

lll'l

L
at

"I.
1..

(I)

57

to the principle of self-identity, Currer Bell is necessar—
ily the sort of object that is Charlotte Bronte. The
statements in our two classes may not have been analogous
in quite the way that the causal theorists thought; but
they are analogous in that they are all necessary de re.
And it is the fact that they are necessary de re that
explains why they are a posteriori. While gold is neces—
sarxily the sort of object that is the element with atomic
Innntoer 79, we only discover that gold is the element with
atcznmic number 79 through experience. And while Currer Bell
is r1ecessarily Charlotte Bronte, we only discover that
(hlxrzrer Bell is Charlotte Bronte through experience.
Yet many philosophers continue to reject de re modal—
itZi-Gass. Consider statements (1)-(3). These statements owe
tileijLJ: necessity to the fact that they express the essence

of? tll‘leir subjects. But many philosophers believe that this

S 0 O O
Ort of nece551ty only makes sense Within some sort of

context. Quine, for example, writes that

Mathematicians may conceivably be said

to be necessarily rational and not nec-
essarily two-legged; and cyclists nec-
essarily two-legged and not necessarily
rational. But what of an individual who
counts among his eccentricities both math-
ematics and cycling? Is this individual
necessarily rational and contingently two-
legged or vice versa? Just insofar as we
are talking referentially of the object,
with no special bias toward a background
grouping of mathematicians as against
cyclists or vice versa, there is no sem-
blance of sense in rating some of his
attributes as necessary and others as
contingent. [19]

58

properties are necessary or contin-

Quine's point is this:
If we are focusing

gent only within a particular context.
then rationality is a

on an individual as a mathematician,
But it is a necessary property within

necessary property.
this context. If we focus on this individual as a cyclist,
But again, it

then two-leggedness is a necessary property.
If we

is a necessary property only within this context.
then it does

wrench this individual from these contexts,

not: make sense to describe either rationality or two-leg-

gedness as necessary.
Quine's point becomes pertinent once we realize that

statements (1 )-(3) are necessary only within a particular

context. Statements (1)-(3) express the essence of their

subj ects only within a context where we are interested in
Without

tilea :internal structure of heat, gold, and water.

tIIj-ES context, we have no reason to suggest that the inter-
sstructure of heat, gold, and water is somehow more

ruaJL
In fact, in some

'1me rtant than their other properties.
0
t1““333: context, we may not even consider these properties

De
cstary. For example, the fact that gold has a particu-

1&1
JE‘ tatomic number is absolutely irrelevant when it comes

tc>
Q conomics .
Or consider statements (4)-(6). These statements

Owe
their necessity to the principle of self-identity.

Mat
§kie suggests that

...these de re modalities are,
broad sense, de dicto after all.

in a very
Though

59

these necessities apply to individual
things and natural kinds, that they so
apply is primarily a feature of the way
we think and speak, of how we handle
identity in association with counterfac-
tual possibility. They reflect implicit
rules for the ascription of identity, for
the recognition of the same person or
thing or stuff or species, in neutrally
described merely possible situations.
The topic of names (and certain general
terms) comes in only because such names
(etc.) are intended to belong to things
(etc.) whose identity is determined by

these rules. [20]
Insofar as we follow certain rules in talking and theoriz-
ing; about Currer Bell and Charlotte Bronte, for example,
thee statement 'Currer Bell is Charlotte Bronte' can be
deassczribed as de dicto. Even though the principle of self-
irieerjtity applies to Currer Bell and Charlotte Bronte as
Obj ects, the fact that this principle applies to these ob-
jects is due to the way we talk and think.
Thus, we can argue that statements (1)-(6) are actu—
‘a1“1-3?’ necessary de dicto. An object, by itself, may have
VaI‘lious properties, but these properties are neither nec-
essary or contingent until we consider those properties
wi th in a particular context. The fact that we choose to
deg Qribe certain properties of an object as necessary with-
1:1 iEL given context reflects our interests and theories
ah><:)“‘;lt that object. The fact that we choose to handle iden—
t1‘1t:-2i? and self-identity in a particular manner reflects the
walEET that we talk and think. But if we decide which proper-

'tjk
EE=ES are necessary, and if we base this decision on our

{Leon
The 0t

are Ill

 

 

60

theories, then necessity is, in a broad sense, de dicto.

The object simply has properties.

Whether these properties

are necessary or not depends upon us, not the object.

I1

V, v .
53%: ‘a

it is

ZaVe

In!

In

 

Chapter Three
In chapter two, we learned that the causal theorists
Categorized the following statements as both necessary and
a posteriori:

(1) Heat is the form of energy constituted
by the motion of atoms and molecules
in solids.

(2) Gold is the element with atomic number
79.

(3) Water is H20.

N0“? ‘vhile this suggestion fits nicely with our intuitions,

it .143 by no means the only suggestion that philosophers

haVe made .

A great many philosophers argue that the reason why
Statements (1)-(3) prove so difficult to classify is be-
cause our distinctions are fundamentally flawed. These

p'l'lli--IIL<:>sophers argue that the logical positivists did not

EEITJE7 Lin misunderstanding the nature of their distinctions;

t:
hey erred in accepting these distinctions in the first

3)
lace. And according to these philosophers, rather than

t -
:L.1:l]<;er around with the traditional distinctions, we should

$3;j‘
Jrr‘l;>ly reject them altogether. And this conclusion becomes

3"
IQE=1F1.more compelling once we adopt a particular theory of

rr1<sa
Q. jfling.

'«I‘hg

Network Theory

1;) For the past thirty years, several philosophers have
3::~‘:
§§ IFNbsed various theories of meaning that are commonly de—
:‘=:~

ii~l3ed as network theories of meaning. Among the

61

 

I1

 

62

Emilosophers who proposed such theories are Churchland
(§gientific Realism and the Plasticity of Mind), Field
("Logic, Meaning, and Conceptual Role"), Hanson ("Causal
Chains"), Harman ("Conceptual Role Semantics"), Kuhn (The
Eggnicture of Scientific Revolutions), Lewis ("Psychophysi-
Cal. and Theoretical Identifications"), and Loar (Mind and
Meaxiing). While these various theories differ in many res-
Pects, they all agree that the meaning of a term is given
bY' ists role and position within a theoretical network.
Evie!) those terms which are commonly defined by means of
necessary and sufficient conditions actually depend upon
some sort of theoretical context for their meaning.
To a certain extent, the claim that the meaning of a
term depends upon its theoretical context is quite plausi-
kDJL‘Ei-— Take the term 'castling' from chess. If we consulted
a chess dictionary, we would find something like the fol—

castling: a move in chess that is ac-

complished by first moving the King two

squares along the first rank in the di-

rection of the Rook that is to be used;

and then moving the Rook to the other

side of the King and placing it on the

square immediately next to the King. [1]
(:)]t:.
1:. ‘\’”li.ously, in order to understand this definition, we need

<::>

. :Ftnow what is meant by 'King', 'Rook', 'square', and

3t:~
ahki.
QB

$§-t::

But if our entire knowledge of castling consisted
12-his definition, then we can hardly be said to under-

I‘JVICi the meaning of the term 'castling'. To understand

 

 

63
'castling', we also need to know a few rules of chess,

particularly those rules that govern castling. To under-

stand 'castling', we should recognize the move when we see
it. To understand 'castling', we should be able to make

time move ourselves. Thus, to understand 'castling', we

neexi to know a fair amount about chess. And as a matter of

facfl:, when we first learn chess, we do not learn a few

terms and rules in isolation. Rather, we are given a great

deéil. of information about the game as a whole from the very
Start.

The network theorists claim that all terms resemble
'castling' in respect to the fact that all terms obtain
t11€3£i¢r meaning within some sort of theoretical context.

{P51]<:<az the term 'group' from algebra, for example. 'Group'
Iléisss

a.very precise definition that is given by means of

ti . . . .
ecessary and suff1c1ent conditions.

A group is a nonempty set that is closed

under a binary operation and that meets
the following conditions:

(i) The associative law holds.
(ii) There is an identity element.
(iii) Every element has an inverse.
155‘;‘

“:; to understand this definition, we first need to know

ugfilil

i3‘“: is meant by 'set', 'binary operation', 'associative
3L.;53L

w.

, 'identity element', and 'inverse element'. By the
hi
lr‘filfia we know all this, we have more than a fleeting ac—
3“
. ‘EE“jLntance with mathematics.

Thus, while we may define

§%
1: <2nap' by means of necessary and sufficient conditions,
L“Sss

:fact remains that these conditions are understood only

 

IA

within

ical"
| A
QERC
as th

varic

the <

I
€35-

full

the:

CC

t.

 

64
Within the context of mathematics.

This view is also plausible when it comes to "theoret-
ical" terms--terms such as 'quark', 'atom', 'force', 'DNA',
'gene', and 'energy'. After all, we describe these terms
as theoretical to emphasize the fact that they occur within
Varrious theories. Yet many theoretical terms, especially

thcuse from physics, are defined in terms of equations. In
the case of 'energy', for example, we have the equation
'e=mc". Couldn't we argue that this equation provides the
Ihllhl. meaning of 'energy'? Not according to the network
t31€3c>rists. They agree with Putnam when he writes
The concept 'energy' is an excellent ex—
ample of a law~cluster concept. It enters
into a great many laws. It plays a great
many roles, and these laws and inference
roles constitute its meaning collectively,
not individually. [2]

Despite our equation for energy, Putnam argues that the
meaning of 'energy' is not given by a single defining law
(3

1:. «E1 single defining characteristic. Rather, it takes its
IT!
ea11ing from its overall role and position within physics.

TE'E)
‘51 term 'energy' is best understood within the overall

<==<::
‘Tth:ext of physics.

Although we may be willing to concede that game terms

EEJI=>
Q o o o
rm their theoretical context, we may not believe that all

terms from mathematics and science obtain their meaning

‘Es:

53::7rns obtain their meaning from their context. For exam—
1;).Ili
. 53: 0' consider observation terms--terms such as 'hot',

=53:
‘:::’Ifit', 'white', or 'sweet'. Intuitively, we tend to think

 

IA

 

 

65

that these terms obtain their meaning from their connection
to a particular sensation. 'Hot' means the sensation hot,
for example. But the network theorists argue that even
these terms obtain their meaning from their role and posi—
ticni within a theoretical network. But how can they jus—
tify this claim?
They could justify this claim by appealing to Wittgen-
SteaiJu's point that even observation terms have public cri—
teriafor their use. If I describe newly fallen snow as
113C1.- for example, few people, apart from my optometrist
aILéi ‘physician, will care what my visual sensation happened
tc> lash Perhaps it was red or perhaps it was white. But
"nC’SSWt; people will simply say that my description is incor-
Ifea¢::t:. This is because we have public criteria for our use
()1? (:olor terms. Taking this point, network theorists can
aJrglsle that public criteria are provided by some sort of
t:lj“53<:retical network.
Churchland supports this claim by means of the follow-
jLaljlel thought experiment. He proposes that we imagine a

EE:]E:
‘Ei<:ies of beings who perceive temperature visually. Ac-

<:=..::>
JCT<ding to Churchland, the world looks to them much as it
:1.

%

‘:3fiks to us in black-and-white pictures taken with infrared

2:‘Tlsitive film. Next, Churchland asks us to imagine that
: *1 Q . . .
se beings speak a language that resembles English in
§\>~
‘Eéiry respect but two: First, their language lacks a color

Q. .
":=€ibulary. And second, the temperature vocabulary is

‘

HIE

vat:

He

 

66

learned visually rather than tactilely. So, Churchland
asks, how should we interpret their terms 'hot', 'warm',
and 'cold'? If we believe that sensation terms obtain
their meaning from their relationship to a sensation,
theni‘we should interpret their terms 'hot', 'warm', and
'ccild' as 'white', 'gray', and 'black'. After all, the
serusation that these beings call 'hot' is the sensation
that we call 'white'.
But such a translation seems incorrect. Our intuition
-iS tihat these beings mean the same thing by 'hot' that we
"Kiéiri by 'hot'. Moreover, Churchland points out that such a
trEtrislation also makes many of their beliefs about tempera-
t“JdI?€e incorrect. Churchland concludes from this that obser-

V511tlion terms do not obtain their meaning from sensation.

He Writes that

The meaning of a term (or the identity
of a concept) is not determined by the
intrinsic quality of whatever sensation
happens to prompt its observational use,
but by the network of assumptions/be-
liefs/principles in which it figures.
Sensations are just the causal middle—
men in the process of perception, and
one kind will serve as well as another
so long as it enjoys the right causal
connections. [3]

‘N7<EE=

Inappen to use tactile sensations to learn our tempera-
:‘:?<e vocabulary. Churchland's beings use visual sensa—

1t; -
.:1‘
‘::>11s to learn their temperature vocabulary. But so long
§§
fih

‘:::'K1t.temperature as these beings, their temperature terms

“ve share the same beliefs, assumptions, and principles

 

have thE
as we st

55.16 me

t' .
Wt?
COTLV inc

drawn 1
cal p0:
to the
failur
tional
netwo:

tions

What

dI' at;

tit
Co:

be

 

67

have the same meaning as our temperature terms. So long
as we share the same theoretical network, terms have the
same meaning.

We may or may not find the network theory of meaning
(xnrvincing. But one reason why philosophers have felt
drawn to this theory is because it is a response to logi-
cal. positivism. Or to be more accurate, it is a response

to tile failure of logical positivism. But the reason for
faiLIJJre of logical positivism is the failure of the tradi-
ti<311ial distinctions. Many philosophers have adopted the

network theory as they gave up the traditional distinc—

tions.

“at Happens to the Analytic/Synthetic Distinction?
As we know from Chapter One, the logical positivists
Ci
ltrial‘flr the analytic/synthetic distinction as follows:
(LA) A statement is analytic if and only
if its truth value is determined
completely by linquistic convention.
(LS) A statement is synthetic if and
only if its truth value is not de—
termined by linquistic convention.

3§L<::
‘<::‘:>rding to the network theorists, if we base the analy-

t:-:i.
‘<:= J’synthetic distinction on the notion of linquistic
(:2‘::.thl
It)

‘\7ention, then we cannot justify a sharp demarcation
53’ 1t:;
3L 'Vveen those statements that we wish to describe as ana—
‘:¥’ ‘it:; -
:LIZ and those that we wish to describe as synthetic.

The logical positivists believe that there are two

68

notions of analyticity. One type can be described as the

logical notion of analyticity. The statements that fall

under this type include statements such as 'Bachelors are
bachelors' and 'Fv-F'. The linquistic conventions that
determine the truth value of these statements are the
conventions that govern our use of logical words and
logical participles. Quine says that

The logical truths, then, are those
true sentences which involve only lo-
gical words essentially. What this
means is that any other words, though
they may also occur in a logical truth
(as witness 'Brutus', 'kill', and 'Cae-
sar' in 'Brutus killed or did not kill
Caesar'), can be varied at will without
engendering falsity. [4]

The

1:11.53

second type of analyticity is what Putnam describes as
linquistic notion of analyticity. The statements that
Iféa-JL I. under this type include 'Bachelors are unmarried men'
and 'Brothers are male siblings'. The conventions that
govern these statements are synonymy relations. The state-
It!
Gatlrlft: 'Bachelors are unmarried men', for example, owes its
1:;3:.
151*t2h to the fact that in English, 'bachelor' is synonymous
w '-
1 t1“: 'unmarried man'.
In 'Two Dogmas of Empiricism", Quine attacks the
JL-Ii‘
Irjl<guistic notion of analyticity. Quine argues that we
Qa
)rjllnot provide an account of synonymy that is not circular.
13‘::>
‘c‘ “9‘7 do we know that 'bachelor' is synonymous with 'unmar-
.:i‘
. EE=<3 man'? We could argue that they are synonymous because

5

Q“LThelor' is defined as 'unmarried man' in the dictionary.

 

69
But as Quine points out, the dictionary does not make
'bachelor' synonymous with 'unmarried man'-—it only reports
that 'bachelor' and 'unmarried man' are synonymous. We
could argue that 'bachelor' and 'unmarried man' are synony-
mous because they are interchangeable. But what do we mean
by "interchangeable"? Do we mean that they are inter—
changeable because they are extensionally equivalent or be—
cause they are intensionally equivalent? When we claim
that 'bachelor' and 'unmarried man' are synonymous, Quine
argues that we are making a stronger claim than that they
are extensionally equivalent. 'Morning star' and 'evening

star' are extensionally equivalent; but they are hardly

'un-

synonymous. So, we want to claim that 'bachelor' and
married man' are interchangeable because they are inten-
sionally equivalent. But how do we know when two terms are
intensionally equivalent? According to Quine, when we say
that two terms are intensionally equivalent, we mean that a
statement that unites the terms, such as 'Bachelors are un-
married men', is analytic. Thus, to justify the analyti-
<Jity of 'Bachelors are unmarried men' by the use of synon-
lnny, we are already presupposing that the statement is
analytic.

Quine attacks the logical View of analyticity in
"Carnap and Logical Truth". Although logical and mathema-

tical statements have, in the past, been described as true

in virtue of their self-evidence, the logical positivists

70
argue that these statements owe their truth to linquistic
conventions. But the linguistic conventions that apply to
these statements do not depend upon relations of synonymy.
Rather, we organize the truths of logic and mathematics
into formal axiomatic structures. We then assign meanings
to the axioms, definitions, and inference rules that belong
to these structures. We stipulate the conventions that
govern our use of these truths.

But Quine argues that this account is misleading. He
points out that the axioms and definitions (and presumably
the inference rules as well) of axiomatic structures are
either discursive or legislative. If they are discursive,
then we have chosen certain truths that allow us to derive
the rest from a pre-existing set of truths. But as Quine
points out,

...this is not truth by convention. The

truths were there, and what was convention—

al was merely the separation of them into

those to be taken as starting point (for

the purpose of the exposition at hand) and

those to be deduced from them. [5]
Neither discursive axioms or definitions can be said to
idlstitute truth by convention. We picked these truths to
Sarve as axioms and definitions, in part, because we al-
ready knew that they were true. On the other hand, legis-
lative axioms and definitions do institute truth by conven-
tion. In the case of legislative axioms and definitions,

we explicitly adopt a convention. We stipulate that cer-

tain axioms and definitions are true; and thus, we can say

71

that they are true by convention.

The problem, however, is that this distinction between
the legislative and the discursive describes only how we
came to accept certain axioms and definitions—-it does not
describe these axioms and definitions themselves. Quine

writes

The distinction between the legislative
and the discursive refers thus to the act,
and not to its enduring consequences...

So conceived, conventionality is a pass-
ing trait... It is a trait of events and

not of sentences. [6]
Although we may adopt certain axioms and definitions on the
basis of legislation, this does not mean that we continue
to accept these axioms and definitions on the basis of
convention.

Scientists, for example, could adopt a particular law

on the basis of legislation. Putnam writes that

The principle 'e=%mv" may have been in-

troduced, at least in our fable, by stip-

ulation; the Newtonian law of gravity may

have been introduced on the basis of in-

duction from the behavior of the known

satellite system and the solar system (as

Newton claimed); but in subsequent devel-

opments these two famous formula were to

figure on a par. [7]
1M2cording to Putnam's "fable", scientists adopted 'e=%mv"
On.the basis of legislation. They stipulated that this
Statement was true. But after they adopted it, they treat—
ed it the same way they treated every other statement with—

in their theory--they treated it as a synthetic statement.

They may have adopted the statement by legislation; but

72
they continued to accept it on the basis of experience.

Since this distinction can apply to statements other
than those of logic and mathematics, we cannot use the
legislative/discursive distinction to justify a sharp
demarcation between those statements that we wish to de—
scribe as analytic and those that we wish to describe as
synthetic. We want to argue that analytic statements are
true by convention while synthetic statements are not. The
problem, as Quine points out, is that this notion of con-
vention only explains how we came to adopt a statement as
true. It does not explain why we continue to accept that
statement as true. In fact, we may use convention to adopt
a scientific law as well as a logical or mathematical axiom
or definition. But if we wish to use the notion of conven-
tion to argue that the latter are analytic while the former
is not, then we have already decided which statements are
analytic and which are synthetic.

Although the network theorists reject the logical
Exasitivists' version of the analytic/synthetic distinction,
ii few, following in the footsteps of Putnam, accept a
IIragmatic version of the distinction. Consider the follow—
ing three statements:

(i) Bachelors are unmarried men.
(ii) Vixens are female foxes.
(iii) Brothers are male siblings.
We tend to feel that these statements are analytic while

the following statements are synthetic:

73

(iv) Bachelors are neurotic.

(v) Vixens are carnivores.

(vi) Brothers are obnoxious.
Even with Quine's arguments against the analytic/synthetic
distinction, it still seems as though the statements in the
first group owe their truth value to the meaning of the
terms that are used within those statements while the
statements in the second class do not owe their truth value
to the meaning of their terms.

Putnam agrees with the logical positivists that state-
ments (i)-(iii) are analytic while statements (iv)-(vi) are
synthetic. He even agrees with the logical positivists
that statements (i)-(iii) are true by linquistic conven-
tion. Putnam wants to argue that analytic statements are
those statements that have one-criterion words as their
subjects and the criterion for those words as their predi—
cates. One—criterion words are those terms that are
defined by means of a single law or a single characteris—
tic. Analytic statements are those statements that connect
ii one-criterion word with its criterion. Putnam argues
tfllat such statements meet four criteria (which I have
‘Paiaphrased):

(1) The statement can be rewritten in the
form 'Something (someone) is an A if
and only if it (he, she) is a B' where
'A' is a single word.

(2) The statement holds without exception;
and it provides us with the criterion

for something to be the sort of thing
to which the term 'A' applies.

74

(3) This criterion is the only one which is
accepted and employed in connection with
the term 'A'.
(4) 'A' is not a law—cluster term. [8]
These four criteria enable us to know when a statement is
analytic.
Statements (i)—(iii) meet these four criteria. Consi-
der (i). When it comes to the first criterion, (i) can be
rewritten as 'Someone is a bachelor if and only he is an
unmarried man'. When it comes to the second criterion, (i)
holds without exception. When it comes to the third cri-
terion, being an unmarried male is the only criterion that
is accepted for being a bachelor. When it comes to the
fourth criterion, 'bachelor' is not a law-cluster term.
The only role it plays in explanations is through its tie
with 'unmarried man'. This is a point that even Quine
recognizes when he writes
One looks to 'unmarried man' as semanti—
cally anchoring 'bachelor' ... sever its
tie with 'unmarried man' and you leave it
['bachelor'] with no very evident social
determination, hence no utility in com-
munication. [9]
'Brother', in its synonymy with 'male
sibling' is essentially like 'bachelor'
with its synonymy with 'unmarried man'.
...it is only the few verbal links that
give the terms the fixity needed in com-
munication. [10]

While statements (i)—(iii) meet these four conditions,

statements (iv)-(vi), in contrast, do not.

But while Putnam resurrects the analytic/synthetic

75

distinction, he only resurrects it as a pragmatic
distinction. It is no longer the distinction that it
was with the logical positivists. Putnam's version of
this distinction is rather limited. For one thing, he
only resurrects the linquistic notion of analyticity.
This means that while we can describe the statement
'Bachelors are unmarried men' as analytic, we cannot
describe 'P v -P' as analytic. For another thing, Putnam's
version depends upon one-criterion words; and by Putnam's
own estimate, there are only a few hundred such words in
English.11 Moreover, a term that is a one-criterion word
at one time may no longer be a one-critrion word at another
time. Putnam argues that 'atom' is an example of a term
that was once a one-criterion word that has now become a
law—cluster term. So, the statement 'Atoms are indivisi-
ble', which philosophers once argued to be analytically
true is now empirically false.1
Nor does Putnam believe that his version of the analy-

tic/synthetic distinction is exclusive and exhaustive.
Z\lthough he believes that some statements are most defin-
itely analytic and other statements are most definitely
SYnthetic, most statements are neither. Rather, analyti—
City can be a matter of degree. Putnam writes

What these statements [the law of conser-

vation of energy, 'there is a past', and

'bachelors are unmarried men'] reveal are

different degrees of something like con-

vention and different degrees of system-
atic import. [13]

76

Quine makes a similar point when he writes that

There can be no doubt that sheer verbal

usage is in general a major determinant

of truth. Even so factual a sentence as

'Brutus killed Caesar' owes its truth not

only to the killing but equally to our us-

ing the component words as we do. [14]
These passages suggest that the analytic/synthetic distinc—
tion falls along some sort of continuum. Those statements,
which are clearly analytic, fall at one end of this contin-
uum while those statements, which are clearly synthetic,
fall at the other end of the continuum. But most state—
ments fall somewhere in between these two extremes. Thus,
we cannot use the analytic/synthetic distinction to divide

all our statements into two exclusive and exhaustive class-

es.

What Happens to the A Priori/A Posteriori Distinction?
Putnam believes that this attack on the analytic/syn-

thetic distinction is important primarily because it simul-

taneously does away with the a priori/a posteriori distinc-

tion.15

Since the logical positivists equate a priori
Eitatements with analytic statements and a posteriori state-
Huents with synthetic statements, it should not surprise us
that the collapse of one distinction leads to the collapse
0f the other. But should we be so quick to dismiss the a
priori/a posteriori distinction? After all, the logical

positivists drew this distinction in terms of experience.

(LP) A statement is a priori if and only

the
easy

the

dete
Whi:
to:

has

tru
is ;
the

MC;

77

if its truth value is not determined
by experience.

(LT) A statement is a posteriori if and

only if its truth value is determined

by experience.
Now it seems as though we could accept this distinction
whether or not we accept the analytic/synthetic distinc-
tion. Even if we reject the analytic/synthetic distinc-
tion, it seems as though we could still argue that certain
classes of statements, say those of logic and mathematics,
are a priori while other classes of statements, say those
of the empirical sciences, are a posteriori. Perhaps the
logical positivists' mistake was not that they accepted the
a priori/a posteriori distinction but that they based it on
the analytic/synthetic distinction. But it is not quite so
easy to divorce the a priori/ a posteriori distinction from
the analytic/synthetic distinction as we might think.

The truth value of every a posteriori statement is
determined by experience. The problem, however, is that
While every a posteriori statement depends upon experience
tn) some degree or other, not every a posteriori statement
lmas quite the same relationship with experience. It is
ifairly easy to see how we use experience to verify the
truth value of the statement 'The pen is in my hand'. It
is not quite so easy to see how we use experience to verify
the truth value of the statement 'Electrons circle the
nucleus of the atom'. The logical positivists believed

that a posteriori statements fell into one of two classes.

78

An a posteriori statement is either an observation state-
ment or a theoretical statement. While we verify or falsi—
fy an observation statement directly by experience, we can
verify or falsify a theoretical statement only indirectly
by experience. The logical positivists argue that when it
comes to theoretical statements, we can only determine
their truth value once we "reduce", or translate, them to
some set of observation statements. We can use experience
to determine the truth value of a theoretical statement
only because every a posteriori statement "is equivalent
to some logical construct based upon terms which refer
to immediate experience."16
But such a reduction proved difficult to carry out.
In fact, as we know, the logical positivists were unable to
reduce theoretical statements to some set of observation
statements. The failure to effect this reduction is
sketched by Hempel in "Empiricist Criteria of Cognitive
Significance". Hempel traces the failure to reduce
theoretical statements to observation statements by means
of complete falsifiability, complete verifiability, and
partial verifiability. He traces the failure to reduce
the theoretical terms within a theoretical statement to
a set of observation terms by means of explicit defini-
tions and reduction rules. He traces the failure to tie
theoretical terms within a scientific theory to experience

by means of reconstructing that theory into a formal

79
axiomatic system. In each case, Hempel shows that this
attempted reduction not only failed to exclude "metaphy-
sical" nonsense but also failed to include various sorts
of a posteriori statements.

Wfiat is interesting, though, is that once the logical
positivists began to reconstruct scientific theories into
formal axiomatic structures, they adopted a holistic posi—
tion. They conceded that we cannot talk about the "experi—
ential meaning" of a given statement in isolation. As
Hempel points out

A single sentence in a scientific theory

does not, as a rule, entail any observa-

tion sentences; consequences asserting

the occurrence of certain observable

phenomena can be derived from it only by

conjoining it with a set of other, sub-

sidiary hypotheses. Of the latter, some

will usually be observation sentences,

others will be previously accepted theo-

retical sentences. [17]
In order to know the empirical implications of a particular
statement, Hempel points out that we not only need to know
other theoretical and observation statements, but we also
need to know the logical and mathematical apparatus of the
theory in which that sentence occurs.

But this holism means that we can no longer talk about
verifying or falsifying an individual statement. If a
theoretical statement entails an empirical consequence only
with the addition of subsidiary statements, then we cannot

conclude that this statement is false if the empirical con-

sequence fails to occur. Perhaps the theoretical statement

80

is false; but perhaps it is a subsidiary statement that is
false instead. Quine argues that since all the statements
that express our knowledge are related to one another by
various logical and causal relations, we cannot verify or
falsify individual statements. Rather, "our statements
about the external world face the tribunal of sense exper-
ience not individually but only as a corporate body."18
Quine argues that every statement within our network is
ultimately justified by the fit of the entire network

with experience. And if a revision needs to be made,

it can occur anywhere within that network.

But without the analytic/synthetic distinction, this
holism also undermines the a priori/a posteriori distinc—
tion. Although they conceded that we use logic and mathe-
matics to derive the empirical consequences of a scienti—
fic theory, the logical positivists believed that logic
and mathematics could not be falsified by an empirical
consequence. If any revision needed to be made, it would
occur elsewhere in the theory. This is because they be—
lieved that the statements of logic and mathematics, as
analytic statements, had no empirical content. These
statements were a priori. But once we do away with the
analytic/ synthetic distinction, we have no justification
for claiming that the statements of logic and mathematics
have any sort of special status within our network.

Unless there is a sharp division between analytic

and s
that

Eaa‘

shat

for

at
149

EECE

boa]

blu1

81

and synthetic statements, we cannot justify the claim
that logic and mathematics are independent of experience.
Many philosophers believe that just as there is no sharp
distinction between logic and mathematics, there is no
sharp distinction between mathematics and physics. Putnam,
for example, points out that

It is worthwhile pausing to remark just

how much of classical mathematics has

been developed for physical application

(the calculus, variational methods, the

current intensive work on nonlinear dif-

ferential equations for a start) and what

a surprising amount has found physical

application. [19]
The boundary between mathematics and physics becomes even
more blurred if we examine the work of two of the most
recent Fields Medalists. Ostensibly, the Fields Medal
is awarded for original work in mathematics. But Vaughan
F.R. Jones won his award for connecting Von Neumann alge-
bras (which are used in quantum mechanics) with knot theory
(a branch of topologY). Edward Witten, who is actually a
physicist rather than a mathematician, won his award for
his work in string theory. Although string theory is a
branch of physics, Witten argues that it will ultimately
flourish as a branch in geometry; and he provides a hint
at what this geometrical foundation will look like with

his topological field theories.20

This suggests that the
boundary between mathematics and physics is rather more
blurred than the logical positivists thought. But since

this boundary is blurred, we cannot justify the claim that

82
one discipline is a priori while the other is a posteriori.

Quine believes that all our knowledge, including that
of logic and mathematics, forms one overall theoretical
network. He writes that

The totality of our so-called knowledge

or beliefs, from the most casual matters

of geography and history to the profoun-

dest laws of atomic physics or even of

pure mathematics and logic, is a man-made

fabric which impinges upon experience.[21]
Quine believes that every statement within our theoretical
network is ultimately justified by the fit of the entire
network to experience. This means that every statement
within our network is, to some degree or other, a poster-
iori. We cannot argue that some statements within our
network are somehow independent of experience. And this
holds for the statements of logic and mathematics as well
as for the statements of empirical science.

While we can distinguish logic and mathematics from
the rest of our knowledge, we cannot make a sharp distinc-
tion. Quine believes that all our knowledge is bound into
some overall theoretical network. At the center of this
network, Quine places logic. Moving away from the center,
Quine argues that we come to mathematics. While logicians
need not rely on mathematics, mathematicians rely on logic.
Quine argues that mathematicians add their distinctive
terms and objects to logic. From mathematics, Quine argues

we move to physics. While logicians and mathematicians

need not rely on physics, physicists need both logic and

83
mathematmatics. Physicists add their distinctive terms
and objects to mathematics. From physics, we move to
chemistry. While logicians, mathematicians, and physicists
need not rely on chemistry, chemists presuppose logic,
mathematics, and physics. Chemists add their distinctive
terms and objects to that of mathematics and physics. From
chemistry, we move to biology and so on.22 As we can see
from Quine's picture, although we can distinguish logic and
mathematics from the rest of our knowledge, this is not a
sharp distinction. Logic and mathematics differ in degree
from the rest of our knowledge, not in kind.

Just as the network theorists resurrect a pragmatic
version of the analytic/synthetic distinction so do they
also resurrect a pragmatic version of the a priori/a pos-
teriori distinction. While every statement within our
theoretical network is ultimately justified by the fit of
the entire network with experience, not every statement
within the network is equally related to experience. Some
statements, say the axiom of extensionality, are farther
removed from experience while other statements, say 'The
pen is in my hand', are relatively close to experience.

In general, the statements of logic and mathematics tend
to be among the statements that are the farthest removed
from experience. In this respect, we can describe them

as a priori. Compared to the other statements within our

network, they are relatively a priori. But this is only

84

a matter of degree and not of kind.

What Happens to the Necessary/Contingent Distinction?
In Chapter One, we learned that the logical positi-
vists draw the necessary/contingent distinction as:
(LN) A statement is necessary if and
only if it must be true—~it is
impossible for it to be false.
(LC) A statement is contingent if and
only if it can be false--it is
possible for it to be false.
Considering that the network theorists reject the logical
positivists' versions of both the analytic/synthetic and
the a priori/a posteriori distnctions, it is not surprising
that they reject this version of the necessary/contingent
distinction as well. The network theorists reject the
claim that it is impossible for any statement to be false.
They believe that it is possible, in principle, for any
statement within our network to be revised or dropped. And
this does not exclude the statements of logic and mathema-
tics.

But just as the network theorists resurrect pragmatic
versions of both the analytic/synthetic distinction and the
a priori/a posteriori distinction, they also resurrect a
pragmatic version of the necessary/contingent distinction
as well. While no statement within our theoretical network

is immune from revision, not every statement is equally

likely to be revised. We protect certain statements from

85

revision. Quine suggests that we could describe these
statements as "necessary". In effect, this is a new notion
of necessity--a notion that is based upon the likelihood of
revision. The more likely a statement is to be revised,
the less necessary it is; and the less likely a statement
is to be revised, the more necessary.

But why would we protect certain statements from revi-
sion? Quine argues that we tend to protect those state-
ments that are central to our network. The statements of
logic and mathematics are central to our network in that
any revision of logic or mathematics also affects physics,
chemistry, and so on. Thus, in the interests of minimal
mutilation, we protect logic and mathematics. Quine writes

...the more fundamental a law is to our
conceptual scheme [theoretical network],

the less likely we are to choose it for
revision. When some revision of our sys-
tem of statements is called for, we pre-
fer, other things being equal, a revision
which disturbs the system least. ...Con-
jectures of history and economics will be
revised more willingly than laws of physics,
and these more willingly than laws of math-
ematics and logic. Our system of statements
has such a thick cushion of indeterminacy,
in relation to experience, that vast domains
of law can easily be held immune to revision
on principle. We can always turn to other
quarters of the system when revisions are
called for by unexpected experiences. Math-
ematics and logic, central as they are to
the conceptual scheme, tend to be accorded
such immunity, in view of our conservative
preference for revisions which disturb the
system least; and herein, perhaps, lies the
"necessity" which the laws of mathematics
and logic are felt to enjoy. [23]

But this same point can apply to statements other than

86
those of logic and mathematics. Lakatos points out that we
also protect certain statements in science. According to
Lakatos, the history of science is actually the history of
scientific research programmes [sic], where each programme
consists of a succession of related theories. Within each
programme, there are certain statements that we protect
from revision. Lakatos describes these statements as the
"hardcore" of the research programme. Whatever changes
need to be made within a programme, these statements will
be held immune. Consider the following three statements:
(1) Heisenberg's Uncertainty Principle:
We cannot measure the position and
the momentum of a particle simultan—
eously with precision.
(2) Evolution occurs by the mechanism of
natural selection.
(3) To every action, there is an equal
and opposite reaction.
(1) belongs to the hardcore of quantum mechanics. (2) be-
longs to the hardcore of Darwinian theory. And (3) belongs
to the hardcore of Newtonian mechanics. Whatever changes
need to be made in quantum mechanics, Darwinian theory, and
Newtonian mechanics, these statements will be held immune.
If we revise these statements, or give them up, we have,
in effect, moved to a new research programme.
While we may view the statements of logic and mathema—
tics as central to our overall theoretical network, we may
view the hardcore of a research programme as central to

that programme. If we were to revise the statements of

logic and mathematics, then we end up by revising our

87

entire network. Similarly, although on a smaller scale,
if we revise the statements that make up the hardcore of

a research programme, then we end up revising the disci-
pline in which that programme occurs. If we give up cer—
tain principles in logic and mathematics, we have, in
effect, moved to a new network. Similarly, if we give up
the hardcore of a programme, then we have moved to a new
programme. The role of a hardcore is analogous to the
role of logic and mathematics within our overall theoreti-
cal network. It is only the relevant body of knowledge
that is different. In the case of logic and mathematics,
the relevant body of knowledge is our entire network while
in the case of a hardcore, the relevant body of knowledge
may be only a particular discipline.

But this means that we can describe certain statements
from science as necessary. Quine, at any rate, is willing
to grant scientific statements the status of necessity when
he writes

In principle, therefore, I see no higher

or more austere necessity than natural

necessity; and in natural necessity, or

our attributions of it, I see only Hume's

regularities, culminating here and there

in what passes as an explanatory trait or

the promise of it. [24]
We may be more willing to revise or give up the hardcore of
a particular research programme than we are willing to re-

vise or give up the statements of logic and mathematics.

We may even concede that the statements of logic and

88

mathematics are more necessary than the statements that
make up the hardcore of a particular research programme.
But once again, this is a difference of degree and not of
kind. Logic and mathematics are necessary for the same
kind of reason as certain scientific statements are

necessary.

Chapter Four

The network theory and the causal theory are both
plausible. Yet, as I have presented them, they are incom-
patible. According to the network theory, every term ob-
tains its meaning from its sense (which is given by its
role and position within a theoretical network); but ac-
cording to the causal theory, some terms obtain their mean-
ing from their reference. Despite this apparent contradic-
tion, many philosophers seek to combine the two theories.
This is not surprising since each theory, to a certain
extent, complements the other.

It would be quite convenient if we could place the
causal theory within some sort of theoretical network. In
chapter two, we learned that the causal theory presupposes
essentialism when it comes to statements such as

(1) Heat is the form of energy constituted
by the motion of atoms and molecules
in solids.

(2) Gold is the element with atomic number

79.
(3) Water is H O.

2
But essences and essential properties are essential only
within some sort of theoretical context. Without this
context, we have no reason to suspect that certain proper-
ties are more important than other properties. Statements
(1)-(3), for example, provide the essence of their subjects
only within a theoretical context where we are interested

in the internal structure of heat, gold, and water. With—

out a particular theoretical network in place, we have no

89

90

reason to claim that the internal structure of heat, gold,
and water is more important than their other properties.

It would also be convenient if we could incorporate
features from the causal theory into the network theory.
If we accept the network theory, as I have presented it in
Chapter Three, then every time we make an adjustment within
our network, we change the meaning of every term within
that network. Since every term is related, directly or
indirectly, to every other term within its network, if we
change the meaning of one term, we change the meaning of
all. But this is preposterous. Most of us tend to believe
that certain terms retain their meaning throughout changes
within our network.

In "The Meaning of Meaning", Putnam argues that
'xpuoos', which occurs in Archimedes's dialect of Greek,
and 'gold', which occurs in our dialect of English, have
the same meaning. We certainly translate 'Xpuods' as
'gold'. But Archimedes's scientific beliefs are not ours.
As Putnam points out, 'xpuoos' and 'gold' not only belong
to different languages but to different networks. If the
network theorists are correct and every term obtains its
meaning from its role and position within a theoretical
network, then 'Xpuods' and 'gold' should not have the
same meaning. But we think that they do. While we admit
that Archimedes's beliefs about Xpuoos differ from our

beliefs about gold, we, or, at least most of us, do not

91

admit that Archimedes means something different by 'xpuods'
than we mean by 'gold'. Putnam states that this is because
'xpuohs' and 'gold' are both natural kind terms; and as
such, they obtain their meaning from their reference. And
so long as 'Xouoos' and 'gold' have the same reference,
they have the same meaning.1

If certain terms within our theoretical network ob-
tained their meaning from their reference, then these terms
would not only retain their meaning throughout changes
within our network but they would also retain their meaning
as we move from one network to another. Thus, it would be
quite convenient if reference had some sort of role within

the network theory.

F¢llesda1= How to Synthesize the Causal Theory and the
Network Theory

Although the causal theory and the network theory are
incompatible as I have presented them, offhand, it does not
seem as though they must be incompatible. After all, the
main point to the network theory is that every term obtains
its meaning from its role and position within a theoretical
network. But this point does not imply that every term
must have the same role. Perhaps our network has different
classes of terms. Perhaps these different classes play
different roles. And perhaps the role that certain classes
play is referential.

In Chapter Three, we examined the epistemological and

92
semantical aspects of the network theory. But there is an

ontological aspect as well. Our theoretical network, after
all, provides us with our ontology. It is our theoretical
network that tells us what exists and what does not. And
among the things that exist, according to our network, are
objects. We analyze the world in terms of objects. Our
knowledge, from ordinary beliefs to scientific theories,
concerns objects of one sort or another.

Objects are rather interesting entities. Follesdal,
in "Essentialism and Reference", argues that objects have
three features that are particularly pertinent to refer-
ence. First, objects have a great many properties, some of
which we know and others of which we do not. Second, many
objects change through time—-an object may have a property
at one time that it lacks at another. Third, objects are
the sorts of things about which we can have false beliefs--
we may believe something of an object at one time that we
dismiss as false at another. Given the importance of ob-
jects within our ontology and given the fact that objects
have these features, F¢llesdal argues that "we should
expect a language to have a category of expressions that
is especially designed to refer to these objects and to
stay with them through all these changes that they and our
beliefs about them undergo."2 In effect, F¢llesdal is
claiming that within our theoretical network we want cer-

tain terms to serve as rigid designators. We want certain

93
terms to obtain their meaning from their reference.

But Follesdal is no orthodox proponent of the causal
theory. According to Follesdal, the causal theorists
discern two questions. There is an ontological question:
How does a term obtain its meaning? And there is an epis-
temological question: How do we come to know the object
designated by a given term? Fpllesdal believes that the
causal theorists are principally concerned with the onto-
logical question. They wish to explain how certain terms
obtain their meaning. Or to be more precise, they wish to
explain the connection between certain referential terms
and their objects. And they explain this connection by
means of the causal theory.3 An object undergoes a
baptism--we decide that a particular name or term will
designate a particular object or a particular kind of
object. After this baptism, the name is spread throughout
the linguistic community. But, as F¢llesdal points out,
this account only settles the ontological question.4 It
explains how I use the term 'Kripke' to refer to Kripke,
for example; but it does not explain how I came by my
knowledge of Kripke.

One problem pops up immediately. While this account
seems to work for proper names, where there is usually some
sort of naming ceremony, it does not work quite so well for
natural kind terms. As Zemach states

... no one knows, nor can ever hope to
know, on what occasions and with res-

94

pect to what objects our ordinary Eng-

lish substance—terms were first utter—

ed. We do not even know whether the

substance that was originally referred

to as "water" was indeed water (i.e.,

was the substance we call "water").

According to this version of Putnam's

theory, it is possible that we, all of

us are utterly mistaken, and what we

call "water" is simply not water. It

is possible that only, say, spittle,

or milk, is water, and nothing else

is. [5]
Zemach's point does not necessarily undermine the rigidity
of the term 'water'. Perhaps Zemach is correct that while
we use 'water' to refer rigidly to water, the originators
of the term used 'water' to refer rigidly to spittle or
milk; but in either case, 'water' is a rigid designator.
His point does raise a problem, however, if we are trying
to explain how 'water' came to refer to water. His point
intimates that perhaps the ontological and epistemological
issues of the causal theory are not as distinct as the
causal theorists have presented them.

F¢llesdal collapses the causal theorist's distinction
between ontological and epistemological questions. He
believes that how we learn to connect a particular term
with a particular object may also explain how that term is
related to the object. He argues that this is because
language is a social institution. "What our names refer
to--and not only how we find out what they refer to--de-

pends upon evidence that is publicly available in situa-

tions where people learn and use language."6 He continues

'OE

sag

Ref

War.
the
Cha

Chal

95

to stress this point when he writes

For both what a term refers to and how
we find out what it refers to depend on
a complex interplay of several factors:
assent to and dissent from sentences is
just one; ostension is another; actions,
including non-linquistic actions, are a
third; and above all, our theories of
how people are likely to go wrong in
their perception and their reasoning are
a fourth. Here interaction within the
whole speech community comes in. Some
people are less likely to go wrong in
certain matters, because they are better
located for perception and observation,
because they are better trained and per-
haps specialists on these matters, etc.
This "linquistic division of labor" was
first observed by Putnam. However, nei-
ther he nor Kripke will probably accept
my view that it actually contributes to
determining the reference and not just to
finding out what the reference is. [7]

 

This suggests that our theoretical network may not only
have a class of terms that play a referential role within
the network; but it also suggests that our theoretical
network may, in turn, play a role in determining the ref—
erence of these terms. Unfortunately, F¢llesdal, in "Es-
sentialism and Reference", does not follow up on this

suggestion.

Referential Terms

F¢llesdal's point is this: Within our network, we
want a class of terms that refer to objects; and we want
these terms to retain their meaning not only throughout the
changes that objects may undergo but also throughout the

changes that our beliefs about these objects may undergo.

96

But Follesdal does not insist that these terms obtain their
meaning from their reference. And for good reason. We can
use a term referentially even if that term does not obtain
its meaning from its reference. In reality, the terms that
we use referentially fall upon a continuum where sense,
which is ultimately tied to our network, gradually takes a
larger and larger component of meaning. Examine the fol-
lowing classes of terms:
(i) Currer Bell, Charlotte Bronte, Jolmo
Lunmga, Mt. Everest, Malvinas Is-
lands, Falkland Islands.
(ii) Mallard, garlic, snow rose, gold,
water, elephant, caribou, reindeer.
(iii) Anas platyrhynchas, Allium sativum,
Hellaborus niger, Rangifer tarandus,
Elephas maximus, Loxodonta elephanta.
(iv) atom, electron, quark, gene.
(v) group, ring, triangle, square.
(vi) vertebrate, marsupial, mammal, coe-
lenterate, chordata.
(vii) bachelor, vixen, brother.
Each class consists of terms that we use referentially.
But not every class consists of terms that obtain their
meaning from their reference. While the terms in class (i)
obtain their meaning from their reference, the terms in
class (vii) do not. In fact, as we move from class (i) to
class (vii), we gradually move from terms that obtain their
meaning from their reference to terms that obtain their
meaning from their sense. As we move from one class to the
next, sense gradually becomes more important.

Class (i) consists of proper names. And when it comes

to proper names, the causal theory is quite plausible.

97

Thanks to F¢llesdal, we can now show this account of proper
names is not incompatible with the network theory. We can
argue that proper names play a particular role within our
network. They are those terms that rigidly designate their
objects; and since they are rigid designators, they obtain
their meaning from their reference. But even when it comes
to proper names, we cannot ignore our theoretical network
entirely. When we use proper names, we implicitly rely on
various metaphysical assumptions—~for example, we assume
that the objects we have named endure through space and
time; we assume that these objects are identical to them-
selves; we assume that these objects cannot be in two
places at the same time; and so on. Indeed, unless we rely
upon these sorts of assumptions, we have little reason to
adopt proper names.

Class (ii) consists of natural kind terms. These
terms, according to the causal theory, also obtain their
meaning from their reference; and we may, in fact, find the
causal theory quite persuasive on this fact. Yet when we
try to explain exactly what constitutes a natural kind, we
begin to see just how difficult it is to divorce natural
kinds from the background knowledge that is provided by our
theoretical network. This is because a kind is made up of
things that we find similar. But the fact that we find
certain things similar suggests that we have some reason

to find them so.

 

98
Initially, it may seem as though we do not need any

background knowledge to divide up the world into natural
kinds. Many philosophers and scientists believe that cer-
tain similarity relations, in particular the similarity
relations that cause us to perceive the members of a natu—
ral kind as similar, are innate. Quine, for example,
writes that

A standard of similarity is in some sense

innate. This point is not against empiri-

cism; it is a commonplace of behavioral

psychology. ... Without some such prior

spacing of qualities, we should never ac-

quire a habit; all stimuli would be equal—

ly alike and equally different. [8]
This claim becomes more plausible once we realize that even
small children and animals divide the world into many of
the same natural kinds as adult humans. Now surely small
children and animals are not relying upon a full blown
theoretical network to divide the world up into these nat—
ural kinds. Rather, it seems more plausible that there is
some sort of innate mechanism that causes even small chil-
dren and animals to perceive some things as similar and
others as not.

But this innate mechanism only takes us so far. Per—

haps the similarities that are due to this innate mechanism

are only superficial. Mayr points out that

The history of taxonomy decisively re-
futes the assumption that similarity is
self-evident and not in need of careful
evaluation. It requires a great deal

of knowledge to be able to look through
superficial similarities and to discount

99

superficial dissimilarities. [9]
Viceroy and Monarch butterflies appear remarkably similar;
yet once we know something about butterflies, we distin-
guish between the two. Male and female mallards appear
quite dissimilar; yet once we know something about birds,
we place them in the same kind. While we may begin with
innate notions of similarity, we also rely upon our theo—
retical network to help us sort things into kinds.

But we are also realists when it comes to natural
kinds. We do not think that they are simply creations of
our minds. Rather, we believe that there is a reason why
things belong to the natural kind that they do; and this
reason is something other than the fact that we perceive
them as similar. Putnam writes that natural kinds are
"classes of things that we regard as of explanatory impor-
tance; classes whose normal distinguishing characteristics
are 'held together' or even explained by deep-lying mech-

anisms."10

In yet another article, Putnam writes that
To belong to a natural kind, something
must have the same composition, or obey
the same laws--indeed, what makes com-
position important, when it is, is its
connection with laws of behavior.--as
model members of the class, and this
composition or these laws are not usu—
ally known when the natural kind term
is introduced, but requires an inordin-
ate amount of investigation to disco-
ver. [11]

It may be true that we initially perceive natural kinds on

the basis of some innate mechanism. But it is also true

100

that we believe that these natural kinds play an explana—
tory role within our network. And we then use our network
to help us place things into kinds. For example, we use
our knowledge of evolution and reproduction to place male
and female mallards within the same kind; and we use the
same sort of knowledge to separate Viceroy and Monarch but-
terflies into two different kinds.

The fact that our theoretical network comes into play
for natural kind terms is not surprising if we remember the
discussion on paradigms in Chapter Two. Although natural
kind terms obtain their meaning from their reference, their
reference is not fixed beforehand. Thus, we choose para—
digmatic examples of the kind, and by characterizing that
example, we can find other members of that kind. But in
characterizing our example, we rely upon our overall know-
ledge. We use our theoretical network. This is not to
deny that reference is the principle component of meaning
for natural kind terms. We may be willing to make minor
adjustments in the reference of a natural kind term on
the basis of our theoretical network; but we would reject
major adjustments.

Class (iii) consists of taxonomic names for biological
species. While biological species match up, for the most
part, with biological kinds, this match up is not perfect.
We often divide a natural kind into two or more species.

This is the case with elephants. We divided elephants, the

101

kind, into two species: African elephants, or elephas max-

 

imus, and Indian elephants, or loxodonta elephas. We may

 

also combine two natural kinds into one species. While we
may originally have thought that reindeer and caribou, for
example, constituted two kinds, we now count them as one

species--Rangifer tarandus. But these cases are relatively

 

minor. For the most part, biological kinds and biological
species match up fairly well.

And this is what we should expect. Quine points out
that "we revise our standards of similarity or of natural

kinds on the basis of second-hand inductions."12

He argues
that we rise above our innate notions of similarity by

...developing modified systems of kinds;

hence modified similarity standards for

scientific purposes. By the trial-and-

error process of theorizing [we have] re-

grouped things into new kinds which prove

to lend themselves to many inductions bet-

ter than the old. [13]
Taxonomy, at the level of species, is intended to be an im—
provement over natural kinds. In sorting out the world, we
begin with natural kinds. But we do not stop with this
initial sorting. We try to make our notion of natural
kinds more precise. And to do this, we rely more upon our
background knowledge.

When we divide things into biological species, we def-

initely rely upon scientific knowledge. We make use of

genetic testing and evolutionary history. We examine the

morphological characteristics of the entities in question

102
and their life—cycle. We study their reproductive tech-
niques. And this knowledge is reflected as we place things
into species. But still, throughout this process, we are
not so much rejecting natural kinds as we are refining
those kinds. And thus, we can argue that the meanings of
the terms in this class are primarily due to reference.

Class (iv) consists of the names for theoretical
kinds. Obviously these terms depend upon our theoretical
network in one respect. Before we accept the existence
of the entities named by these terms, we must first accept
certain theories. Yet once we adopt these entities within
our ontology, the terms that name them function much like
natural kind terms. These terms are rigid designators.
Although our beliefs about atoms, electrons, quarks, and
genes are continually changing, we use the terms 'atom',
'electron', 'quark', and 'gene' to refer to atoms, elec-
trons, quarks, and genes respectively. The meaning of
these terms is given by their reference.

But while reference may provide the major component of
meaning for these terms, sense also provides a component.
Consider the following passage from Putnam:

... I do not see much point, for example,
in saying that someone is referring to
Quine when he uses the name 'Quine' if he
thinks 'Quine is a Roman emperor' and that
is all he "knows" about Quine; unless one

has some beliefs which are true or approx—

imately true, then it is at best idle to
consider that the name refers to the bear—
er of one's idiolect. [14]

103
When it comes to proper names, my intuitions are not the
same as Putnam's. My intuition is that the person who mis-
takenly believes that Quine is a Roman emperor still refers
to Quine when he uses the name 'Quine'. The reason why the
statement 'Quine is a Roman emperor' is false is because
'Quine' refers to Quine; and governing the Roman empire is
not one of Quine's accomplishments. But when it comes to
the terms for theoretical kinds, my intuitions are with
Putnam. When it comes to theoretical kinds, we had better
have some beliefs that are approximately true.

Consider the Greek term 'arouos' and the English term
'atom'. We certainly translate 'aTouos' as 'atom'; but do
these two terms have the same meaning? Do Leucippus and
Democritus mean the same thing by 'aTouos' that we mean by
'atom'? My intuition is that they do not. This intuition
may seem rather strange considering that I agreed with Put—
nam that 'Xpuoés' and 'gold' have the same meaning. And I
agreed with Putnam that these two terms have the same mean-
ing even though Archimedes's beliefs about xpuoos may dif-
fer from our beliefs about gold. Thus, it seems rather
strange that I willing to argue that 'xpuohs' and 'gold'
designate the same natural kind but 'aTouos' and 'atom'
designate two different theoretical kinds.

Yet once we accept F¢llesdal's claim that how we learn
to connect a particular term with a particular object also

explains how that term is related to this object, this

104

intuition becomes less strange. Our ontology includes both
natural and theoretical kinds. But a theoretical kind does
not enter within our ontology in quite the same way as a
natural kind does. Natural kinds precede theory. We ac-
cept the existence of the kind first and then develop a
theory about that natural kind. But theory precedes
theoretical kinds. We usually add theoretical kinds to

our ontology as hypothetical entities. We accept their
existence before we prove their existence. Yet how do we
prove the existence of a theoretical kind? We prove the
existence of a theoretical kind once we find an entity

that fits the description given by our theories. Perhaps

a term for a theoretical kind functions as a rigid designa-
tor; but it can only serve as a rigid designator once we
fix its reference. And we do this by means of the theories
within our theoretical network.

This seems to suggest that 'aTouos' and 'atom' desig-
nate different theoretical entities. Leucippus and Demo-
critus argued for the existence of a particular kind of
entity. This entity came in various sizes and shapes. It
was indivisible. It was the smallest unit of matter. And
it built up everything else in the world. Yet while Leu-
cippus and Democritus posited the existence of this entity,
they never found a referent. They never found an indivisi-
ble unit of matter that came in various sizes and shapes.

In contrast, modern proponents of the atomic theory not

105
only posited the existence of a particular kind of entity
but they also found the referent.

This point, by itself, is not sufficient to prove that
'aTouos' and 'atom' designate two different kinds. After
all, we may be inclined to argue that Boyle, Newton, and
Dalton meant the same thing by 'atom' that we do although
they too failed to find a referent. But when we compare
our atomic theory with the atomic theory associated with
Leucippus and Democritus with our atomic theory, it seems
as though we have two different research programmes. Con-
sider the following statement:

(4) An atom is an indivisible unit of matter.

If we compare atomic theories, this statement belongs to
the hardcore of Leucippus's and Democritus's research pro-
gramme. Moreover, within their programme, this statement
is not only true, it is analytically true. In their pro-
gramme, the term 'aTouos' was a one-criterion word.15 Now
in our version of the atomic theory, (4) hardly constitutes
part of our programme's hardcore. We not only believe that
this statement is synthetic, we believe that it is false.
Given that we have two different research programmes, and
given that our research programmes determine our ontology,
it does not seem unreasonable to argue that 'aTouos' and
'atom' designate two different theoretical kinds.

The terms in group (v) designate mathematical objects

and structures. These terms are particularly interesting

106

in that they straddle Putnam's distinction between one-
criterion words and law-cluster words. Consider the terms
'triangle' and 'group'. Each term has a definition.
(a) A triangle is a plane figure with an
area enclosed by three straight lines.
(b) A group is a non—empty set that is
closed under a binary operation that
meets the following conditions:
(i) The associative law holds.
(ii) There is a unit element.
(iii) Every element has an inverse.
These terms resemble one-criterion words in that their
reference is determined fully by their sense; and this
sense is given by these definitions. But these terms
resemble law-cluster terms in that they take part in a

great many mathematical and physical laws.

The terms in group (vi) designate taxonomic entities
other than species. There are two activities associated
with taxonomy: microtaxonomy and macrotaxonomy. In micro-
taxonomy, taxonomists distinguish one species from another
while in macrotaxonomy, they group species together into
higher taxa. Mayr argues that

... there is a drastic difference between
the species taxon and the higher taxa.
Higer taxa are defined by intrinsic char-
acteristics. Birds is the class of fea-
thered vertebrates. Any and all species
that satisfy the definition of "feathered
vertebrates" belong to the class of birds.
An essentialist (typological) definition
is satisfactory and sufficient at the
level of the higher taxa. [16]

While at the species level, reference is the primary com-

ponent of meaning, at the leve of the higher taxa, sense

107

becomes the primary component of meaning. The reference
of these terms can change without affecting their meaning.
Consider the term 'coelenterate'. The usual definition
of 'coelenterate' is the following:
The phylum of coelenterates are those
invertebrates that possess a radially
symmetrical body with a saclike inter-
nal cavity.
Twenty years ago, this phylum included combjellies. These
days, it does not. Yet while the reference of 'coelenter-

ate'

has changed, its meaning has not. This is what we
should expect when sense provides the predominant component
of meaning.

The terms in class (vii) are one—criterion words. We
use these terms to refer to whatever objects satisfy their
criterion. So, the sense of these terms determines the
reference. After all, the meaning of 'bachelor' does not
change one whit as various bachelors marry. While the
reference of 'bachelor' is continuously changing, its

meaning does not. And this is because its meaning is

determined by its sense.

What Does This Do To The Traditional Distinctions?
Although we can now argue that reference provides the
meaning for certain classes of terms within our network,
this has little impact on either the analytic/synthetic
distinction or the a priori/a posteriori distinction.

These distinctions remain much as they have been drawn by

108

the network theorists. We can continue to argue that
certain statements within our network, those that meet
Putnam's four criteria, are analytic. And we can continue
to argue that certain statements within our network are a
priori relative to other statements within our network.

So long as we continue to treat the analytic/synthetic
distinction and the a priori/a posteriori distinction as
pragmatic distinctions, this synthesis has no impact.

But this synthesis has a major impact on the neces-
sary/contingent distinction. The advantage of this syn-
thesis is that it allows us to keep the more plausible
features of the various accounts of necessity that we
have examined up until now. The disadvantage of this
synthesis is that our notion of necessity becomes more
complex. We can no longer produce one overall all-encom—

passing account of necessity.

Types of Necessity
We would like to have one all-encompassing account of
necessity. But consider the following statements:

(a) A bachelor is an unmarried male.

(b) A vertebrate is an animal with a
backbone.

(c) A triangle is a plane figure with
an area enclosed by three straight
lines.

(d) f=ma.

(e) The snow rose is the Hellaborus
niger.

(f) Currer Bell is Charlotte Bronte.

(9) Gold is the element with atomic num-
ber 79.

109
While we can argue convincingly that each statement is nec-
essary, we cannot use the same argument to justify the nec—
essity of each statement. Each statement may be necessary;
but they are necessary for different reasons.

But this does not mean that we need seven accounts of
necessity. Consider statements (a) and (b). Statement (a)
belongs to the class of "analytic" statements, a class that
includes statements such as:

(a) A bachelor is an unmarried male.
(a1) Vixens are female foxes.
(a2) A brother is a male sibling.
Statement (b) belongs to the class of taxonomic identities
for higher taxa, a class that includes statements such
as:
(b) A vertebrate is an animal with a
backbone.
(b1) Marsupials are mammals with exter-
nal abdominal pouches.
(b2) The phylum of coelenterates are
those invertebrates that possess
a radially symmetrical body with
a saclike internal cavity.
Now while these statements belong to different classes, I
believe that they are necessary for much the same reason.

While there are certain differences between these two

classes, there are also similarities. These similarities

become apparent once we examine the chart on the following

page:

110

(a) (b)

 

analytic A bachelor is an unmar— I A vertebrate is an ani-
ried male. mal with a backbone.
' one-criterion I law—cluster

figure 2

In chapter three, we learned that 'bachelor' is a one-cri-
terion word. We also learned that (a) is analytic because
its subject is a one-criterion word and its predicate gives
that criterion. It would be convenient if we could argue
that 'vertebrate' is also a one-criterion word. Unfortun-
ately, it is not. Just consider the following statements:
(b3) Vertebrates have gill slits in their
pharnyx at some stage in their life—
cycle.
(b4) Vertebrates have a digestive tract.
(b5) Vertebrates reproduce sexually.
(b6) Vertebrates share a common evolution—
ary history.
As we can see, 'vertebrate' occurs time after time in
biological theories. But while we cannot argue that
'vertebrate' is a one-criterion word, statement (b) seem-
ingly resembles (a)-(a2) more than it resembles (b3)-(b6).
The fact that (b) resembles (a) is not surprising.
This is because 'vertebrate', while it is not a one—criter-
ion word, resembles 'bachelor'. We use both terms to
designate certain classes of objects. Within our network,
it is convenient to have a term that refers to unmarried

males; and within our network, it is convenient to have a

term that refers to animals with a backbone. But while

111
'bachelor' and 'vertebrate' have a referential use within
our network, they obtain their meaning from their sense.
Before we categorize something as a bachelor or as a verte-
brate, that thing must first meet certain conditions. To
be a bachelor, it must be an unmarried male; and to be a
vertebrate, it must be an animal with a backbone. Although
'vertebrate' may not be a one-criterion word, there is one
criterion that something must meet before we categorize it
as a vertebrate. And in that respect, 'vertebrate' resem-
bles 'bachelor'.

When it comes to statements (a) and (b), the logical
positivists give the correct analysis for their necessity.
Statements (a) and (b) are analytic. And because they are
analytic, they are necessary. Borrowing Aristotelian term-
inology, we can argue that (a) and (b) express the nominal
essence of their subjects. They express the "essence" of
their subjects in that they give the meaning of the terms
'bachelor' and 'vertebrate'. And so long as these terms
are defined in terms of a single criterion, and so long as
(a) and (b) state this criterion, (a) and (b) will be true.
They can only be false if we change the meaning of 'bache-

lor'

and 'vertebrate'. Thus, their analyticity not only
proves their necessity but gives the explanation for their
necessity.

It is tempting to make the same argument for

statements (c) and (d). It is tempting to argue (c) and

112
(d), along with (a) and (b), express the nominal essence
of their subjects-~while (c) gives the meaning for 'tri—
angle', (d) gives the meaning for 'force'. It is tempting
to argue that (c) and (d), along with (a) and (b), are
analytic. And it is tempting to argue that (c) and (d)
are necessary in virtue of this analyticity.

This is certainly the sort of argument that the logi-
cal positivists would make. And in the case of (c) and
(d), such an argument is plausible. In reference to (c),
we already know that the logical positivists believe that
the statements of logic and mathematics are all analytic.
And while we have yet to note it, any quick perusal through
a physics textbook quickly reveals that definitions have
a role in science. Quine remarks that

The less a science has advanced, the

more its terminology tends to rest upon

an uncritical assumption of mutual un-

derstanding. With increase of rigor

this basis is replaced piecemeal by the

introduction of definitions. The in-

terrelationships recruited for these

definitions gain the status of analytic

principles, what was once regarded as a

theory about the world is reconstructed

as a convention of language. [17]
It is true that when the logical positivists reconstructed
scientific theories into axiomatic structures, they argued
that certain statements were analytic—-certain statements
within the structure were true by definition. And such

statements, according to the logical positivists, would be

necessary in virtue of this fact.

113

But if we wish to make this argument, we soon run into
a problem. What happens when we change our theories? As
we know from chapter three, statements within our scienti-
fic theories can change their status. A statement that we
accept as a definition at one time, we can accept or reject
on empirical grounds at another time. Putnam says that

The distinction between statements nec-
essary relative to a body of knowledge
and statements contingent to that body

of knowledge is an important methodolo-
gical distinction and should not be jet—
tisoned. But the traditional philoso-
phical distinction between statements
necessary in some eternal sense and
statements contingent in some eternal
sense is not workable. The rescuing

move which consists in saying that if a
statement which appears to be necessary
relative to a body of knowledge at one
time is not necessary relative to the
body of knowledge at a later time, then
it is not really the same statement that
is involved, that words have changed
their meaning, and that the old state-
ment would still be a necessary truth if
the meanings of the words had been kept
unchanged, is unsuccessful. The rescu—
ing move which consists in saying that
such statements were only mistaken to be
necessary truths, that they were contin-
gent statements all along, and that their
'necessity' was merely 'psychological' is
just the other side of the same blunder.
For the difference between statements that
can be overthrown by merely conceiving of
suitable experiments and statements that
can be overthrown only by conceiving of
whole new theoretical structures-~some-
times structures, like Relativity and
Quantum Mechanics, that change our whole
way of reasoning about nature--is of log-
ical and methodological significance, and
not just of psychological interest. [18]

Putnam is making at least two points. His first point is

114

that a statement may be necessary for some reason other
than meaning. If we were to give up statement (a), for
example, then 'bachelor' has changed its meaning; but if

we were to give up (d), perhaps the meaning of 'force' has
changed. But if we were to give up (d), we would also have
modified our entire system of physics.

Putnam's second point is particularly pertinent for
the network theory. If we reject (d), this does not mean
that its necessity was only psychological. It could still
be necessary relative to a certain body of knowledge. This
is an important point. Compare (c) and (d) with (a) and
(b). As we can see from the following, one difference is

readily apparent.

 

Trivial Central
(a) (C)
A bachelor is an unmarried A triangle is a plane figure
male. with an area enclosed by

three straight lines.

(b) (d)
A vertebrate is an animal f=ma.
with a backbone.

 

 

figure 3

While (a) and (b) play only a trivial role within our theo-
retical network, (c) and (d) play a relatively important
role. They are either central to our network as a whole or

they are central to a particular research programme within

115

our network. And because they are central, the network

theory gives the best account for their necessity.
Consider (d). Lakatos argues that (d), as one of

Newton's three laws of dynamics, constitutes part of the

'9 Now while

hardcore of Newtonian gravitational theory.
(d) appears to be a definition, not every statement that
constitutes part of the hardcore of a research programme

need be a definition. Consider the following:

(d ) Evolution occurs by the mechanism of
natural selection.

In chapter three, this statement was described as part of
the hardcore of Darwinian theory. But this statement is
not a definition. In fact, if we only accepted the logical
postivists' account of necessity, we may even argue that
this statement is contingent. But if we consider the
importance of this statement for a particular theory of
evolution, we can argue that it is necessary relative

to that theory. And we can make the same argument for

(d).

The network theorists also argue that the statements
of logic and mathematics owe their necessity to the central
role that they play within our network. But if we examine
various statements from logic and mathematics, this claim
is a little disingenuous. This is because the statements
of logic and mathematics fall into various sub-divisions.
There are definitions:

(c) A triangle is a plane figure with

116

an area enclosed by three straight
lines.

(c1) A group is a nonempty set of ele-
ments that is closed under a binary
operation and that meets the follow-
ing conditions:

(i) The associative law holds.
(ii) There is a unit element.
(iii) Every element has an inverse.

There are axioms:
(c2) For all well—formed formulas, p and
q, the following holds: p+(q+p).
(c3) The axiom of infinity.
There are theorems:

(c4) 1+1=2.

(C5 ) (A&B)+A.

(c2 ) Every map on a flat surface or sphere

can be colored with no more than four
colors.
And there are metaphysical principles:
(c7 The principle of identity.
4) The principle of excluded middle.

(c8 ) The principle of contradiction.

(C10) The principle of self-identity.

(C11) The principle that any statement that
follows logically from necessary state-
ments is itself necessary.

Now do we really want to argue that every single one of
these statements owes its necessity to the central position
that it holds within our theoretical network. After all,
however useful (c6) may be for map-makers, is it really
central to our overall network? And while (c3) is one of
the axioms in Zermelo—Fraenkel's system, does it really
play an important role in physics and chemistry?

One problem is that the notion of centrality is ambig-

uous when it comes to logic and mathematics. A statement

117

may be central to our overall network; or it may only be
central to logic and mathematics as two of the disciplines
within our theoretical network. Consider (c3)--the axiom
of infinity. This axiom, as one of the axioms in the Zer—
melo-Fraenkel axiom system, plays a central role in mathe-
matics. In general, it is one of the axioms that we use to
derive other mathematical truths. In particular, it is the
axiom that guarantees the existence of infinite sets. If
we consider mathematics solely as an intellectual disci-
pline, then we could argue that (c3) is necessary only be—
cause it constitutes part of the "hardcore" of mathematics.
Now consider (c4). Since this statement is derived from
other truths of logic and mathematics, we cannot argue that
it is part of the hardcore of mathematics. But if we con-
sider our network as a whole, (c4) plays a far more impor—
tant role than (03). When it comes to balancing our check-
books, which statement do we use? (c3) or (c4)? If we
consider the role that mathematics plays within our network
as a whole, then we can argue that (c4) owes its necessity
to its central position.

To further illustrate this point, consider the "meta-
physical" principles. Where Quine places the truths of
logic at the very center of our network, I would place
these principles instead. But if we consider logic solely
as an intellectual discipline, then these principles are no

more important than any other statement of logic. Yet if

118

we consider logic as part of our overall network, these
statements play a fundamental role. Consider statements
(c7)-(c9)--the so-called laws of thought. Copi argues that
while these laws, considered merely as logical tautologies,
are no more important than any other tautology,

... the three laws of thought can be re—
garded as having a certain fundamental
status in relation to truth tables. As
we fill in subsequent columns by refer-
ring back to the initial columns, we are
guided by the Principle of Identity: if
a T has been placed under a symbol in a
certain row, then in filling out other
columns under expressions containing
that symbol, when we come to that row

we consider that symbol still to be as-
signed a T. In filling out the initial
columns, we put either a T or an F, be-
ing guided by the Principle of Excluded
Middle; and nowhere do we put both T and
F together, being guided by the Princi-
ple of Contradiction. The three Laws of
Thought can be regarded as the basic
principles governing the construction

of truth tables. [20]

The laws of thought can also be regarded as having a cer-
tain fundamental status in relation to proofs. Proofs by
contradiction, for example, implicitly rely upon the Prin—
ciple of Excluded Middle. If we reject the Principle of
Excluded Middle, as certain intuitionistic mathematicians
do, then we reject indirect proofs. Rejecting such proofs
can have a ripple effect throughout our entire network.
For example, the classical proof for the law of trichotomy
(which states that every real number is either zero, posi-
tive, or negative) is an indirect proof. This particular

law plays a fundamental role in calculus and analysis. But

CE

0:

119

calculus and analysis are important for physics as well as
mathematics. Thus, we cannot reject the Principle of Ex-
cluded Middle without affecting a great deal of our overall
network.

As a matter of fact, however, the statements that are
central to logic and mathematics as particular disciplines
within our network and the statements of logic and mathema-
tics that are central to our overall network are related in
a certain respect. The axioms of logic and mathematics are
often discursive. We choose certain "truths" of logic and
mathematics as a basis from which we can derive other
"truths". These statements are central to logic and mathe-
matics as disciplines. They constitute the "hardcore" of
logic and mathematics. But we would not choose these par-
ticular statements as our basis if they did not enable us
to derive truths that are central to our overall network.
We are not likely to choose an axiom system in mathematics,
for example, if it did not allow us to derive '1+1=2'.

While logical positivism best accounts for the neces-
sity of statements (a) and (b), and while the network
theory best accounts for the necessity of statements (c)
and (d), the causal theory best accounts for the necessity
of statements (e) and (f). Statement (e) belongs to a par-
ticular class of taxonomic statements-—the class of taxono-
mic identities for species. This class consists of

statements such as

120
(e) The snow rose is the Hellaborus
niger.
(e1) Mallards are members of the species
Anas platyrhynchas.
(e2) Garlic is Allium sativum.
Statement (f), too, belongs to a particular class of state-
ments. It belongs to the class of singular statements that
contain two rigid designators. This class consists of
statements such as
(f) Currer Bell is Charlotte Bronte.
(f1) Jolmo Lungma is Mt. Everest.
(f2) The Malvinas Islands are the Falk-
land Islands.
In fact, if we do not accept the causal theory, we are not
likely to describe these statements as necessary.
They certainly do not owe their necessity to the fact
that they play a central role within our network. The
statements in both classes are rather trivial. But these

statements are not trivial in quite the same way as state—

ments (a) and (b). Consider the following:

 

 

 

sense reference

(a) (e)
A bachelor is an unmarried , The snow rose is the Hell-
man. aborus niger.

(h) (f)
A vertebrate is an animal Currer Bell is Charlotte
with a backbone. . Bronte.

figure 4

In each case, where 'S' is a statement, we are committed to

121

the truth of the following: 'S+DS'. If S is true, then it
is necessarily true. But in the case of (a) and (b), we
are also committed to the truth of S. Statements (a) and
(b) may be trivial; but we are firmly committed to their
truth. But in the case of (c) and (d), we are not so com-
mitted to the truth of S. We are quite willing to concede
that perhaps we made a mistake in identifying Currer Bell
with Charlotte Bronte; and we are quite willing to allow
botanists to change the taxonomic name of the snow rose.
If (c) and (d) are true, they are necessarily true; but we
are willing to admit that they could be false.

This comes back to the fact that while 'bachelor' and
'vertebrate' obtain their meaning from their sense, 'snow
rose', 'Hellaborus niger', 'Currer Bell', and 'Charlotte
Bronte' obtain their meaning primarily from their refer-
ence. We are committed to the truth of (a), for example,
because we define 'bachelor' as 'unmarried man'. While we
could make the case that (e) is the taxonomic definition of
'snow rose', the fact remains that 'snow rose' does not
literally mean 'Hellaborus niger'. Nor does 'Currer Bell'
literally mean 'Charlotte Bronte'. While 'bachelor' obtains
its meaning by its verbal connection to 'unmarried man',
'snow rose' and 'Hellaborus niger' obtain their meaning
by their connection to a particular kind of plant.

Surprisingly enough, while (e) and (f) are trivial,

their necessity is ultimately tied to the account of

122

necessity given by the network theory. As we recall, the
causal theorists admit that statements (e) and (f) ulti-
mately owe their necessity to the principle of self—iden—
tity. Statements (e) and (f) are necessary only because
the principle of self-identity is necessary. And why is
this principle necessary. It owes its necessity to the
fact that it is central to our network. Thus, while (e)
and (f) are not central to our network, their necessity

is linked to our network by our reliance upon a particular
metaphysical principle.

Logical positivism accounts for the necessity of
statements (a) and (b). The network theory accounts for
the necessity of (c) and (d). The causal theory accounts
for the necessity of (e) and (f). Which theory accounts
for (g)? Intuitively, we believe that (g) owes its neces—
sity to the fact that it is an Aristotelian essence. Now,
of our three accounts of necessity, which accounts for
Aristotelian essentialism. The problem is that while we
may believe (9) is necessary, none of the accounts of
necessity that we have examined explains Aristotelian
essentialism. If we want to describe (9) as necessary,

we need an additional account of necessity.

Chapter Five
In Chapter One, I began with a relatively modest goal.
I only wanted to show that the traditional analysis of
necessity, which equated necessary statements with a priori
statements, was inadequate. This was because certain
statements, namely,
(1) Heat is the form of energy constituted
by the motion of atoms and molecules
in solids.
(2) Gold is the element with atomic number

79.
(3) Water is H O.

2
are both necessary and a posteriori. By the end of Chapter
One, my only goal was to modify this traditional analysis
so that it could account for these statements. But if the
discussion is chapter four is correct, which I believe it
to be, then no slight modification will do.

In Chapter Four, I made two claims: First, I claimed
that necessary statements fall into at least seven differ—
ent classes:

(1) "Analytic" statements.
(ii) Taxonomic identities for the higher
taxa.
(iii) Logical and mathematical truths.
(iv) The hardcore of a scientific research
programme.
(v) Taxonomic identities for species.
(vi) Singular identity statements contain—
ing two rigid designators.
(vii) Aristotelian or "metaphysical" essences.
Second, I claimed that we cannot reduce these seven classes
to just one epistemically or semantically homogeneous class

of statements. If the statements in these various classes

123

124

are necessary, then they are necessary for different rea-
sons. In Chapter Four, we accounted for the necessity of
the statements that fell into the first six classes. But
we have yet to account for the necessity of those state—

ments that fall into the seventh class. In this chapter,
we shall finally discover why statements (1)-(3) are nec—

esary.

Why We Need An Additional Analysis of Necessity
Intuitively, we feel that statements (1)-(3) are nec-
essary because they express some sort of Aristotelian es-
sence--they express the "real" or "metaphysical" nature of
their subjects. Yet none of the analyses of necessity that
we have seen up until now accounts for Aristotelian essen-
tialism. If we wish to argue that statements (1)—(3) owe
their necessity to the fact that they express the essence
of their subjects, then, as we shall see, we need an addi-
tional analysis of necessity.
Why is logical positivism inadequate? Consider the
following two statements:
(2) Gold is the element with atomic number
(4) ngachelor is an unmarried man.
These statements resemble each other in that each seemingly
expresses the essence of its subject. Yet this resemblance
is superficial. Statement (4) only expresses the nominal

essence of its subject while (2) expresses the

125

"metaphysical" essence of of its subject. (4) only tells
us what a bachelor is by stating the definition for the
word 'bachelor'. (2), in contrast, tells us what gold is
by giving us its internal constitution--it does not tell us
what gold is by merely stating the definition of the word
'gold'. If it did, then someone who fails to know that
gold is the element with atomic number 79 would also fail
to know the meaning of 'gold' just as someone who fails to
know that a bachelor is an unmarried male fails to know the
meaning of 'bachelor'. But this is not so. And this sug—
gests that while (4) owes its necessity to some sort of
linquistic practice, (2) does not. Thus, any account of
necessity that justifies necessity solely in terms of lin—
quistic conventions and practices, such as logical positi-
vism, will be inadequate.

Why is the causal theory inadequate? The causal
theorists, at the very least, concede that statements (1)-
(3) express Aristotelian essences. But this is only be—
cause the causal theory presupposes essentialism. Consi—
der once more the discussion from Chapter Two. Since
'heat', 'gold', and 'water' obtain their meaning from
their reference, they are rigid designators. As it so
happens, certain definite descriptions, such as 'the form
of energy constituted by the motion of atoms and molecules
in solids', 'the element with atomic number 79', and 'H20"

are also rigid designators. These particular descriptions

126

pick out the same kind or the same phenomena in every pos—
sible world. Yet these descriptions refer rigidly only
because they happen to express the essence of their sub-
jects. The causal theorists, in effect, are presupposing
essentialism when they argue that statements (1)-(3) are
necessary. But as Putnam points out, presupposing essen-
tialism is not quite the same thing as justifying essen-
tialism. If we examine the account in Chapter Two, Putnam
writes that

The difficulty is that Kripke individ-

uates objects by their modal properties,

by what they (essentially) could and

could not be. Kripke's ontology pre-

supposes essentialism; it cannot be

used to ground it. [1]
Thus, if we wish to justify essentialism, we need some
analysis of necessity in addition to the causal theory.

Why is the network theory inadequate? Offhand, it

appears as though the network theory can fully account for
Aristotelian essentialism. We certainly can use the net—
work theory to argue that statements (1)-(3) are necessary.
We can argue that thse statements either constitute part
of the hardcore of a research programme or they follow
logically or causally from some hardcore. Consider (2),
for example. We know from the periodic table of elements
that gold is the element with atomic number 79. And if
anything counts as the hardcore in chemistry, surely it

is the periodic table of elements. While neither (1) nor

(3) constitute part of the hardcore of a research programme

127

in either physics or chemistry, they follow causally from a
hardcore. The problem, however, is while this account can
explain why statements (1)-(3) are necessary, it does not
explain why they are essential. We have many research
programmes in science, each of which has its own hardcore.
But not every statement that makes up the hardcore of a
particular programme expresses an essence. The network
theory, by itself, does not distinguish between those
statements within a hardcore that express an essence and
those statements that do not. Thus, if we wish to argue
that statements (1)—(3) are essential, we will need some

analysis of necessity that goes beyond the network theory.

What Is An Essence?

How do we distinguish between statements that express
only a nominal essence and statements that express an Aris-
totelian essence? Take the following two statements:

(2) Gold is the element with atomic number

(4) Agbachelor is an unmarried man.
Intuitively, we believe that these two statements are dif—
ferent. But as we noticed in the preceding section, we
have yet to develop an analysis of necessity that justi-
fies this intuition.

How should we develop an analysis that allows us to

distinguish between statement (2), which we believe to be

an Aristotelian essence, and statement (4), which we

be

58

di

96

A:

H‘-

128

believe to be a nominal essence? We could suggest, as we
did in chapter two, that an Aristotelian essence is neces-
sary de re while a nominal essence is only necessary de
dicto. Yet many philosophers find fault with this sug-
gestion. Marcus, one such philosopher, insists that

... what has gone wrong in recent discus-

sions of essentialism is the assumption

of surface synonymy between 'is essenti-

ally' and de re occurrences of 'is nec-

essarily. [2]
And Marcus has good reason for finding fault with this as-
sumption. While every essence or essential property is
necessary de re, not every prOperty that is necessary de re
is essential.

There are all sorts of properties that are necessary

de re. Just consider the following:

(a) the property of being self-identical.

(b) the property of being some thing.

(c) the property of having some property.
Examine property (a). We may be tempted to argue that this
property is necessary de dicto since it can be reformulated
as a truth of logic, namely '(x)(x=x)'. But it is also
necessary de re. For any object or any entity, it is the
sort of thing that is necessarily self-identical. Yet
while these properties are necessary de re, they hardly
constitute the essence of anything. As a matter of fact,
these properties are vacuous.

These properties are vacuous because they apply to

everything. And properties that are true of everything

129

cannot be Aristotelian essences. Bennett writes that

Being an entity is a necessary property

of everything, i.e., a transcendental

property. ... Essential properties sort

the entities of which they are true in

some fashion. [3]
Bennett's point becomes particularly apt once we realize
that essences must meet two conditions: First, the essence
of some thing must be what that thing is; and second, the
essence of some thing must also distinguish that thing from
everything else. If we know the essence of a particular
thing or kind of thing, we not only know what that thing is
but we also know how to distinguish that thing from every-
thing else. Statements (1)—(3) meet both conditions.
Statement (2), for example, expresses the essence of gold.
Gold is the element with atomic number 79. And it is this
fact, the fact that gold has a particular atomic number,
that separates gold from everything else. Vacuous proper-
ties, such as (a)-(c), fail the second condition. If
everything is necessarily self-identical, for example, we
can hardly use this property to separate one thing from
another.

If we cannot equate essentialism simply with de re
necessity, couldn't we equate it with non-vacuous de re
necessity? After all, we distinguish vacuous from non—vac—
uous properties all the time. But then we are faced with

the following question: How do we distinguish non-vacuous

properties from vacuous properties? It is one thing to say

130

that we make a distinction. It is another thing to explain
how we make the distinction.

Some philosophers answer this question by arguing that
there are certain conditions that an essential property
must meet. Marcus, in "Essential Attribution", and Teller,
in "Essential Properties: Some Problems and Conjectures",
discuss two such conditions. Both Marcus and Teller situ—
ate their discussion within the context of modal logic.
They carefully mention that their conditions occur within
a particular system of modal logic (Marcus places her con-
ditions within S4 while Teller places his within 35' as
interpreted by Kripke semantics. Interestingly enough,
both Marcus and Teller provide the same two conditions.4
And interestingly enough, we can discuss these conditions
apart from their modal context.

In their first condition, Marcus and Teller rule out
certain vacuous properties. Marcus argues that "implicit
in essentialism is that an object has attributes necessar-
ily that are not necessary to other objects."5 To capture
this intuition, Marcus provides the following condition:

(1) (3x)UPx & (3x)-UPx.
Teller, who also accepts this condition, argues that it
prevents a trivialization of the notion of an essential
property since it excludes those properties that hold
necessarily of everything and those that hold necessarily

of nothing.6 In any event, it excludes vacuous properties

131

such as (a)—(C).

For their second condition, Marcus and Teller both
wish to emphasize a particular feature about Aristotelian
essentialism. According to Marcus,

Aristotelian essentialism takes it that,
if anything is a man or mammal, it is so
necessarily. These are not properties
that anything can have per accidens. The
same strong conditions extends to proper—
ties (e.g., rational-animal) which are
definitive of a kind (e.g., man). [7]
Marcus attempts to capture this feature about Aristotelian
essentialism in the second condition which states:
(ii) (x)(Px+E]Px).
Teller argues that this condition captures a generalized
form of the idea that whether a thing has P or not makes
a difference in the very identity of that thing.8

Statements (1)-(3) meet both conditions. Consider
(2). There is something which necessarily has the property
of being the element with atomic number 79, namely gold;
and there is something which does not have this property
necessarily, namely everything else. Thus, the first con-
dition is met. And since the property of being an element
with atomic number 79 constitutes the very identity of
gold, anything that has this property has it necessarily.
Thus, the second condition is met.

There are at least two problems with this account.

Teller describes the first problem as the problem of

boolean combinations. He believes that any property that

132

meets these two conditions can be described as a minimally
essential property. But, as he points out,

... any union or complement, and so any

boolean combination of minimally essen-

tial properties will again be a minimally

essential property except for those pro-

perties which in some possible world hold

of everything or hold of nothing. [9]
Why is this a problem? Teller notes that one purpose of
essentialism is to characterize natural kinds. But while
"the property of being a tiger picks out a natural kind,
and so does the property of being an electron. But the
property of being a tiger-or-an-electron does not pick out
a natural kind."1O Yet if we wish to argue that the pro-
perty of being a tiger is a minimally essential property
and the the property of an electron is another minimally
essential property, then so is the property of being a
tiger-or-an-electron. Our two conditions do not rule out
these sorts of properties.

Nor do these two conditions rule out every vacuous
property. While they may do away with properties (a)-(c),
they do not do away with other sorts of vacuous properties.
Compare the following two properties:

(d) the property of being Charlotte Bronte.
(e) the property of being the element with
atomic number 79.
Although we may be willing to concede that the property of
being the element with atomic number 79 is the essence of

gold, we are not so willing to concede that the property of

being Charlotte Bronte is the essence of anything. Yet

133

both properties meet Marcus's and Teller's two conditions.
Consider property (d). This property certainly meets the
first condition--there is something that is necessarily
Charlotte Bronte, namely Charlotte Bronte; and there is
something, namely the rest of us, that does not have this
property necessarily. And this property meets the second
condition—~if something is Charlotte Bronte, then it is
necessarily Charlotte Bronte. Yet this property, the
property of being Charlotte Bronte, is vacuous. It is a
vacuous property despite the fact that it only belongs to
one entity. This is because an essence is supposed to
meet two conditions. If we know the essence of something,
then we not only know what that thing is but we also know
how to distinguish it from everything else. Property (d)
meets these two conditions only in the most trivial sense.

So, once more, we are left with the following ques—
tion: How do we determine which properties are vacuous and
which are not? If we compare vacuous properties with non—
vacuous properties, one answer comes quickly to mind. Cer—
tain properties play a role within our theoretical network.
Such properties include

(e) the property of being the element with
atomic number 79.
(f) the property of being a tiger.

Other properties play only a trivial role, if even that,
within our network. Such properties include

(g) the property of being a tiger-or-an-
electron.

134

(d) the property of being Charlotte Bronte.
Despite the fact that Goodman is concerned with induction,
this seems rather reminiscent of his distinction between
the predicates 'green' and 'grue'. We accept 'green' as
a predicate but we reject 'grue'. We believe that 'green'
is projectible for inductive inferences while 'grue' is
not. Goodman argues that this is because 'green' is en-

trenched.11

'Green' plays some sort of role within our
network while 'Grue' does not. Much the same point can
be made about properties (e) and (f). These properties,
in contrast to properties (d) and (g), are entrenched
within our network.

This suggests that while we cannot reduce essentialism
to the hardcore of research programme within our theoreti-
cal network, neither can we ignore theoretical context.
When we claim that statements (1)-(3) express the essence
of their subjects, we are making a claim about the role
that these statements play within our network. When we
describe certain properties as essential, we are making a
claim about the role that that these properties play within
our theoretical network. Rather than attempting to charac-
terize these properties in terms of some list of necessary
and sufficient conditions, it is more useful to character-

ize these properties in terms of the role that they play

within our network.

135
Essence and Explanation
We do not want to divorce essences and essential

properties from our theoretical network. This is because
essences and essential properties are essential insofar as
they play a particular sort of role within our theoretical
network. But what is this role? Teller suggests that
they play an explanatory role. He writes that

... a natural kind is the extension of

some property that plays a significant

role (or a signficant explanatory role)

within a true (general, explanatory, ...)

theory. I will call such a property an

explanatory essential property, or, more

euphonically, an explanatory essence,

the suggestion being that such a proper-

ty plays a key role in explanations, or

in an explanatory theory. [12]
The advantage of this move is that it enables us to distin-
quish vacuous properties, such as the property of being
Charlotte Bronte, with non-vacuous properties, such as the

property of being the element with atomic number 79. Non-

vacuous properties have an explanatory role within our

1 fill-'37

network while vacuous properties do not.
Teller is by no means the only philosopher who con-

nects the notion of essences and essential properties to

 

some sort of explanatory role. When it comes to gold, for
example, Sober writes that

...we want to require that essences are
necessary properties that play a certain
causal (and hence explanatory) role. ...
That an atom possess a given atomic num-
ber causes it to have numerous other
properties. [13]

136

And Copi writes that

...within the context of [our scientific

theories] that property [of being the

element with atomic number 79] is funda-

mental. From its chemical formula more

of its properties can be inferred than

any other. ... To the extent to which

one small group of properties of a sub-

stance can serve as a basis from which

its other properties can be causally

derived, to that extent we can be jus-

tified in identifying tha group of pro-

perties as its real essence. [14]
These passages reiterate Teller's point. We choose a
particular property to serve as an essence if it plays an
explanatory role within our theoretical network. The
existence of this property explains the existence of other
properties. An essence is a property that explains why a
particular object or kind has the properties that it has.
We can use the fact that gold has a particular atomic
number, for example, to explain why gold has various other
properties. In that sense, its atomic number serves as
its essence.

But an explanatory essence need not be an Aristotelian
essence. An explanatory essence is a property that allows
us to explain the existence of other properties. But this
property need not be an Aristotelian essence. It need not
be an Aristotelian essence even when we discuss natural
kinds. Natural kinds, as we know from Chapter Two, are not
alike. This difference becomes quite pronounced once we

compare biological species with physical elements. The

explanatory essence for the latter is an Aristotelian

137

essence. The explanatory essence for the former is not.
These days, scientists reject any role for Aristotelian
essentialism in biology. This is ironic since, as Teller
points out,

Historically, one place where theorists

tried to put doctrines of essential

properties to work was in attempts to

develop a theory of natural kinds, or

real species, in application to biologi-

cal classes. [15]
But whatever the historical record may be, biologists
gave up essentialism. This is basically for two reasons.
The first reason is that they have yet to find an essence
for any biological species. And the second reason is
that the theory of evolution does away with the need for
essentialism.

Let's examine the first reason. Suppose a biological
species has an Aristotelian essence. This means that there
is some property that belongs to every member of that spe-
speies and only to the members of that species. This
property will then be the defining characteristic of that
species just as its atomic number is the defining charac-
teristic of gold. Sober argues that this is a constituent
definition--the essence of a species is defined in terms of he
the properties of its members.16 But we have yet to find
such a property for any species. Just think about the
various sorts of candidates for such a property. The es-

sence of a species cannot be a morphological property. For

while members of a species are morphologically similar, we

138

may decide not to exclude entities from that species which

lack that particular morphological property. The essence

of a species cannot be a phenotypic property. For once

again, we may decide not to exclude entities from that

species which lack that property. And an essence of a

species cannot be a genotypic property. This is because,

as Sober points out, the genetic variation among a given

population is prodigious; and it is highly unlikely that

there is a particular gene that every member, and only the

members, of a given species possesses.

The fact that biologists have failed to find essences

for species is no problem considering that the theory of

evolution does away with the need for essences in biology.

There are various reasons cited for this. Some scientists,

and philosophers, argue that this is because essentialism

is committed to the view that species are static. Others

argue that the essentialist is unable to account for the p.

continuity of nature. For his part, Sober argues that
... evolutionary theory has removed the
need for providing species with constitu-
ent definitions; population thinking pro-
vides another way of making species sci-
entifically intelligible. This consider-
ation, coupled with the principle of par-
simony, providing an additional reason

for thinking that species do not have
essences. [18]

 

r-

Sober goes on to argue that in evolutionary theory, the
population is the unit of organization, not the organisms

that make up that population. We then treat the population

139

as an entity in its own right, subject to its own forces
and obeying its own laws. And we can do this without re—
course to essentialism.

But while biological species may lack an Aristotelian
essence, they do not lack an explanatory essence. But this
explanatory essence, as Sober insists, is not in terms of a
constituent definition. Rather, it is a historical explan-
ation. In the case of species, explanatory essences con-
cern historical lineages. Morphological, phenotypic, and
genotypic properties of a species are all discussed within
the framework of a historico-evolutionary explanation.

This framework is sufficient to provide an explanatory

essence; but it is not an Aristotelian essence.

What is an Aristotelian Essence?

What is our problem? We can distinguish essential
properties from vacuous properties by their explanatory
role; but how do we distinguish essential properties from
explanatory essences? The problem is that when we describe
some property as essential, it plays an explanatory role;
but not every property that plays an explanatory role is an
Aristotelian essence. So, what is unique about essential
properties?

To answer this question, it is useful to examine
statements (1)-(3) one more time.

(1) Heat is the form of energy constituted
by the motion of atoms and molecules

140

in solids.

(2) Gold is the element with atomic number
79.

(3) Water is H O.

2
Apart from expressing an Aristotelian essence, what do
these three statements have in common? One feature that
they share is that they are all associated in one way or
the other with the atomic theory. Thus, to understand why
these statements express an Aristotelian essence, we need
to first know something about atoms and molecules.

The atomic theory is a rather interesting theory. We
can argue that ontologically and metaphysically it is our
most fundamental theory. We can argue that it is our most
basic theory. And this is because everything in the uni-
verse is made out of atoms. This is not a minor point.
Feynman, one prominent physicist, emphasizes its impor-
tance when he wrote that

If, in some cataclysm, all of scientific

knowledge were to be destroyed, and only r
one sentence passed on to the next genera- F
tions of creatures, what statement would ;
contain the most information in the few- :
est words? I believe it is the atomic j
hypothesis ... that all things are made '

of atoms... [19] 5

The atomic theory provides the ontological building blocks

 

for the entire universe. And thus, it is our basic theory.
This should give us some idea of what an essential
property must be.
(E1) A property of an object or kind is
essential if and only if it is an

explanatory essence of that object
or kind in our basic theory.

141

On our present view, the atomic structure (and by exten-
sion, the molecular structure) of something is its essence.
Thus, heat is essentially the motion of atoms and molecules
in solids; gold is essentially the element with atomic num-
ber 79; and water is essentially H20. But this notion of
an essential property is not just any explanatory essence.
It is the explanatory essence in our basic theory. The
advantage of this move is that it not only explains how an
essence is tied to our theoretical network; but it also
shows why the more general criterion of being part of the
hardcore of just any theory does not do. It must be the
hardcore of our basic theory.

Yet we quickly run into the following problem. While
we believe that everything in the universe is constructed
from atoms, we also believe that atoms, in turn, are con—
structed from subatomic particles. But this seems to
suggest that elementary particle theory and quantum mechan-
ics are even more basic than atomic theory. After all, if
everything in the universe is made from atoms, and atoms
are made from elementary particles, then everything in the
universe is made from elementary particles. Thus, it seems
as though elementary particle theory and quantum mechanics
are ontologically and metaphysically our most fundamental
theories; and as such, these theories constitute our basic
theory. Yet even so, we still believe that statements (1)-

(3) express the essence of their subjects. But this means

142

that we do not equate essentialism with the explanatory
essence of our most basic theory.

Perhaps we could argue that we have many basic theor-
ies. After all, elementary particle theory and quantum
mechanics hardly do away with atomic theory. Feynman
writes

The theory of chemistry... was summarized
to a great extent in the periodic chart of
Mendeleev ... All these rules were ulti-
mately explained in principle by quantum
mechanics, so that theoretical chemistry
is in fact physics. On the other hand, it
must be emphasized that this explanation
is in principle. ... It turns out to be
very difficult to predict precisely what
will happen in a given chemical reaction;
nevertheless, the deepest part of theore-
tical chemistry must end up in quantum
mechanics. [20]

...it is going too far to say that quantum

mechanics has given a precise understanding

of the periodic table. It is possible, how—

ever, even with a sloppy approximation--and

some fixing--to understand, at least quali-

tatively, many chemical properties that show

up in the periodic table. [21] r:

We can View quantum mechanics as a supplement to the perio-
dic table of elements rather than as a replacement. It
provides us with a "deeper" understanding of the periodic

table of elements; but it does not replace it. In a cer-

 

I"

tain respect, these theories do not compete with each
other. It is just that one theory gives us a deeper under-
standing of the other. We still believe that everything in
the universe is constructed from atoms. It is just that

elementary particle theory and quantum mechanics give us a

143
better understanding of what atoms are. Thus, it seems
as though we could argue that we have more than one basic
theory.
In light of this, we may wish to modify our notion of
an essential property.
(E2 ) A property of an object or kind is

essential if and only if it is the

explanatory essence of any basic

theory.
We can argue that statements (1)-(3) still express the
essence of their subjects so long as they belong to any
basic theory--it need not be our "deepest" basic theory.

But this leads to another problem. Do we really want

to argue that an explanatory essence in any basic theory
is going to give us an essence? Perhaps not. Elementary
particle theory and quantum mechanics can certainly pro—
vide an explanatory essence of physical elements; but it
is not clear that they give an Aristotelian essence.
Moreover, it seems as though we could also argue that

certain theories in biology are also basic theories.

Certainly some theories in biology are ontologically and

 

metaphysically more fundamental than others. For example,

we could make a serious case that genetic theory and evolu-
tionary theory are, in some sense, more basic than other
theories in biology. But we already know that biologists
have rejected essentialism.

This suggests that we still need something more for

a property to be an Aristotelian essence. But what could

144
this be? Rosenberg argues that

...calling all the entries in the periodic
table "elements" is based on their all shar—
ing a type of property; that is, they all
have atomic structure. What distinguishes
one element from another is the difference
between their atomic structure, the number
and arrangements of electrons, out of which
their atoms are uniformly composed. Now it
is because all the elements have a common
type of structure and differ in the particu-
lar structures of this type that they mani-
fest that can justify the taxonomy of the
periodic table. ... If the elements did not
have a common structure, then they would
constitute a heterogeneous collection of
items. [22]

Rosenberg's comments suggest that we believe that something
has an Aristotelian essence if this essence concerns its
inherent structure. We believe that
(E3) A property of an object or kind is

an essential property if and only

if it is the explanatory essence of

any basic theory; and its explanatory

role concerns the inherent structure

of that object or kind.
An Aristotelian essence is a structural property that plays
an explanatory role.

If something has an essence, then we are talking about
the role that its structure or form plays in our theoreti—
cal network. In the case of physical elements, we not only
can explain what these elements are in terms of their in—
herent structure but how to distinguish one element from
another in terms of this structure. In the case of biolo-

gical species, the structure or form of the kind is not

relevant. Rather, we use a historical analysis to explain

145
what a species happens to be and how we can distinguish
that species from any other species.

Perhaps, one day, physicists will develop theories in
which the form or structure of physical elements are no
longer relevant. At that time, we may give up essential-
ism even for physical elements. But for the time being,
we believe that the explanatory essence of physical

elements can be given in terms of Aristotelian essences.

Notes

Chapter One

1. Leibniz, "Monodology," in Leibniz Selections, edited by
Philip P. Wiener. New York: Charles Scribner's Sons. pp.
539-540; Hume, "An Enquiry Concerning Human Understanding,"
in The Empiricists. Dolphin Books. Garden City: Doubleday
and Company. p. 322.

 

 

2. Leibniz, New Essanyoncerning Human Understanding,
translated by Alfred Gibbon Langley. New York: The Macmil—
lan Company. 1896. p. 143; Hume, A Treatise of Human Na-
ture, edited by L.A. Selby-Bigge. Oxford: Clarendon Press.
pp. 13, 15, 70. Leibniz believes that demonstrative know-
ledge ultimately originates from self-evident propositions
while Hume argues that even these self-evident propositions
(or intuitions) can be traced back to impressions.

 

 

 

3. Leibniz, New Essay Concerning Human Understanding, p.
14; Hume, "An Enquiry Concerning Human Understanding," pp.
322-323. Hume limits demonstrative knowledge to logic and
mathematics while Leibniz expands it to include various
truths from ethics, theology, and metaphysics.

 

4. Leibniz, "Monodology," p. 539.

5. Hume, "An Enquiry Concerning Human Understanding," pp.
322-323.

6. ibid. p. 322.
I

7. Leibniz, "Monodology,' pp. 539, 540.

8. Hume, "An Enquiry Concerning Human Understanding,"
p. 323.

9. ibid. p. 324.
10. Leibniz, "Letter to Arnauld," in The Philosophical

Writings of Leibniz, edited and translated by Mary Morris.
New York: E.P. Dutton and Company. 1934. pp. 57.

 

 

 

11. Leibniz, "Monodology," p. 539.

12. Kant, Critique of Pure Reason, p. 48.
13. Kant, Prolegomena to Any Future Metaphysics, translated

by Lewis White Beck. Indianapolis: Bobbs-Merrill. 1950.
p. 14.

 

14. Kant, Critique of Pure Reason, translated by Norman

 

146

147

Kemp Smith. New York: St. Martin's Press. 1965. p. 49.
15. ibid. pp. 42-43.

16. ibid. p. 43.

17. ibid. p. 43.

18. ibid. p. 44.

19. ibid. p. 49.

20. ibid. pp. 52-53.

21. Ayer, Language, Truth, and Logic, New York: Dover
Publications. p. 72.

 

22. ibid. p. 77.

23. ibid. p. 84.

24. Aristotle, "Posteriori Analytics," in The Collected

Works of Aristotle, translated andbedited by Richard
McKeon. New York: Random House. 92 25-30

 

 

Chapter Two

1. Putnam, "The Analytic and the Synthetic," in Mind,
Language, and Reality. Cambridge: Cambridge University
Press. 1975. p. 36.

 

2. Kripke, Naming and Necessity, Cambridge: Harvard Uni-

 

versity Press. 1972. pp. 37, 39. f.
3. ibid. pp. 35, 36. g
4. ibid. pp. 35, 36. '
5. ibid. p. 143.

6. ibid. p. 140. L.

 

7. Hull, "A Matter of Individuality," in Conceptual Issues
in Evolutionary Biology, edited by Elliott Sober, Bradford
Books. Cambridge: M.I.T. Press. 1984. p. 640.

 

 

8. Putnam, "Why There Isn't A Ready-Made World," in Realism

and Reason. Cambridge: Cambridge University Press. 1983.
p. 220

 

9. Putnam, "Reference and Truth," in Realism and Reason.

 

148
Cambridge: Cambridge University Press. 1983. p. 71

10. ibid. p. 73.

11. Kripke, Naming and Necessity, p. 138.

 

12. Sober, The Nature of Selection. Bradford Books. Cam—
bridge: M.I.T. Press. 1985. p. 164.

 

' 97b 5—15.

13. Aristotle, "Posterior Analytics,‘
14. Hull, "A Matter of Individuality," p. 638.

15. ibid. p. 638.

16. ibid. p. 638.

17. ibid. p. 638.

18. Hahn, "Logic, Mathematics, and Knowledge of Nature,"

in Logical Positivism, edited by A.J. Ayer, New York:
The Free Press. 1959. p. 152.

 

19. Quine, Word and Object, Cambridge: M.I.T. Press. 1960.
p. 199

 

20. Mackie, "De What Re is De Re Modality," Journal of
Philosophy (71) September 19, 1974, pp. 560-561.

 

 

Chapter Three

1. Horton, Dictionary of Modern Chess, Philosophical Libra-
ry. New York: Bonanza Books. pp. 28-29.

 

2. Putnam, "The Analytic and the Synthetic," p. 52.

3. Churchland, Scientific Realism and the Platicity of
Mind, Cambridge: Cambridge University Press. p. 15.

 

4. Quine, "Carnap and Logical Truth," in The Ways of Para-
dox. Cambridge: Harvard University Press. 1975. p. 110.

 

5. ibid. p. 116.
6. ibid. p. 119.
7. Putnam, "The Analytic and the Synthetic," p. 45.
8. ibid. p. 65.

9. Quine, Word and Object, p. 56.

 

149

10. ibid. p. 56.

11. Putnam, "It Ain't Necessarily So," in Mathematics,
Matter, and Method. second edition. Cambridge: Cambridge
University Press. 1979. p. 238.

 

 

12. Putnam, "The Analytic and the Synthetic," p. 68.
13. ibid. p. 35.
14. Quine, "Carnap and Logical Truth," p. 108.

15. Putnam, "Two Dogmas Revisited," in Realism and Reason.
Cambridge: Cambridge University Press. 1975. p. 88

 

16. Quine, "Two Dogmas of Empiricism," in From a Logical

Point of View. second edition. New York: Harper and Row.
1961. p. 20.

 

 

17. Hempel, "Empiricist Criteria of Cognitive Signifi-
cance," in Aspects of Scientific Explanation. New York:

The Free Press. 1965. p. 112.

 

18. Quine, "Two Dogmas of Empiricism," p. 41.

19. Putnam, "What is Mathematical Truth," in Mathematics,
Matter, and Method. second edition. Cambridge: Cambridge
University Press. 1979. pp. 75-76.

 

 

20. Peterson, "Award-Winning Links Twixt Math and Physics,"
Science News, (138) August 25, 1990. p. 119.

 

21. Quine, "Two Dogmas of Empiricism,‘ p. 42. r

22. Quine, Philosophy of Logic. Foundations of Philosophy :
Series. Eaglewood Cliffs: Prentice-Hall, Inc., pp. 99, y
100. 3

 

23. Quine, Methods of Logic. third edition. New York:
Holt, Rinehart, and Winston. 1972. p. 2, 3.

 

 

24. Quine, "Necessary Truth," in The Ways of Paradox, L
Cambridge: Harvard University Press. 1975. p. 76.

 

Chapter Four

1. Putnam, "The Meaning of 'Meaning'," in Mind, Language,

and Reality. Cambridge: Cambridge University Press. 1975.
pp. 235-238.

 

 

150

2. F¢llesdal, "Essentialism and Reference," in The Philo-
sophy of W.V. Quine, edited by Hahn and Schlip. The
Library of Living Philosophers. La Salle: Open Court.
1986. p. 107.

 

 

3. ibid. p. 109.
4. ibid. p. 109.
5. Zemach, "Putnam's Theory on the Reference of Substance

Terms," Journal of Philosophy, (73) March 11, 1976. p.
123.

 

6. lelesdal, "Essentialism and Reference," p. 109.
7. ibid. p. 110.

8. Quine, "Natural Kinds," in Ontological Relativity and
Other Essays, New York: Columbia University Press. 1969.
p. 123.

 

 

9. Mayr, Principles of Systematic Zoology, New York:
McGraw Hill, Inc., 1969. p. 201.

10. Putnam, "Is Semantics Possible?" Mind, Language, and
Reality. Cambridge: Cambridge University Press. 1975.
p. 139.

 

11. Putnam, "Reference and Truth," p. 74.

12. Quine, "Natural Kinds," p. 128.

13. ibid. p. 128.

14. Putnam, "Explanation and Reference," Mind, Language,

and Reality. Cambridge: Cambridge University Press. 1975.
p. 203.

 

 

15. Putnam, "The Analytic and the Synthetic," p. 68.

16. Mayr, "Species Concepts and Their Applications," in
Conceptual Issues in Evolutionary Biology, edited by
Elliott Sober. Bradford Books. Cambridge: M.I.T. Press.
1984. pp. 534-535.

17. Quine, "Truth By Convention," in The Ways of Paradox.
Cambridge: Harvard University Press. 1975. p. 77.

 

18. Putnam, "It Ain't Necessarily So," in Mathematics,
Matter, and Method. second edition. Cambridge: Cambridge
University Press. 1979. pp. 248-249.

 

 

151

19. Lakatos, "Methodology of Scientific Research Pro-
grammes," in Methodology of Scientific Research Programmes,
edited by Worrell and Currie. Cambridge: Cambridge Univer-

sity Press. 1978. p. 48.

 

20. Copi, Introduction to Logic. seventh edition. New York:
Macmillan Publishing Company. 1986. pp. 308-308.

 

Chapter Five

1. Putnam, "Why There Isn't a Ready-Made World," p. 220.

2. Marcus, "Essential Attribution,"

Volume (68) April 8, 1971. p. 193.

Journal of Philosophy,

 

3. Bennett, "Reference, Essentialism, and Modality," Jour-
nal of Philosophy, (66) October 16, 1969. p. 487

 

4. As a matter of fact, Teller acutally provides three
conditions; but since his first two conditions are simply
conjuncts of Marcus's first condition, I am going to treat
his two conditions as one.

5. Marcus, "Essential Attribution," p. 196.
6. Teller, "Essential Properties: Some Problems and Conjec-

tures," Journal of Philosophy, Volume (72) May 8, 1975.
p. 236.

 

7. Marcus, "Essential Attribution,‘ p. 198.

8. Teller, "Essential Properties: Some Problems and Conjec-
tures." p. 236.

9. ibid. p. 237.
10. ibid. p. 237.

11. Goodman, Fact, Fiction, and Forecast, Indianapolis:
Bobbs-Merrill Company, Inc. 1965. p. 82, 95.

 

12. Teller, "Essential Properties: Some Problems and Con-
ll

jectures, p. 242.

13. Sober, The Nature of Selection, Bradford Books.
Cambridge: M.I.T. Press. 1985. pp. 164-165.

 

14. Copi, "Essence and Accident," in Naming, Necessity,
and Natural Kinds, edited by Stephen Schwartz. Ithaca:
Cornell University Press. 1977. pp. 189-190.

 

 

152

15. Teller, "Essential Properties: Some Problems and Con-
jectures," p. 238.

16. Sober, "Evolution, Population Thinking, and Essential-
ism." Philosophy of Science, (47) 1980. pp. 347, 355

 

17. ibid. p. 380.
18. ibid. p. 360.
19. Feynman, Leighton, and Sands, The Feynman Lectures

On Physics. Volume 1. Reading: Addison-Wesley Publishing
Company. 1963. p. 1-2.

 

 

20. ibid. p. 3‘1.

21. Feynman, Leighton, and Sands. The Feynman Lectures
On Physics. Volume 3. Reading: Addison-Wesley Publishing
Company. p. 19-14.

 

 

22. Rosenberg, The Structure of Biological Science, New
York: Cambridge University Press. 1985. pp. 201-202.

 

BIBLIOGRAPHY

Aristotle, "Posterior Analytics." In The Collected Works
of Aristotle. Translated and edited by Richard McKeon.
New York: Random House. 1941.

 

 

Ayer, Alfred Jules. Language, Truth, and Logic. New York:
Dover Publications, Inc.

Bennett, Daniel. "Essential Properties." Journal of Phil-
osophy. (66). October 16, 1969. pp. 487-499.

 

Churchland, Paul M. Scientific Realism and the Plasticity
of Mind. Cambridge: Cambridge University Press. 1979.

Copi, Irving M. "Essence and Accident." In Naming, Neces-
sity and Natural Kinds. Edited by Stephen P. Schwartz.
Ithaca: Cornell University Press. 1977. pp. 176-191.

 

. Introduction to Logic. seventh edition. New York:
Macmillan Publishing Company. 1986.

Feynman, Richard P, Robert B. Leighton, and Matthew Sands.
The Feynman Lectures On Physics. Volume 1. Reading: Add-
ison-Wesley Publishing Company. 1963.

. The Feynman Lectures on Physics. Volume 3. Reading:
Addison-Wesley Publishing Company. 1963.

F¢llesdal, Dagfinn. "Essentialism and Reference." In Tee
Philosophy of W.V. Quine. Edited by Lewis Edwin Hahn
and Paul Arthur Schlip. The Library of Living Philoso-
phers. La Salle: Open Court. 1986. pp. 97-114.

 

153

154

Goodman, Nelson. Fact, Fiction, and Forecast. second edi-
tion. Indianapolis: Bobbs-Merrill Company, Inc. 1965.

 

Hahn, Hans. "Logic, Mathematics, and Knowledge of Nature."
In Logical Positivism. Edited by A.J. Ayer. New York:
The Free Press. 1959. pp. 147-164.

 

Hanson, N.R. "Causal Chains." Mind (64). 1955. pp. 479-
487.

Hempel, Carl G. "Empiricist Criteria of Cognitive Signifi-
cance: Problems and Changes." In Aspects of Scientific
Explanation. New York: The Free Press. 1965. pp. 101-
122.

 

 

Horton, Byrne J. Dictionary of Modern Chess. Philosophical
Library. New York: Bonanza Books. 1969.

 

Hull, David L. "A Matter of Individuality." In Conceptual
Issues in Evolutionary Biology. Edited by Elliott Sober.
Bradford Books. Cambridge: M.I.T. Press. 1984. pp. 623-
645.

 

 

Hume, David. "An Enquiry Concerning Human Understanding."
In The Empiricists. Dolphin Books. Garden City: Doubleday
and Company, Inc. pp. 307-340.

 

 

. A Treatise of Human Nature. Edited by L.A. Selby- ‘
Bigge. Oxford: Clarendon Press. 1968.

Kant, Immanuel. Prolegomena To Any Future Metaphysics.
Translated by Lewis White Beck. Indianapolis: Bobbs-
Merrill. 1950.

 

 

. Critique Of Pure Reason. Translated by Norman Kemp
Smith. New York: St Martin's Press. 1965.

 

Kripke, Saul. Naming and Necessity. Cambridge: Harvard
University Press. 1972.

 

. "Identity and Necessity." In Naming, Necessipy

155

and Natural Kinds. Edited by Stephen P. Schwartz,
Ithaca: Cornell University Press. 1977. pp. 66-101.

 

Lakatos, Imre. "The Methodology of Scientific Research
Programmes." in The Methodology of Scientific Research
Programmes Philosophical Papers Volume I. Edited by
John Worrell and Gregory Currie. Cambridge: Cambridge
University Press. 1978. pp. 8-101.

 

 

Lauden, Larry. Progress and Its Problems. Berkeley: Univer-
sity of California Press. 1977.

 

Leibniz, Gottfried. "Letter to Arnauld." in The Philosophi-
cal Writings of Leibniz. Edited and translated by Mary
Morris. New York: E.P. Dutton and Company. 1934. pp.
57-61.

 

 

. "Monodology." In Leibniz Selections. Edited and
translated by Philip P. Wiener. New York: Charles Scrib-
ner's Sons. pp. 533-551.

 

 

. New Essays Concerning Human Understanding. Trans—
lated by Alfred Gibbon Langley. New York: The Macmillan
Company. 1896.

 

Mackie, J.L. " De What Re Is De Re Modality?" The Journal
of Philosophy. (71) September 19, 1974. pp. 551-561

 

 

Marcus, Ruth Barcan. "Essential Attribution." Journal
of Philosophy. (68) April 8, 1971. pp. 187-202.

 

Mayr, Ernst. Principles of Systematic Zoology. New York:
McGraw—Hill, Inc., 1969.

 

. "Species Concepts and Their Applications." In Con-
ceptual Issues in Evolutionary Biology. Edited by Elliott
Sober. Bradford Books. Cambridge: M.I.T. Press. 1984.
pp. 531-540.

Peterson, Ivar. "Award-Winning Links Twixt Math and
Physics." Science News, 138 (August 25, 1990). p. 119.

 

 

156

Putnam, Hilary. "The Analytic and the Synthetic." In Mind,
Language and Reality. Cambridge: Cambridge University
Press. 1975. pp. 33-69.

 

. "Explanation and Reference." In Mind, Language,
and Reality. Cambridge: Cambridge University Press.
1975. pp. 196-214.

 

 

 

. "Is Semantics Possible?" In Mind, Language, and
Reality. Cambridge: Cambridge University Press. 1975.
pp. 139-152.

 

. "It Ain't Necessarily So." In Mathematics, Matter,
and Method. second edition. Cambridge: Cambridge Univer-
sity Press. 1979. 237-249.

 

 

. "The Meaning of 'Meaning'." In Mind, Language,
and Reality. Cambridge: Cambridge University Press.
1975. pp. 215-271.

 

 

. "Reference and Truth." In Realism and Reason. Cam-
bridge: Cambridge University Press. 1983. pp. 69-86.

 

 

. "Two Dogmas Revisited." In Realism and Reason.
Cambridge: Cambridge University Press. 1983. pp. 87-
97.

 

. "What Is Mathematical Truth?" In Mathematics, Matter
and Method. second edition. Cambridge: Cambridge Univer—
sity Press. 1979. pp. 60-78.

 

 

. "Why There Isn't a Ready-Made World." In Realism
and Reason. Cambridge: Cambridge University Press. 1983.
205-228.

 

 

Quine, W.V. "Carnap and Logical Truth." In The Ways of
Paradox. Cambridge: Harvard University Press. 1975.
pp. 107-132.

 

. Methods of Logic. third edition. New York: Holt,
Rinehart, and Winston, Inc., 1972.

 

 

157

. "Natural Kinds." In Ontological Relativity and
Other Essays. New York: Columbia University Press. 1969.
114-138.

 

 

 

. "Necessary Truth." In The Ways of Paradox. Cam-
bridge: Harvard University Press. 1975. pp. 68-76.

 

 

. Philosophy of Logic. Foundations of Philosophy
Series. Eaglewood Cliffs: Prentice-Hall, Inc., 1970.

 

. "Truth By Convention." In The Ways of Paradox.
Cambridge: Harvard University Press. 1975. pp. 77-106.

 

 

. "Two Dogmas of Empiricism." In From a Logical Point

of View. second edition. New York: Harper and Row. 1961.
pp. 20-46.

 

. Word and Object. Cambridge: M.I.T. Press. 1960.

 

Rosenberg, Alexander. The Structure of Biological Science.
New York: Cambridge University Press. 1985.

 

Shapere, Dudley. "Reason, Reference, and the Quest For
Knowledge." Philosophy of Science. (49) pp. 1-23

 

Sober, Elliott. "Evolution, Population Thinking, and Essen—
tialism." Philosophy of Science. (47) 1980. pp. 350-
383.

 

. The Nature of Selection. Bradford Books. Cambridge:
M.I.T. Press. 1985.

 

Teller, Paul. "Essential Properties: Some Problems and
Conjectures." Journal of Philosophy. (72) May 8, 1975.
pp. 233-248.

 

Wittgenstein, Ludwig. Philosophical Investigations. 3rd
edition. New York: McMillan. 1968.

 

158

Zemach, Eddy M. "Putnam's Theory on the Reference of Sub-
stance Terms." Journal of Philosophy. (73) March 11,
1976. pp. 116-126.