ik—-------_--_-_.-.-

SCEENTfiHC LAWS AND NECESSiTY

fiiswwion for the Efiegree of £331. 3.
MCHEGAN STATE WEVERSITY
MARY, W RYEé'N SW?!

1973

«I

LIBRAR y

IMHijganlStaUc
University

This is to certify that the

thesis entitled

SCIENTIFIC LAWS AND NECESSITY
presented by
Many Nrynn Smith
has been accepted towards fulfillment

of the requirements for

Ph.D. degree in PI’I'i'IOSOph!)l

 

 

 

Date " 3/

\6
Km

0—7 639

 

g amomc av ‘g
; HUM; & SUNS'
, BUUK BINDIRY NC.

I LIBHARY BINDERS
SIMIGN‘IT. IICIICII

 

ABSTRACT
SCIENTIFIC LAWS AND NECESSITY
By
Mary Nrynn Smith

The failure of regularity theorists, Hume's descendants in
recent philosophical literature, to formulate a non-necessity view of
law statements in a first-order extensional language has led us to
investigate the possibility of construing the law statements of science
as contingent intensional metalinguistic statements which have nomic
necessity. In our metalinguistic, necessity theory of laws, the laws
of science are quasi-legislative reports which impute that certain true,
first-order, non-strict, universal statements on the object level of
language reporting past, empirical conjunctions of properties or events
are to be considered as determinations (complete or part al) of a given
property or event. When a property or event is determined by another
property, event, or set of properties and events, the latter have been
inputed by the scientific community as identity marks of the former.
Law'statements which legislate a necessary connection between properties
and events are themselves, qua laws, contingent. However, law statements
may, at times, become necessary when they function within scientific work
as a priori statements. There are also lawful statements, non-vacuously
deducible from intensionally contingent law statements, which are

analytic and therefore necessary.

 

\2'

up. In.
0. ~ (Out!

0.
.'\

IE

I I. on,

? II;

' . b .5
x

"' .Q-v.

l- .

II . v '

”“l.'0p. .5
’ \

"' -~ 6..

..‘ ‘0'..-
.

’ \

9-; .'.'-
1;“ : ,.
II. I. .
."- 'S..
0. ‘v.‘
9

'-

‘P:. N

u, "' 1.
0..
g' . 3 'p!
I. '

I~.

.
.\ '§.,'
.

.r‘

 

Mary Nrynn Smith

Nonetheless we have not concluded, as other writers have
concluded, that a priori law statements and/or analytic lawful state-
ments are non-empirical unfalsifiable or self evident truths. Rather
we have concluded from an extensive investigation of the notions of
"analyticity," "a prioricity," “necessity," "presuppositions," and our
reconstructions thereof that, among other things, all presuppositions
are only relatively necessary. As such, they can be falsified if the
true statements entailing them become falsified. In fact we propose a
relative notion of necessity which characterizes all the uses of the
term “necessity" that have found wide currency in modern philosophical
literature. In our view, this relative necessity is the necessity of
both a priori statements (identified prematurely by Kant as necessary
presuppositions of any experience whatsoever) and analytic statements.

Despite the influential view of Willard Van Orman Quine that the
analytic-synthetic distinction should be abandoned rather than recon-
ciled with the fact that sentences change their degree of falsifiability
over time, we have also formulated a theory of meaning and analyticity
in which non-enduring, particular statements are analytic or synthetic
on a given occasion. Analytic statements are thus transient, necessary
relative to the definition statements which entail them, and falsifiable
when these definitions are changed. Analytic statements are also
empirical'when the terms within them refer to things in the empirical
world.

We have concluded, therefore, that law statements are inten-

sional, contingent metalinguistic statements about the empirical world

 

0'"

..... pp
\

 

'Oll
9_\_ _
' '1!

w
'.‘~ . ..
.. _
II. ~ -
u

‘-
'0‘...
l

' on I

.

e

:"O‘n.

I. ‘

I.

 

 

Mary Nrynn Smith

which legislate that members of the scientific community as well as
those who follow their work should consider certain relationships
between properties and events as necessary relationships which are to
hold until further notice.

This explication of nomic necessity has required a radical
reassessment of several classical philosophical distinctions and the-
ories. However, it seems fitting that philosophy of science follow the
example of science by considering a shift from old ways of thinking to
new in the face of a persistent anomaly-~the existence of apparently

nomically necessary statements which are nevertheless about the

empirical world.

SCIENTIFIC LAWS AND NECESSITY

By
Mary Nrynn Smith

A DISSERTATION
Submitted to
Michigan State University

in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Philosophy
1973

'1

52

Q‘

5.
If}
It

if

In the process of completing this dissertation, I received

Gen

ACKNOWLEDGMENTS

aid from the members of my dissertation committee, particularly my
chairman, Richard Hall. I am also exceedingly grateful to my husband,
Norman, whose encouragement, domestic cooperation and assistance in

obtaining research materials made this entire project possible.

*****

ii

TABLE OF CONTENTS

Chapter Page
I. THE PROBLEM OF LAN STATEMENTS . ............ 1
Two Positions . . ....... . ....... l
The Regularity Theory and Objections ......... 2
The regularity view of lawlike statements . . 2
Objections to the regularity view of the law
statements of science ............. 3
The vacuosness objection ............. 4
The hypothetical cases objection ......... 5
The mere coincidence objection .......... 8
The accidental generalization objection ..... 12
The subjunctive conditional objection ...... 17
Conclusion .................... l8
Necessity Views and Objections ............ 2]
The necessity view of Ewing ........... 21
The rationalist position of Joseph ........ 23
Empiricist objections to rationalist necessity
theories .................... 26
Hume's Arguments and Rejoinders ........... 29
The Humean view of causal connections ...... 29
Hume's a priori argument ............. 30
The novel object objection to Hume's a priori v
argument .................... 31
One defense of Hume ............... 35
A second possible defense of Hume ........ 38
The single case objection ............ 43
Correlation of events objection to Hume's
a priori argument ............... 48
The crucial experiment objection ......... 5l
Hume's conceivability argument .......... 54
The psychology objection to Hume's conceivability
criterion .................. 54
The imaginary facts objection to the
conceivability argument ............ 56
Hume's induction argument ........... . 59
Induction and the idea of causal connections
based on custom ................ 60
Belief and probability .............. 6l

iii

Chapter Page

The presupposition objection to Hume's

induction argument ............... 61
The historical certainty status objection . . . . 66
The certainty and proof objection . . . . . . . . 69
Hume's impressions argument ........... 75
Our arbitrary assumption objection to Hume's

impressions argument .............. 76
The idea of necessity from external impressions

re-examined .................. 80
Causal connection externally given--Max Black . . 82

II. THE KANTIAN VIEW OF CAUSALITY ............. 88
Historical role ................. 88
Kant's argument ................. 91
Kant's premises and conclusion .......... 95
The minimal interpretation of Kant's argument . . 97
Criticism of premise I .............. 100
Objections to premise III ............ 102
Further defense of premises III, IV and V . . . . 105
Misunderstandings of premise V .......... 115
Further objections of non—sequitur ........ 118
Another objection of non-sequitur ........ 126
Other arguments in the "Second Analogy" ..... 129
Summary ..................... 137
III. THE ANALYTIC AND SYNTHETIC DISTINCTION ......... 142
A Classical Distinction and Objections ........ 142
Historical role ................. 142
Kant's analytic-synthetic distinction ...... 144
The number of Kantian criteria .......... 146
Kant's containment criterion ........... 147
The vagueness objection to Kant's containment

criterion ................... 148
The form objection ................ 151
The variability objection ............ 154
Kant's second formulation of his A-S

distinction .................. 156
The epistemological objection to Kant's

identity criterion ............... 158
Kant's third formulation the non-ampliative

criterion ................... 159
Kant's fourth formulation the empirical

independence criterion ............. 159
The ambiguity objection ............. 160

iv

Chapter Page
Objection to Kant's fifth Leibnizean

‘formulation ' .................. 162
Kant's last formulation of the A-S distinction
the explication criterion ........... 164
Directions of a reconstruction .......... 166
A Distinction Reconstructed ............. 168
The meaning criterion--our proposed
reconstruction ................. 168
The containment criterion, a precursor of the
meaning criterion ............... 169
The meaning criterion avoids the form,
epistemology and perspective objections . . . . 173
The meaning criterion and the Leibniz
criterion .................. 175
The meaning of "meaning" that is meant ...... 177
The metalinguistic theory of meaning circumvents
criticism ................... 186
Objections to the metalinguistic theory of
meaning .................... 189
Meaning statements ................ 194
Analyticity ................... 199
Definition .................... 200
Types of linguistic meaning statements ...... 203
Lexical definition ................ 204
Lexical definitions according to the
metalinguistic theory of meaning ........ 205
Stipulative definition .............. 209
Synonymy ..................... 211
The reconstruction of the A-S distinction . . . . 214
Criticism of the A-S distinction ......... 216
IV. THE A PRIORI-A POSTERIORI DISTINCTION ......... 225
Kant's second distinction ............ 225
Are there really two Kantian distinctions? . . . . 230
The a priori criteria .............. 230
Necessity .................... 231
Kinds of necessity ................ 235
The kinds of necessary propositions which are
relatively necessary .............. 240
Relative necessity ................ 243
The necessity of a priori statements ....... 251
Presuppositions ................. 251
Objections to Strawson's theory ......... 257
The weaker sense of "presuppose" ......... 260
Assumptions ................... 264
Alternatives ................... 266

Chapter Page

Kant's proof of an a priori principle ...... 279
Can a priori statements be synthetic? . . . . . . 284
Views on the synthetic a priori . . . . ..... 289
V. A NEW NECESSITY THEORY OF LAW STATEMENTS ........ 305
Preface to our theory .............. 305
Another problem for necessity views ....... 307
Law statements .................. 314
Properties which can serve to identify ...... 323
The form of law statements ............ 326
Law statements as imputations .......... 333
Our theory circumvents objections to
regularity view ................ 337
A view of science . . . . . ........... 341
Essentialism ................... 349
HORKS CITED .......................... 354

vi

CHAPTER I

THE PROBLEM OF LAW STATEMENTS

Two Positions

For several decades, philosophers of science have discussed the
logical status of scientific laws. But there has been little agreement
(among their answers to the following two questions:
1. What is the complete logical form of laws of science or

lawlike statements?

2. What are the differences between them and other kinds of

statements?1

Two major positions have emerged from the controversy generated
by the above questions. One position, commonly called the regularity
view, represents empiricist thought on the subject of laws. Its origins
can be found in the writings of David Hume, and it has been the dominant
position of twentieth century thinking on this subject until recent
years. The second position is a rationalist position, which, for want
of a better name, we shall call the necessity view. Early writers of

this century such as A. C. Ewing and Brand Blanchard pr0posed forms of

__

1R. S. Walters, "Laws of Science and Lawlike Statements,"
EncycIODedia of Philosophy, lst ed., IV, '414.

 

r O‘-

I 0 .9"!

-
OI! I u
u
. A
Q; r;
'5 u‘v
I
. s;
.‘r ',
ch U.
n - o
\v' '1 ~
I II- -

 

V“- p
u'\ .
O

In:
\
I» .

.
-pou..
-

1 r
i

 

‘-._

 

 

the necessity view in opposition to Hume's philosophical descendants,
but their view was met with little sympathy. However, the dominancy of
the regularity view over the years has led to a thorough philosophical
scrutiny of it; and now its drawbacks are considered so numerous and
serious that philosophers are attempting to formulate an acceptable
necessity view once again. As Nicholas Rescher notes, present day
writers merely disagree as to how the necessity which is an aspect of

scientific law statements should be characterized.‘
The Regularity Theory and Objections

The regularity view of lawlike statements2
There are numerous statements in the literature which describe

the basic tenet of what we (as well as others) call the regularity view
of law statements. Consider the following assertions by David Hume,
A. J. Ayer, and Hans Riechenbach, respectively:

Similar objects are always conjoined with similar. Of

this we have experience. Suitably to this experience,

therefore, we may define a cause to be an object,

followed by another, and where all the objects, similar

to the first, are followed by objects similar to the
second.3

lNicholas Rescher, Scientific Explanation, p. 103.

2We shall refer typically to law statements, for every law
statement is at least lawlike; whereas, not every lawlike statement
is a law statement.

3Essential Works of David Hume, ed., by Ralph Cohen, p. 100.
'Hfis shallibe cited henceforth as HUme, Enquiry.

 

 

 

apu — v. f I O. .
o u I are ‘ in . I p\d a _ .v F
1.- .t o . 1 . u a 3‘ b,» P .\v u u . . e . F
o u .1. up. .1. b 3. o I. 9-. :u .4: o . :-
pus f P n a... m ”o g n P v I I y .u‘ . . .
v n . I i O u -\v A n ~\.u us.
my a - W p . .g. . .u.
I
. m. 2.. :w ... F. .. .. . ....

 

 

 

. . a proposition expresses a law of nature when it
states what invariably happens. Thus, to say that
unsupported bodies fall, assuming this to be a law of
nature, is to say that there is not, never has been and
never will be a body that being unsupported does not
fall.

. by a causal law2 the scientist understands a
relation of the form if - then,3 with the addition
that the same relation holds at all times. . . .“

We have called the common content of the above passages--that
laws describe regular sequences of events--the basic tenet of the
regularity view, because this initial view has been expanded upon by
various regularity theorists in attempts to defend their thesis from
sundry objections. In order that we may unravel the full formulation
of the regularity theory as it now stands in philosophical discussions,
we shall consider the objections which have been raised against the
initial tenet above and the evolution of the regularity view which
these objections have initiated.

Objections to the regularity view of
theTlaw statements of science

 

The objections which have been raised in opposition to the

basic tenet of the regularity view are numerous and sometimes involved.

 

1A. J. Ayer, The Concept of a Person, p. 220.

 

2Various writers discuss the general problem of law statements
in terms of only causal law since causality was the subject of Hume's
traditional discussion.

3No clue is given as to whether this "if-then" allows for just
indicative conditionals or subjunctive as well.

I'Hans Reichenbach, The Rise of Scientific Philosophy, p. 157.

 

 

 

I u I h
s i 7 II n ‘ .v
. 1 .n. It . . .u. .. . p
1 on» n b . u 6 .
I I .- . . P a u r In. .4. a
a 1. O v n.- a I A. .0-
i i. . . . ~ .
n n n. u c .I 6.0;. ..\. u.\o ..1. .1

 
 
 
 

 

 

We shall consider five major objections at some length. They are the
following:

1. the vacuousness objection,

2. the hypothetical cases objection,
the mere coincidence objection,

the accidental generalization objection,

01-be

the subjunctive conditional objection.

The vacuousness objection

It is important to note that, in the context of this first
objection as well as in that of all the others, law statements such
as "All 5 is P" are interpreted by regularity theorists as equivalent
to general implications (material implications) of the form (x)
(Sx 3 Px). Indeed this interpretation is the prime target of many
of the objections advanced against the regularity view generally.
The objection of vacuousness is levelled precisely against this
material implication interpretation of true lawlike generalizations,
since such material implications are true whenever their antecedent
is false. It thus turns out that "All unicorns have a single horn"
is a universal truth which may be considered a law via the regularity
view.1 If it is suggested, however, that lawlike statements be written
as equivalent to "(x) (Sx 3 Px) - (3x) Sx”; the vacuosness objection
can be circumvented. Yet as Ayer notes, this mode of circumvention

is undesirable in that "there are certain cases in which we do wish

 

lAyer, op. cit., pp. 222-223.

 

9... a”:

h I-
b

'Vr.‘

V "-p

‘. 0-.

I0. I
n " -
". . .

‘1.

.0. .
r
, y
1
c..
u p,
’ \ ‘

 

 

 

to take general implications as expressing laws of nature, even though
their antecedents are not satisfied."1 We often find laws in science
which consist of terms which denote ideal entities. An example is
Newton's law that a body on which no forces are acting continues at
rest or in uniform motion along a straight line.2 If in fact there
are no bodies free 0f forces acting upon them, this law might be said
to be vacuously true.3 One might take the option of viewing such a
law as a statement of possibility such that, if there were any bodies
on which no forces were acting, then they would behave according to
Newton's law. However, it is agreed by some regularity theorists that
the positing of a possibility-interpretation for such a law is incom-
patible with the basic-tenet-interpretation of laws proposed by the
regularity theory.“ It is precisely because the regularity theory
excludes the introduction of modal terms such as 'possibility' in
characterizations of lawlike statements that it has met with so many

objections; another of which is the hypothetical cases objection.

The hypothetical cases objection
As we have just seen, considerations of the initial and somewhat
non-critical vacuosness objection have already involved defendants of

the regularity view in a discussion of possibilities or hypothetical

 

1Ibid., p. 223.

 

“Ibi ., p. 227.

nlplt ‘

’\v

..... '

I... l

r. ‘ '

. ‘
v

 

\
'5' '0
II' P.
I,-
u
a
n:-
I..
'0

III
a

.
I
l"
u-

 

 

cases with respect to the analysis of lawlike statements. But the
raising of the possibility of possibility talk, brings the regularity
theorist into very dangerous territory where objections abound. As
Ayer rightly notes:

What we want to say is that if there were any bodies on
which no forces were acting then they would behave in the
way that Newton's law prescribes. But we have not made
any provision for such hypothetical cases : according to
the view which we are now examining, statements of law
cover only what is actual, not what is merely possible.1

William Kneale, who is not a regularity theorist, focuses upon
the hypothetical cases objection in much the same way:

If on the strength of our records we suggest that there
is a law of nature that all dodos have a white feather '
in their tails, we say in effect that, if there had been
any dodos other than those mentioned in our records, they
too would have had a white feather in their tails. But
an unfulfilled hypothetical proposition of this kind
cannot be derived from a proposition which is concerned
only with the actual. A contingent universal proposition
can always be expressed in the form "There are in fact no
a things which are not B," and from such a proposition it
is impossible to deduce that if something which was not
in fact a had been a it would also have been 8.2

It is clear from the above comments that no one who adopts the
basic tenet of the regularity theory; as expressed by Hume, Ayer, and
Reichenbach; can speak about law statements pertaining to hypothetical
cases. In view of this, it would seem a wise course for such theorists
.to avoid the introduction of possibility talk into their discussion.

C. D. Broad offered a solution to the vacuosness objection which was

designed exactly to follow such a course, i.e., to avoid direct

 

1Ibid., p. 223.

2William Kneale, Probability and Induction, p. 75.

“NONI":

I‘l
‘

a. I. -—
..a \po—

II o.»

I-On' o
" \
I...

:K- o .,
"'Io .,
0|

1': :1.
. ‘ . .

-

'“‘ 5.,
—

'1-
.-

 

b ‘ .
\ “:
l
s. ‘p
.I. _
.
0“
.’..-A
.1 -‘
."
.I‘I .-
I‘.‘: .
a.-
..“ ‘a-
-:
. v
1 .
0' _
9 .

 

confrontation with the problem of possibilities or hypothetical cases.
In his article "Mechanical and Teleogical Causation,“ Broad proposed
that seemingly vacuous laws such as Newton's force free body law be
thought of as referring not to hypothetical objects or events but "only
to the hypothetical consequences of instantial laws."1 In this manner,
Newton's law is construed as implying that there are instantial laws
about non—force free bodies which, together with the proposition that
there are force free bodies, entails the conclusion that such force-
free bodies continue at rest or in uniform motion along a straight line.
The conclusion, existentially interpreted, will be false; but it will
follow as a necessary consequence of the premise and law since the
premise will also be false. Ayer is dubious about the merits of this
clever solution; but even if the vacuosity objection is successfully
thwarted without resort to possible cases, there are other non-vacuous
laws which seem to involve an analysis of lawlike statements in a dis-
cussion of hypothetical cases. These are functional laws such as 'x =
Fy,‘ where "the range of the variable y covers all possible values of
the quantity in question."2 A law such as this cannot be regarded as
just the compendious assertion, as Ayer says, of all its actual in-
stances since the number of instances in the rage of y which are not
actual is infinite. Ayer refers to the law which relates the volume
and temperature of a gas under a given pressure as an example which

would have the form 'x==Fy.' Ayer writes:

 

1Cited by Ayer, op. cit., p. 223.
2Ibid., p. 224.

But now it is not to be supposed that all these values
are actually to be found in nature. Even if the number
of different temperatures which specimens of gases have
or will acquire is infinite, there still must be an
infinite number missing. . . . The formulation of the
law in no way indicates which the actual instances are.
. . . As asserting with regard to any given value n of
y that either n is not realized or that there is a
corresponding value m of x? This is the most plausible
alternative, but it makes the law trivial for all the
values of y which happen not to be realized. It is hard
to escape the conclusion that what we really mean to
assert when we formulate such a law is that there is a
corresponding value of x to every possible value of y.1

It appears that our consideration of functional laws brings
us squarely back again to discussion of possibilities. These laws
aside, however, there are still other important reasons why the
regularity view always seems vulnerable to the hypothetical case
objection one way or another. The other important reasons are in

fact objections three, four and five.

The mere coincidence objection

The mere coincidence objection arises most easily in connection
with causal type law statements. There are alterable cases of regular
sequence which are not regarded by anyone, except the superstitious

perhaps, as cases of regular causal sequence. It may happen that walk-

 

ing under a ladder is always followed by bad luck. But surely this
regular invariable sequence or constant conjunction (as Hume might say)
is not considered a causal law.

The mere coincidence objection can be easily avoided if quali-

fication is added to typical statements of the regularity view such

 

1Ibi ., p. 225.

 

 

—
I‘ U
'0- In.
‘ D
U . u.
I
u . ..
I- u-
o . .

‘ 0.0..

 

r"--.
I b
5‘ V‘ .
1. .
‘ . -l
' l
'u.
~ .
.
.‘ I
‘ '0.
'-
o ‘n.
I . \
I..‘
a
I.’. .
.
.. .
o
..
0- -P.'
\ . -
~
I.'
a

‘ a
._V
I“
I
- i
x
u. ‘
. n
a
\I ‘-
c .3.
.
n
I
h- 1
a. I
s -
‘Q
.-
9
.
I I ‘.
' h
5

r-.
- .
-‘~
.
i
.c
.
. "fi
-
.
A ..
. I
' \

 

 

that the regular sequences referred to are those that occur during the
justifying stage of a law's biography. A regularity theorist could

then rejoin to the mere coincidence objector that lawlike statements

are not compendious assertions of any old regular invariable sequences
of events but of those invariable sequences of events that occur in
connection with experimentation. This selectivity of constant con-
junctions might eliminate mere coincidences from the stock of universal
assertions of actual fact which are considered lawlike. Though such an
amendment is easily added to formulations of the regularity theory, it
is a mistake to think that such an amendment is unnecessary--to think
the automatic status of any invariable sequence of events one of law,

an absurd strawman of the regularity view. Quite the contrary, all
three of our representative regularity theorists: Hume, Ayer, and
Reichenbach, are in need of such a clarifying amendment in their dis-
cussions. Hume's treatment of cause is somewhat ambiguous, but com-
mentators are generally agreed that he speaks of constant conjunctions
with reference to the discovery of causes.1 While constant conjunctions
may lead to the confirmation of a causal law, they do not, in and of
themselves, lead to the discovery of a cause as opposed to the discovery

of a mere coincidence.

 

1Hume's discussion is ambiguous in that he may mean by the
phrase "discovery of the cause" the discovery of a causal law which
is actually the confirmation of a statement that such and such is the
cause of such and such.

10

We find it difficult to suppose that Ayer would deny the need
for such an amendment to certain of his words which are misleading in
this regard:

0n the view which we have now to consider, all that is
required for there to be laws in nature is the existence
of de facto constancies. In the most straightforward
case, the constancy consists in the fact that events,

or properties, or processes of different types are
invariably conjoined with one another.1

So too, the formulation of the regularity view of Hans Reichen-
bach is decidedly in need of such a clarifying amendment. Reichenbach
devotes a chapter of his book The Rise of Scientific Philosophy to dis-
cussion of causal law statements in terms of the regularity theory. In
this chapter, "The Laws of Nature," Reichenbach curiously leaves out all
reference to experimentation as a discovery tool in the search for law-
like regularities and claims that repetition is all that distinguishes
the causal law from a mere coincidence. Reichenbach claims that Hume's
view is generally accepted now (1951, the time of his writing) by the
scientist:

To say that the electric current causes a deflection of
the magnetic needle means that whenever there is an
electric current there is always a deflection of the
magnetic needle.2

Reichenbach maintains further that when we refuse to accept one
event as the cause of another, it is due to the fact that the observed
coincidence, the shaking of a theatre when an explosion appears on the

screen, occurred only once. Since it only happened once, it is known

 

lAyer, op. cit., p. 222.
2Hans Reichenbach, op. cit., pp. 157-158.

11

to be a mere coincidence but if the theatre ppulgf always shake when

an explosion is visible on the screen, then there would be a causal
relationship between the explosion on the screen and the shaking of

the theatre. Though it can hardly be Reichenbach's intent, his
description seems to reduce the discovery of causal connections to

the act of counting rather than to the act of experimenting followed
perhaps by counting.2 We say followed perhaps by counting, for it is
generally accepted that there are crucial experiments which are under-
taken in order to confirm a causal law, and in these cases, one never
counts past one. But this last consideration brings us beyond the mere
coincidence objection which may be handled by the proposed amendment to
another objection which we might call the repetition objection.

Obviously law statements based on crucial experiments cannot be
regarded as the economical compendia of actual invariable sequences of
events or as a shorthand denoter of regularities. However, Arthur Pap
raises a further interesting objection in this regard.

Arthur Pap notes that in order to carry out his program of
collecting repetitive sequences of similar events, Reichenbach must
decide at each occurrence of the repeating sequence whether a particular
antecedent event (c) and a particular consequent event (e), belong to
classes of events C and E, such that every member of C is followed by
some member of E. Unless the class C is specified, the regularity view

has no clear meaning in view of the plurality of causes. If one narrows

 

lHans Reichenbach, 0p. cit., pp. 157-185.

2Reichenbach does not explain his use of "would." Elsewhere in
his book on pp. 97 and 175 he does mention experimentation in passing.

12

the class of antecedents enough to ensure that each instance is
relevantly similar to the preceding instances, he is bringing the
regularity theory away from its basis of recurring similar sequences
more and more to a unit class or one specific sequence.l He explains:

By doing this the credibility of the generalization will

reach a maximum if A is narrowed down to a unit-class

comprising as its only member the causal antecedent

under discussion!

The problem revolves around the regularity view phrase,
"exactly similar circumstances," which one sometimes finds in regu-
larity view literature. Pap's criticism is that the only antecedent
exactly similar to the antecedent A is the antecedent A itself. In
light of this criticism, it is the singular occurrence of an event

sequence which should indicate with the most certainty that the sequence

was a causal one between the antecedent and the consequent.

The accidental generalization objection

The accidental generalization objection is similar to but not
quite the same as the mere coincidence objection. Ayer, among many
others, notes that the accidental generalization that all the ciga-
rettes now in his pocket are made of Virginia tobacco and the lawlike
generalization that the planets of our solar system move in elliptical
orbits are both formulated as '(x) <I> x 211 x' on the regularity view.
The first generalization, while not a mere coincidence, (Ayer likes

Virginia tobacco) is, nevertheless, merely a generalization of fact.

 

1Arthur Pap, "Regularity Theory of Causation," Journal of
PhiIOSOphy, XLIX (1952), 659.

 

P o

.u- :-
l '

.v- .u.
i I

9.. an.
a c I-

3
uh.

I
.4-

 

 

 

 

 

|
I I1 Illll L. . .. ~
. . . . . L .. . r .~. —~. :-

u s .4. ~ r—.\ f D ’ .~§
he .01. n... ~ art “no 4 P ix - ~d .\. .u. ._
n . . c

d s. s 6 MN I. I u .n- s n o as. b... u. agh-

. O O . u u .
”\n .0. o“ u - MC .. u. Mfr .... a . e . .-

13

However, the material implication formula which expresses this
generalization of fact also expresses the generalization of law about
planets. In search for a formal difference between the two generali-
zations Ayer says:
. . there seems to be a sense in which the generality

of what we are calling generalizations of fact is less

complete. They seem to be restricted in a way that

generalizations of law are not. Either they involve

some spatio-temporal restriction, as in the example of

the cigarette case, or they refer to particular indi-

viduals.1

These last words of Ayer suggest a proposal, the restrictedness

solution, for distinguishing generalizations of fact from lawlike gener-
alizations. The restrictedness solution provides that generalizations
of fact are simply conjunctions of particular statements which defi-
nitely refer to individuals. In this sense, accidental generalizations
may be considered restricted while lawlike statements are not. However,
such a proposal will not solve the problem posed by the accidental gen-
eralization objection for there are accidental generalizations which do
not refer definitely to individuals. An example of this latter kind of
sentence is "All the pears in this basket are ripe" or "All members of
the Greenbury School board for 1964 are bald."2 As Hempel indicates,
the sentences just mentioned as examples of non-lawlike statements do
not refer definitely to any individuals and therefore are of essentially

generalized form just as one might want to say lawlike statements are

those which are of essentially generalized form. It is also the case

 

1Ayer, 0p. cit., p. 226.
2Carl Hempel, Aspects of Scientific Explanation, p. 339.

14

that accidental generalizations can always be written so as to exclude
any reference to specific places or times;1 and, conversely, it is
possible that lawlike statements can be expressed so as to contain
reference to particular individuals or to specific places and times.2
To eliminate accidental generalizations, we would have to stipulate
that no general sentences are lawlike which happen to have only a
finite number of instances. Restrictedness thus is now identified
with a finite extension of the terms in an accidental generalization.
However, this qualification cuts too deeply again for it excludes some
lawlike statements of science which are concerned with a finite number
of objects such as some laws derivable from those of celestial mechan-
ics.3 The tack of proposing a restrictedness difference between acci-
dental generalizations and lawlike statements seems doomed unless one
adopts yet another construal of restrictedness in the manner of Popper.
According to his interpretation of restrictedness, a universal statement
is restricted, not merely because its extension is finite, but because
its extension is closed. As Walters explains:

The normal kind of restriction in universal statements is

restriction to a specified finite space-time region; with

certain other assumptions it is sufficient to say that the

class denoted is closed, although it may not entail any
particular number.“

 

1Ayer, op. cit., p. 226.

2Nelson Goodman, Fact Fiction and Forecast, p. 78.

3Hempel, op. cit., p. 340.
l'Walters, op. cit., p. 413.

 

DC
a

l

 

 

. . w 1
c . u. o u - u o - I . . ..- u.. p ..~ . .. c .
P .4. - .oo : w :- auy . 9 sun :5. p .c 9 .s. ah~ . .— . . . .s. . .
. «I. 1‘ n\- a 9 n n u n r a V n O a - 0 v ! .1; 1.. .0 n - .uu
. e o I . - .~. I u u .
u'.\ elI - e P n A. k 6 u a. u .Ii
-
5 I A 3.51- 0? r e. u to a .D «01. a‘ u

Dir:0.

"act.-
to

15

K. Popper in his article "A Note On Natural Laws and So-Called
'Contrary-To-Fact Conditionals'"1 criticizes an argument advanced by
critics of the regularity view, particularly critics of the view's
construal of law statements as material conditionals. Popper summarizes
the argument he considers faulty as:

Natural laws may be written simply as universal statements,
such as "All planets move in ellipses." But if we formu-
late them in this form, then we do not give full expression
to what is meant--we do not fully cover our use of natural
laws. If we wish to give some expression td‘tfieir logical
peculiarities, we should rather write something like “All
planets must move in ellipses" or "Planets necessarily move
in ellipses" in any case, we should indicate that natural
laws are logically stronger than ordinary universal state-
ments; an ordinary universal statement always follows from
the natural law, and it can replace the natural law for
many purposes, but it is not equal in logical strength to
the law and should be distinguished from it.2

Popper notes that the above argument is supported by the claim
that subjective conditionals follow from law statements, but that
"corresponding subjunctive conditionals do not follow from such uni-
versal statements which do not express natural laws."3 In answer to
this argument, Popper maintains that the fact that subjunctive condi-
tional statements follow from law statements but corresponding ones do
not follow from accidental generalizations is not due to a difference
in the logical form of these two types of statements but rather to a
vacillation concerning the extension of their respective terms. Using

the example "All men are mortal," written as (3.0) "All A's are B's,"

 

1Karl Popper, "A Note on Natural Laws and So-Called 'Contrary-
To-Fact Conditionals,'" Mind, LVIII (1949), 62-66.

2Ibid., p. 62.
3Ibid.

. C
,..
o v'

...-
o'

of -

f.

u I V

.oOt-f;

 

13.25

I'D
q ‘
tl'
I

.1.
(I?

II‘)
n
I

ll!

II.
t I

I"

 

Fe
U
s.
.Q
\
-
I-
“
I P.
.‘_
'-
-~
I. ~
. n
\

 

l6

Popper claims we can always deduce the indicative conditional "If x is
one of the A's then x is one of the 8's" and also the subjunctive con-
ditional (3.1) "If x were one of the A's then x would be one of the 8's."
But we can never deduce (3.2) “If x were added to the A's then it would
be one of the 8's." By "added to," Popper claims to mean some operation
corresponding to the class-addition {a} + A. Popper using another
example-universal "All my friends speak French" claims that we cannot
deduce "If Confucius were a friend of mine, then he would speak French"
because 'this latter statement means: "If Confucius were added to the
people I call my friends, then he would speak French.“1 Popper goes
on to say:

Similarly, we can now see that, from "All planets move in

ellipses," we can of course deduce "If the moon were a

planet it would move in an ellipse." (For this means:

"If the moon were one of the planets, or belonging to the

class of planets, then. . . ." But we cannot deduce "If

we extend the class of things covered by the name “planet"

so as to include the use», then she would move in an

ellipse lsicl. . . .2 We thus find that we make a mistake

because we neglected the extensional or class-aspect of

our terms; we did not see that in the case of type (3.1),

we kept our terms extensionally constant while in the case

of type (3.2) we assumed that the extension of our terms

may vary (which is, of course, impermissible in exten-

sional logic).3

It appears that the main thrust of Popper's answer to critics

of the regularity view is that subjunctive conditionals of a certain

sort 33g deducible from accidental generalizations just as subjunctive

 

1Ibid., p. 64.
2Popper may be guilty of petitio principi in these lines.

3Ibid.

l7

conditionals of this same sort are deducible from law statements.
However, this is not the point at issue. As Ayer says:

Thus one cannot say that generalizations of fact do not

entail subjunctive conditionals, for they may very well

contain dispositional predicates: indeed they are more

likely to do so than not: but they will not entail the

subjunctive conditionals which are entailed by the

corresponding statements of law.1
The subjunctive conditional objection

The argument Popper intends to defeat can still be raised if

one claims, as its advocates do, that there are, nevertheless, subjunc-
tive conditionals which assert that the moon would move in an elliptical
orbit not only if it were numerically identical with one of the planets,
but also if it belonged to the class of planets intensionally.2 Such
subjunctives are not deducible from accidental generalizations simply
because the intensional aspects of the terms in such accidental gener-
alizations are not considered while their extensional aspects are. When
I say "All my friends speak French" I can enumerate who my friends are
but perhaps not what makes each of them my friend. There is no clue in
such a statement as to whether or not speaking French is a requisite
pr0perty that any x must have in order to be a friend of mine. On the
other hand, this is precisely what is indicated by a natural law as
Popper himself recognizes:

Accordingly, the phrase "If x were an A . . ." can be

interpreted (1) if "A" is a term in a strictly universal

law, to mean "If x has the pr0perty A . . ." but it can
also be interpreted in the way described under (2); and

 

1Ayer, op. cit., p. 229.

2"Intensionally" here means defined according to prOPEVty-

 

|
'l--"

ct'

U'I

 

 

18

(2), if “A“ is a term in an “accidental" or numerically
universal statement, it must be interpreted “If x is
identical with one of the elements of'A."1

Popper concludes:

. there is no need to admit that the two types of
statements are different in import or in logical form,
except in the structure of their terms which, in the
case of natural laws, may be called "strictly universal
(or qualitative or perhaps intensional) terms," and, in
case of numerically universal statements," numerically
universal (or enumerative or individual or singular or
particular or perhaps extensional) terms.2

Though Popper chooses to regard the difference between
restricted and unrestricted universals as not one of logical form,
but as one of the structure of their terms; his opponents might well

fail to see how the differing structure of terms is not in some sense

a difference in logical form.

Conclusion

The characterization of law statements in science as unre-
stricted universals seems to introduce possibility talk into the dis-
cussion and to thereby debilitate once again the regularity theory
characterization of laws. As Ayer notes:

If generalizations of law do cover possible as well as
actual instances, their range must be infinite: for while
the number of objects which do throughout the course of
time possess a certain property may be finite, there can
be no limit to the number of objects which might possibly
possess it: for once we enter the realm of possibility

we are not confined even to such objects as actually exist.
And this shows how far removed these generalizations are

 

1Popper, op. cit., p. 65.
2Ibid.

.
‘IOI'c
I

u. .

\h
0--

 

 

 

. -.
u

. .

. I
I r

 

 

 

 

 

 

19

from being conjunctions: not simply because their range

is infinite which might be true even if it were confined

to actual instances but because there is something absurd

about trying to list all the possible instances.1

Popper says as much himself elsewhere in his article "A Note

on Natural Laws," for he claims that a strict universal "can never be
defined by enumeration of the elements which belong to them."2 If laws
are considered as strict (unrestricted) universals, it is difficult to
see how they can be subject to the basic tenet of the regularity view.
However, it may be the case that advocates of the regularity theory
would accept the qualification that laws of science are unrestricted
universals, as Popper seems to, but not that laws are logically
necessary. Such an advocate might argue, as Walters suggests, that
statements about possible instances stand in the same relation to
unrestricted universal law statements as do statements about actual
unobserved instances. Kneale claims that regularity theorists who take
this route consider laws as expressive of two conditions: "Either it is
a law of nature that every A thing is B, or there has been or will be
somewhere at some time an A thing that was not or is not B."3 In short,
they are committed to the view that every natural possibility, i.e.,

every state of affairs not excluded by a law of nature, must be realized

somewhere at sometime.“ Indeed to avoid such a commitment Popper made

 

1Ayer, op. cit., p. 227.
2Popper, op. cit., p. 65.

3William Kneale, "Universality and Necessity," British Journal
for the Philosophy of Science, XII (1960-1962), 98.

“Ibid.

 

 

 

 

20

the following suggestion in the appendix to his Logic of Scientific
Discovery of 1959. Kneale quotes from Popper:

Popper concedes that something more must be done to
distinguish laws from accidental generalities and
PY‘Oduces the following definition of natural neces-
slty:

A statement may be said to be naturally or physically
necessary if, and only if, it is deducible from a
statement function which is satisfied in all worlds

that differ from our world, if at all, only with
respect to initial conditions.1

Supposedly natural necessity is different than logical necessity
in that the former avoids talk of possibility while the latter does not.
However, as Kneale points out:

For to say that a statement function is satisfied in
all worlds that differ from the actual world, if at all,
only with respect to initial conditions is to say in
Effectthat it holds for all possible worlds that con-
tain instances of the same attributes and relations as
are exemplified in the actual world and of these only;
and what holds for all possible worlds is obviously
necessary.2

Apparently the regularity theorist is inevitably forced to
introduce intensional aspects such as necessity into his analysis of
law statements. From considerations of the above objections to the
regularity theory and their attempted answers, it is easy to see that
the main intent of such a view is to avoid characterizing the law
statements of science as logically necessary. The chief reason for
avoiding this identification is the belief that the laws of science

are a posteriori or empirical statements which are always subject to

n

H
H
U
.1.
Q

 

 

o';pu .

"l-- l
\-

.'v' i

1" "‘ i

 

 

 

 

21

refutation. Laws are deemed contingent and based on observation
(directly or indirectly). Since contingency and necessity are
traditionally thought of as contradictory concepts, laws if contingent
cannot be necessary. As prelude to our analysis of the law statements
of science, we will briefly consider some necessity characterizations
of law statements and the objections that they have elicited. Among
those objections, we shall no doubt find the beliefs mentioned above

which have motivated expositions of the regularity view.
Necessity Views and Objections

Necessity view of Ewing_

Unlike the regularity theorists, A. C. Ewing claims that
“besides regularity we must introduce the notion of determination and
necessity" into discussion of causality1 such that "the effect not
merely does but must follow the cause, and this depends on the specific
nature of the cause as such."2 He claims that this necessity is analo-
gous to that "underlying valid inference." In elaboration, Ewing says:

Where a conclusion follows logically from a premise,
this must be because the fact expressed by the premise
is so connected with the fact expressed by the conclu-
sion that the former could not possibly occur without
the latter occurring. This is logical necessity. The
theory according to which the connection between cause
and effect is the same as or very like that of logical
necessity may be called the rationalist or the entail-
ment theory of causation.

 

1A. C. Ewing, fundamental Questions of Philosopby, p. 162.

2Ewing directs his discussion toward Humean thought and he
therefore speaks about causality instead of law statements.

31bid., p. 162.

22

The above passage, though-vague, is the most detailed

explanation of the entailment theory that Ewing gives. He maintains
in his writings on the subject that causation is not just entailment,1
and he recognizes, as well, that "entailment" is a word more properly
used in reference to a relationship between sentences yet he says "we
can after all make legitimate inferences from cause to effect. How
could we do this if the cause did not in a very important sense entail
the effect? The relation need not be exactly the same as the entailment
which occurs in formal logical reasoning, but it must at least be analo-
gous to it in the important respect that it justifies the conclusion."2
The precise characterization of this analogous entailment is left un-
formulated. Ewing at best proposes only the germ of an entailment
theory about causation, and one which promises confusions between claims
of necessary connection in nature and necessary connections between
statements which describe nature. Ewing himself was aware of the like-
lihood of such confusion, for, as he emphasized:

It does not indeed follow that the relation of causality

is simply identical with or a species of the relation of

logical entailment, indeed this seems to me certainly

false; but it does follow that this is one of the rela-

tions which together make up the complex relation of

causality.3

Another necessity theorist N. Maxwell endeavors to give a more

explicit account of this entailment by saying:

 

1I i ., p. 163.
2Ib‘l .

U’
Q

G.

3A. C. Ewing, Idealism: A Critical Survey, p. 168.

23

At the time of the occurrence of event E , there exists

that which can only be completely descriSed by proposi—

tions P1 lsicl, which logically imply propositions P2

Isicl, that state that E2 occurs subsequently. This

says that there exists a logically necessary connection

between E,, and E2. The logical connection is between

propositions not events lsicl.1

This last passage betrays the confusion Ewing warns against.

Maxwell's use of "the" and "a" logical connection implies, as his
passage suggests, but one logical connection--between pr0positions
not between events. But he arbitrarily asserts mid-paragraph that
the logical connection which he believes to hold between propositions
holds also between events. Surely it does not seem that the same
(numerical) connection can hold between two sets of relata at the same
time.2 A more compelling rationalist, or necessity, view is expressed
by Mr. H. W. 8. Joseph who speaks only about objects in nature and not
about sentences describing those objects. However, the clarity and
detail of his view permit easy translation into "sentence about object
talk"; and Ayer raises objections to Joseph's theory on the level of

talk about sentences.

The rationalist position of Joseph

In his book The Foundations of Empirical KnoWledge A. J. Ayer
summarizes, in direct quotation, the view of Mr. H. W. 8. Joseph which
Ayer takes to be representative of the kind of view we are here pro-

posing. He quotes from Joseph's An Introduction to Logic:

 

1Nicholas Maxwell, "Can There Be Necessary Connections Between
Succeesive Events?“ British Journal for Philosgpby of Science, XIX
1967 , p. 2.

2This is the least that can be said by way of objection.

 

0
Pl
5
.
|
o
l
l
a
u
‘I
A!
'\
‘1
U‘
n
I
U
1
U
f
p
O
-
n
I
I
Q
fl
U
Q
.
I
a
\l
'.
‘1
I
O
3":
a ' .
'11,.
i
‘91.
. \
u
0’. .
.-
5 ‘ '
' c
a. 1-
‘1'
u
a
I"
I
It.
-_:
n
1 ‘p
I'u.
.I0
0 ‘T
'-
C.
. _-.
-

 

 

24

. . if a thing (a) under conditions (c) produces a
change (x) in a subject (5), the way in which it acts
must be regarded as a partial expression of what it is.
It could only act differently if it were different. As
long therefore as it is (a) and it stands related under
conditions (c) to a subject that is (5), no other effect
than x can be produced; and to say that the same thing
acting on the same thing under the same conditions may
yet produce a different effect is to say that a thing
need not be what it is. . . . But this is in flat con-
flict with the Law of Identity. A thing, to be at all,
must be something and can onl be what it is. To assert
a causal connexion between (a) and (x) implies that (a)
acts as it does because it is what it is; because in fact
it is (a). So long therefore as it is (a), it must act
thus, and to assert that it may act otherwise on a sub-
sequent occasion is to assert that what is (a) is some-
thing other than the (a) which it is declared to be.1

Ayer objects first that Joseph is treating all general proposi—
tions of law as if they were logically necessary. Hence, the evidence
which was formerly taken to establish the existential proposition that
this is an instance of A2 will not be sufficient to establish it once
the connotation of A is widened by making it logically necessary that
every instance of A should, in the relevant conditions, be conjoined
with an instance of X. Unless one has reason to believe that every
instance of A is not logically but factually conjoined in conditions
such as these with an instance of X, there can be no reason to believe
that this really is an event of the kind A in the revised sense.

Ayer is misguided in that it does not appear to be Joseph's

claim that the statements of law which are, in some not yet fully

 

1A. J. Ayer, Foundations of Empirical Knowledge, p. 208.

2Note Ayer's change of 'a' in the quoted passage to 'A' as he
talks not of a thing 'a' but of a predicate A.

“warrior
-v ' ‘ V '

 

c F o . LI 7 mt 2.— a S l d ‘L- .r r,
. .u.\ I. a c» hr” C PM a. I F 5 N c
nub.» “flu E AJ ht ‘A—‘h A." 0.1. I” v . U YFIU J a .W . P u A v r O
. . v. r .C r Pm 11 r .. I S E D; .C . U .4. IR...
.A S p w a: I H 1 I a w :u S 11 AU V e a: N S {-1 n . .1.
n e .T S a . a -flu . e r .1 a: a r. S. 24 .- ~ 1
3. C. 3w a 11 F.” t ,u S S a: at F r. 1 1
P I.” S 9: A: 1‘ :I o‘! P P: I.. r. .I th. U5. 0 “I T O...
I a t a -\. u a a”. t. .a\uv #1» n .6.» u a ..\u . . n I;
.H L. n P wl. 2. v e . p . . u R...- M-. a...
. . n v I .Ia u w MC T. n v.5. pl. .1 ... I I.

25

explained sense, definitional are logically necessary. Only further
statements based upon these laws such as "if this is an instance of A,
X must occur" are logically necessary; and we know instances of X to
be instances of A on the basis of laws which contingently claim a
connection between A and X. As Hospers writes:

When we say, "Water must boil at 212°F, we do not have

a statement which is necessary in and of itself, such as

"A is either a rose or not a rose." Rather, the "must"

is spoken only by way of logical inference from some

statement of law in which boiling at 212°F is taken to

be a defining characteristic of water.1

Ayer goes on to argue that if Joseph's argument is accepted

it would not be possible for us, as it is at present, to abandon a
"causal law without making a change in our usage of words."2 It is
more inconvenient, he claims, to be constantly altering the meaning of
words than merely to discard a series of empirical hypotheses. Ayer
continues that it would not be prOfitable to adopt the practice of
making all causal expressions definitional unless we had very good
reason to believe that the concepts that were yielded by these defi-
nitions would continue to be applicable to the empirical facts. But
this is exactly the case, is it not, when we formulate not empirical
hypotheses but laws? We do so only after very good inductive evidence
and the weighing of pragmatic reasons. Perhaps by accident, Ayer uses

the phrase "empirical hypotheses" interchangeably with "causal laws"

when in fact the two are not interchangeable. Empirical hypotheses are

 

1John Hospers, An Introduction to Philoso hical Analysis, p. 282.
We shall present a theory re: this necessity in Ehapter III.

2A. J. Ayer, The Foundations, p. 202.

.n.

. .' . I
. a
je’l'l: elf-‘a'

 

 

26

not definitional and can always be discarded with no change in the
meaning of their predicates, even when their predicates are used in
new hypotheses. 0n the other hand, there are terms such as "mass"
which occasionally do undergo a change in meaning as new laws replace
old ones in which this term occurs.

While these last statements of Joseph and Hospers indicate
a plausible avenue of analysis, there is still need for a detailed
explicit necessity theory which will characterize law statements and
explain nomic necessity. But before we propose the details of our
necessity theory, it will be well first to examine some general objec-
tions to necessity theories which have been raised by empiricists in
the past. As the name of our theory suggests, we intend to formulate
a characterization of law statements which analyzes their necessary
aspect in a manner consistent with the fundamental empiricist belief
that the law statements of science express contingent non-self—evident
truths.
Empiricist objections to rationalist
necessityIIheories '*

The following are some of the main empiricist criticisms of
rationalist necessity views such as that of A. C. Ewing:

l. We cannot prove any such law (causal) a priori but
only establish it as an empirical generalization.

2. In cases of a priori reasoning we attain certainty,
but in cases of causal reasoning only probability.

3. we cannot prove any logical connection between cause
and effect . . . as regards the physical world at least.1

 

1Ewing, Fundamental, pp. l64-l66.

 

ftfii
d

‘35: in tre .

. uni". 51‘-
£.:VI'U‘ \ I

3's]? refer ‘.C

 

1:. P ;‘;P On
:rv wr'&l~a=.

   

8

3“; “

a”. ‘ .
*5)“. .5“ u

27

Arguments in support of the three objections above can be
found in the writing of David Hume, the father of modern empiricism.
Objection (l) is supported by arguments in Hume's writing which we
shall refer to as Hume's a priori argument and his conceivability
argument. Objection (2) is supported by Hume's induction argument,
as we shall call it, and objection (3) by his impressions argument.
We shall now present the details of these four arguments and examine
each critically. During the course of our examination we shall seek
to establish the following claims concerning Hume's arguments:

Re: the a priori argument.

a. The cause of an effect and the effect of a cause can,
in some cases, be known independently of any obser-
vation of their constant conjunction or of even one
single conjunction as one might accomplish in the
case of a totally novel object. We shall call this
our novel object objection.

b. The cause of an effect and the effect of a cause can,
in some cases, be known or discovered via the observa-
tion of just a single conjunction ofithat cause and
effect. He shall call this the single case objection.

c. There are constant conjunctions of events which are
not directly pronounced to be causal sequences. We
shall call this our mere correlation objection.

d. There are causal laws which are not confirmed by the
observance of the constant conjunction of_the cause
and effect to which they pertain but by one single
crucial conjunction. This we shall call the crucial
experiment objection.

 

 

Re: the conceivability argument:

a. Hume's conceivability criterion is psychological.
Appropriately we shall call this our psychology
objection.

b. If matters of fact are known via empirical observa-
tion, hypothetical empirical claims cannot be based
on the use of a non-empirical criterion such as the
conceivability criterion. This we shall call our
imaginary facts.

28

Re: Hume's induction argument:

a. Belief in the uniformity of nature is, in fact, not
derived by inductive empirical observation but is
presupposed. This is our presupposition objection.

b. The belief in the uniformity of nature may not be
a probable belief if it is not based on empirical
observation. This objection shall be discussed
as the historical certainty status objection. ‘

c. Hume suggests that in some cases of causal reason-
ing we attain certainty. He also talks of proofs
about matters of fact. This is our certainty proof
objection.

Re: Hume's impressions argument:

a. Hume's assertion that all simple ideas are directly

derived from simple impressions is an unproved

assumption. This will be our arbitrary assumption

hypothesis.

b. There are those who claim that causal connections

are known primitively by observation. We do not

necessarily agree with this claim nor are we inter-

ested in talking about causal connections in nature.

But for sake of full treatment, we shall discuss

this causal perceptibility objection.

After discussion of the above rejoinders to Hume's four
arguments: the a priori argument, the conceivability argument, the
inductive argument and the impressions argument: we will begin recon-
structions of the terms "analytic," "a priori," and "necessity." We
will argue in Chapters III-and IV not only that logical relations may
hold between statements concerning matters of fact but that due to
these logical relations, some empirical statements entailed by others
may be known true with the necessity typical of deductive inference.
We shall argue in Chapter V that the law statements of science, though
contingent assertions about the empirical world, nevertheless give rise
to corresponding definitional statements; and these definitional state-

ments are much like axioms one might find in a mathematical axiomatic

 

. -

 

Hal 5

I." ‘
..- O

 

- .h. r C» n I \.l v a r r .0. I
r& ' l .v u r a II ‘ I. a A s h I .O b r ‘ v
2... :h C. r v 6.... ”J y... s: in u v--. .5. r\v :—
v ~13 F I -P n. A Yul 9|- FV F S 8 lhl \ .I .8 § v ..‘ I .- I.
9 VI ‘A‘ o i .q A. .v *1. 95 ac n\.v A r
P e . . . . ".1. r ll F. :L . t .3
IVI I}. d I. ‘4‘ 1'! rl. .V. a. .w;.
- I 3 a \ .. At» ‘4- An. A . AI: IV a n
E r. . a .' .1. a r. a v .
o ..5 .2! A Q. t. ‘1‘ I § I '
MT. u... .f .1. Lu. “I .umu . . .r.-. v.1 z. .

 

29

system. As such these factual definition statements1 can entail other
empirical statements with deductive certainty within a given scientific

paradigm.2

Hume's Arguments and Rejoinders

 

The Humean view of causal connections
As we have seen earlier in our discussion of the regularity view,
Hume implicitly reduced causal law statements to generalizations of fact

when he gave the following definition of "cause" in the Enquiry Concern-'

 

ing_Human Understanding:

 

. . . we may define a cause to be an object followed by

another and where all the objects, similar to the first

are followed by objects similar to the second.3

It is this definition which gives the crux historically of the

regularity view that laws express the constant conjunction of events.
As we look about us, we notice that C is regularly followed by E and
we say that C is the cause of E. But we notice onJy invariable
sequences and we do not perceive any necessary connection between
. C and E such that the one should always follow the other.

Hume based his above claim on a thoroughly empiricist episte-

mological theory which today strikes us as exceedingly simplistic, but

 

1In Chapter III we shall explain these quasi-reportive legis-
lative statements in full.

2See The Structure of Scientific Revolutions by Thomas S. Kuhn
for discussions of sCientifiEparadigms.

3Hume, op. cit., p. 100.

AU

r\ .
.HJ
. v
A:

. V

'0-“f

n on."
.E

a

 

 

 

 

30

let us closely examine eachpoint of Hume's claim with regard to his

definition of cause.

Hume's a priori argument

Hume first explains what he means by a priori by discussing
two kinds of statements, i.e., those that express relations of ideas
and those that express matters of fact. He says:

All the objects of human reason or enquiry may naturally
be divided into two kinds, to wit, Relations of Ideas and
Matters of Fact . . . that the square of the hypothenuse
is equal to the square of the two sides is a proposition,
which expresses a relation between these figures. . . .
Propositions of this kind are discoverable by the mere
operation of thought without dependence on what is any
where existent in the universe. . . .Matters of fact

. . are not ascertained in the same manner; nor is our
evidence of their truth however great of a like nature
with the foregoing. The contrary of every matter of fact
is still possible; becauSe it can never imply a contradic-
tion and is conceived by the mind with the same facility
and distinctness as if ever so conformable to reality.
That the sun will not rise tomorrow is no less intelligible
a proposition and implies no more contradiction than the
affirmation, that it will rise.

It is, of course, only reasonings about the relation of ideas
which can be discovered a priori. Reasonings concerning matters of
fact seemed to Hume to be founded on the relation of cause and effect
which is the only relation by which we can go beyond the evidence of
our memory and senses. But Hume goes on to claim that:

The knowledge of this relation is not in any instance
attained by reasonings a riori :but arises entirely
from experience, when we 1n , that any particular

objects are constantly conjoined with each other.
Let an object be presented to a man of ever so strong

 

1Ibid., p. 62.

- A'A rn€rfi
fr. v‘i-s !

' ‘ ,_
'31 ”1"- I

 

 

 

31

natural reason and abilities; if that object be entirely

new to him, he will not be able, by the most accurate

examination of its sensible qualities, to discover any

of its causes or effects.1

In support of the above claim, Hume goes on to say that in

pronouncing the effect of an object without benefit of past observation
the mind can only invent or imagine some event which it ascribes to the
object as its effect; and this invention must be entirely arbitrary.
Thus, even with "the most accurate scrutiny and examination," the mind
can never possibly find the effect in the supposed cause.2

The novel object objection to Hume's
apriori argument

 

It has already been noted that Hume maintained that if a man is
presented with an entirely new object, he cannot by the most accurate
examination and scrutiny of its sensible qualities discover any of its
causes or effects. Without the benefit of past observation, the mind
can only invent possible candidates as the effects or the causes of
effects and this is entirely arbitrary.

While Hume's language seems straightforward enough, one can yet
only wonder what thoughts are really being conveyed within this great
clarity of expression. Are not the words "examination" and “scrutiny"
connotative of empirical observation? Is Hume saying that without past
observation, present observation is useless in determining the causal

properties of an object or the effects which it might issue; or is he

 

1Ibid., p. 63.
2I id., p. 65.

 

 

 

saf'g tr” ;

UV"

 

"SEI'CES of

I I‘ll

:vat v
‘I ‘I I
.-.S: .. 1e. '3 .

‘ .I «1]
Jal'll] Cl

I“

1".‘A- .P"
" “'35 w I-

32551.6".6';
TIME": ‘a‘."
"'0 IFECK r

‘ v
'3' a {32.5

EU: uf 5

d flip. 1
143e‘VE:
. I
“'9 5.“.

 

s
\ I
"l Je .
Q.‘.‘
<- 1
F.
j 511
.l
s "
.
J: i‘gu.‘
v i ‘
"I..
'- uf‘

32

saying that accurate examination and scrutiny of an object are not
instances of observation?

If we have never seen a bowling ball in all our life and
suddenly, some Monday morning, one appears on our doorstep, is it not
possible that by close examination and scrutiny we can tell that the
ball will dent the top of our wooden dining table even though we should
not toss this heavy, hard object upon it? Though we have had no previ-
ous experience with bowling balls, we have noted in the past that
billiard balls dropped on thin glass crack the glass. We have also
seen wrecking crews knock down buildings with a large ball suspended
from a cable. Surely then Hume must mean, not only that a man must be
devoid of any past experience of a particularly newly presented object,
but also in this case that the man must be devoid of all experience of
balls and wooden tables and sheets of thin glass. But is it not also
the case that, after rising from a cushiony chair, one often notices a
depression or imprint left in the cushion which looks in shape and
de51'9" much like that portion of the body which was just in previous
contact with the chair? The memory of this type of incident might lead
one to the conclusion that the bowling ball will damage a wooden dining
table. Are we then, on Hume's terms, to eliminate all past observations
0f ourselves sitting and for that matter lying down or standing as when
we leave our footprints in mud, sand and for some special feet, cement?
It must be then that Hume is asking that a given man, presented with a
tOtally new object, be devoid of al_l_ past experience entirely if he is

t° be completely incapable of making any causal appraisal of a novel

Object ,

. w
V‘-

6r
by

Int-F.
'
u-fUV'
, o
r .a'
I»
.‘p‘ n; [-I'
:1 vi

\
.11
D
'5

 

F I r l O. n o n...
at. I 6 v ‘1 n v nru .Vul.. —
.n PI. Ale J - ..
. 6 u I. ,4 u .o . 1F.
. .u. a: a a C» .o s is J
P u .a‘ .Flv n F-.. .
-\- S r .n. 1 l 6 s a u
.4. I I Au» 0 u an» ..
Vb.» I... .. .\. r .t\ o a »
11. v .I
_. n . .6" .p . .r. w- . .

 

 

 

.33

Thus our novel object criticism of Hume's a priori argument
is that the cause of an effect or the effect of a cause. even with
respect to an object that is totally unfamiliar, can be known inde-
pendently of the observation of their constant or single conjunction.
There may be associations or logical connections between ideas (even
between ideas which pg copies of sense impressions) which are in the
mind of observers though these logical connections are not themselves
ideas which are copies of impressions of logical connections (rubber.
bands etc.) in the empirical world. It is often the case, even in
science, that certain logically connected ideas are chosen as laws
instead of other logically connected ideas because the former permit
more deductions about the causes and effects of novel phenomena than
' the latter. Despite his a priori argument (the name of his argument,
not the kind) seemingly to the contrary, Hume recognizes this possi-
bility himself:

T'Wi ll here be worth our observation, that the past
experience, on which all our judgments concerning cause
and effect dependi may operate on our mind in such an
i"Sensible manner as never to be taken notice of, and
may even in some measure be unknown to us. A person,
who stops short in his journey upon meeting a river in
his way, foresees the consequences of his proceeding
1"Olr‘ward, and his knowledge of these consequences is
conveyed to him by past experience, which informs him
01’ such certain conjunctions of causes and effects.

”12 can we think, that on this occasion he reflects on
any past experience, and calls to remembrance instances
that he has seen or heard of, in order to discover the
Effects of water on animal bodies. No surely; this is
not the method in which he prOceeds in his reasoning.

e idea2 of sinking is so closely connected with that

\

1It is not obvious whether psychological or logical.

2Italics mine.

 

 

 

a' ' r
db. 5 UI
f .‘
r
A. c.(

‘I I
"_“'.Qn ,
.x' . :S 1’:

5f]! '1 A .

"~15 “a

1‘.

‘Ssue Mr 3

(I..: ‘fr‘l
I11. 7
' ‘ UV “

 

“ J: Qre

I: :irrqt 4

a,

‘ "C "an.
Vi

.‘A

‘\ F .

.HE M..Or
‘
.1

 

It:

 

34

of water, and the idea of suffocating with that of
sinking, that the mind makes the transition without
the assistance of memory . . . experience may produce
a belief and a judgment of causes and effects by a
secret Operation, and without being once thought of.1

Analogous with our bowling ball example, this river may be the
first a man has ever seen. But while scrutinizing only its sensible
qualities the man may presume the effect of the drowning of some animal
bodies via past experience with pools or lakes as an effect that might
issue from this novel object, the river. The man has not seen one
single conjunction of the impression of a river and the impression of
mmeone drowning in it. But he does have an idea of water which, in
Hume'51words, is insensibly connected with an idea of drowning as is
fineidea of heaviness and solidity with the idea of damage in our bowl-
ing baTl example. Thus if by a priori reasoning Hume means reasoning
done before any observation or experience whatsoever, his claim that
one canruat discover the cause or effect of an object is rather trivial,
for who would dispute it? But let us pursue this examination of his
a priori argument.

Might not accurate examination and scrutiny consist of experi-
mentation? Webster's Dictionary says of "scrutiny" that it is critical
Obse"vation or minute inspection. "Examination" is defined as the act
0f examining or state of being examined; a search or investigation, or
a tEStlrig by an appropriate method.2 It would seem, thus, that it can
be infe"red from Hume's own choice of words that the effects or causes

\
1David Hume, A Treatise of Human Nature, pp. l03-lO4.

ZNebster's New Collegiate Dictionary, l3th ed., pp. 286, 76l.

 

"irr

""

 

 

.2

-. 1"I;_

 

OD dA-dq ;‘
1’ I
R ‘5". '

I. 0’ .0.
1 - JSL :n

'¢:“r :rH1
Inn-- in.

Pr '.
't 35 3'2?

I

I I
In
It)
\

E. L; '
u. .‘r the

“In. '
«:3'5;6.
I.

3's

. .’!-
u \ 'gp,
' h
F‘v‘
' r
:r- .r
w. J: 4
Y
n ,-
o
b
'-
prod
‘ ‘. 'A‘-C.
\ r
Ubv¥
I‘

 

{M
u .
v q"
Hr In, '
U
5L
'9-

35

of an object cannot be acquired by experimentation with the object

on a given occasion by the man in the street. But we are hard put

to decide if this merely is an expression, on Hume's part, of a lack

of trust in the wit of the ordinary man, or an underestimation of the
breadth and scope of this man's past experience, or a stipulation that,
when an ordinary man examines a novel object very accurately, he must
only stare at it very, very hard and ignore any cues of size or solidity

that he may notice.

One defense of Hume
It might be proposed in Hume's defense that he did not consider
experimentation within the bounds of accurate examination and scrutiny
at least in everyday life. Perhaps Hume recognized a strict distinction
between the casual observations of everyday life and the experimental
observations made by scientists in their laboratories. Such an inter-
pretation would render Hume's words in the quotation on pages 30-3l
of this paper more cogent, but unfortunately such an interpretation
is not one derivable from Hume's writing either on the whole or in part.
In close reading of Hume, the absence of reference to experi-
mental observation is striking. It is almost always the case that
Hume's examples of observation instances are of the observations of
everyday life, e.g., we observe constant conjunctions of heat and flame,
snow and cold, day and night. As commentator A. P. Cavendish notes,
Hume's idea of experimentation in his own inquiries is not one of

scientific experimentation as moderns think of it:

 

HE 161

' 6
am :52.

' ' ‘.
_:r '10,.
‘:1e ex:
fave c."
:a;.' inc.
were:
m 15

"mpg...

w' lbhn
‘.‘-3r of
£
13:3"

n
fir. A.
V” v'Ju"

u

 

 

 

 

 

 

 

36

Hume tells us he is going to study "the Science of Man“

and that the object of this science is "to explain the
principles of human nature.“ The method he proposes is
"the experimental method," and this involves the perfor—
mance of careful and exact experiments. These experiments
may include either introspective observation of mental
phenomena, or the objective observation of human behavior.
Hume is very cautious about introspection because, he says,
"reflection and premeditation would so disturb the opera-
tion of my natural principles, as must render it impossible
to fOrm any just conclusion from the phenomenon." He
concludes that "we must therefore glean up our experiments
in this science from a cautious observation of human life

. . . by men's behavior in company, in affairs, and in
their pleasure." All this makes it look just as if he is
going to do what we should now call experimental psychology.
BUt if we read Hume's works, we do not find any accounts of
"careful and exact experiments," nor do we find any "cau-
tious observations of men's behavior." There is something
that Hume calls an "experiment," but this is a procedure
all his own and it is invariably introspective in charac-
ter. There are observations of human behavior but they

are not accounts of a particular man's behavior in care-
fully specified circumstances. They are accounts of
certain general features of human behavior, which are

in fact obvious to everybody.1

It might be objected that the above passage indicates only that

in the study of human behavior Hume fell short on an experimental ideal

that theoretically was not ready for implementation in his time. How-

ever, the passage indicates more than this; for it shows Hume's willing-

ness to use the term 'experiment' in reference to everyday observations.

In fact it is the all inclusive term "experience," not "experiment,"

which Hume consistently uses to refer to that which is the great guide

of causal inference.

There are some commentators who seem to disagree with

Cavindish's appraisal of Hume's use of "experiment." Laird claims

 

1A. P. Cavendish, David Hume, pp. l9-20.

 

 

.13”

 

 

W in ..,

i..", b .

... ' a.
5&6"? .:

.. ‘ . .-
r . "n'
erJ ' r

f -
r::‘~:n nr
5 , V

1 I
‘7“ « rc...
' 'r‘ 532.
g

pun”), ‘
:1... ,6!" 55

Gr 0.";
IA.
“Vin, :-

 

 

 

 

37

that, in the Treatise (p. 332ff), Hume "conducted psychological
experiments with great precision according to the formal rules of
natural philosophy."1 However, if one but turns to pp. 332ff of the
Treatise, one finds, as Cavendish describes, only introspective‘a
priggi_reasonings about human behavior which Hume refers to as
experiments. In the section entitled "Experiments to confirm this
system," Hume presents in his second experiment the following:

Thus suppose, I regard a stone or any common object,

that belongs either to me or my companion, and by that
means acquires a relation of ideas to the object of the
passions; 'Tis plain, that to consider the matter of a
priori, no emotion of any kind can reasonably be expected.
For besides, that a relation of ideas operates secretly
and calmly on the mind, it bestows an equal impulse
towards the opposite passions of pride and humility, love
and hatred, according as the object belongs to ourselves
or others; which opposition of the passions must destroy
both, and leave the mind perfectly free from any affection
or emotion. This reasoning a priori is confirmed by
experience. No trivial or vulgar object, that causes not
a pain or pleasure, independent of the passion, will ever,
by its property or other relations, either to ourselves
or others be able to produce the affections of pride or
humility, love or hatred.2

Further support of the Cavendish claim and the one expressed
herein that Hume used the word 'experiment' (where he did use it at
all) in an everyday observation sense can be found in a quotation from
Hume that Laird himself selects to discuss elsewhere in his book Humeig

'Philosophy of Human Nature. In the quotation Hume refers to historical

 

events as experiments!

 

1John Laird, Hume's Philosophy of Human Nature, p. 142.

2David Hume, Treatise, p. 332.

 

 

 

Its the
3:: Jr‘.
1r. 3)) I
{aw-5,.
01.1335
swirgs
IETS, l.’

|! .
'fl F+‘
.‘II 5'.

ml :'
the Si’é
:e::'£-=
3‘36" E)

“‘FFGN.
”V by. .

 

'2‘ ..

a! A

" ‘ of 9
5|

3!-

‘: 1’ I 9

:3: _
s h“?-
it')’ 5’“
I.“
l rc. A
J v.»-
::Q ‘
r: -’ ’ ‘
|J PP:
.7 l
1n r. .
S.F.1
‘ a ‘7',

 

38

Its chief use (history) is only to discover the constant

and universal principles of human nature by showing men

in all varieties of circumstances and situations, and

furnishing us with materials from which we may form from

our observations, and become acquainted with the regular

springs of human action and behavior. These records of

wars, intrigues, factions and revolutions, are so many

collections of experiments, by which the politician or

moral philosopher fixes the principles of his science, in

the same manner as the physician or natural philosopher

becomes acquainted with the nature of plants, minerals and

other external objects, by the experiments which he forms

concerning them.1

It is obvious in light of his own usage of the term "experiment"

that Hume, more often than not, regards "experimentation" as equivalent
not only with "observation“ but with the observation of the facts of
everyday life. However, while this conclusion undercuts the first line
of defense that might be offered in Hume's support, there is a second

line of defense which must still be considered.

A secondgpossible defense of Hume

It might be conjectured that if "scrutiny" and "accurate
examination" in Hume's work include operations that life within the
realm of even everyday kinds of observation, some types of scrutiny
are only of sensible qualities while others (which reveal causes and
effects) are observations of motion, solidity, extension, and other
primary qualities. Therefore, Hume is really maintaining that scrutiny
of only the sensible qualities of a novel object, such as the bowling
ball in our example, cannot result in knowledge of its effects; but

scrutiny of the motion of the bowling ball as it is drapped on a dining

 

1Laird, op. cit., p. 264.

39

table together with scrutiny of the impenetrability of each object, ball
and table, on contact can result in knowledge of one of the effects of
the bowling ball, a dent. In other words Hume is stating that observa-
tion of the mere sensible or secondary qualities of an object is not
enough to enable anyone to make a determination of possible effects.

While the second defense of Hume above seems to be a very
plausible interpretation of Hume, nothing could be further from the
truth.

In Hume's theory of knowledge the traditional secondary qual-
ities such as color and taste, which Hume calls "sensible qualities,"
are not mere sensible qualitiesat'all.~ Rather they are on the same
footing, if not a better one, as the traditional primary qualities of
solidity and extension to name a few.

Unlike his predecessors,1 Hume denies that the distinction
between primary and secondary qualities is disclosed by the senses.

As Ralph Church says in'Hume's Theory of The Understanding:
Thus he (Hume) proceeds to urge that the primary qual-
ities may not be known in separation from the seondary.

Since motion must be that of a body, motion is not

imaginable apart from extension and solidity. Extension
is not found apart from colour. Hence extension may not

be known apart from that quality. . . . Apart from both
color and hardness, no one of "the primary qualities
chiefly insisted on" may be so much as imagined. The
distinction in question, therefore would put beyond the
possible content of belief the very qualities whose
independent and permanent existence it was-designed to
explain.2

 

1With the notable exception of Berkeley.
2Ralph Church, Hume's Theory of The Understanding, p. 99.

 

40

Laird too states Hume's words:
When we reason from cause and effect, we exclude
that neither colour, sound, taste, nor smell have a
continu'd and independent existence._ When we exclude
these sensible qualities there remains nothing in the
universe, which has such an existence.1
In the Treatise, Hume's discussion of sensible qualities is
one which aims to challenge the representative theory of perception.
He says:
First, 'Tis easy to observe, that tho' bodies are felt
by means of their solidity, yet the feeling is a quite
different thing from the solidity; and that they have
not the least resemblance to each other.2
Hume can be understood to be arguing in connection with sensible
qualities that if one establishes external entities by saying that
perceptions must be caused by something, the non-permanent character
of our perceptions suggests an external world of discrete non-enduring
objects which causes them. Therefore, advocates of the causal repre-
sentative theory of perception cannot thereby establish a belief in the
continued and independent existence of body. But Hume's argument
(pp. 230-23l) seems self refuting since Hume has declared that a
sensible quality cannot automatically be thought to be like the hidden
external cause of it and thus an advocate of the representative theory
of perception can claim that discontinuous perceptions are caused by

permanent independent external objects.

 

'Laird, op. cit., p. 154.
2Hume, Treatise, pp. 230-231.

(I)

:IE‘S CIS

l
.50 15w- #6
v v-‘h ’-

As tc ‘
train

.
NPR-In 1

Ch...
‘11::5'
:rre.ie
cream

”A.
'J-wy
T" u‘
.; 5" I

u. _

K

‘rcc :r
the or
be We

 

 

 

 

 

 

 

41

But this view of sensible qualities has nothing to do with
Hume's discussion of particular causes and effects within the realm
of perceptions. Hume says this himself:

As to those impressions, which arise from the senses,
their ultimate cause is, in my opinion, perfectly
inexplicable by human reason, and twill always be
impossible to decide with certainty, whether they arise
immediately from the object, or are produced by the
creative power of the mind, or are derived from the
author of our being. Nor is such a_guestion anyway
material to our present purpose. we may draw inferences
from the coherence of our perceptions, whether they be
true or false; whether they represent nature justly, or
be mere illusions of the senses (p. 84, Treatise).

When Hume does speak about sensible qualities in his discussion
of particular causes as in the passage quoted on page 7 of this chapter,
Hume's words inescapably lead to the conclusion that if causes and
effects are discoverable at all it must be via observation of only
sensible qualities whether in an experimental or nongexperimental
situation.-

Cavendish again discusses the doctrine of hidden powers which
Hume was arguing against. It was held by some thinkers still in Hume's
day that in order to know that bread will nourish us we need only know
as certain that it possesses a certain hidden power in the same way as
the whiteness of the bread. It was thought that we might come to see
the power in the same way as the whiteness of the bread. This power
was regarded as a quality of bread just as its whiteness. Cavendish
continues:

Hume's objection is simply that it is a false account.
Suppose we consider the statement that salt is soluble

in water. It is agreed that we can know this only by
experience. But what exactly is the experience? We

 

:re:::.;te
:near‘es e
11:: intere

L51' M N

 

 

ls-

l’,‘

42

see the salt immersed in water, and then we see it dissolve.

We do not see anything about the salt that necessitates its

dissolving; all we see is that it does in fact dissolve.

And even if we perform the experiment a large number of

times, we never see anything more. We never see in addi-

tion the power that makes the salt dissolve.1

Cavendish explains that philosophers of Hume's period were

preoccupied with epistemological problems rather than explanatory
theories except for the explanatory theory of hidden powers.2 Hume
was interested in the question of "how do we know bread will nourish
us," and not in the question "why bread will nourish us." We observe
only a constant conjunction of people eating bread and people being
nourished. 'However, each conjunct is a collection of sensible qual-
ities. Only sensible qualities of bread and people are observed and
the effect of bread-~nourishment (a healthy looking moving body) and
the cauSe of nourishment-~bread can be known via this constant conjunc-
tion of sensible qualities. It is this interpretation, not that of the
second:proposed‘defense, which is consistent with Hume's philosophical
views. James Wilber's words seem to sum up this compatibility of our
interpretation of Hume's words and Hume's actual words. Wilber writes:

When thinking in a Newtonian analytic view such as in

his preliminary analysis of the idea of cause and effect,

Hume has regarded experience as a collection of impres-

sions. Though the collection contains an order within it,

an order arising from both temporal sequence and associa-

tion of various kinds of impressions with others, this

order is not regarded as anything beyond the collection

itself. Coupled with his empiricism. it is this view of

experience that has given rise to the Pyrrhonist diffi-
culties Hume has already encountered. To speak of

 

'Cavendish, op. cit., p. 70.
2Ibid., p. 67.

43

experimental reasoning, on the other hand, is to speak of

something involving reason as well as experience and sug-

gests that the "man of experience," in order to fulfill

his capacity for experimental reasoning must have acquired

certain beliefs and habits.1

In view of all of the above, it is quite unclear what it is

that Hume is really asserting or means to assert about a priori
reasoning from mere observation of the sensible qualities of a novel
object. The answer may lie in pinpointing some possible confusion in

Hume's writing.

The single case objection

Without giving Hume any benefit of doubt, his words with respect
to the discovery of a cause or effect are, if intelligible at all. sheer
nonsense. However, with the benefit of a doubt, it is possible that
Hume is guilty of a subtle confusion between the discovery of particular
causes and effects on particular occasions with the discovery of causal
laws. We can discover, on a particular occasion, that the dent in our
table is being caused by the dropping of a bowling ball upon it. But,
that does not mean that one can confirm from the one instance a causal
law, which expresses the fact that bowling balls dropped on wooden
tables of certain specifications, will always dent the tops of the
tables.

Hume's possible confusion of the discovery of a particular
everyday cause with the confirmation of a causal law could easily stem

from an on-going vagary of meaning surrounding the term "cause." Since

 

'James Wilber and Harold Allen, The Worlds of Hume and Kant,
p. 66.

44

the advent and development of science, the chief concern about causality
has had to do with the causal relations expressed in general laws. But
not all the law statements in science are causal law statements, and

not all causal statements are lawlike causal statements. If these two
distinctions are kept in mind, the confusion over the terms "cause,"
"causal," and "causality" should soon dissipate.

Now when a given cause is to be discovered, one can often deter-
mine the cause of an event even when it occurs for the very first time.
In terms of Hume's discussion, the effects of an object can be deter-
mined after close examination even if the object is totally new. In
the example of the bowling ball, we can probably tell just by looking
at it that this object will no doubt dent a dining table if dropped upon
it. Should a doubt exist as to this effect, we have merely, within the
bounds of close and accurate examination and scrutiny,1 to drop the ball
‘on the table. If a dent appears immediately thereafter in the table top,
almost anyone present can conclude that dropping the ball on the table
has caused the dent. This judgment is made even by those persons who
have never before observed a bowling ball or its being dropped on a
dining table top. By what ratiocination is this discernment of the
cause of the dent possible? Ducasse offers a very plausible answer.

He states:
. . . the causality relation is one that obtains between

three terms of a perfect experiment--the relation, namely,
between a given state of affairs S and two changes in it;

 

1Here we mean scrutiny of experimental sensible qualities in the
manner of experimentation that Hume's use of “experiment" implies.

 

. “Al(l~

 

oxe of
E at a'
triaci:
, .

1. 1r. S
1718 ('
COHEN
By and
'25.?
frat:
an al

CCCuf'E

 

 

45

one of them a change C at a time Ti’and the other a change
E at an immediately sequent time T2; To say that this
triadic relation obtains between S, C and E is to say that
C in S is etiologically both sufficient and necessary to E
in S (i.e., both cause of and condition of E in S) and
conversely, that E in S is etiologically both necessitated
by and contingent upon C in S (i.e., both effect of and
resultant of C in S). This definition of causality, being
framed in terms of but one occurrence of the sequence in S
of C and E, does not require the supposition that it ever
occurred before or ever will again.

Ducasse is of course pointing out something that is so obvious
it is usually overlooked in discussions of causality. In the everyday
situation of driving a car, it is easily noticed that occasionally, in
the course of normal driving, the car sometimes turns to the left and
at other times to the right. If the cause of this turning of the car
is unknown, the driver may compare the situation at time t1, before the
car starts to turn, with the situation S at time t2 of the car, as the
car starts to turn. The conditions necessary for the car's operation
at both times seem to be the same. The ignition is on, the engine is
turning over, the gas pedal is depressed. But in S at t2, there is one
change or difference of condition. In S at t2, the driver's hands are
turning the steering wheel left or right, i.e., C occurs. Since this
condition can be narrowed as the only new element in the situation by
which we may in fact distinguish S at t1, from S'at t2, it is concluded
that commencing to turn the steering wheel with my hands to the right or
the left in S at t,, thereby initiating S at t2, is the cause of the car

turning to the right and left, respectively, of its own power. The

 

1Curt John Ducasse, Causation and the Types of Necessity, p. l38.

 

 

arcing 0f
31:51) ce 1
C'taming

0r.
recessary f
a‘s: tw: t
s.r“1‘cient
'2) itself,
:36 car to
n‘3e the i
1‘6 left.

Fr:
the his":
"Ether t“.-
:r::1.r1;-d 1.
)Cfi‘ng f:

I
if. 1
".H: ‘

he 1'

h

a I“ ‘
‘-~5>T:

 

hs“

 

46

turning of the car, E, occurs only after S at t2, and not S at t1,
but since there is no difference between S at t1, and S at t2, except
C (turning the steering wheel), E must be due to C in S at t2.

On the view of Ducasse then, all the conditions of S at t1 are
necessary for the car to turn, but they are not sufficient. One must
also turn the wheel. Hence, turning the wheel is also necessary and
sufficient together with the conditions of S at t1, to E. Of course C
by itself, i.e., not in S, though necessary, would not be sufficient for l
the car to turn of its own power, for if the steering wheel is turned
while the engine is off, the car will not turn to either the right or
the left.

Practically there are some problems in applying this method in
the determination of causes, for it is sometimes not clearly known
whether the only change observed is in fact the only change that really
occurred in the situation. But this is precisely what one should be
looking for when close examination or experimentation is made. It is
both the ideal goal of experimental science and common sense to control
all possible variables, with the exception of one, in the search for a
particular cause. If the uncontrolled variables reduce to more than
one, then we must conclude only that E was caused by one or several of
the changing variables.

Arthur Pap has criticized the above definition of causality,
proposed by Ducasse since it supposedly makes the words "cause" and
"effect“ applicable to only concrete events. Ducasse vindicates his

theory with the gentle reminder that the instances of given kinds of

47

events always consist first of concrete events. As every state of
affairs that exists and every change in it that occurs is fully con-
crete, the initial application of "cause" to the concrete situation is
a requirement which any correct analysis of causality would have to
satisfy. Statements of causal laws can be interpreted with no dif4
ficulty in terms of the concrete analysis here presented.
. . the rough causal generalization: "Throwing bricks

at windows ordinarily causes them to break" would be

explicated as: "Any concrete change that is an instance

of the kind "throwing a brick at a window" does under

concrete circumstances that are an instance of the

"ordinary“ kind, cause a concrete change that is an

instance of the kind "breakage of the window."1

It is at this point, the transition of particular cause talk

to causal law talk, that we may easily return to the words of Hume.
Hume has already said, it may be remembered, that knowledge of cause
or effect in an object cannot be obtained through close inspection or
examination when past experience of the object is lacking. It has just
been shown that such a claim on one interpretation is simply untrue and
on a second interpretation trivial to the point of being nonsensical.
However, it may be that we can construe Hume to be really arguing that
one cannot confirm2 a causal law concerning an object without the aid
of past experience. In other words, as he himself says, knowledge of

cause and effect (in the sense now of causal law not concrete particular

causes and effects) cannot be attained by a priori reasonings but only

 

1Ibid., p. l4l.

2Hume-speaks of the discovery of causes. We are giving his
words a loose interpretation to include the confirmation of a lawlike
causal connection in a last attempt to salvage his argument.

 

 

‘or age

i‘vE'CS CO'

.3 ‘IPF
seam. .:

33.52531 I

we 3:32

"~17 “.55
~11, _
Kill, bu:

 

 

;:;:P. t
"b 0.

'1':
sir
.Io.
the ¢
:1.

.1“H ‘bf

48

from experience when we find that any particular objects are constantly
conjoined with each other.' Thus only in the light of the experience of
events constantly conjoined experimentally do we confirm a causal law
concerning them. He describes this step as one, from the notice of a
constant conjunction of an object and its effect to the belief such that
other objects which are similar in appearance will be attended with
similar effects. However, Hume uses elementary sense data examples of
the constant conjunctions which presumably are the inductive basis for
confirming causal laws.- More than once, does he refer to heat and flame
or snow and cold, when referring to the constant conjunctions of objects
and their effects. It is difficult to talk of such mundane conjunctions
in discussing the topic of causality especially causal laws in science
at all; but for the sake of argument, let us discuss a low level con-
stant conjunction of past experience which presumably is the basis for
the confirmation of a particular causal law.

Correlation of events objection to
Hume's a priori argument

 

Many philosophers including Hume discuss the constant conjunc-
tion of day and night. If Hume is to be taken at his word, one con-
cludes on the basis of this constant conjunction in all previous
experience, that night will be the effect of day, or day will be the
effect of night or night will be the cause of day or day will be the
cause of night. To handle this rather circular situation, let us assume
that the first people to observe this constant conjunction began experi-

encing the world in the early daylight of morning. On a popular

 

 

 

553K173."
1 4 .
"'IES V121

23.535 1"1' 2"

d

'._: P. .fi 1
8': rd: LU”

3‘ a'ct'er

:‘cr of :11

 

 

 

49

assumption that causes precede effects, it would be quite correct on
Hume's view for these people to formulate the causal law that day
causes night, since Hume claims repeatedly that we are determined just
by custom to expect one appearance, i.e., flame from the appearance

of another, heat, after recurrent observations of the constant conjunc-
tion of the two objects. But surely there is no law that day causes
night nor could there be. Rather, on noticing this regular sequence,
we find a question, not an answer as Hume supposes. Why does light
come and go?

The layman, operating in the Ducasse theory fashion, might look
for a change in the daylight situation which could possibly account for
the disappearance of sunlight. 'He notices that in autumn, the trees
lose their leaves gradually and the days lose their light gradually
sooner and sooner. Here, then, is another invariable sequence—-trees
losing leaves, days becoming shorter. But still we are not disposed to
generalize this sequence into causal law for not all trees shed their
leaves. ‘Though perhaps, this is why we always have some daylight.1 Is
there a predicament in this last discernment? Probably not. The only
constant conjunction of tree with light phenomena ever noticed so far,
is that shady trees block the presence of light beneath them. On this
account we must conclude that the more leaves trees have, the less light
our days have. But exactly the converse is true with respect to shorten-
ing days and deciduous trees. Thus, there must be some other change C

in S which causes E. Alas, as we now know, the change C has been found

 

1The evergreen trees cause the winter days to be as short as
they are.

 

 

 

 

'.C :e 3 VB

argle of t

"f- m";

fire the
:e‘re a c.
an . c .
liflytll] tl

3; :3! ar:

v: 12.1.:

":EP‘ESte:

Ir. -'d 6"...
.. 1. ‘5

”A

L"? a a

{D

i
.. L
’ a

a”
“:1,”
=1.
.'

P I.‘
' ‘
~21:i
8".

 

 

50

to be a variance in distance between the earth and the sun and the
angle of the earth's axis in the situation S of the earth revolving
around the sun, that causes the days to shorten. It is the change C',
the rotation of the earth, in the situation 5', the earth revolving
around the sun which causes night and day.

Obviously there were many constant conjunctions experienced
before a causal law was confirmed in this example. Indeed, it is
doubtful that there is actually a causal law which states the cause
of day and night for it is the common man, rather than the scientist,
who would ask this particular causal question.1 The scientist is
interested more in asking what is the pattern of movement of the earth
around the sun expressed quantitatively and his equation for this,
though a physical law, would not be a causal law. Bertrand Russell was

so impressed by this last point, that in Mysticism and ngjc, he takes

 

the extreme position that there are no longer any causal laws in science;
and therefore, that it is a waste of time to still try to analyze what

is meant by "cause" and related terms.2 That all laws are not causal

has already been mentioned, but it seems doubtful that there are abso-
lutely no causal laws in science or even formulas which can be rewritten
in causal form. Whatever the case, we must begin our discussion with
causal laws in order to discuss any other kind since most historical
discussions, such as Hume's, which still influence present views,

including Russell's, dwell on causality.

 

1Scientists today are asking some other interesting questions
concerning this phenomenon.

2Bertrand Russell, Mysticism and Logic, pp. l803208.

51

The foregoing example‘of the conjunction of day and night,
turned to causal law, should indicate that if Hume was talking about
the observance of causal conjunctions directly leading to the confir-
mation of laws, he was wrong. As a matter of fact, the observance of
constant conjunctions in the past often tell us nothing but merely
raise questions as to causes. It is usually only observations of
constant conjunctions occurring during experimentation that provide
the basis for the confirmation of causal and non-causal laws alike.

In th‘i 3 experimental sense, Humeis almost right in saying that men
will not infer from one case of a constant conjunction or more properly
one Conjunction what they will from a thousand, though the one be in no
"35’ d1 fferent from them.- Though most of Hume's discussion never refers
t° exDerimentation, it'is true that experiments are usually1 repeated
many times before a hypothesis is confirmed and changed in status to a
law. But there are cases, it is generally allowed, of crucial experi-
ments where the singular conjunction of events decides the case within

the realm of lawlike statements.

Mucial experiment objection

It will be remembered thatwe have already noted that Hume in
at least one instance concedes the point of. the crucial experimental
obj€2Ction Hume claims in theTreatise (pp. 104, lbs) that, "we may
attain the knowledge of a particular case merely by one experiment,

p"‘Ovided it be made with judgment, and after a careful removal of all
\

lHere "usually" means somewhat less than almost always.

 

 

 

l- a .
GE..n..‘AV

'llg.

"5: FCC:

 

11".. -

V v.1 1:

'3. . .

' we. '7‘]

:E'o“

v' ‘A.

'. IPA
‘E

 

52

foreign and superfluous circumstances." However, Ducasse suggests that
while Hune's "Rules ‘by’which to judge of causes and effects" (p. l73),
may seem to be Hume's guide for isolating "foreign and superfluous
circumstances" in a crucial experiment they are not, since these prin-
ciples "as he (Hume) states them assume knowledge of single facts of
causation.“

When Hume claims by the first three principles, which are really
defirii tions, that the cause and effect must be contiguous, the cause
"“1511 proceed the effect, and that there must be constant union betwixt
the Cause and effect; he presumes already that the cause is already
know11.. For how can one judge if the cause is contiguous, precedent,
Md constantly united with the effect unless one knows what the cause
is before judging if it, the cause, satisfies these conditions. Yet
th"Oughout his writing Hume maintains that causal inference (knowledge
01' a cause) p_1;_i_s_e_§_ from'experience of constant conjunctions.

Laird, in his book, makes the further criticism that by Hume's
0"“ view, he has no right'to'advance any rules for use in an affair
Wh‘C*\ is "sensitive" and "alOgical."2 Laird contrasts the rules with

that which is alogical for "Hume said (p. 175) that they (the rules)
Supp] ied all the 'logic' he intended touse" in this matter.3
Thus it appears that our crucial experiment objection to Hume's

a
pr“iori argument is relevant since Hume does not really seem to allow

__‘y‘~‘_‘ ..
1Ducasse, op. cit., p. l3.
2Laird, 0p. cit., p. l40.

3Ib‘ld.

53

for thi s possibility or'recognizexitsdestructiveimport for his
constant conjunction view of the confirmation of causal laws. It

also seems that the crucial experiment objection to Hume's a priori
argument, together with our other three objections, indicate that the
first objectionl empiricists raise to necessity views of causal laws

is ambiguous if not utterly ill-taken.

It will be remembered that the first empiricist criticism of

a necessity view of law'statements was that we cannot prove any law
(causa1) a priori but only establish it as an empirical generalization.
This ‘i s essentially the claim of Hume's a priori argument that causes
or causal laws can only‘be'discovered or perhaps justified by the con-
stant repeated conjunction ‘of similar events. Yet our four objections
indicate that whether Hume was talking about particular causes, causal
laws, ‘ their discovery or justification his claim is unsupported. Some-
times we can know that some particular event will follow another without
the Previous experience of the two events constantly conjoined (our
“0"91 object objection). Sometimes too a cause can be discovered via
experience of just a single conjunction of it and its antecedent (our
Single cases objection). Conversely there are oft repeated constant
conjunctions of events in our experience which we never come to consider
Iawful relationships (our mere correlation objection). It is also the
case that even causal laws can beconfirmed by a single conjunction of

ev
e" ts (our crucial experiment objection). We thus conclude that if
\
1See page 26.

flTFi—‘T-i

L

“a

 

 

 

Okanf‘
3111‘.- '3

..-.' '..'
E..!"..if

.1 P P-
a? .‘. L.
‘

 

 

 

54

anything is wrong with a‘necessity theory of laws it is not the first
empiricist criticism that laws are empirical and therefore in no way
a priori. However Hume does offer another argument in support of this

empi ri cist criticism which we shall now consider.

Hume' s conceivability argument

Hume's conceivability argument is best expressed in the follow-
ing Question which he asks. Even if motion in a billiard ball A, in
direct line of a moving‘billiard ball 8, should be suggested to me
as the result of their contact or impulse; may I not conceive that a
hundred different events might also follow from that cause? In sum,
he con cl uded'that every'cause and effect are totally distinct from one
anOtherand that neither'can be inferred from the other without obser-

vation and experience.

conce1 va ty-cr ter1on
It must be apparent by now that Hume is talking about cause

i“ a most confused way. Where it appears that he is speaking of the

”Mom man'susage of the term, Hume is wrong and superfluous since

it 'i S no man's option to consider objects without any benefit at all
Of past experience that might be helpful. But where it appears that
Hum may be discussing the epistemological basis of causal laws, he is
ta] k‘ing about the practice of some science that does not use any exper-
imentation as we think of experimentation and confirms sense data type

la
“'3 such‘as heat is always followed by flame and perception of snow is

 

55

followed by perception of cold.1 If anything. Hume is describing the
scient‘i fic practices of cave men. With nothing to work with but the
observations of such sense data conjunctions as snow is followed by
cold, 1 t must indeed be quite easy for Hume to imagine or conceive
that heat or other unexpected effects might follow snow in their
Perceptions. On the other hand, if custom were as strong a factor in
men's thinking as Hume claimsit is, how is it so easily conceivable
that. Geffects not bolstered by constant conjunction, or even a single
eehjtlrlction, e.g., the sun will not rise tomorrow, are possible?
Either custom impels us or it does not. Zabeeh indicates that Hume's
concee'i vability criterion employed a notion of contradiction which was
base<1 on associations of custom and was therefore a psychological, not
a ‘053‘i cal, type of contradiction. As Zabeeh says:
The criterion of conceivability itself was explained in
terms of the law ofcontradiction, and the notion of
contradiction was explainedin terms of psychology and
not syntax. In Hume's system, the substitution of psy-
chology for logic is inescapable since there is no place
for formal relations in the world of contingent facts,
\Ii 2., in the world consisting only of loose and separable
Eseense-data connected together in our mind by psychological
I‘inks of association . . . his criterion, namely, customs
<>I~ habits, cannot perform its important function, if it is
(Inly psychological and not a linguistic criterion. Unless
‘fige register the customary association of ideas with words
‘In forms of definition or rules, we cannot properly use
Saeneral words in language.2
\
1"Sensation" seems a more appropriate word than "perception."

127‘ 2Farhang Zabeeh, H_u_me Precursor of Modern Empiricism, pp. ll2,

“we w

56

William Kneale criticizes the kind of associations to which
the conceivability criterion “is applied, i.e., the‘psychological
association of sense impressions. As Kneale says:

It is clear, however, that, except in certain parts of
psychology, the laws ‘of'nature formulated by plain men
and scientists are not about-sense, and image, but about
sticks and stones, pieces of metal, elephants, and stars,
a1 1 of which can be perceived but certainly not sensed.
The most plausible example which might be offered to
“i 1lustrate Hume's account of natural laws would be the
Proposition that lightning causes thunder. But even
here reflection shows that Hume's account is inadequate.
Nilen I speak of lightning, I do not mean a vivid visual
Tmpression occurring as an element in my experience, but
rather a public event which can be perceived by other
People or photographed Similarly, when I speak of
thunder, I do not mean the noise impression which some-
121 mes though not invariable, follows a bright flash in
my experience, but rather a public event which can be
perceived by other people or recorded by machine. My
i niagination of a bright visual sensum nOt followed by
a loud auditory sensum does, indeed, enable me to know
1:l1at I might experience the one without the other. For
While imagining this course of experience I can recog-
n 1ze its possibility. But imagination is powerless to
Drove a lack of connexion between lightning and thunder.

 

The 1'ma inar facts ob'ection to the
margumen

Psychological connections between sense data and ideas aside,
it does not seem that the imagination holds sway over logical connec-
tions that might obtain between ideas, and Zabeeh notes that Hume
himself wams against the flights of imagination when one deals with

phi 1 osophy and science. ‘As Zabeeh quotes again from Hune:
\
lKneale, Probability, p. 79.

not *Imagination can only tell us what is psychologically possible
logically possible.

 

 

'1‘ I “M‘EIVMIJ

 

m

#r;‘

01"

03.1?
One

 

5 b.

in. r
it"fir ,;

-I‘l b |
' 1T I

E 1. 5:-
r; ._ |
J :XJLr.
I': . '
* 2r: .

57

He must pardon children,- because of their’age; poets of

their fancy; But what-excuse'shall we find to'justify

our philosophers'in‘so-signal 'a weakness? (flights of

the imagination)..(Treatise, p. 226).1

One reason for warning against the flights of the imagination

is. as Laird suggests, "we have no right to assert that chimeras or
centaurs could exist in-nature simply because we can dreamily fabricate
such monsters.“ As Laird continues:

ljtztne concluded incontinently that "the idea of existence

'15; the very same with the idea of what we conceive to be

existent. To reflect on anything simply and to reflect

(Dru it as existent, are nothing different from each other.

That idea, when conjoin'd with the idea of any object,

makes no addition to it. Whatever we conceive, we con-

Cfeive to be existent. Any idea 'we'please to form is the

‘I dea of a being‘and-theidea of a being is any idea we

IJ'I ease to form" (Treatise, p. 66).3

Despite the above passage, Hume makes it very clear that he

does not think that any knowledge of causes and effects whatsoever, can
be derived by a priori reasoning (such as using a conceivability crite-
rion we suggest). ' ~Yet very-quickly does 'he follow with a pronouncement
about causes and effects by means of his a priori conceivability crite-
Vl°“«- The effect, the sun will not rise tomorrow, will not occur. Thus
either of these effects may-beactual in the future. But is this not a
Claim about a particular empirical event, which is made not on the basis
0f e><perience (since the sun has never failed to rise), but on Hume's a

prion—1 judgments that the sun's not rising is a possible event? What
\

1Zabeeh, op. cit., p. 58.

2Laird, op. cit., p. 80.
31b1d.‘

 

 

 

 

 

r‘y", 'na

 

58

right has Hume, "on his'own principles of strict empiricism. to be making
claims about possible-future‘empirical matters which arise from the a
priori conceivability machinations of his mind! It might be conjectured
in defense of Hume thatin‘considering matters of fact as only the
SUbJ'ect of probable belief the lack of certainty always permits a
counter possibility if probability and possibility are considered
Opposi te ends on a'continuumofbelief. However, we shall illustrate
in the next section, on induction, that in fact Hume considered some
matters of fact the subject-of-certain not probable belief. This cer-
tain'ly is not-demonstrativecertainty, but one to which Hume neverthe-
less ‘appliesthe term-"proof." As Laird'says precisely, "intuitive or
demonstrative'certainty'impliedthatthe opposite was inconceivable.“
In the cases of beliefin-matters-of fact to which Hume is willing to
apply a non-demonstrative sense of "certain" it is not clear whether
one can believe in the possibility of opposite matters of fact or not.
Nevertheless as our psychology and imaginary facts objections
indi Cate Hume's conceivability argument offers neither a logical nor
empi Y‘i cal standard for judging the possible. Thus Hume's conceivability
ar'gument lends no support to the first empiricist criticism of a neces-
sity view of laws, i.e., that laws are in no way a priori. We shall
therefore nextconsider Hume‘s induction argument, the basis for the
secGrid empiricist criticism of'necessity views of law statements that
in tbases of a priori reasoning we attain certainty, but in cases of

c
a”Sal reasoning only probability.
\
‘Ibid.. p. 85.

 

 

"'se to 1
Page the

't an 1,.

iv In

 

 

 

 

59

Hume's induction argument

Hume next'makes perhaps “his most‘important‘point which gave
rise to the traditional problem of induction. He notes that all men
make the following inference, and'he asks by what process of reasoning

it could possibly be made:

At least it must be acknowledged that there is here a
consequence drawn by‘the-mindg'that there is a certain
Step taken; a process ofthought, and an inference, which
wants ‘to be explained.“ These two propositions are far
from being the'same, "I havefound that such an object
has always been attended with such an effect and I fore-
see, that other objects, which are, in appearance, similar,
W‘i ll be attended with similar effects. . . . The connexion
be tween these propositions is not intuitive. There is
r‘equi red a medium, which may enable the mind to draw such
an inference, if indeed‘it be drawn by reasoning and argu-
ment. What that medium is, I must confess, passes my
Cfomprehension; anditis incumbent on those to produce

1 t , who assert, that it really exists and is the origin
01" all our conclusions concerning matters of fact.1

The inference described in Hume's discussion above is not to be
found . he claims, in thearguments of-demonstrativereasoning. But if
this ‘i nference is based on experience, we are involved in a circular
av‘Elm'flent, for the uniformity of nature is then at once derived from
a“‘Qumcents of matters of fact-and supportive of arguments of matters
0f 1="=l<:t.2 All inferences fromexperiencesuppose that the future will
mseMble the past and 'that'similar‘powers will be conjoined with similar
sensi ble qualities. Thus, no arguments from experience can prove this

rese'WJlance of the past to the future since all these arguments are

\

1Hume, Enguiry, pp. 68-69.

2This has come to be known as the old problem of induction.

60

founded on the supposition of that resemblance. Hume stresses that
nature may change such that the future will not resemble the past in

this next passage:

In vain do you pretend to have learned the nature of
bodies from your past experience. Their secret nature
and consequently all their effects and influence, may
change without any change in their sensible qualities.
Thi s happens sometimes, and with regard to some objects.

Why may "013 it happen always and with regard to al ‘
ObJ'ects?1

Induction and the idea of causal

connection Based on custom

Hume concludes finally that the principle on which the inference
from similar sequences in the past to similar sequences in the future is
basecj is custom or habit. After recurrent observations of the constant
conjun ction of two objects, i.e., heat and flame, we are determined
just by custom to expect the one from the appearance of the other.

Hume claims that it is for this reason that we draw an inference from
a thousand instances which we are not able to draw from one instance
that i s in no respect different from the thousand. No man can infer
that every body will move when impelled by another if he has only seen
one body move after being impelled by another. When one has found in

many instances that two objects,2 i.e., heat and flame, have always

bee” conjoined together; if heat is again presented to the senses,

cus tOm carries the mind to expect the second concomitant, flame.
\

1Hume, Enquiry, p. 72.

2It is Hume who calls heat and flame, objects.

61

Belief andéprobabi l i ty

Hume characterizes belief as a more intense and steady conception
than that which attends the mere fictions of the imagination.‘ This
steady manner of conception, called belief, arises from a customary
conjunction of the object with something present to the memory and
senses. Causation is but one of the principles of connection among
three, the other two being Resemblance and Contiguity; and beliefs
arising from the relation of causation are at some times probable,
claims Hume.2 Where the past has been regular and uniform without
exception, we expect and believe that the event will occur with the
greatest assurance; butwhere different effects have been found to
follow from causes which are to appearance exactly similar, we determine
the probability of the event, for all the various effects occur to the
mind and enter into its determinations.3

The presupposition objection to Hume's
induction argument

Hume's chief contribution with respect to the discussion of
causes is perhaps the curiosity he expresses about the inference that
men make from the-statement that such a cause has always been followed
by such an effect in the past to the generalization that all similar
causes will be succeeded by similar effects in the future. If objects

hurled into the air today fall to earth, nature will not change such

M _. ‘_... ___

1Ib'id., p. 80.
2Ib'id., P. 81.
’Ibid., p. 87.

 

 

n A“

Upt 5!,
n‘ I
'LJE .

NS:
um"

an‘

 

 

62

that tomorrow a similar object hurled into the air will fly upward

even higher instead of falling back to earth.' Hume defines this
liniformity of nature principle as 'instances of which we have nad no
experience 'resemble'those of which we have had experience' (Treatise,
lap. 89, l04ff).1 But problematically, while this presupposition of
riature's uniformity supports inductive inferences, the presupposition
'itself according to Hume is derived from experience. Hume wrote in the
‘rreatise (p. 105) that the general uniformity of nature 'has established
“itself by a sufficient-custom| in 'millions of experiments.‘2 As Laird

notes :

For, in accordance with the relational theory of general-

ization he (Hume) had set forth regarding abstract ideas,

he held that what is general is always derived from what

is specific, and is only a way in which the specific is

considered to represent and summarize a multitude like to

itself. General uniformity, therefore, was, like specific

uniformities, only an instance of associative transition,

"inference" or expectation.3

The Humean view described above, however, gives rise to what is

usually called the traditional or old problem of induction which is that
specific inductive arguments about causes and effects cannot be justi-
fied by a belief in the uniformity of nature if such a belief is itself
Justified by specific inductive arguments about causes and effects even
in 'many million' of experiments. The charge posed by the traditional
problem of induction is thus one of circularity. However, this circu-

larito, criticism does not arise unless it is agreed with Hume that the
\

1Ib'id., p. 106.
2I id., p. 107.
3I id., p. 106.

 

63

principle of the uniformity of nature is a conclusion based on
experience. If the uniformity of nature is a conclusion of deductive

reasoning it may well serve as justification for specific inductive

inferences about causes and effects. Applying his conceivability
cri terion Hume ruled out the possibility that the uniformity of nature

is a conclusion of deductive reasoning, for he maintained that a sudden

change in nature's habits was perfectly conceivable.

Oddly enough, however, Hume regards the principle of the
Uniformity of nature to be a presupposition as well as being a belief
eStablished by experience. Flew says in quoting from Hume, Abstract l5:

All probable arguments are built on the supposition that
there is this conformity betwixt the future and the past,

and therefore can never prove it.1

The appearance in the above quotation of "supposition,“ a common

Synonym for "presupposition," suggests that Hume in fact thought belief

in the uniformity of nature a presupposition of causal .inferences. But

it would seem that such a belief or any belief2 must be either presup-

DOsed or derived inductively from experience but not both.3 We counted

that the uniformity of nature principle is presupposed and not derived
Via associations through custom or habit from experience as Hume sug-

gests in the former passage of his view on page 62.

\

lA. Flew, Hume's Philosoplm of Belief, p. 64.

2We mean specifically any belief about matters of fact in a
Humean context. See below.

31h Chapter IV we shall explicate “presupposition" and show how
presuppositions are deductively established though they may very well

be about matters of fact.

64

Hume talks of constant conjunctions of similar causes followed
by similar effects in the past which give rise to the belief that nature
is uniform. But as Laird notes:

Strictly, however (he admitted) since all impressions
were distinct existences, he should have spoken of
impressions of kind x, and should have also said that
in most of our causal inferences we proceeded by analogy
rather than by exact resemblance, dealing with x' fol-
lowed by y' , and with x" followed by y" rather than
sinply with x followed by y. Again, the resemblance

or analogy "admits of many different degrees" Treatise

142, and scientific gentlemen might need much insight,

and more faith, to assert that there was any analogy
at all.

Contrary to the above claim, it might be premised that men are
$0 structured by nature that they have an affinity to see particular
Similarities in the enpirical world. Such an affinity might be con-
sidered a natural instinct of some sort. But clearly if the above
Quotation from Laird is correct, such a natural instinct view is not
one that Hume himself held. In agreement with Hume's view, there are
those who say that there are no exactly similar events and indeed the
non-exact resemblances or analogies with which scientists work "admi t
of many different degrees."2 If it is true, as Hume suggests, that
Sci entists may need more faith than insight to assert that there is any
analogy at all, part of this faith might well be a presupposition that
nature is uniform. We presuppose the uniformity of nature when we take
an event E2 to be like a previous event El due to some sort of resem-

b1 ance that we think we find between them by analogy. The finding of

\_

1Laird, Op. cit., p. 91.
2Ib'id.

65

such an analogy, however, testifies to our faith that things or events
have not changed and that there is some analogy to be found.- Once it

is admitted that events are not exactly similar but only somewhat sim-
ilar, perhaps similar in certain respects, it may be that these respects
themselves are not exactly similar but somewhat similar; and the door is
open wide for as much insight and faith as men can muster for the obser-
vational occasion of finding analogous similarities between events past,
present, and future. Perhaps right from the first, the uniformity of
nature had to be assumed to be true, in order that criteria for deci-
sions of similarity between events could be made, i.e., so that some
constant conjunctions of similar events could even be perceived as
constant conjunctions at all. It is, therefore, our surmise that the
uniformity of nature is an assumption which may be made or thought true
before any selection among 'various sequences of events can be made on
tnie basis of some event similarity. If one does not suppose that there
are any similar events one would not look for similar aspects in non-
exactly similar events in order to determine which events are the same
("same" here means "exactly similar for our purposes").1 The view that
tflie uniformity of nature is presupposed precludes the circularity that
arises when it is held that belief in the uniformity of nature is itself
based on the past experience of uniformi ties. Circularity is avoided
r|<>t because as a presupposed belief the uniformity of nature is arbi-

tr‘arily assumed to be true but because it may be a presupposition which

\‘

1The use of quotation marks for the definiens phrase is in
keeping with the theory of meaning which we pr0pose in Chapter III.

66

is deductively entailed by premises known to be true.1 An example of
such an argument is discussed at length in Chapter II where we treat
Kant's proof that the universality of causality is a necessary pre-
supposition of talk about events. In Chapter IV we explicate in full
our theory of presuppositions and show how they can be considered
statements true necessarily, i.e., deductively proved. Such statements
are thus legitimate justifying premises for other arguments. However,
before moving on to these considerations, let us still consider evidence
which supports our claim that belief in the uniformity of nature is pre-
supposed and not based upon particular inferences of similarity among

events.

The historical certainty status
objection ~

Assuming nature is uniform, there is yet the possibility that,
the predictions from past to future may be unfulfilled. An expected
effect may not come about or some event may occur when its usual ante—
cedent is absent. It is now widely held that the laws of science,
causal and non-causal alike, are only inductively probable rather than
deductively certain. Hume states the reason for this inductive credit-
ability gap when he claims that it is only in vain that anyone pretends
to have learned the nature of bodies from past experience. "Their
secret nature and consequently all their effects and influence may

change without any change in their sensible qualities."2

1In Chapter IV we will propose a theory of presuppositions which
will clarify this claim.

2See p. 60.

67

This passage is particularly strange in the context of modern
science which is theoretically elaborate and has much to say about the
nature of things which is beyond the realm of sensible qualities. Talk
of all beyond sensible qualities being secret from our knowledge seems
to be a description once again of only a very young science. 0n the
other hand, it must be remembered that Hume is an empiricist, indeed a
phenomenalist at times, and thus via his impressions argument all that
is not sensible is for him hidden from us forever even though we may
speculate and theorize continually about the reality that lurks beyond
our sense perceptions. It is therefore the case that we can think we
know the nature of an object but be wrong somewhere in our theorizing
if there is some change among sensible qualities such that those sen-
sible qualities which were predicted by our theory fail to materialize.

Hume's statement that the hidden nature of things may change
while all their sensible qualities appear the same appears paradoxical.
Certainly the changing of nature's physical structure, as Hume describes
it, poses no problem to the scientist or ordinary man, for they have no
way of knowing that this hidden nature has in fact changed since the
change is hidden and what is not hidden according to Hume, the sensible
qualities, have not changed.

There are times of course, when the sensible qualities of things
do change. We observe unexpected meter readings, or colors in chemical
mixtures, etc.; and this is when our laws have, as it were, run amuck.
Does Hume's strange passage (p. 60) refer perhaps to this situation of

the failure of an established law? If in the passage he is describing

this ki
Flute in
scienti
irCng,

:cnfim
right, I
had the!
aiike, (

and car:

 

Cliteiva

 

i .
aitlcns

 

 

 

68

this kind of scientific setback, we notice behavior counter to what

Hume indicates. In the famous cases of the failure of particular laws,
scientists have said not that nature had changed but that the law was
wrong. Their previous calculations and data which suggested and/or
confirmed the law were in error. None of them say that the law was
right, but that it failed to predict accurately because in fact nature
had changed! Hume stresses that all men, scientist and ordinary man
alike, can conceive of the situation in which nature actually changes
and certain effects that are expected fail to occur. But if this is so
conceivable, why do none of us ever suppose that all our laws and calcu-
lations are or were correct when confirmed and that nature has violated
them by changing physically. Why did Kepler say that Tycho Brahe had
had inadequate instruments with which to observe the crucial parallax
needed to indicate that Copernicus had been right--the earth revolved
around the sun. Kepler must have had able powers of imagining and
conceiving and could have said just as well that Brahe was correct

in concluding that the earth was the center of the universe; but that
sometime after Brahe's true conclusion, the earth actually changed
places with the sun; and when he, Kepler, investigated nature later,

he found that it had changed. This brings us to a decisive point, which
is, that the uniformity of nature cannot be a belief based on experience;
fer scientists, even eminent ones, tenaciously hold on to their belief
that nature does not change even when experience indicates either that
nature changed or that our laws are wrong. This point is so important

that it bears repeating. Every time a law of science is repudiated as

69

false or inaccurate, scientists have refrained from interpreting as

a physical change an anomalous experience which is indicating just as
strongly that all carefully formulated laws are correct and that nature
every once in a while physically changes, showing herself quite clearly
not to be uniform! We are thus led back to our previous claim that the
uniformity of nature is presupposed and not based on conjunctions of
similarities in experience. Scientists always assume nature to be
uniform whenever they talk of similar or same events. In the face of
falsifying evidence they do not regard this belief as falsified but

rather a belief in some particular law.

The certainty andgproof objection
Consider the fbllowing passages from Hume's writings:

There are some causes, which are entirely uniform and
constant in producing a particular effect; and no instance
has ever yet been found of any failure or irregularity in
their operation. Fire has always burned, and water suffo-
cated every human creature; the production of motion by
impulse and gravity is an universal law, which has hither-
to admitted of no exception. But there are other causes,
which have been found more irregular and uncertain; nor
has rhubarb always proved a purge, or opium a soporific

to everyone, who has taken these medicines. It is true,
when any cause fails of producing its usual effect,
philosophers ascribe not this to any irregularity in
nature; but suppose that some secret causes, in the
particular structure of parts have prevented the
operation. . . .1

Note that in the last six lines, Hume describes a situation in
which philosophers have given up neither their belief in the uniformity

of nature nor their belief in particular laws when expected observations

 

1Hume, Enquiry, p. 86.

70

fail to occur. Rather suppositions of initial conditions are made so
that neither of the other beliefs will have to be abandoned. He goes

on:

Our reasonings however, and conclusions concerning the
event are the same as if this principle had no place.
Being determined by custom to transfer the past to the
future, in all our inferences; where the past has been
entirely regular and uniform, we expect the event with
the greatest assurance, and leave no room for any con-
trary supposition. But where different effects have '
been found to follow from causes, which are to appear-
ance exactly similar, all these various effects must
occur to the mind in transferring the past to the
future, and enter into our consideration, when we
determined the probability of the event . . . we must
assign to each of them a particular weight and author-
ity, in proportion as we have found it to be more or
less frequent.1

This is a most interesting passage; for it is clear that in this
quote Hume does not realize just how problematic induction is as a
method of inference. His words indicate that he does not believe that
all inductive conclusions are at best only probable. He says that con-
junctions that have exceptions result in laws that are only probable
whereas those that are constant without exception result in laws that
are certain. It is apparent that Hume as many men of his time, thought
that some laws are certainly true. Hume, thus, could think of their
failure to hold only in terms of an actual change in nature. Hume
believed, at least in this passage above, that the belief in the uni-
formity of nature was only probable; but that laws based on constant

conjunction without exception were believed to be certain. This

 

1Ibid., p. 87.

 

 

O':‘I'O
' nb~v
uulpln I
| v

‘A'. ‘ -
Uri-n K

."'.~
bl-

 

.'

 

 

71

exegesis of Hume's words makes an interpretation that he posed the
traditional problem of induction a very confusing interpretation indeed
since the old problem of induction does not arise unless one claims that
both kinds of belief (in the uniformity of nature, and laws) are only
probable. The probable but not certain aspect of both these types of
belief is what gives rise to the circularity when one tries to base one
belief with certitude on the other. On the other hand, Hume did not
believe that all laws were certain; for some clearly had exceptions.
Yet his words in this regard (see above) indicate that he is talking
about probability more in keeping with the frequency theory of proba-
bility than with any inductive probability.

Laird notes thataccording to Hume the causal inference, even
after experience of constant conjunction, could not be an affair of
knowledge. It belonged to probability.1 But Hume was also to use the
term "probable" in confusing ways. Laird states:

Hitherto, he said, he had, following Locke, used the
term "probability" to designate any assent other than
"knowledge" (p. 129). Common sense, however, would
distinguish between what is only probable (i.e.
conjectural and uncertain) from causal expectations
(which were utterly convincing). Hume therefore
proposed a threefold division into (rational) "knowledge,"
(causal or experimental) "proofs" and (conjectural prob-
abiiities). He had also to note that there was a mathe-
matical calculus of the probabilities both of "causes"
and of "chances."2

Further Laird maintains that Hume g1g_distinguish a logical

from an illogical kind of inference, from experience or between

 

'Laird, op. cit., p. 87.
2Ibid., p. 90. '

Trev

The

'.dlced

forser

find of dis

fight be c:

:a'nct 13::

in the azcv

7“ tie fcll

AS we h
WES the
Of a be
In Cite
to be n
This tr
taleed,
was ass
Nev

I
reascn
If not
Sieaxin
CUEWCQ
tier: of
0?‘ "red
Strati‘
The
59971) L
Certair
tion) a
“'9” tr

Only ii

Th‘

 

 

72

prejudiced conviction to conviction based upon repeated experience.

The former was unphilosophical or unreasonable "probability." This
kind of distinction has a modern flavor and reminds us of work that
might be done within the realm of inductive methodology. However, we
cannot ignore the three-fold division and Hume's terminology described
in the above passage. Laird describes the confusion in Hume quite well
in the following words:

As we have seen the causal problem according to Hume,

was the problem of the extension beyond actual experience
of a belief, which simulated but was not actual experience.
In other words, there was a transition of the mind, felt
to be necessary, but going beyond what had been observed.
This transition Hume also called "inference" and he main-
tained, in the end, that the "transition" or "inference"
was associative and not an affair of "reason."

Nevertheless he was quite prepared to speak of causal
"reasonings," and even of causal "proofs." Such language,
if not regarded either as a concession to common ways of
speaking or as a plain inconsistency involved the conse-
quence that there was a species of causal reasoning and
even of causal logic, provided that such “proof," "logic"
or "reasoning" was neither the intuitive nor the demon-
strative "reason" of knowledge or scientia.

The word "proof" in this connexion was (as we have
seen) used to indicate that the causal inference yielded
certainty (in the sense of complete assurance or convic-
tion) and so had to be distinguished from "probability"
when the latter term was understood to mean what was
only likely and was not completely assured.1

This last sentence above, together with remarks made by Ralph
Church on the subject, suggest that our initial supposition that Hume
regarded probability to be of a statistical type is correct. Church

says:

 

1Laird, op. cit., Pp. 104-105.

Tne analyS'I
15 so far :
ity of W
the matter I
aga‘n to be
not a W5;
in the pro:
ideas invo?
returreFCe
irrl‘ESS‘IOFS

This c:1
is: contradict
afied the old

It seer.
iesvibing the
learning it.

ri- e
se .0 passage

 

‘1. Or it is p
325:“) hold of ‘

:i: a stroncer

actsail

y DI‘ESEF

luau
..c 'de

d as ccr
ar‘w

.' (HEP, w'nc
:-£3=‘I

~51! failir

5;ng ”1 .
vibe Wit?

' £9 Hume had
":9 1e assuur
--i

Y .

‘4
1» '
0-“ VIEW!

\

1 .
Cnurc?

73

The analysis given of belief in the probability of causes

is so far the same as that already given of the probabil-
ity of chances that Hume feels under no obligation to take
the matter any deeper.~ In this connection, however, it is
again to be noticed that the probability in question is

not a property of any or all of the contents of a belief

in the probability of causes. In all such cases the various
ideas involved will be images not of impressions whose
recurrence is uniform and constant, but rather of contrary
impressions.1

This confusing use of "probability" by Hume seems to obscure if
not contradict the insight he has when discussing what is typically
called the old problem of induction.

It seems more so then that Hume was on the verge of actually
describing the traditional problem of induction rather than actually
describing it. Perhaps it was an insight that was unsteady, giving
rise to passages that illustrated it and other passages that obscured
it. Or it is possible that his successors, who were able to seize
steady hold of this insight largely through the suggestive work of Hume,
read a stronger discussion of the problem into Hume's writings than he
actually presented. In the early stages of science, certain laws were
regarded as correctly formulated and discovered for all time. It was
only later, when many major laws, gravitation, etc. were thought to be
possibly failing in various ways, that people began to think that laws,
even those without any yet known exceptions, were at best only probable.
Since Hume had already proposed that the uniformity of nature was a
probable assumption based on experience alone, philosophers began to

hold both views concomitantly, i.e., that the uniformity of nature, as

 

1Church, op. cit., p. 199.

.271 a5 13's we
.as Give” MW
'32: as 55'9”“.
rustle and we
a'eys held tia'
'rcsctian might
"C',:tive laws '
'atre is unifOI
minus expel"

Thus the
2‘ 1a state’en'.
3.5.51 reasoning
tienargunent, i
:2jecti3n to Hui:
‘3‘th status ob
iesszposed by s
'at‘csaie for If

a.sai
states-er:

 

“It iiiself did

u'.‘
‘
3{fin

111159 to C

 

74

well as laws were only probable truths; and the problem of induction
was given explicit philosophical attention. Had philosophers noticed
that as scientists began to suspect all their laws as only inductively
probable and were ready to give them up on occasion, they nevertheless
always held that nature was uniform no matter what; the old problem of
induction might never have arisen. The inference made in particular
inductive laws could be anchored to the certain presupposition that
nature is uniform which is true for the scientist in spite of the
anomalous experience of some unexpected non-similarity.

Thus the second empiricist objection to any necessity type view
of law statements that in a priori reasoning we attain certainty but in
causal reasoning only probability fails since its support, Hume's induc-
tion argument, is subject to serious objections. Our presupposition
objection to Hume's induction argument as well as our historical cer-
tainty status objection indicate that the uniformity of nature is
presupposed by scientists and as such it may provide a deductive
rationale for inductive practice. Hence it is possible that inductive
causal statements are well grounded and perhaps certain or necessary.
Hume himself did not always write as if inductive practice was circular
giving rise to only probable laws. This is clearly reflected in his
usage of the words "certain," "probable" and "proof." On occasion Hume
explicitly regards causal conjunctions without exceptions as certain.

It is also the case that if the uniformity principle is a pre-
supposed truth about the empirical world it serves as an example of at

least one particular empirical statement which is certain. If laws can

STE'OW be cer

"Hater. secc

31's i'tress

Trev-r
bled:-

‘er :reir cau
ecs his discu
xviion that
;::.z-t H F‘
' :r - 9 ed
'itn‘lng but cc;
1C :h‘flk Of an
“Sealer exi
"ecessary COrne
‘C'CE-‘trates hi
"*1 there is a
.. .4 ,

‘ “9 traced

: sh _
hire;

‘ ~-Ses OVEr

WF'ESents t
1' we (10 r
‘1”de I
OtAEr,

 

y
I.;ere E:

  

75

somehow be certain some may be necessary or a priori despite the

ill-taken second empiricist criticism of necessity law theories.

Hume's impressions argument

 

Though Hume has already stated that effects are totally distinct
from their causes and cannot be discovered through them a priori, he
ends his discussion in the Enquiry with a further argument against the
position that there is some necessary connexion between a cause and its
effect. He bases his claim only on the assumption that all ideas are
nothing but copies of our impressions and that it is impossible for us
to think of anything which we have not antecedently felt either by our
internal or external senses.1 Since there is no external impression of
necessary connexion on which to base the idea of necessary connexion, he
concentrates his argumentation against the popular Opinion of his time
that there is an internal impression of necessary connection. Many of
Hume's contemporaries thought that the idea of necessary connection
could be traced to an internal impression which arises from reflecting
on the operations of one's own mind and on the command which the will
exercises over the organs of the body and the faculties of the soul.
Hume presents three arguments against this thesis which are briefly:

1. We do not know the connection between body and soul
and we do not know the power of the one over the

other.

2. There are organs of the body which are not moved by
any action of our will.

 

1Hume, Enquiry, p. 89.

 

3. The
not '
ard i
CSEFE
lig‘l':

For v
session of
2‘ air mind 0
3:1} by '78th
EZE'da't, arc
‘5 My what 1!
:‘stai definit
eject fcllowe

a h"

32,». to tha'

 

76

3. The immediate object of power in voluntary motion is

not the member itself of the body but certain nerves

and animal spirits and it is impossible that this

operation which is still so mysterious and unintel-

ligible should be known by any inward sentiment.1

For very similar reasons, we cannot claim that any internal
impression of necessary connections arises due to the voluntary action
of our mind over itself, according to Hume. Thus, "the mind is carried
only by habit upon the appearance of one event, to expect its usual
attendant, and to believe that it will exist."2 Necessary connection
is only what we associate together in our mind. Therefore, an addi-
tional definition of cause, other than the one already cited, is "an
object followed by another and whose appearance always conveys the
thought to that other."3
Our arbitrary assumption objection to
Hume's impressions argument

Hume based his definitions of cause on an interesting, episte-
mological theory. 'The cardinal principle of his pure empiricism was
that knowledge consists of ideas and there is no simple idea which is
not based on some previous sense impression.“ But this principle was
ad0pted by Hume as doctrine without proof. As Kemp Smith writes:

Hume, who was sceptical--so it was about almost every-

thing else, has yet been so uncritical as to erect the
elaborate body of argument which constitutes the Treatise

 

'Ibid., pp. 91-92.

 

N
H
U"
do
0.

u

p..99.

 

(D
0—.
U’
J.
D.

0

p. 100.
p. 54.

 

.8
0—0
9'
do
0.
I

on a fad:
exarine,
hirself i
I: the safe i
hctcricus
ard ccnfu
he need e
Ewing
“[51:15 SEC:
"IE P'ovides
Ziiiienges at.
W“ from a cc
the idea Of (16
‘1." have; but H
mm“! des
Tcassert,
0n the QTCI
”Ct. €01“.me
5'49“] tha

1.0 the Win

 

 

77

on a foundation which he has not been concerned to

examine, and to the unreliability of which he has

himself though allunconsciously, been chief witness!1
In the same vein Passmore says:

Notoriously, his exposition of that theory is careless

and confused; . . . because he felt that on this point

he need expect no serious criticism.2

Ewing points out that Hume's principle that all ideas are copies

of simple sense impressions was nothing but an arbitrary maxim for which
Hume provides no proof. In attempted support for his assumption, Hume
challenges anyone to produce a simple idea which is not based or did not
come from a corresponding sense impression. The obvious candidate was
the idea of necessary connection, which many people declared themselves
to have; but Hume ruled this counter-instance out in a way that Ewing
succinctly describes:

To assert, as Hume does, the validity of the principle

on the ground of failure to produce an idea that does

not conform to it, and then, later in the argument when

such an idea namely the idea of causality, is introduced,

to deny that it is a real idea, because it does not conform

to the principle is a breach of the most elementary rules

of logic.3

It is now acknowledged, even by many Humeans, that the episte-

mological apparatus which led Hume to disavow the idea of necessary
connection as based on anything but subjective feeling is a shaky

foundation, indeed, on which to build one's theory.

 

1Cited by J. A. Passmore,_flume's Intentions, p. 89.
2Ib'id.

3A. C. Ewing, Kant's Treatment of Causality, p. 7.

 

 

‘
II" ’

Ducass
:fneCESSary C
géfng hi5 d6:
if the first 0
of "if-HESS" 9
155 was ”0t d‘
:eaeen causes
rate; but tha‘
‘:.-nd in sentir
22' an internal

It stri
ems (see p;
car as derived
tenchsion is s
‘3.“ the fact t

:an have n:

 

'I’ e'g-a the E
Q. C.

3":“rd I
" and thlr
I. h '
I asainst tr

'in.
~ 3 certain

.Va: inveren t

 

TEC‘fM

“Ely knc

\

IHUFE’

 

78

Ducasse notes that Hume himself speaks as though he has an idea
of necessary connection when he gives his definition of “cause." After
giving his definition (see p. 29) Hume adds, "Or in other words where
if the first object had not been the second never had existed."1 Talk
of "if-ness" gives hint of an idea of necessary connection. However,
Hume was not denying that we do have an idea of a necessary connection
between causes and effects, which he himself could not totally oblit-
erate; but that this idea is illegitimate since its origin is to be
found in sentiment or feeling which is neither an external impression
nor an internal impression of an admissible sort in his empiricist
system.

It strikes one immediately, after reading Hume's three major
reasons (see pp. 75-76) for concluding that no idea of causal necessity
can be derived from reflection on one's own voluntary actions, that his
conclusion is somewhat of a nonasequitur. Surely, it does not follow
from the fact that the mind-body interaction is in detail unknown, that
one can have no awareness of the action taking place, which would result
in, e.g., the effect of a voluntary arm lift. In The Mind and Its Place
In Nature, C. D. Broad pr0poses a variant of this objection to Hume's
second and third points. He claims that Hume's objections are valid
only against the contention that we know immediately that a volition to
make a certain movement is the sufficient condition for the happening of
that movement. Hume's arguments are irrelevant to the belief that one

immediately knows that the volition is a necessary condition for the

 

1Hume, Enguiry,-p. lOO.

1:;enlfig Cf

reeves and it;
:n5 not HVCV;
'eSITt. that ‘
:t'ect‘jen hi1
DST-‘39" 5633’ E
:e:.een cause

urctis nat ;
5:.e'son nas a
that ire p
izsft'on in v

”at.

Itis sir;‘
‘Is no close
5'C the no.
desire and

It wou‘

::r':n6" h
"VUn bE‘r

 

 

79

happening of just that movement at that particular time. That our
nerves and muscles must also be in the right state indicates only that
there are other necessary conditions for a particular event. We are
thus not wrong in thinking that we know, without having to wait for the
result, that the volition is a necessary condition of the movement. In
connection with Hume's first point that we do not know the connection
between body and mind, Broad maintains that there is a connection
between cause and effect in cases of volition and voluntary movement,
which is not present in other cases of causation. When it is said that
a person has a volition to move his arm, there is involved in this say-
ing that the person also has an idea of his arm and an idea of the
position in which he wants his arm to be.1 Broad concludes from this
that:

It is simply silly in view of this fact to say that there

is no closer connexion between the desire to move my arm

and the movement of my arm than there is between this

desire and the movement of my leg or my liver.2

It would seem from Broad's description that the necessary

connection between cause and effect in voluntary actions is somehow
in the mind, perhaps an internal impression. One thinks of moving his
arm and at the same time, he thinks of the position in which his arm is
to be. It is difficult, however, to describe in what way this thinking

constitutes evidence of a necessary connection between the two events;

but even if it could be outlined more precisely than it is by Broad,

1C. D. Broad, The Mind and Its Place in Nature, p. 102.

 

2Ibid., p. 102.

there rerain
be given cor?
iris explana
mg inanim-
me: I rrcve r
:1 vclition a
maid it eve
:‘Iiiard ball
1271 would al
‘4‘? W‘ee‘ne of
3' aware-fiess .

This l
’23er no uti‘
{'19 €an 0f ”6
if: effeCt r9]
Sijer the Vlee-n

{NWT

F—u.em of ine

'0 iw.

      

 

’80

there remains another problem. -Should the idea of necessary connection
be given complete and adequate explanation in cases of voluntary actions,
this explanation cannot be extended by analogy to cases of causality
among inanimate objects. All that might be said by analogy is that if,
when I move my arm, I become aware of some necessary connection between
my volition and the following position of my arm, then a billiard ball,
should it ever wiggle its number or perform whatever voluntary action a
billiard ball could be capable of in a vivid imagination, the billiard
ball would also notice a necessary connection between its volition and
the wiggle of its number, that is, if the billiard ball had a will and
an awareness of what it willed.

This whole line of thought is preposterous and bodes of abso-
lutely no utility in solving the problem of explaining the origin of
the idea of necessary connection which accompanies the idea of the cause
and effect relations of inanimate objects. Let us then go on to con-
sider the views of those who would take a more direct approach to the
problem of inanimate cause.

The idea of necessity from external
impressions reexamined ‘—

In his book entitled Symbolism, Alfred North Whitehead contends
that both Hume and Kant regarded causal efficacy "to be the importation,
into the data, of a way of thinking or judging about those data." But
it is his own view that causal efficacy rather than sense data is the
given in primitive experience. It is not the case that the world as it
is given in sense presentation is "the aboriginal experience of the

lower organisms, later to be sophisticated by the inference to causal

efficacy.
nicn is dcr
iata is only
:ecexions .-
‘t'essed by

ire-:iate ser

a‘ - .
a. t'e f'utUrE

 

 

81

III

efficacy. 0n the contrary, it is first the causal side of experience
which is dominating; and a phenomenalistic awareness of transient sense
data is only the result of abstraction from the given. Our present
perceptions are determined by the past in ordinary awareness when we are
impressed by causal efficacy. 0n the other hand, vivid enjoyment of
immediate sense datanotoriously inhibits apprehension of the relevance
of the future.2 The former type of experience is, in Nhitehead's words,
vague, haunting, and unmanageable, while the latter is handy and easy to
produce at will. This latter is the superficial product of complexity
which halts at the present and is concerned only with the show of things.
But the former is to again use a ready word of Whitehead, heavy with the
contact of the things gone by which lay their grip on our immediate
selves.3 Present events are subject to the limitations laid upon them
by the actual nature of the immediate past.

If Nhitehead's account sounds in some way mystical or perhaps
poetical, well it might, for he also has direct recourse to the poetic,
"Pereunt et imputantur"--"The hours perish and are laid to account.M
The "pereunt" refers to the world of passing presentations while the
"imputantur" refers to the world of presentation as it is disclosed in

its causal efficacy.

 

1Alfred North Whitehead, Symbolism, p. 49.
2Ibid., p. 42.
3Ibid., p. 44.

 

“An inscription on old sundials in 'religious' houses.

 

Intere
yffieuit to S
e‘ectreiatlc
.enehzatie.
mes craracte
‘sin terms of
eaeagreed th
:.tthat araly
egreicgical
we?! to say
"ficetes nct

3.537 i‘ n e
‘a‘ :.‘:Ofl ECL

U

\"'K

In his

“J Cid"

inary
50P€thl

fling

3'1 . .
-Ias brlnqi

" ””9 5 dOQr
SEEMr.
~bt case

"“695er to a

;;‘.nf ‘
aCL one

I
2:: -
N'

C" W

S ”Order

.55
‘ 5“ He‘d».

\E
Max B
Isidfie‘y _.

(I)

 

{I
i“

82

Interesting as this portrayal of causality may seem it is
difficult to settle the controversy that rages around the cause and
effect relation by maintaining that causal efficacy is primitive,
unanalyzable, and directly known. This is especially the cause when
one's characterization of the external impression of causal efficacy
is in terms of words such as "heavy" and "haunting." Hume might well
have agreed that we are aware of causal efficacy when we view the world,
but that analysis reveals this awareness to be based merely on the
psychological associations of the mind. It does not prove Hume wrong
merely to say he is. If Hume does not concede Whitehead's point it
indicates not that Hume is wrong but perhaps only less poetical.

Causal connection externally given--
Max Black

In his article "Making Something Happen" Max Black describes
many ordinary actions of human subjects as perfectly clear cases of
making something happen. These perfectly clear cases include actions
such as bringing the glass nearer, closing a window, opening a drawer,
turning a doorknob, or sharpening a pencil.1 These events are such
clear-cut cases of making something happen that Black contends it would
be absurd to ask for any evidence that the man made the window close
if in fact one actually saw him do it. Black seizes upon the above
cases in order to establish preanalytically what constitutes a paradigm

case of making something happen. By focusing on paradigm cases of

 

Max Black, "Making Something Happen,“ Determinism and Freedom,
ed. by Sydney Hook, p. l5.

 

 

:amg meta
:iiiasc;rers

since the net
:3! ;ass fI‘CE'.
TIE scphisti.
:"tg teapen '
$1550 what we
‘frst ate: we

:icated vocab.

9'1}

Thtflcause
iTIOthn of
CO’tEPDCM
55’: tile 6.;
In Ex;'
zzrie‘qunce of
ct ”me-"taken

re: .
,er 'Cbid ha

He adds

131%
“9 ”O‘ied

Tits, '
.1 7t, 50mg

“

.9 .~
lir‘dow H0“.

 

an
we: .
act‘IOn ;

 

 

 

83

making something happen, Black hopes to counter the claims of those
philosophers who state that it is impossible fOr anything to be a cause
since the notion of cause is self-contradictory. Black claims that we
can pass from the homespun language of making something happen to the
more sophisticated language of "cause" and "effect," for to make some-
thing happen is to cause something to happen. But in order to under-
stand what we mean by "cause" and "effect," we must labor to understand
first what we mean by the precausal language in which the more sophis-
ticated vocabulary is embedded. Thus Black isolates the following
criteria of application of "cause" in the paradigm case:

The cause was a free act of*a person, the effect was a

motion of an inanimate object, the cause and effect were

contemporal (operative through the same time interval)

and‘the Effect was a necessary consequence.1

In explaining later how it is that the effect is a necessary
consequence of the cause, Black states that the cause is conditionally
necessary for the occurrence of the effect such that if a person P had
not undertaken his action A, of closing a window for example, the window
never would have closed.

He adds of course that it is "untrue to say that a glass would
not have moved as it did unless P had made it do so, for if P had not
moved it, some other person might have done so. What we mean is that
the window would not have moved by itself; that is if P had not per-

formed action A, or some other action resulting in the glasses moving,

the glass would have remained stationary. In other words, though all

 

1I id., p. 21.

 

czrer featon
the giass wi
s‘t.ation fm
.1 'a;;eneo’, (
an: the same

rizh Black's

09-.
C

. .his exar
IrC’Ete, par
ing this part

In this c
idradlgm i
occasion ‘
inducing
been delil
love, we I
no: bef'
P did do

thus mean
Iaintaini
d CCtemo
that the

Charfitter
mdltim
talk abOU
musua],

E159 0f Ca'
”heme

It is

M it Very

 

84

other features of the setting remained unchanged, if A does not occur,
the glass will not move. To explicate the necessary aspect of the
situation further, Black states that "we say that M had to happen when

A happened, only if M would always ensue, given an unchanged setting

 

and the same concomitant."1 This last phrase is curious in comparison
with Black's major contention. Elsewhere in the article he contends
that his example of the use of “cause" is one of its applications to
concrete, particular cases much as the analysis of Ducasse. In stress-
ing this particularity, Black says:

In this cursory examination of some features of a
paradigm of making something happen, I have had little
occasion to refer to any "constant conjunction" between
producing action and induced motion. The omission has
been deliberate . . . in order to be sure that P made 0
move, we need only look. The verifying situation is
right before our eyes. To establish conclusively that
P did do such and such and that 0 was moving thus-and-
thus meanwhile . . . I do not say we should be right in
maintaining that A made M happen whenever an action and
a cotemporal motion ‘are contiguous. In using the
language of "making something happen" we take for granted
that the episode in view has a special and appropriate
character. Should we be challenged to specify these
conditions in full detail, we should eventually have to
talk about constant conjunctions; and in deciding in
unusual, unfamiliar or abnormal settings whether the
use of causal language is appropriate, prolonged
inductive investigations might be needed.2

It is interesting to note that, in the above passage, Black
makes it very clear that his paradigm cases of cause or making something

happen are individual cases of particular causes working in particular

 

W

1Ibi ., p. 25.
2Ibi .

a.

Q

 

 

s‘taations t
ieson in hi

an: drop:6d '
:‘a high ri:
:aracfge exa:
ionews the <
hot but in

that he nairi
:a.se happens
g‘ien an unch
2'5: in the p
E“ect there

WSW struck

:efig so strut
tee picttre w
for: of the
’eot.‘ Thus
9150*. that t'

Wham” o

L“?
Lui- descri be,

“its.

 

Wow \'
-- '“S‘ar'
lag

85

situations that may never again be repeated, e.g., there was only one
person in history who was ever murdered by someone, a causal agent,

who dropped a dumbbell on the victim's head from the fortieth floor

of a high rise building. However, in isolating the features of his
paradigm example, Black includes the point that the effect necessarily
follows the cause and clarifies the necessary relation not in particular
terms but in terms of a repetition of instances. It will be recalled
that he maintains that to say an effect has to happen when a certain
cause happens is only appropriate if the effect would always ensue
given an unchanged setting and the same concomitant. It appears then
that in the paradigm case of'a particular cause engendering a particular
effect there is not an element of necessity such that one would say the
person struck by a barbell falling from the fortieth floor must die.

We may be able to determine what killed the man should he die after
being so struck, but it does not appear that any necessity talk enters
the picture with regard to cause-effect relations until general condi-
tions of the particular occurrence are spelled out by repeated experi-
ment.1 Thus Black's paradigm perhaps bears all the criteria mentioned
except that the effect necessarily follows the cause. An adequate
explanation of the necessity which is involved in causal language is
best described in terms other than in this basically Humean view of

Black's.

 

7'

1By Black's own admission (see quotation on p. 84), we cannot
say that P made 0 happen unless we have seen repeated conjunctions of
0 following an action of P. Though Hume might say "do not" instead of
"cannot" Black is adapting a Humean type view.

The f
in the “it“
necessary COR
:firit‘lve da‘.
oreitnin our
sizh szecolat
"‘ tree are
ir nature. W.
Tau: staterent:
in scientific
393% the cons
moi for the

liner II ,

We be?

 

 

 

86

The fOregoing objection to Hume‘s impressions argument expressed
in the writing of Broad, Whitehead, and Black has been that somehow a
necessary connection between events is perceived or sensed. Some
primitive datum of a necessary connection lurks in the empirical world
or within our own bodies which can be observed by us. Interesting as
such speculation may be, it is not our concern in this writing to decide
if there are observable or even unobservable connections between events
in nature. We are concerned rather in providing a necessity theory of
law statements which will clarify the use of the terms “nomic necessity"
in scientific discourse with respect to these statements. We shall now
begin the construction of our necessity theory as we discuss Kant's
proof for the necessary truth of the general laws of causality in
Chapter II.

We begin the exposition of our necessity theory of law state-
ments with a treatment of Kant's argument for several reasons. We shall
find in Chapter II that Kant provides a deductive proof, in the "Second
Analogy," for the truth of the general law of causality (every event has
a cause). Not only is GLC (the general law of causality) true but it is
regarded by Kant as an a priori synthetic truth and a necessary presup-
position of experience. However, this assessment of GLC poses a problem.
It is a long-standing belief among empiricist philosophers that a priori
truths are necessary, self-evident truths which are independent of
observation and experience. It is also a long held belief that syn-
thetic truths are contingent and in some sense (an ambiguous one as

we shall see in Chapters III and IV) factual.

 

We h
of-caases an
"fan, can be
:zje:t, its
.‘amai leOw
direct o:ser'
itcvei otjec
55' be kfiown
EiP'ESSive 0*.
2'5} can déd;
:EWticai s

We wi
.3"Gricityll a
”5 prim-kits,
fixer-:56 woeth
35?:th te

:75 C539 the:

 

tfijel we SU:

.5; 0f Causa‘l

‘3':

amen

W that e.

 

87

We have already argued-in this present chapter that knowledge
of causes and effects, particularly on the every day level of observa-
tion, can be known independently of the experience of a particular novel
objec1:, its causes and its effects. Thus it is possible to gain or have
factual knowledge in a particular situation without any specific and
~ direc1: observation of what comes to be known, the particular effect of
a novel object. There is then a sense in which some factual information
can be known a priori. We think that this realization is particularly
expressive of modern scientific practice in which scientists say that
they can deduce true statements of fact, even factual laws, from
theoretical statements far removed from direct observational evidence.

We wish then to examine in succeeding chapters whether "a
Prioricrity" and "empirical facticity" are contradictory terms. Since
"a Pritaricity" is used interchangeably with "necessity," we shall also
examine whether "necessity" and "empiri cal facticity" may not be con-
tradictory terms. If these two terms are not contradictory it may be
the case that some factual claims are known not with probability but
with certainty as when they are presupposed by premises known to be
true. lde suggest that this is true of a claim such as GLC (the general
13‘" 01“ causality) and so it seems appropriate to begin now by discussing
a.necessary presupposition deductively proved by Kant which is consid-
ered a priori, synthetic and which makes a claim about the empirical

world that every event has a cause.

It we
sakered Kan:
tar: himseif
2.551 tonne:
year.1

Kant

x'rection an

 

CHAPTER II

THE KANTIAN VIEW OF CAUSALITY

Historical note.

 

It was Hume's view of the character of the causal nexus which
awakened Kant from his "dogmatic slumber" sometime around 1772, but
Kant himself subscribed to part of the Humean view, viz., that the
causal connection was not a logical relation of ideas, prior to that
year.1

Kant came to realize the synthetic character of the causal
connection around l763 when he wrote his treatise on Negative_9uantity.
Kant drew the distinction between logical and real opposition by noting
that logical analysis reveals that the predicate "eternity," when
applied to God, must exclude that of mortality, but that only experience
can reveal that the movement of one body should prevent the movement of
another. In real opposition, the conflicting tendencies both remain
present in the object and only cancel each other's effects while in
logical opposition the predicates cancel each other. Logical opposition

is impossible in real existents, for nothing that exists can be self-

 

1Lewis W. Beck, ”Editor's Introduction," Prolegomena To Any
Future Metaphysics, Immanuel Kant, p. XI.

88

 

" .ofll

 

ts-ntradictcr‘
ard real gro

I see ve
ground a
it is di:
containei
groord t!
corse:ae!
following
can follt
the actua
t0 P-y trt
its natJi
CQFSEQU'er
in mien
the grcur

In 0r
2'3)” restar-
affix” Dost
iffg WhiCh
”'iit win
Qifs t'eatr
1" In Ka

dOes
princ

 

 

2' H9 a:
to tr

It in

U

(.

(”ant ,
‘1’; f1

“d5 ca

 

89

contradictory.‘ Similarly, Kant made a distinction between logical
and real ground, fbr he says:

I see very well how a consequent is deduced from its
ground according to the law of identity namely because
it is discovered through analysis of concepts to be
contained in the ground. . . . I call the first kind of
ground the logical ground, because its relation to the
consequent is logical, i.e. can be clearly realised as
following from the law of identity . . . how something
can follow from something else otherwise . . . I call
the actual ground, because while this relation belongs
to my true concepts there is no judgment possible about
its nature. . . . A relation of the logical ground and
consequent always constitutes an affirmative judgment
in which the predicate is the consequent and the subject
the ground. In the actual relation of cause and effect
this is never the case.2

In one of the "Reflexionen" published prior to l762, Kant
further remarks that we can only know the connection of cause and
effect a posteriori, not a priori. There are two important historical
points which need mentioning at this juncture since they may shed light
on what will soon be seen by the reader as possible ambiguities in
Kant's treatment of causality. These points are:

l. In Kant's Inaugural Dissertation of 1770 he still
does not think it necessary to prove the general

principle of causality, and. . . .

2. He applies this general principle without question
to things - in - themselves.3

It is not until the writing of the Critique of Pure Reason that

Kant finds causality to be valid only for experience in time, for

 

1A. C. Ewing, Kant's Treatment, p. 32.
2Ibid.. pp. 33-34. (Ewing quotes from Kant.)
31b1d., p. 36.

 

52.:cinctiy s:
In the [3:
are discc
ence but
depends c
in all se

“fit. the wri

‘iea that th

at: that per(

The c

m may

IsaIization i

“3’ 6feiricai

IeI‘ ,
‘ as meta,

‘E‘EISE- It

W the Synt

55555”! in‘.

(he
' ODDDSes

WWW in

fl": In accc

“' ‘35 this t.

 

90

without it, such experience would be impossible. As A. C. Ewing,
succinctly summarizes in his book Kant's Treatment of Causality:

In the Dissertation, the a riori concepts like cause

are discoveredby abstrating'from all sensible experi-

SESSnBEtOSCEfiE“lflitt‘éhi'ietrJ—C'éitlniirii‘i‘2.3 253331“

in all sensible experience.1
From the writing of the Dissertation onward, Kant was to develop the
idea that the categories presuppose perception for their application
and that perception of objects presupposes the categories.

The change in Kant's view between the Dissertation and the
Critigue may well have been stirred by the work of Hume. With the
realization that the causal law could be justified neither logically
nor empirically, Kant knew that the foundation of natural science as
well as metaphysics was in jeOpardy; and Kant set himself to their
defense. It is in pursuit of this defense that Kant departs from Hume.
From the synthetic nature of causal laws, Hume concluded that the only
necessity involved in the concept of causality was psychological. But
Kant opposes Hume at 8793-794 in the Critigue when he says, ”Hume was
therefore in error in inferring from the contingency of our determina-
tions in accordance with the law, the contingency of the law itself."2
It is thus that Kant seeks to reply to Hume by demonstrating the nec-

essary truth of the general law of causality itself which particular

empirical laws presuppose. Graham Bird in his book Kant's Theory of

 

‘Ibid., p. 36.

2Graham Bird, Kant's Theory of Knowledge, p. 165. (Bird notes
this reference.)

 

(scaledoe cit
__.__L—
in the (Tract

It is the
any cause
is rather
of the c:
as introc
or as ex;
without 1'
objective

 

 

91

Knowledge—cites a description of the causal law given by Wittgenstein
in the (Tractatus 6.32):

It is the “form of a law" and does not itself express

any causal relation between particular phenomena. It

is rather a conceptual truth which determines the sense

of the concepts "cause" and "event" and may be regarded

as introducing the concept "cause" into our experience

or as expressing its function in that experience, for

without it the discrimination of an event and of an
objective time order would not be possible.1

Kant's argument

Kant presents his view of causality in the ”Second Analogy" in
the Critique of Pure Reason. A. C. Ewing claims in his book Kgntflg
Treatment of Causality that Kant gives six separate proofs of the
category of causality. However, Ewing considers five of the proofs
stated in the second edition to be essentially the same while the fifth
proof in this edition is viewed as different from the other five, and
Ewing treats it as a separate argument. Though close scrutiny suggests
to this writer that Kant hasreally given only one main proof which he
states in various ways, the passages which Ewing delineates as separate
proofs are those most quoted and studied by students of Kant; and so,
they will be presented here in order to familiarize the reader with the
“Second Analogy” in some depth.

Ewing states that the five similar proofs of the category of
cause turn on the connection between objectivity and necessity. Since

Kant maintains that our representations can tell us nothing about

 

1Ib'id.

 

grings-in-t
a now it i
intime. T:
hing desigt

Eat as
happenie
entitIe

oniy as

upon 8 I
me do.
folio-«s

the stre
of this

tower dc
O'der ir
3p;reter
order a;
0f a hou
0f the r
BEIOW an
maniioid
19ft 0r

CeptIOng
Ute nani
Elent tn
the Perc
fCIIOW u

z: is in on

We su

M.’
3". act

jitleo

‘¥~9f,

 

 

 

92

things-in-themselves but only about phenomena, the question arises as
to how it is that we distinguish between objective and subjective order
in time. The answer is given by Kant in the following passage which
Ewing designates as Kant's first proof.

But as I also note, in an appearance which contains a
happening (the preceding state of the perception we may
entitle A, and the succeeding B) B can be apprehended

only as following upon A; the perception A cannot follow
upon B but only precede it. For instance, I see a ship
move down stream. My perception of its lower position
follows upon the perception of its position higher up

the stream, and it is impossible that in the apprehension
of this appearance the ship should first be perceived
lower down in the stream and afterwards higher up. The
order in which the perceptions succeed one another in
apprehension is in this instance determined, and to this
order apprehension is bound down. In the previous example
of a house my perceptions could begin with the apprehension
of the roof and end with the basement, or could begin from
below and end above; and I could similarly apprehend the
manifold of the empirical intuition either from right to
left or from left to right. In the series of these per-
ceptions there was thus no determinate order to connect
the manifold empirically. But in the perception of an
event there is always a rule that makes the order in which
the perceptions (in the apprehension of this appearance)
follow upon one another a necessary order.1

It is in this way that a distinction is made between apprehensions of
objective succession and other apprehensions.

Ewing states that the second proof, which begins at 8239-241,
runs along the same lines with the additional reminder that causation
cannot be proved by empirical induction because of its necessary

character. Kant fbrmulates the case a third time by saying:

 

1Immanuel Kant, Critique of Pure Reason, trans. by Norman Kemp
Smith, p. 221 (Al92-8237).

 

thectii
another

entitle

arises r
itseif,

subjecti
of the n
reiatior
what dig
resaits

rtie, ar
one spec
as our r
order as
0tlectiv
Imust a
5 Dositi
be othem

The .
S‘Eifind Edi-t1]
“53' can bog,
PES’ESQEtati“

I Deiceii
that they
01: which
mnnectlr
not the v
the prod.
fete-Win;
Wt mac:

 

 

93

Objective meaning cannot consist in the relation to
another representation (of that which we desire to
entitle object), for in that case the question again
arises how this latter representation goes out beyond
itself, acquiring objective meaning in addition to the
subjective meaning which belongs to it as determination
of the mental state. If we enquire what new character
relation to an object confers upon our representations,
what dignity they thereby acquire, we find that it
results only in subjecting the representations to a
rule, and so in necessitating us to connect them in some
one specific manner; and conversely, that only in so far
as our representations are necessitated in a certain
order as regards their-time-relations do they acquire
objective meaning. . . . I apprehend an object to which
I must ascribe a certain determinate position in time--
a position which in view of the preceding state, cannot
be otherwise assigned.1

The fifth proof in the first edition and the one added in the
second edition are both discussed in terms of synthesis by Ewing, and
they can both be understood via the following quotation which also is
representative of the sixth argument in the second edition.

I perceive that appearances follow one another, that is,
that there is a state of things at one time the opposite
of which was in the preceding time. Thus I am really
connecting two perceptions in time. Now connection is

not the work of mere sense and intuition, but is here

the product of a synthetic faculty of imagination, which
determines inner sense in respect of the time-relation.
But imagination can connect these two states in two ways,
so that either the one or the other precedes in time.

For time cannot be perceived in itself, and what precedes
and what follows cannot, therefore, by relation to it, be
empirically determined in the object. I am conscious

only that my imagination sets the one state before and

the other after, not that the one state precedes the other
in the object. In other words, the objective relation of
appearances that follow upon one another is not to be
determined through mere perception. In order that this
relation be known as determined, the relation between the
two states must be so thought that it is thereby determined

 

1Ibidu PP. 224-225.

 

as neces
which of
the revs
it a ne:
test the
and in 1
cause an
latter 1
that may
be perce
empirica
inscfar

therefor
as like»
rierce,

tie law.

The
afferent fri
iii the fl [‘5 a

If. then
tiere‘on
E”? Drec
iSlllce I
through

01‘ empir
aWEdi‘an.
fusceedil
fiie Dla
“19mm,
hem aCC
"‘7‘ empi,»
:10” of

 

 

 

94

as necessary which of them must be placed before, and
which of them after, and that they cannot be placed in
the reverse relation. But the concept which carries with
it a necessity of synthetic unity can only be a pure con-
cept that lies in the understanding, not in perception;
and in this case it is the concept of the relation of
cause and effect, the former of which determines the
latter in time, as its consequence--not as in a sequence
that may occur solely in the imagination (or that may not
be perceived at all). Experience itself-~in other words,
empirical knowledge of appearances--is thus possible only
insofar‘as we subject the succession of appearances, and
therefore, all alteration, to the law of causality; and,
as likewise follows, the appearances, as objects of expe—
rience, are themselves possible only in conformity with
the aw.

The other proof for causality, which Ewing considers to be
different from those presented so far by Kant, is the fourth, stated
in the first edition, and the fifth in the second edition:

If, then, it is a necessary law of our sensibility, and
therefore a formal condition of all perceptions that

the preceding time necessarily determines the succeeding
(since I cannot advance to the succeeding time save
through the preceding), it is also an indispensable law
of empirical representation of the time-series that the
appearances of past time determine all existences in the
succeeding time, and that these latter, as events, can
take place only insofar as the appearances of past time
determine their existence in time, that is, determine
them according to a rule. For only in appearances can
we empirically apprehend this continuity in the connec-
tion of times.

Understanding is required for all experience and for its
possibility. Its primary contribution does not consist
in making the representation of objects distinct, but in
making the representation of an object possible at all.
This it'does by carrying the time-order over into the
appearances and their existence. For to each of them,
(viewed) as (a) consequent, it assigns through relation
to the preceding appearances, a position determined a
priori in time. Otherwise, they would not accord with

 

1I id., pp. 218-219.

 

tine its
the posi
time is

of posit
apnearar
must det
ard take
wards, t
confcrri
certaire

antes tr
Pfifluces
Gus com.
is net w
ilEfEln

 

 

95

time itself, which (in) a priori (fashion) determines

the position of all its parts. Now since this absolute
time is not an object of perception, this determination
of position cannot be derived from the relation of
appearances to it. On the contrary, the appearances
must determine for one another their position in time,
and make their time-order a necessary order. In other
words, that which follows or happens must follow in
conformity with a universal rule upon that which was
contained in the preceding state. A series of appear-
ances thus arises which with the aid of the understanding,
produces and makes necessary the same order and continu-
ous connection in the series of possible perceptions as
is met with a priori in time—-the form of inner intuition
wherein all perceptions must have a position.1

Kant'sgpremises and conclusion

but

Ewing's speculation notwithstanding, these passages reveal

one central argument embodied in the following premises and

conclusion.

I. "The apprehension of the manifold of appearance is always
successive. The representations of the parts follow upon

one another? (8234).

II. "How things may be in themselves apart from the representations

through which they affect us is entirely outside our sphere of
knowledge" (BZ35-Al90).

III. In spite of I and II above, we have to show what sort of a

connection in time belongs to the manifold in the appearances
themselves,2 for we are conscious of an objective sequence in

time, i.e., events (paraphrase of BZ35-Al90).

 

yet.

1Ibid.. pp. 225-226.

zThese appearances would not be synthesized by the imagination

I‘i’.

 

make

IV.

96

"Every apprehension of an event is a perception that follows
upon another perception" (BZ37-A192). This premise follows
from:

a. “that something happens, i.e., that something or some state
which did not previously exist comes to be, cannot be per-
ceived unless it is preceded by an appearance which does
not contain in itself this state" (BZ36-Al91).

b. "For an event which should follow upon an empty time, that
is a coming to be preceded by no state of things, is as
little capable of being apprehended as empty time itself"
(3237-Al92).

"I also note in an appearance which contains a happening (the

preceding state of the perception we may entitle A, and the

succeeding B) B can be apprehended only as following upon A;

the perception A cannot follow upon B but only precede it.

The order in which the perceptions succeed one another in

apprehension in this instance is determined and to this

apprehension is bound" (8237-Al92).

"In the perception of an (event) there is always a rule that

makes the order in which the perceptions in the apprehension

of this appearance follow upon one another a necessary order.

The objective succession will consist in that order of the

manifold of appearance according to which in conformity with

a rule the apprehension of that which happens follows upon the

apprehension of that which precedes" (BZ38-Al93).

 

In
truth or ia
the premise
net irply c
Desettled;
pretations
ll?! it, t?‘.
fitemreta:
V‘s"! Hell
‘iiemr‘etat
that Kant 1"
far the mir

sion that ~;

re'v'fidis an
ate

‘it Cdusa

 

 

y‘Et t

97

In judging this argument, one must of course judge both the
truth or falsity of the premises and its deductive validity. Even if
the premises are regarded as true, we may still conclude that they do
not imply causal necessity;1 for the matter of validity also needs to
be settled; and this is not easily done since there are varying inter-
pretations of the premises and conclusion which vary its meaning and,
with it, the validity of the proof. Before considering the various
interpretations that have inspired criticism, let us consider what
might well be called "the minimal interpretation" since it is the
interpretation which makes the narrowest claims and proves the least
that Kant may have wanted to prove about causality. There is support
for the minimal interpretation in the text and it does yield a conclu-
sion that is implied by the premises. As a valid demonstration, it
reveals an interesting argument which can perhaps be extended to prove
a little more of what some critics claimed Kant tried to demonstrate
about causality.

The minimal interpretation of Kant's
ar ument

The first, second, and third premises state what Kant believes
to be the dilemma of the perceiver. We have only perceptions to be the
dilemma of the perceiver. We have only perceptions which are successive
and tell us nothing about things-in-themselves. Yet we observe that, in

fact, perceivers do discriminate among perceptions, i.e., that some

 

1This last phrase 'causal necessity' is left purposely vague at
this point since its clear meaning changes under the varying interpreta-
tions yet to be considered.

 

:erce;.ions
(mile other
we and t
t'a.gh the
are merely
estream an

:e roving d

3'73.” SUCCES!

 

 

 

 

 

98

perceptions constitute an event or objective succession in an object
while others do not. In one case, we see the basement window of a
house and then the roof and say that nothing is happening to the house
though the perceptions are successive. No event is taking place. We
are merely viewing the house parts. In another case, one sees a ship
upstream and then the ship farther downstream. This boat is said to
be moving downstream though these perceptions are also nothing more
than successive. Kant's problem stated in premise III then, is how

is this discrimination possible? What do perceivers do in order to
call the latter successive sequence an event and the former successive
sequence not an event? xPremise IV is a partial description of that
situation in which the term 'event' is applicable.‘ Faced with two
successive state perceptions, they must differ if we are to allow that
anything has happened or changed. If two state perceptions were the
same, there would never be talk of a change or event taking place. But
alas, there are sequences of successive but differing state perceptions
which are said to be events, i.e., the case of the ship; and there are
other sequences of successive but differing states which are said not
to be events or cases of change in the object, i.e., the case of the
house. In premise V, Kant gives the final criterion which makes the
descrimination between the two successive sequences of differing state
perceptions possible. In the first sequence, i.e., that of the ship

moving downstream, the order of perceived states is determined while

 

1If the linguistic bent of phraseology seems too modern an
interpretation of Kant's thought, see 8238-A193 on p. 222 of N. K.
Smith trans.

in tie sec
and M
rated an
:e'ceiver'
in which A
'Ja‘. a det
Kant tails

mason, 51"

"539 Dene

975': of ti-

nt” 1, 0a.
"3“ the t6
Mining 1
32:95 Carr

‘9‘.»
1“ a ,

 

 

99

in the second sequence the order is not determined. If A precedes B
and we cannot on our part perceive B to precede A, the sequence is
termed an event. If B precedes A due to no perceptual actions on the
perceiver's part, it is termed an event but not the same event as that
in which A precedes B. Finally, the conclusion states the deduction
that a determined order requires something that determines it which
Kant calls a rule, the law of causality, or the rule of sufficient
reason, since the subject_perceiver has already been excluded as the

determinant. It is not important that we know what the cause is which

 

affects a moving ship but merely that there is some cause. It is not
the case either that state A which precedes state B is the cause of
state B. The state of being upstream is not the cause of the ship
moving downstream. Thus premise V must be understood to be making a
determined order claim for events even when all the possible constituent
state perceptions are in fact not perceived by the perceiver. In the
event of the ship moving downstream, constituent perceptions x, ship
upstream, and y, ship downstream, are perceived; but possible percep-
tion 2, oarsman paddling ship, is not so perceived._ It also appears
from the text that the perceiver need not know the specific cause
obtaining in a given situation but only that if the order of perceived
states cannot be changed by any action on the subject's part qua per-
ceiver then it is determined from withOut and there must be a rule of
necessity. This is reminiscent of Descarte's description of adventi-
tious ideas. At any rate, Kant demonstrates that some causal factor is

operative though not that it must be known in any particular case. He

also concl
succession
in the cas
{as in the
st; rcvin

As-
Kant's arg.

arch have

['iilClSW :
\

Ewi
5'3Jent, y
Siar-t With
Successive,
“a.” a 5.
me’ We:

iifeiiatei

fife SUCCESS

 

”I
~r7‘ESenta.

ifst‘ént a.

iiitatw'

long

it): other

 

 

100

also concludes that events constitute objective succession, i.e.,
succession of states in the object of perception be it apparent (as
in the case of my dreaming about the ship moving downstream) or real
(as in the case of my wakefully seeing, under normal conditions, a
ship moving downstream which is also seen by my comrades).

Assume that the above minimal interpretation is faithful to
Kant's argument for the moment and consider the following objections

which have been levelled against the truth of the various premises.

Criticism of premise I

Ewing claims that four out of the five statements of the
argument, which he delineates, argue from objectivity to necessity and
start with the assumption that all apprehension of the manifold is
successive.‘ It segm§_that Kant is not only claiming that there is
always a succession in time involved in our experience but that we can
never immediately apprehend the co-existent or permanent. We can only
immediately apprehend the successive, and thus all our representations
are successive and never co-existent. But at 8234 Kant says, "My
representations of the co-existent are treated as always successive
just like my representations of the objectively successive." Ewing
concludes that Kant denies the possibility of distinguishing the co-
existent and the sequent, in any single case, by simply having repre-
sentations that co-exist, as opposed to representations that follow

each other, on the ground that apprehension is always successive.

 

1Ewing, Kant's Treatment, p. 218.

 

In
:syciolcgy
we are cor:<

:‘or. involi

 

 

101

In opposition, Ewing believes that the findings of modern
psychology clearly indicate, in cases of introspection, that "what
we are conscious of even as an object of explicit attention or cogni-
tion involves co-existent elements and is not a mere one dimensional
succession.”1 The elements range from slight sensations, which, as he
says, help to constitute the general feeling tone of a given attended
situation; to objects of concentrated attention. Ewing goes on to claim
that introspection, far from revealing the merely successive, never does
so at all, for he characterizes the merely successive as the absolutely
simple, a mere line without breadth.2 Whether Kant or Ewing is right
about the complexity of representations is not only a question of
psychology but also quite beside the point.

Even if one could perceive the bottom of the house while attend-
ing to the roof, Kant could redesignate his example of a simple percep-
tion as a perception of the whole house A at one end of the block and
construct his subjective sequence by comparing this perception to
another perception of the whole house B at the other end of the block.
Presumably the second perception of house B could not be had, even
faintly from one's side vision, as house A is being attended to by the
perceiver. It is this possibility that Ewing probably considers when
he concedes that his objection does not seriously damage Kant's line of
argument. The deduction that Kant makes is not dependent on his assumed

premise of a merely successive manifold, and his argument is even stated

 

1Ibid., p. 84.
2Ibid.

 

En
twat the d
vc'Ives ca:
distinctio-
at our per
successive
cation, b
succession
exitliie of
fitatlon 01
*Etce, Pier
"29 with
wi;59d pr!
55CCESSlve

anEther.

.I I
H. ‘
i

1". G

“CtiCts
\
Pr

 

“’35 car

fie SJCCES

 

. I
J‘i'tn 0f F:

354g.,.

-*~lee

\zx

 

102

without this assumption in the passage which he added in the second
edition.1

Ewing recognizes that the second analogy is an attempt to show
that the distinction between objective and subjective succession in-
volves causality. There can be no doubt, Ewing thinks, that such a
distinction is made and that it is not determined by the actual order
of our perceptions even though not all perceptions in his belief are
successive. At times, the co-existent can be given in a single per-
ception, but there still are other times when the status of objective
succession is denied to successive perceptions, as in the counter-
example of viewing houses A and B. There are times too when the deter-
mination of objective co-existence is made about successive perceptions.
Hence, premise I can be maintained and even slightly revised to harmo-
nize with Ewing's insight without weakening Kant's argument. The
revised premise is: The apprehension of the manifold is sometimes
successive. The representations of the parts then follow upon one

another.

Objections to premise III

Prichard makes the claim that co-existence or permanence and
change can be directly perceived and that if the manifold is successive
the succession is of event and non-event perceptions. He questions the
truth of premise III by saying that Kant's problem of distinguishing

objective sequences or events from merely subjective successions simply

 

1Kant, op. cit., p. 2l8 (3233).

 

ices not i
mediate
when Sci t!
also raise
criticisr

GI
out that 1
Dricnard i
the order
he say dis
sate obje:
water was
5i?! state
ti conside
a“WS o
Utimteij.
1: seaside
Clfipiex e.

“a other

"Eta Pd 5

 

 

103

does not arise because we apprehend the time order of events
immediately.‘ We perceive that Jones crosses the finish line first
when Smith loses a race and Jones wins the race. This objection is
also raised by Lovejoy and is similar to, but stronger than, the
criticism just discussed in connection with premise 1.2

Graham Bird, in his book Kant's Theory of Knowledge, points
out that the difficulty with Prichard's objection is simply that, while
Prichard is speaking about the order of events, Kant is speaking about
the order of states or state perceptions that go to make up an event.
He may distinguish events in terms of the orders of two states in the
same object. In example, Bird suggests that the question of whether
water was frozen or ice melted is settled by different orderings of the
same states of fluidity in the water. Thus Kant has ground for choosing
to consider the order of states rather than the order of events. The
analysis of complex events, such as Jones winning the race from Smith,
ultimately depends upon simpler events; and Kant has "naturally chosen
to consider the simplest events on which the discrimination of more
complex events depends."3 The simple events depend of course not on
any other events but only upon the order of states in an object. If
Prichard argued further that when we do at least perceive directly the
order of states, Kant's problem would become a problem also fbr him

because the question of which sequence of perceived states is to be

 

‘H. A. Prichard, Kant's Theory of Knowledge, pp. 294-295.

2Arthur 0. Lovejoy, "On Kant's Reply to Hume," Kant; Disputed
Questions, ed. by Moltke S. Gram, p. 297.

3Bird, op. cit., p. l57.

an.“

 

 

:er'i’ed ar
stresses

taseniy s

"T?’C:;ariai
4‘ . A q

M '(‘J‘did r
"o.se roof
ifitemstefl
# Iiv

'C ‘DET‘CI

LB flip-”ea re

 

 

104

termed an event, on occasion, would once again arise. Bird in addition
stresses that it is not even a problem of apparent events being mis-
takenly supposed for a real event in premise III, but rather a problem
of when it is appropriate to speak of an event at all, be it apparent or
real. In support of his contention, Bird emphasizes premises IV and V.
If a ship is first seen in one position and then in another downstream,
we can describe what was perceived as the event of a ship's moving down-
stream whether the ship has actually moved or not. It would still be
appropriate to call such a succession of perceptions an event. However,
it would never be appropriate to call the successive perceptions of‘a
house roof and a house basement an event. As Bird says it, Kant is
interested only in the inference “I perceived A and then perceived B"
to "I perceived the event A-B" and not in the inference from "The event
A-B appeared to take place" to "The event A-B really took place." The
premise of the former does notpresuppose the vocabulary of events; for
after all, that is the point at issue; and Kant's problem is genuine.
Bird suggests that Kant's talk of the transition from subjective
to objective succession may have given rise to the belief among many
critics that Kant is attempting to justify inferences from the way things
appear to happen to the way in whiCh they happen objectively. However,
Kant is trying to indicate the difference between discernment and non-
discernment of an event rather than between recognition and mis-
recognition of an event.‘ As Bird points out, it is possible to mis-
recognize an event only if it is possible already to discriminate

between events and other kinds of phenomena. To quote:

 

 

Hnat
SuCCe
guisn
of a;
The infer
aquestic
in questi:
sorts.
B.
inCMeri
“9': how

:Tt‘EEE‘. an

 

 

105

What distinguishes an actual event from a perceived

succession of actual states in an object also distin-

guishes an apparent event from a perceived succession

of apparent states in an object.‘
The inference from state-—descriptions to event--descriptions involves
a question about the meaning of the term 'event' which is presupposed
in questions of the criteria for distinguishing real from apparent
events.

Bird does well to stress a close reading of premises IV and V

in order to understand exactly what Kant's problem is in premise III,
i.e., how to distinguish objective from subjective succession. However,

Bird's remarks are not quite complete enough to adequatelydefine Kant's

problem and the ambiguity which his critics find in it.

Further defense of_premises III, IV and V

In keeping with the Kantian spirit it must be remembered that
Kant is talking always about ascriptions of objectivity within the
manifold of appearance. He uses 'appearance‘ in the “Second Analogy"
as he would the term 'phenomena,‘ i.e., to name the categorized content
of empirical, sensuous intuition.2 Had Kant been more precise, he would
have called attention to the fact that ascriptions of objectivity are
really two-levelled. First one must separate objective successions from
subjective successions in order to correctly identify a change of state
in the object of appearance, such as ice melting or a ship moving down-

stream, from merely changing perceptions of an object which is not said

 

‘Ibid., p. 158.

2This is not the way Kant typically uses 'appearance' elsewhere,
when it is used to name the lowest level of representation, something
akin to sense data.

tn be coal
is accorfp'
ceivers d‘
treat of a
tivity di!
34‘. both i
«at is ot
tatter of

its?) tta‘.

Cistinctiq
I'Eé‘iiy he:
[flied at:
“My cla
if an Obie
“ the] a?

big a
a“: ‘S. fl

 

 

 

106

to be changing, such as the house of his example. Once this distinction
is accomplished, one must further discern the criterion by which per-
ceivers discriminate real events from imagined ones, as when one might
dream of a ship moving downstream. As Bird suggested, the first objec-
tivity distinction must be made before the second one can be attempted.
But both distinctions are made all the time by people in determining
what is objective in the usual sense of the term 'objective.‘ As a
matter of fact, the first objectivity distinction is made so spontane-
ously that the question of objectivity as it is usually discussed in
philosophy revolves almost exclusively around the second objectivity
distinction. It was for this very reason that Prichard denied that Kant
really had a problem in premise III, as just discussed. Thus, Kant has
called attention to an interesting aspect of the usual problem of objec-
tivity claims by showing that even imaginary events involve application
of an objectivity criterion, i.e., a rule of necessary sequence in time.
if they are to be known as imaginary events as opposed to imaginary non-
events. To focus on this first aspect of objectivity, viz., this first
distinction, is indeed Kant's aim in the “Second Analogy," as Bird
concluded. Hence, it is this understanding of premise III, i.e., of
Kant's problem, which has been incorporated into our minimal interpre-
tation of the argument of the "Second Analogy." But in fairness to
Kant's critics, let us explore the possibility that there is a broader
interpretation which can be put on Kant's argument, which suggests that
Kant is not only trying to solve the problem of how to make the first
distinction with regard to objectivity but the second distinction as

well.

Ll
tent to bl
tivity an:
hGh-Sengi'
aeoncius«
1m cause
”Oi'EVERts
Curse, l5
tie militia
‘1 *0uld
LEV-3:03, Vi
Ethiustn
Lz‘ev‘oy he
fianaliy c
0‘ the "c:
thePrinci
Le‘inizfi
he,
R6350':
Petite
becah.

F‘Ere |
in si-

 

 

107

Lovejoy, in his article “On Kant's Reply to Hume," interprets

Kant to be arguing for causal necessity as the ground between objec-
tivity and subjective dreams or hallucinations and accuses Kant of a
non-sequitur.‘ Lovejoy notes that premises IV and V do not imply such '
a conclusion. The most that premises IV and V can imply, he claims, is
that causal necessity is the criterion for distinguishing events from
non-events be they in veridical perceptions or dreams. The latter, of
course, is the current conclusion that we have already incorporated into
the minimal interpretation of Kant. Supported by the admonition of Bird,
it would seem that in this assumption there has been no mistake; and
Lovejoy may be wrong to identify Kant's confusing terminology as a
confusiOn in his argument. The matter is not so simple; however, for
Lovejoy has reason to believe that Kant actually lifted (either inten-
tionally or absent-mindedly) a different argument from the earlier work
of the "dogmatic phil050pher" Wolff, who attempted to supply a proof of
the Principle of Sufficient Reason which was lacking in the work of
Leibniz.“ Lovejoy quotes Wolff as saying:

. . . a sufficient proof will appear when we show that it

is through this principle alone (Principle of Sufficient

Reason) that the distinction between reality and dreams

between the real world and Schlaraffenland arise.

Because everything has its sufficient reason for existing

there must also always be a sufficient reason why changes
in simple things succeed one another in one way and not

 

‘This objection could have been discussed with the other
criticisms of non-sequitur which will follow later in this paper, but
it also seems to follow naturally topics treated just prior to this
section which challenge the truth of premises III, IV and V.

2This is only one side of a historical controversy. Ewing
strongly disagrees, that Kant gig_totally divorce himself from Wolff--
see pp. 23-24.

in anc
other
order
recast
from d

aitera

Lc

id fian a

taming t'r

an percep

it the se'

Ln'

In.

 

 

108

in another and also why their alterations follow each
other in one way and not in another. . . . Now since an
order of this kind is not found in a dream . . . one
recognizes clearly from this that truth is separated
from dreams by means of order . . . the order of the
alteration of things.‘

Lovejoy claims that while Kant sets out to produce the above
Wolffian argument, he in fact confuses it with his own argument con- '
cerning the problem of distinguishing between perceptions of change
and perceptions of permanence whether the perceptions be objective
(in the sense of the second distinction) or purely illusory.

Unfortunately the strength of Lovejoy's claim that Kant was
confusedly trying to present Wolff's argument as his own depends heavily
on how Kant meant to use the terms 'objective' and 'subjective,‘ and
this is of course the very point at issue. The text, however, does
supply strong clues as to Kant's intentions. Throughout the "Second
Analogy," Kant uses the phrase 'objective succession' in apposition
with the term 'event'; and nowhere does he use the phrase 'objective
event.I This clearly indicates that Bird's claim is correct and that
the minimal interpretation of the argument given here while the least
that can be said can at least be said to accurately represent Kant's
thought. But given the dominant use of 'objective' in philosophy to
describe the problem centering around the second objectivity distinc-
tion, it would be surprising if Kant, who was calling attention really

to a different use of the term 'objective,‘ should not slip into the

common or orthodox use of the term at isolated points. But careful

 

‘Lovejoy, op. cit., pp. 29l-292.

. ading
(ant can

at 5247:

Here
even‘
have
of pg
SO‘ 1
drear

 

 

109

reading indicates that there is only one sentence in the text where
Kant can be even possibly accused of making such a slip. It occurs
at 8247:
Were it not so, were I to posit the antecedent and the
event were not to follow necessarily thereupon, I should
have to regard the succession as a merely subjective play
of my fancy; and if I still represented it to myself as
something objective, I should have to call it a mere
dream.
Close inspection however of this passage reveals two things:
l. An accusation that Kant is confusing at 8247 his own argument
for objectivity distinction l and Wolff's argument for objec—
tivity distinction 2 may be hasty.
2. Lovejoy's criticism that if Kant is trying to ground distinc-
tion 2 on his own premises for distinction l, he is guilty of

a non-sequitur may be ill-taken.

In considering the second point, it must be noted that Lovejoy
bases his claim of a Kantian non—sequitur on the supposed fact that
necessary sequence is neither the primary nor the only criterion by
which objectivity distinction 2 is made. Certainly it seems plausible
for us to agree with Kant at this point that necessary sequence is the
only criterion by which objectivity distinction l is made. But even
if Kant has confusedly used premises that imply distinction l for an
implication of distinction 2, there may still be a non-non-sequitur
(if you will) claim to be made against Lovejoy's contention.

Lovejoy states that: .

Practically the most important criterion of the objectivity

of a perception, at the time that I am having it, is its
agreement with the experience of other men about me. But

it is
applie-
utette
family

a.

a.
their fell
S’l'; mi'ir
:nterion

ecessary

Ff. of he
is that he
was opih
rot the
”temper
iii 3?". d .
‘3‘ U9 wi
31?. ;

(E'Ce;

‘ .
if,"

suing,

 

 

110

it is true that if the test of social currency cannot be

applied, I amlikely to test the perception by seeing

whether it conforms to the rules of causal sequence uni-

formly exemplified in the phenomena of my past experience.‘

Quite on the contrary, however, it seems that people rarely ask

their fellows what they (their fellows) are seeing as they look at a
ship moving downstream or a tower in the distance. Rather the primary
criterion for objectivity in both its senses is the discernment of
necessary sequence. Only when a perceiver cannot make such a discern-
ment of necessary sequence is he likely to ask another observer what it
is that he is seeing. On being asked, is this second fellow in Love-
joy‘s opinion to ask a third person in order to answer the first; or
is not the second fellow‘s answer hopefully given in light of his own
discernment of some necessary sequence? If the second perceiver must
ask yet a third, somewhere at the end of the query line someone must
come up with the answer on the basis of some necessary sequence of his
own perceptions or the questioning will go on ad infinitum. Thus, if
anything, Kant's criterion is the only criterion for ascribing objec-
tivity in either sense of the word, whether immediately or ultimately.
A case can be made then to deny that Kant is guilty of a non-sequitur
even if necessary sequence is used by him as a criterion for objectivity
distinction 2. But a careful look at the only passage in the "Second
Analogy" which might suggest this use of the criterion reveals that Kant

is fully aware that he is pp§_arguing for a ground that will enable the

distinction between apparent and real events to be made. As he says at

 

‘Ibid., p. 305.

E247when the
'I cannot cont
still regardir
to regard the
as part of a c
Kant is mindft
“‘59 the tern
iverlooked fir
sonsistent in
:oriusion in h
“‘5 Practice

Based
‘t‘C‘ ‘5 broad
flan, V12” til
rather than th

tummy . Ewi

In dc‘”9 So, E

 

 

 

111

3247 when the criterion he is trying to establish cannot be applied,
"1 cannot conclude that an event has taken place and if I insist on
still regarding it as an event or something objective I should have
to regard the event as a dream." This positing of something objective
as part of a dream reveals, perhaps better than any other passage, that
Kant is mindful of the two levels of objectivity claims and that he is
using the term 'objective' in an unusual manner to refer to the often
overlooked first level objectivity claim. The "Second Analogy" is also
consistent in this usage, and Kant does not appear to be guilty of any
confusion in his argument but only of using unorthodox terminology. In
this practice he is neither the first nor the last guilty philosopher.
Based on an understanding of the conclusion of Kant's argument,
which is broader than that expressed in the present minimal interpreta-
tion, viz., that Kant has proved there are particular necessary rules
rather than that there is one necessary rule, i.e., the general law of
causality, Ewing considers another criticism of Kant made by Prichard.
In doing so, Ewing reworks Kant's argument just enough to make it a
valid proof of just the kind ofobjectivity Lovejoy said it could not
prove. The hint of this has already been stated.
Ewing quotes Prichard as saying that:

He (Kant) is anxious to show that in apprehending A-B,

as a real or objective succession we presuppose that

they are elements in a causal order of succession.

Yet in support of his contention he points only to the

quite different fact that where we apprehend a succes-

sion A-B, we think of the perception of A and the per-

ception of B as elements in a necessary but subjective
succession.‘

 

‘Ewing, Kant's Treatment, p. 90.

" I fl‘xm'wi

 

 

Hith s
is still adec.
retains a rep'
demnstrated J
{whether actae
Al inplies a r
hfer one stat
which cause A?
Céita cannot t

Since any caus

a causal law c

.' 1
“me the Perce

lsrt‘ori admit
the existence

:E'Ceptas and

9“
id

"95 0f the
we ”WM at

6‘“.
.11,“ the act

SQW‘W in .

‘i’cention an.
t‘r v
0” the

Stine
1" "Entl'On

\

1.
”\id.

|
|
F

 

112

With some bolstering of Kant, Ewing thinks that Kant's proof
is still adequate enough to please a realist palate.- If'a realist
retains arepresentative theory of perception, it can be logically
demonstrated that the possibility of inferring a perceptum Bl
(whether actual or possible) by causal reasoning from another perceptum
Al implies a necessary connection (a connection which enables us to
infer one state from another) between the states of the physical objects
which cause Al and Bl, respectively. Necessary succession in our per-
cepta cannot be really separated from necessary succession in the object
since any causal law which can be discovered connecting percepta implies
a causal law connecting the states of the physical objects which deter-
mine the percepta or'percipibilia.‘ A realist, Ewing insists must a
fortiori admit some causal lawsconnecting our percepta if he deduces
the existence of physical objects in order to account causally for our
percepta, and any such law seems to imply a causal law connecting the
states of the physical objects which determine the percepta. Even if
the realist abandons the representative theory of perception and iden-
tifies the act of perceiving with the object perceived and so with
something in the physical world, he must still distinguish between
perception and imagination; and Ewing agrees with Kant that this dis-
tinctipn can only be made by the introduction of necessity.

On the other hand, if the realist takes the subjective idealist

stance mentioned above, the realist objection fails its mark altogether;

 

‘Ibi ., p. 92.

for the dis‘“
the reaiis‘ dz
that the idea
Coordinatlng
aercipient at
9mg causa1
that for any 1
arrive at an '
different P9“
in this refer!
between them 5
between the d‘
The idea of a
septa involves
Ewing
:hesent minim:
orthodox sensel
the minimal i'|
terion for obkl
ieductive sup;I
He call events.

Ewing

 

at" .
wining to .

\'

 

 

113

for the distinction between perception and object has vanished. If
the realist does not collapse the two into one, he will have to admit
that the idea of any definite physical object can only be reached by
coordinating the experiences of different percipients and of the same
percipient at different times. this being possible only via a thorough-
going causal determination in the object. In conclusion, Ewing argues
that fer any theory of perception, representative or direct, we cannot
arrive at an idea of any particular physical object except by referring
different percepta to it, either as its effects or its states; and to
do this referring, it is necessary to assert a necessary connection
between them so that physical states which form intermediate links
between the different percepta must be themselves causally connected.
The idea of a physical object derived from correlating different per-
cepta involves thorough-going causal connections.1

Ewing here is extending the premises and conclusion of the
present minimal interpretation to imply claims of objectivity in the
orthodox sense of the term. But one could answer Prichard in view of
the minimal interpretation that Kant was not trying to provide a cri-
terion for objectivity level 2 so much as he was trying to provide
deductive support for causal necessity with an argument about what
we call events.

Ewing answers another objection posed by Prichard, i.e., that

according to Kant the same representations are related both as physical

 

1Ibi ., p. 95.

 

IL

 

and as psychi
the object a;
of the proof
other staterie
Ewing thinks
of our expel-1"
Kart's trans
oortent. A

resentatiOn

:ortent. Vi

C0“tent. it

”5t Private

 

114

and as psychial events; that is as our apprehensions and as part of

the object apprehended.~ While there is such talk in early formulations
of the proof of causality, Kant can also be accused on the basis of
other statements of an excessive dualism between act and content; and
Ewing thinks it feasible to identify a physical object with the content
of our experience and not with the act of perception or representation.‘
Kant's transcendental self is the act; and the empirical self, the
content. A physical object constituted of a complex of acts of rep—
resentation would be, in Kantian terms, a set of categories without
content. Viewing representations in the sense of a mental image, i.e.,
content, it can only be said that they are private images. but public
not private pr0perty. 'The identity of the same representations related
both as private imagesand objects is qualitative not numerical.“ On

a subjectivist view, there is nothing real for any representation of
mine to stand in a relation to but minds and their other representations
although by a logical fiction it may be thought as an attribute of a
physical object connected with other such objects. 0n the more objec-
tive view, physical objects have reality as a different class of being
from private representations and-are not composed of the latterr In
neither case do private representations exist both as parts of real
physical objects and as psychial events in the life-history of indi-

viduals.3 Identifying the physical world with "representations" in

 

the sense of
mly that to
Aphysical c“:
that we cool:
of it, but t"

:rivate menta

fSllowing ar;
W lrreversi
35:8Ctive SU:
«595 the DC:
“5'9 Certaiv
by thinking TI
5'? neCeSsar‘l
“hip “DStrj
wind mUSt 1”WI

 

115

the sense of being a "public object" of knowledge or experience means
only that to us; it is nothing except as object of human experience.
A physical object Can only be conceived as an object represented so
that we could not say what it would be apart from all representation
of it, but this does not mean that the physical world is composed of

private mental images.

Misunderstandings ofpremise V
Enough said about objectivity and subjectivity, consider the
following argument raised by Jonathan Bennett against premise V. Taking
the irreversible order of successive sequences which are said to be
objective successions (premise V) as his starting point, Bennett con-
fuses the positing of causal necessity with a predetermined sequence
where certain initial conditions are necessary. He misunderstands V
by thinking if A occurs B will always follow because those conditions
are necessary whenever A occurs that are said to cause 8. If one sees
a ship upstream, one can expect to see the ship downstream next because
wind must fill the sail of-a ship upstream. He gives the following
counter-example to this mistaken notion that he has about premise V.
I saw a long boat being rowed out of the harbour;
which if Kant's analysis is right entails not just (a)
that my visual states did occur in a certain order but
that (b) I could not have had them in any other order.
But since the coxswain of the boat was under orders
from me, I could have secured for myself the spectacle

of the boat being back-paddled stern foremost into the
harbour.1

 

1Jonathan Bennett, Kant's Analytic, p. 222.

 

This
Bennett perh
the boat wou
oasu. Kant
tation after
any particul
boat to foil
or the wind
iresent for ‘
stream and Cl
19.19?- aSk KaJ
55.39058dl‘y 5%

In h
Etchting dis:

Kant Cou
Slfillar i
liar p9“
SPOHded 4
harDOUr !
EVEnt. I
by Order:

NE? of
the fem“
percenti
affal r5 3
about thfi

ASir

““SCJSsed by

 

116

This counter-example hardly rates comment for surely none, but
Bennett perhaps, could possibly say that the coxswain's back paddling
the boat would not result in the boat going back into the harbour pari
pasu. Kant never suggested that we had to perceive any given represen-
tation after seeing a boat upstream other than what might follow from
any particular causal influence. The only time we would expect the
boat to follow downstream would be when the oarsmen paddled it there,
or the wind filled its sail, or when appropriate causal conditions were
present for its moving downstream. Surely if the ship caught fire up-
stream and our next perception was of a capsized ship upstream, we would
never ask Kant why we did not instead see the ship downstream as he
supposedly said we must.

In his article "Kant's Second Analogy of Experience," N. A.
Suchting dismisses this criticism by saying that:

Kant could hardly have claimed that I could not have had
similar perceptions in any other order, for clearly Slflk
ilar perceptions in the reverse order would have corre-
sponded to the event of the boat's being paddled into the
harbour sternwards. But this would have been a different
event. And the difference between my bringing this about
by orders to the coxswain and my bringing about a different
order of perceptionsof the parts of the wall is that in
the former case I can bring about a different order of
perceptions only by causally influencing the state of
affairs itself, whilst in the case of the~wall I can bring
about the difference by causally influencing myself alone.1

A similar objection raised by Schopenhauer against premise V is

discussed by Ewing. Schopenhaure claimed, in opposition to premise V,

 

1“. A. Suchting, "Kant's Second Analogy of Experience," Kant
Studies Today, ed. by Lewis-W. Beck, p. 330.

that ore eve
Kant's descr
Kant identif
objection, E
that B most
that the seqi
or causes"1 ’
Kant's words
IT"? than or
inaction fn
8234, Kant C;
mm the a
W itself 1
names Kant
ExiTples 0f &

V v
as ~
Hy

gECthQ Se:
Ema" tenet
ab:

rut B necess
that the Ken:
3. .

t. .

lThiS

‘Ewlr

 

' 117

that one event may succeed another without being caused by it. On
Kant's description, night would have to be the cause of day, since
Kant identifies objective and causal sequence. In answer to this
objection, Ewing states that Kant has not proved or sought to prove
that B must be causally determined by A if B succeeds A but rather
that the sequence A-B "must necessarily be determined by some cause

"1 and that any given event is causally determined by. in

or causes
Kant's words, "some as yet indeterminate correlate."2 Kant points out
more than once that particular causal laws can only be discovered by
induction from particular experiences, even though Ewing notes that at
8234, Kant does speak as though all objective sequence were causal.
However the actual proof of causality at 8239, 243, and 244, would not
lend itself to thisinterpretation. In addition, Ewing notes that the
examples Kant uses, such as the ship going downstream are not given as
examples of cause and effect even though they are given as examples of
objective sequence. Against the similar objection that causality in
Kantian terms is merely necessary sequence and that, therefore, to say
that B necessarily succeeds A, is to say that A causes B, Ewing states
that the Kantian view is more correctly, that under the given conditions,

B necessarily succeeds A, i.e., that B necessarily succeeds A plus the

sum total of other relevant conditions. Specifically he says:

 

1This excluded perceiver qua perceiver causes.

2Ewing, Kant's Treatment, p. 85.

It is tn
C, D, so
that B n
tatively
tions of
strictly
imply tn

tanther ot‘e
In a

ave been re
hzactive va
:iH-A. Suci
0f Kant's ar
Necessar‘
Event-pg
ity of A \
Indetemjrl
Sethting thev

' ' - the
EVEnt E l
SEQbEnce
of these
EVQnt 0::

:"3'3 prEMlSe
' ' - the
pr§5U9303
this id65
15 that c

 

118

It is the task of science to suggest other conditions
C, D, such that we can formulate a law to the effect
that B necessarily succeeds A, C, D (or often quanti-
tatively varies in a certain proportion to the varia-
tions of A, C, and 0).. the Kanatian view is not,
strictly that B necessarily succeeds A for that would
imply that B must always succeed A.

Further objections, of nonasequitur

In addition to the foregoing claims against the premises, which
have been refuted, there are other criticisms which challenge the
deductive validity of Kant's argument'such as the following presented
by N. A. Suchting. Suchting formulates premise V and the conclusion

of Kant's argument as:

Necessarily if A and B are the constituents of a certain
event-perception, then the temporal order of perceptibil-
'ity of A and B is what may be variously described as
"determined, " "determinate," in "conformity with a rule"
"necessary. "2

Suchting then adds an interpretation of premise V given by Graham Bird
who elaborates that the above mentioned necessity is:

. the logical necessity that to apprehend a certain
event E is just to apprehend a certain fixed determinate
sequence of states so that to apprehend any other sequence
of these states would be by definition to apprehend an
event other than E.

From premise V (its meaning explained by Bird) one concludes:

. the idea of a determinate order between two states
presupposes that of something which determines it and
this idea of a determinate or reason for such an order
is that of a cause.

 

1Ib‘ld., p. 88.
2Suchting, op. cit., p. 329.
3Bird, op. cit., p. l55.

119

Suchting argues that if the above faithfully reproduces Kant's
train of thought then the argument rests on a pun. He asks:

How could one get from the proposition that events are
logically determinate in the sense that they are con-
stitutedby a determinately (ile. definite specifically)
orderedisequence ofistates to the conclusion that this
sequence is causally‘determined, i.e., such that the
determinate order in question is due to some causal
relation . . . the fallacy may also be (i) "Necessarily
if A and B are the constitutents of a certain event-
perception then the temporal order of A and B is
necessary" that (ii) "If A and B are the constituents
of a certain event-perception then necessarily, if Ai
then B,“ the latter conclusions then being detached.

 

In other words, Suchting asks how is what he calls the logical
necessity as described by the underlined above (premise V as interpreted
by Bird) linked to causal necessity. Suchting thinks that Kant has con-
fused logical with causal necessity and that he is guilty of a non-
sequitur. Suchting is correct to reject the quoted passages as a non-
valid argument, but in light of the minimal interpretation, herein
presented of Kant's argument, Suchting is wrong to accept Bird's
elaboration of premise V as an adequate reflection of Kant's idea. The
fallacy thus lies with Bird and not with Kant. Indeed Bird's sentence
is an anomaly in his own article and does not reflect his own thoughts
on Kant's argument. This sentence is inconsistent with the rest of
Bird's discussion and can only speak of some temporary confusion on
his part.

Surely A followed by perception B, if an event, is a different
event than B followed by A. That is logically the case if one honors

 

1Suchting, op. cit., p. 332.

 

the law Of 1
different at
because a 56
as an event

event sequei‘

by perceiver

tecause of s
vrfie this c
non-event se
that one ca:
It seems uni
accept it wi
'ot‘rer' and
I'itten in t
”sad:

Kant Spe
certain
fixed, d

in"! othe
defim' ti

This

19*:

\§~

531'” wh

PEra
chSEry

tlrace the

h:
.s - '
amine:

 

120

the law of identity.1 Thus, it is logically necessary that A-B is a
different event than B-A. But this is not Kant's point simply. It is
because a sequence is in determinate order at all, that we identify it
as an event whether A-B or B-A, or any other, as opposed to some non-
event sequence. The ordered sequence whose order cannot be altered
by perceivers qua perceiver is determined, it seems plausible to say,
because of some agency which there is tradition to call causal. But
while this causal determinacy or necessity separates the events from the
non-event sequences it is also logically necessary as in any identity
that one causal event cannot be itself and some other at the same time.
It seems unlikely that Bird on a reassessment of his statement would
accept it without the word 'fixed' inserted in the fourth line between
'other' and 'sequence' and the insertion 'as E' after 'E.‘ If it were
written in this way the error would probably not arise. It then would
read:

Kant speaks of the logical necessity that to apprehend a

certain event E, as E, is just to apprehend a certain

fixed, determinate sequence of states so that to apprehend

any other fixed sequence of these states would be, by

definition, to apprehend an event other than E.

This statement more clearly reveals the operation of logical

necessity when speaking of successive perceptions that are.causally
necessary. If Bird were to be consistent in his writing, he would

embrace the following interpretation mentioned by Suchting later in

his article: “that if A-B constitute an event, then the order of

 

1I am indebted to Gerald Massey for the reminder that this claim
is true unless of course time is circular.

121

apprehension is, given certain standard circumstances, independent

of us. But Suchting will not even accept this corrected interpretation
of V because it does not imply C1 which (as luck would have it) is given
the following erroneous interpretation by him at the end of his article.
Suchting thinks that C claims that A and B are causally related, and
even given V, it is possible that A should have occurred in a causally
determined way, but that the later B should have been causally undeter-
mined. Bird's erroneous interpretation of V did not together with the
other premises imply the conclusion C, and now the corrected interpreta-
tion of V together with the other premises does not imply Suchting's
erroneous interpretation of C. Suchting is of course wrong in his
appraisal of C for the very reasons already stressed in this connection.
Kant's chief example does not suggest that even if A and B are two per-
ceptions whose order is determined the determinate or causal agent is A.
Kant would claim A rarely is the cause of B since the cause in many
cases is completely unknown. It is this insight that Suchting overlooks
in both his criticisms. ‘Kant's three claims: (l) that the past neces-
sarily determines the present to be present; (2) that event A-B neces-
sarily determines the order of A and B by identity not to be that of
event B-A; and (3) that the order of A and B is necessarily A-B if the
subject cannot perceive them in differing order in the present succes-
sion of perceptions, are distinct and independent claims which should
not be confused with each other nor interpreted as dependent on one

another.

 

l"C" here stands for the words “the conclusion."

122

We come now to perhaps the biggest vagary of all in interpreting
Kant's conclusion. It was hinted at earlier in the historical note,
that at first, Kant did not think it necessary to prove the general law
of causality, i.e., every event presupposes a cause.1 When Kant speaks
in the conclusion of Va ”"19 that makes the order in which the percep-
tions follow upon one another a necessary order," what rule is he talk-
ing about? Is he arguing for the Rule of Sufficient Reason or for the
specific laws of science, particular necessary rules? The former is the
alternative chosen for our minimal interpretations but as already men-
tioned, Kant speaks sometimes in a manner which suggests that by neces-
sary rule he means particular causal laws:

A191-BZ36:

. . appearance can be represented as an object dis-
tinct from them2 only if it stands under a rule which
distinguishes it from every other apprehension and
necessitates some one particular mode of connection
of the manifold.

Kant also speaks often as though he is only arguing for the

Rule of Sufficient REason or the general Law of Causality:

BZ46-A201:

This rule, by which we determine something according to
succession of time, is that the condition under which an
event invariably and necessarily follows is to be found
in what precedes the event. The principle of sufficient
reason is thus the ground of possible experience.

 

1Bird calls attention to this phrasing, by citing Kant at
8240, 263 and 289.

2"Them" refers to the representations of apprehension.

123

BZ47-A202:

The principle of the causal relation in the sequence of
appearance is therefbre.also valid of all objects of
experience . . . as being itself the ground of the
possibility of such experience.

If one bases his interpretation of Kant's conclusion on the
latter passages, the minimal interpretation presented herein will be
adopted. Under this interpretation, Kant's argument is a deductively
valid one since the premises imply the general law of causality. Events
are irreversibly ordered sequences of perceptions, such that every
happening (effect) presupposes a cause. This conclusion, however, is
a much weaker claim than the claim which is based on the former passage
above. To prove that a given happening must have some cause is not the
same as proving* that a given happening will always have the same cause
according to a particular necessary rule.l Lovejoy states the differ-
ence well:

. . a proof of the irreversibility of the sequence of
my perceptions in a single instance of a phenomenon is
not equivalent to a proof of the necessary uniformity

‘of the sequence of my perceptions in repeated instances
of a given kind of phenomenon. . . .

It is indeed true that the distinction just made, between
the sequences of perceptions that are "determined by" or
dependent upon the voluntary focusing of our attention
and those that are determined by or independently given
in the object already assumes the principle of causality.
But . . . the point could not be used to prove what Kant
desires to prove . . . that in that realm of externally
caused or non-volitional changes, all phenomena must

follow one another according to a rule of uniform and
'necessary' connection.2

 

1Lovejoy, op. cit., p. 302.
2Ib‘id.

124

In agreement with Lovejoy, it must be concluded that Kant's
premises are inadequate to demonstrate the latter,1 which perhaps was
in fact the conclusion Kant aimed to support. However, there may be a
way, as Ewing discussed previously in connection with the realist crit—
icisms, to do a little underpinning of Kant that will provide an argu-
ment for the claim that there are; at least in each specific present
situation, particular laws or rules which obtain in some sequences of
representations. Kant provides the first step by proving the general
law of causality. To know that E is an event we must know that, as a
happening, E involves a necessary sequence of states some of which may
not be perceived, e.g., the room grows warm though the heating plant is
out of view. But while perceivers make the distinction between event
and non-event sequences of perceptions, it should be noted that per-
ceivers also make the distinction between events that are similar or
the same qualitatively, and different. When a room becomes warm on
Tuesday just as it did on Monday, the two incidents or happenings are
regarded as qualitatively similar events. In both cases, the effects
may be perceived and compared for similarity. But it is unlikely that
the determinant in each case, i.e., the heater tripping on, is perceived;
yet the claim and the belief that the two cases are instances of the
same qualitative event type is often held; and this indicates the pre-
sumption that the causes are the same since event identification is

constitutive of discernment of its cause and effect as well as precedent

 

1See * on preceding page.

125

perceptions which specify the surrounding circumstances of the event.1
Whether this is an epistemologically wise assumption is a question that
has often been raised; but nevertheless, as Kant might state, much of
experience would be incoherent without it. Unlike Kant, we must stress
that perceptions need not necessarily be related or constitute experi-
ence in only one particular way. ‘But as long as such claims as the
above are made, and experience is unified in this particular manner;

a presupposition of necessary laws is required in the sense of being

a prerequisite without which M (this particular unified manifold) would
not be possible. With this extension of Kant's argument, the stronger
claim for particular necessary rules can be partially supported. But
there remains a problem. 'Though a particular necessary rule must be
presupposed to obtain in each given situation, if a perceiver is to
experience in that situation an event as the same event he experienced
before in a previous situation,2 this is still not the answer that Kant
wished perhaps to give to Hume; for it is not the same statement as:
"There are particular necessary rules which will always be presupposed
past, present and fUture, if events are to be identified as qualita-
tively the same as previous ones." In other words, we have learned

since Kant's mistake’ that the manifold may be unified in the future

 

1One might want to say that A-B on Friday is the same qualita-
tive event as A'-B' on Saturday. But if 0 caused A on Friday and 6
caused A' on Saturday would we want to say that B and B' were the same
happening even if they followed A and A', respectively?

2It will be remembered that we discussed this possibility in
Chapter I.

3Kant'believed for instance that Euclidean postulates would
never be abandoned by physical theorists.

126

in a different manner or by a different means than it is now. Thus,
with this drawback of the broader interpretation which argues for
particular causal laws, let us confine ourselves to the minimal
interpretation which will allow us to gleen an interesting and valid

proof of the general law of causality from Kant's "Second Analogy."

Another objection of non-sequitur

Bennett accuses Kant of a non-sequitur in concluding that every

objective process is totally governed by causal laws. To quote:
"X and Y could not have occurred in the order Y - X"
entails 'Given that X and Y happened non-synchronously,
they had to happen in the order X - Y' but it does not
entail "Given that x happened, Y had to follow." The
rule which forbids a professor to precede the Vice-
Chancellor in a procession does not forbid him to Opt
out of the procession altogether.‘

Bennett's criticism can be construed as a plea for statistical
laws and events covered by them. It is said that the exact position
and momentum of electrons cannot be predicted in quantum theory; and
therefore, not all events, in Bennett's words, are totally governed by
causal laws. It is perhaps true that, since Kant's time, the term
'event' is applied, if not by the common man at least by quantum
physicists, when weaker criteria than those specified in Kant's premises
obtain. But it is by no means a foregone conclusion that this is the
case. Looking at just the premises mentioned, an event would still be

noticed only when a change of state occurs; and it would seem also that

again the order of perceptions would not depend merely on conditions of

 

1Bennett, op. cit., p. 221.

127

the subject such as eye movements. Perhaps an electron does not have

a predictable position and momentum at all times; but surely when a
certain phenomenon occurs, i.e., an event takes place or some change

is observed, the quantum physicist like any other scientist seeks to
explain it in terms of some necessary rule albeit a necessary statis-
tical rule. Here the term"necessary" 5 being used in the Kantian
sense as applicable when the order of this observation in relation to
other preceding observations or perhaps non-observable states was
determined independently of the subject. It is this fact which prompts
the scientist to formulate his statistical laws rather than merely to
take a vacation or have his glasses changed when the unexpected is seen.
This is not to confuse talk of necessity in the Kantian sense with talk
of a determined universe where every state X which is causally connected
with state Y must be followed by Y everytime it (X) occurs. Kant is not
arguing for this claim as Bennett supposes he is; but rather he is say-
ing that if and when state Y does occur and we as subjects cannot manip-
ulate the order of its occurrence, then our perception is of an effect
that has necessarily followed some cause X.l Thus most clearly Kant
should have called his proof the proof of the general Law of Effectual-
ity rather than the Law of Causality. If X is perceived 100 times and

Y is perceived 50 of those times to follow X through no subject related
perceptual efforts, then there are 50 events. In all 50 events, the

presence of Y followed by X in a determined order.

 

1In certain passages (8247) Kant does speak as though he is
saying Y must follow X if X occurs, but his total argument does not
yield such a claim.

128

This important objection will be discussed more fully in
succeeding chapters; but for the present, consider the way Bella Milmed
sums the matter up:

If the objection is raised today that the phenomena of
quantum mechanics have nevertheless nullified causality
as a universal principle, it should be noted that they
have also, in the same sense nullified continuity as a
universal principle; so that the association between the
two is strengthened rather than weakened. On the other
hand, the subatomic phenomena of quantum mechanics are
not part of our sensory experience, but are introduced
conceptually in order to provide causes for some rather
puzzling observations. Statistical laws and discontinu-
ity are introduced on the conceptual level in order to
maintain universal causality (and, in that sense, con-
tinuity) on the level of experience. Kant's concept of
causality, it would seem, could be revised and expanded
to include explanations of this sort which he could not
have anticipated. . . .‘*

In like manner, Gerd Buchdahl in ”The Kantian Dynamic of Reason"
stresses the point in the following manner:

Confusion is for instance easily created by an expres-
sion like "nature is subject to law." Normally this
denotes our belief in the universal prevalence of the
realm of determinism; or less sweeping, of the possi-
bility (at least in principle) of everything being
subject to natural laws. It is however quite clear
that the argument of the "Second Analogy" does not
support such conclusions. For according to nature
regarded as a collection of objective states of affairs,
it only means that the possibility of each of these
states presupposes the injection of the concept of an
indeterminate causal nexus. However such a situation
might be quite compatible with the absence of a network
of empirical laws or any laws whatsoever. The argument
has certainly nothing to contribute (by way of basic
guarantees) to this latter possibility however much it
may suggest it.2

 

1Bella Milmed, Kant & Current Philosophical Issues, p. 54.
*Many would disagree with the strong words "in order to maintain
universal causality."

2Gerd Buchdahl, "The Kantian 'Dynamic of Reason' with Special
Reference to the Place of Causality in Kant's System," Kant Studies
Today, ed. by Lewis W. Beck, pp. 356-357.

129

Other arguments in the "Second Analogy"

In addition to all the previous criticisms which have been
considered, there are yet two remaining attempts to criticize Kant's
conclusion on the basis of supposedly different arguments, which are
taken to be representative of Kant's thought in the analogy by again:
Jonathan Bennett and W. A. Suchting.

Jonathan Bennett contends that there is an underlying "ordering
argument" in Kant which surfaces at a few obscure passage points. The
ordering argument attempts to establish the category of cause by dealing
with the way in which subjective data themselves are established, i.e.,
with how one knows in what order certain subjective data, including the
occurrence of intuitions, have happened. The explicit "object-process“
argument, which Bennett claims runs through at least the five proofs
that Ewing finds similar, deals with only one aspect of the way in which
subjective data--'I have had such and such intuitions, in such and such
an order'--are brought under objectivity concepts.“1 Bennett bases his
posit of a Kantian subliminal ordering argument on passages such as the
following:

In this case, therefore we must derive the subjective
succession of apprehension from the objective succes-
sion of appearances. Otherwise the order of apprehen-
sion is entirely undetermined, and does not distinguish
one appearance from another. Since the subjective suc-
cession by itself is altogether arbitrary, it does not

prove anything as to the manner in which the manifold
is connected in the objective realm.2

 

1Ewing, Kant's Treatment, p. 223.

?8ennett, op. cit., p. 22l.

130

Bennett thinks that Kant's use of 'arbitrary' indicates that
there is some question as to what even the subjective succession
actually is; and the first sentence of the passage indicates Kant's
view that, to know in what order one's inner states do or did occur,
one must have knowledge of an objective realm. As Bennett puts it:

. . . the whole passage hints at a problem not about the

relation between one's subjective and objective stories

but about the establishment of one's subjective story and

in particular its temporal ordering.1
In explanation, Bennett cites comments in Kant's "First Analogy" about
time, viz., that 'time cannot be perceived in itself.‘ Bennett under-
stands Kant to be saying that the date of an event, which is experienced
is not a perceptible feature of it. Kant says, "I experience it at a
particular time, but I do not perceive that it has the feature of occur-
ring at that time."2 Since the date is not perceptible, the date is not
recollectable either. To recall when X event (the noon whistle)
occurred, one must recall other surrounding events: the face of the
clock, the state of one's stomach, etc. An event's date is thus a
logical construct out of its temporal relations with other events, and
this applies to subjective as well as objective events in the past.

Extending his theory from the recollection of the dates of past
events to the ordering of past events, Bennett quotes the following
passage from Kant:

All empirical knowledge involves the synthesis of the

manifold by the imagination. This synthesis is always
successive, that is, the representations in it are always

 

 

131

sequent upon one another. In the imagination this sequence
is not in any way determined in its order, as to what must
precede and what must follow, and the series of sequent
representations can indifferently be taken either in back-
ward or in forward order.1

When the imagination conjures up past episodes one by one, the
question may well arise as to the order in which they actually occurred.
To recall that X preceded Y, we must be able to appeal to objective
considerations not just as a prerequisite for having any working concept
of the past but in support of that particular recollection.

Bennett can think of three counterexamples to Kant's supposed
suggestion. One involves simply recalling that X preceded Y because Y
followed X very quickly at the time that they occurred. The other
objection is that one may recall that X preceded Y if a continuous
sequence of happenings beginning with X and terminating in Y took place,
or one may recall that X preceded Y by recalling a time when one expe-
rienced Y while recalling one's experience of X. In conclusion, Bennett
remarks that most orderings of the imagination do not occur in any of
the three ways mentioned above and that the ordering argument does bring
out one more way, a fourth, in which the appeal to objectivity is neces-
sarily involved even in our talk about our own inner states.

It is not obvious that the recollection interpretation of past
objective events subjectively can really be put on the passages quoted.

When Kant speaks of the synthesis of the manifold wrought by the
imagination, he is talking of the role of the imagination in producing

 

1I id., p. 226.

132

images which permit the application of the concepts of the categories
to appearances in order that anything at all can be given to the mind
as a representation during present acts of sensation. Thus while
Bennett's argument is interesting, it is doubtful that Kant ever
intended to speak about the problem of recollection when he wrote the
passages quoted above. Aside from this problem, the analysis runs the
risk of confusing a rather simple argument that Kant gives to prove
causality only to take the question beyond the point that is needed.

It would certainly confuse matters if the subjective succession that
Bennett orders consisted of recollections of previous subjective states
of the kind that Kant chiefly talks about when he uses the phrase sub-
jective succession. We would then be trying to recall if at first we
saw the roof of the house or if our eyes were first affixed to the
basement windows. On the other hand, if the proofs of the analogy are
to show that reference to other objective appearances actually enable
us to order recollections of objective or subjective perceptions in a
present subjective sequence, these proofs must still and firstly enable
us to distinguish between those sequences of appearances which are
objective and those which are not. This is, after all, the primary
issue of the "Second Analogy": to demonstrate that only by reference
to necessary rules can we discern those appearances which constitute an
objective sequence. If this can be done, then Bennett's analysis of how
past events can be ordered in a subjective sequence can be carried
through. But this is merely a by-product of a Kantian victory for the
general Law of Causality which must be won before this secondary gain

can even be attempted.

133

The bulk of passages intriguing to Bennett comprise what Ewing
catalogs as the sixth argument in the second edition. The sixth argu-
ment, however, is one of the main-line Kantian arguments for causality
and it is perhaps the argument right before it, the fifth in the second
edition and the fourth in the first edition, which has seemed the most
puzzling and unique to critics. It is this argument which has appeared
to be singular to them and unrelated to Kant's main proof of causality
which runs through all the other arguments.

After challenging the main Kantian thesis on causality, W. A.
Suchting discusses the anomalous fifth argument in the second edition
in Kant's "Second Analogy of Experience." The lines of this argument
have already been quoted; and therefore, only Suchting's comments need
be mentioned at this point. Suchting argues that the passage (Al99-200,
8244-245, p. 225) may be interpreted along the following lines.

a. Necessarily if A and B are the constituents of an

event perception, then A and B occupy successive

places in objective pure time.

b. Necessarily, the relation between A and B reproduces
empirically the relation between the successive

places in pure time that they occupy.

c. Necessarily, successive parts of pure time are
necessarily connected.

d. Therefore, necessarily the relation between A and B
is the relation of necessary succession in time.
But the latter is just the causal relation. There-
fore necessarily, if A and B are the constituents of
an event-perception, then A and B are causally
connected.1

lSuchting, o . cit., PP. 333-334.
.JL_.___

134

Suchting claims, as Ewing does in his writing, that Kant has
only demonstrated a truism to the effect that "successive periods of
time constitute a series in which no term can bear the same relation
to that which precedes it as to that which follows it," circular time
aside "and the relation between predecessor and successor is never
reciprocal." But obviously this determination of subsequent periods by
precedent periods of time is quite different from causal determination.
If the above premises are taken to be an accurate rendition of Kant's
argument, it must be agreed, with Ewing and Suchting, that Kant is at
least guilty of a non-sequitur in this aberrant fifth argument. How-
ever, it is very doubtful that Kant should be interpreted this way when
one but looks carefully at this argument. Kant only begins by talking
of perceptions in general, that the preceding are indeed before the
latter. Then he moves on quickly, still in the first sentence of the
quoted passage, that perceptions as events can take place also only
insofar as the appearances of past time determine all existences in
time, i.e., "determine them according to a rule.“ It is the phrase
'as events' which is crucial. In light of the first four arguments of
the analogy, it should be remembered that Kant has described events in
a technical way and that Suchting's first interpretative premise should
read:

a. Necessarily, if A and B are the constituents of an
event—perception, then A and B occupy successive

determinate places in objective pure time.

Likewise premise (c) should read:

c. Necessarily, successive parts of pure time are

necessarily connected in determined and undeter-
mined sequences.

135

It is only the very first few words of the passage that speak
about perceptions in general, i.e., those successive representations
that are synthesized into the unified manifold. The rest of the passage
speaks about the subclass of perceptions which are thought to be events
or objective successive sequences. When the imagination has synthesized
the manifold, events are known to be events and as such they constitute
a network of successive relations among themselves in time; the succeed-
ing naturally determined to follow the preceding. There is then a two-
fold discussion of determination in this argument: the trivial one
whereby the past, present, and future are determined in respect to each
other to be what they are, viz., past, present, or future, and the
causal determination which a priori gives the condition needed by the
imagination to synthesize all the representations, both non-determined
sequences and determined sequences, into a larger successive and time-
determined manifold of representations. This is much of what Bennett
discussed with regard to his recollection hypothesis, and Ewing sees
this possibility also when he says at the end of his criticism of this
argument that "Continuity as usually understood involves the coherence
of innumerable shorter times in one and the same time, and it is diffi-
cult to see how this continuity could be realised in phenomena without
a causal connection between past and present."1 It is exactly this
simple point, it appears, that Kant decided to mention in this passage
as he did in (A2l0-2ll, 8256). It is not that he spoke of mere succes-
sions of past to present and then thought they represented by their

 

1Bennett, op. cit., p. 75.

On..-

136

trivial sequence determinate nature a causal necessity. But rather
that a continuity of past to present sequence, determinate as it is,
is only possible if we are capable of sizing up perceptions as events
or non-events. This is not an unusual interpretation of Kant if it be
thought of as in some way a causal theory of the nature of time, for
Suchting himself notes that many philosophers including H. Mehlberg,
H. Scholz, G. J. Whitrow, and A. Grunbaum, to name a few, have claimed
that Kant offered exactly this kind of theory in the tradition of
Leibniz.1

There is one other important reason to think that Ewing's and
Suchting's interpretation of Kant, were it correct, indicates that Kant
is not only wrong but also seriously inconsistent. If A is said to
precede B in time; it is a truism in some sense that A determines B to
succeed it. However it seems very unlikely that Kant would claim on
this basis that A causally determines B to occur and succeed it because,
as has been stated previously, Kant never claimed that it was even
usually representation A which can be known to be the cause of repre-
sentation B. On the contrary, Kant thinks that most causes are simul-
taneous with their effects; and thus after the occurrence of perception
A, it is likely that a cause C, if you will occurs which may or may not
be perceived along with perception B. Kant's oft-repeated example of

the ship upstream which is thought to have moved downstream is an event

 

1Suchting gives the references for these philosophers, respec-
tively, as: "Essai sur la theorie causale du tempsr'Studia Philoso hica,
l (1935), 135; "Eine Topologie der Zeit im Kantischen Sinne," Dialectica,
9 (1955), 73; The Natural Philosophy of Time (London, l961), p. 273; and
Philosophical Problems of Space and Timel(London, l963), p. l79.

137

with percepts A and 8. But Kant is not claiming that A caused B, i.e.,
that the ship's being upstream caused the ship to next be downstream.
Rather the cause, perhaps a hidden oarsman, may not be seen at all; or
the cause may be completely unknown. It was merely Kant's contention
that since the representations A and B were not determined in order by
the subject they must have been by something else, viz., a cause. Thus
it is unlikely that Kant would pin causal necessity merely on the
necessity of a past perception A necessarily followed by another, 8,

in virtue of A's preceding 8 in time since Kant never thought that every
or even many preceding perceptions in an event caused the succeeding
perceptions in the event. It would appear likely, in view of all that
has been said presently about the fifth argument of the second edition,
that far from being anomalous it is much like all the other arguments
of the analogy except that it makes one additional point about the con-

tribution of events or causal sequences to the continuity of time.

Summary
After careful study of the "Second Analogy," we maintain that

Kant poses but one central argument concerning the necessary aspect of
causality, which is embodied in the following premises and conclusion:

I. "The apprehension of the manifold of appearance is always
successive. The representations of the parts follow upon
one another" (8234). (Corrected version changes 'always'
to 'sometimes'.) See page lOl.

II. "How things may be in themselves apart from the representations
through which they affect us is entirely outside our sphere of
knowledge" (8235-Al90).

138

III. In spite of I and 11 above, I have to show what sort of a
connection in time belongs to the manifold in the appear-
ances themselves, for we are conscious of an objective
sequence in time, i.e., events (paraphrase of 8235-A190).

IV. "Every apprehension of an event is a perception that
follows upon another perception" (8237-A192).

This premise above follows from:

a. "That something happens, i.e. that something or some
state which did not previously exist comes to be,
cannot be perceived unless it is preceded by an
appearance which does not contain in itself this
state" (8236-Al9l).

b. "For an event which should follow upon an empty
time, that is a coming to be preceded by no state
of things, is as little capable of being apprehended
as empty time itself" (8237-A192).

V. "I also note in an appearance which contains a happening
(the preceding state of the perception we may entitle A,
and the succeeding B) B can be apprehended only as follow-
ing upon A; the perception A cannot follow upon 8 but only
precede it. The order in which the perceptions succeed one
another in apprehension in this instance is determined and
to this apprehension is bound" (8237-Al92). "In the per-
ception of an (event) there is always a rule that makes
the order in which the perceptions in the apprehension
of this appearance follow upon one another a necessary
order. The objective succession will consist in that
order of the manifold of appearance according to which
in conformity with a rule the apprehension of that which
happens follows upon the apprehension of that which
precedes" (823B-Al93).

The meaning of the above statements varies with the acceptance
of any of three major interpretations: the "minimal interpretation"
presented herein and the two broader interpretations posed by Arthur 0.
Lovejoy. The minimal interpretation maintains that Kant argues for the

general Law of Causality, i.e., "Every event presupposes a cause," by

demonstrating that the Law of Causality is a necessary presupposition

139

of perceivers if the distinction between subjective perception-
successions and objective perception-successions (events) is to be
made. Under this minimal interpretation, the argument contends that
perceivers employ the concept of cause in their organization of the
manifold of appearance.

The first broadening of this minimal interpretation is effected
by the view of Lovejoy, who states that Kant confusedly seeks to make
the distinction between imaginary and real events via the presupposition
of causal determination. However, close inspection of the "Second
Analogy" reveals only that Kant consistently uses the term 'objective'
in an unorthodox manner to speak of the distinction between event and
non-event successive sequences of perceptions and we have labelled this
distinction, distinction l. Nowhere does Kant speak of objectivity in
the usual sense of distinguishing between real and apparent events which
we have called distinction 2. However, Ewing's discussion of the real-
ist criticisms of Prichard support the contention that, if Kant had
attempted to argue for distinction 2, his premises could be elaborated
to accommodate this aim. The present writer also suggested that the
presupposition of causal connection is the primary criterion for judging
perceptions to be objective in sense 2 as well as sense l. Only when
this criterion cannot be applied does the puzzled perceiver turn to his
fellows for their assessment--the criterion thought to be first and
foremost by Lovejoy. At any rate, Lovejoy's first broad interpretation
is not acceptable since it does not concur with Kant's text. The second

broad interpretation which Lovejoy discusses is that Kant's conclusion

140

is a statement not merely about the general Law of Causality but of
particular necessary rules which hold uniformly and universally in
nature. This interpretation which is supported by various passages
of the "Second Analogy" involves Kant in a non-sequitur. For this
reason, the second broad interpretation is rejected in favor of the
minimal one, which finds support in the text but does not produce a
non-sequitur.

An informal argument was presented further to show how the
minimal interpretation of the conclusion could be slightly extended
to cover particular necessary rules as they are employed in specific
situations when perceivers identify events as qualitatively the same.
Still this additional argument does not support a conclusive reply to
Hume, that there are necessary rules which apply in every situation,
past, present, and future. There may be situations such as those in
quantum theory where we would not be able or willing to say that certain
present events are the same as some others past because we have not been
able to predict their occurrence except probabilistically.

We see thus far then that there is a true claim GLC (every
event has a cause) which is deductively entailed jointly by five
premises. This claim, though about the empirical world, was not
established by observation of constant uniformities as Hume maintained
but by true empirical premises of a different sort which entailed it.
In that GLC conveys empirical information, it is contingent. In that
it is established deductively and not inductively, it is not probably

true but certainly true. Some might say it is necessarily true. Is it

141

analytic? Is it a priori? Is it contingent and factual? Is it
synthetic? Kant said that GLC was both a priori and synthetic just
as our claims concerning it might suggest. But in view of a long
held empiricist tradition in philosophy, we are with Kant guilty of

a contradiction in terms. Thus we shall begin now in Chapter III to
explicate some of these terms. We shall examine what may be meant by
true necessarily. Hopefully our exposition will enable us to decide
if synthetic analytic truths can be necessary and if they can also

express facts about the empirical world.

CHAPTER III

THE ANALYTIC AND SYNTHETIC DISTINCTION

A Classical Distinction and Objections

Historical note

Prior to Kant's writing in the eighteenth century, there was
present in the philosophical works of both the rationalists and empir-
icists a division of truths into logical truths and contingent truths.1
The former, called "truths of reason" by the rationalist Leibniz and
"relations of ideas" by the empiricist Hume, expressed necessary and
universal truths which could be established by the law of contradiction.
Both Hume and Leibniz thought that all mathematical truths were of this
kind. Contrasted with these a priori truths were contingent truths
which Hume and Leibniz, respectively, called "truths of fact" and
"matters of fact.“2 These contingent truths were learned from expe-
rience and as such were a posteriori.

Not until Kant did anyone challenge the exhaustiveness of this
classification of truths. It was his belief that mathematics, natural
science and metaphysics contained judgments that were a priori and at

the same time not truths of reason based on non-contradiction. These

 

1Lewis W. Beck, "Editor's Introduction," Prolegomena, p. xiv.

2Ibid., p. xv.

142

143

were truths which Kant called a priori and synthetic. It was Kant's
task in the Critique of Pure Reason to discover how synthetic a priori
truths were possible. As was mentioned at the end of Chapter II, the
principle of causation which Kant sought to establish by his argument
in the "Second Analogy" was considered by Kant to be a prime example of
an a priori synthetic truth. As such, the law of causality is only one
of the laws given by the understanding which is necessary if there is to
be any objective experience.1 These laws of the understanding are not
derived from experience but are a priori principles of the understanding
which apply to all our experience. It is thus an a priori law that all
changes of phenomena occur according to the rule of necessary connection
of cause and effect. But while we know a priori that every event in
experience does have a cause, knowledge of what particular cause is
operating can be gotten only by experience.2

Kant sought to establish synthetic a priori judgments by showing
that some purely rational judgments, i.e., a priori truths, could be
regarded as synthetic. He found his first examples in mathematics which
he claimed produced its concepts by intuition.3 However, Kant faced a
dilemma in his last precritical treatise, for while he saw that space,
the foundation of all geometrical forms, must be itself a primitive

intuition he also declared that space had a "reality proper to itself."

 

1Ibid., p. xvi.
2Ibid., p. xvii.

3Kuno Fischer, A Commentary on Kant's Critique of the Pure
Reason, p. 30.

144

If space had a reality proper to itself it must be given to the mind
from without and thus geomatry as well as mathematics generally must

be empirically given and therefore not a priori.1 In his advance from
l768 to 1770 Kant was moved to regard space as a pure intuition, a form
of the pure reason, and this gave Kant the hithertofore missing reason
needed to explain how mathematical truths could at once be both syn-
thetic and a priori.2 This was the development whereby Kant broke with

the position of Hume and began his Critigue.

Kant's analytic-synthetic distinction

In his Critique of Pure Reason, Kant describes the distinction
between analytic and synthetic judgments in several ways. In the first
characterization which he gives of his distinction at (810) it appears
that Kant is discussing only judgments of a subject-predicate form. He
says:

Either the predicate 8 belongs to the subject A, as some-
thing which is (covertly) contained in this concept A; or
8 lies outside the concept A, although it does indeed
stand in connection with it. In the one case I entitle
the judgment analytic, in the other synthetic.

At the line beginning at A7 in the first edition Kant gives a
second characterization of analytic truths by saying that analytic
judgments (affirmative) are "those in which the connection of the
predicate with the subject is thought through identity" while synthetic

judgments are not. Kant suggests a third characterization in the second

 

 

 

145

edition by saying that synthetic statements can be thought of as
ampliative since the subject concept does not yield the predicate by
any extraction of analysis. Analytic statements add nothing to the
subject through the predicate but merely break the predicate up into
constituent concepts that have all along been thought in it. Thus
analytic judgments can be considered explicative.

Seemingly unsatisfied that he has clarified his distinction,
Kant discusses it yet in three more ways. At Bll of the second edition
Kant reflects on the connection between analytic and synthetic judgments
and experience. He says:

Judgments of experience as such are one and all synthetic.
For it would be absurd to found an analytic judgment on
experience. Since in framing the judgment, I must not

go outside my concept, there is no need to appeal to the
testimony of experience in its support. For, befbre
appealing to experience, I have already in the concept

of body all the conditions required for my judgment1

(the judgment that bodies are extended).

There is no need to appeal to experience in the case of analytic
statements for as Kant goes on to say "I have only to extract from it
(a subject concept like body) in accordance with the principle of con-
tradiction the required predicate and in so doing can at the same time
become conscious of the necessity (necessary truth) of the judgment."2
The necessity of these judgments is thus a fourth characterization of

of analytic judgments. Analytic judgments are necessary in that they

 

1Immanuel Kant, Critigue, p. 49.
2Ib'ld.

146

accord1 with the principle of non-contradiction, i.e., the denial of
analytic judgments yields a contradiction and this resulting contra-
diction might be considered yet a fifth characterization of analytic
judgments.

The sixth and last description which Kant gives of these judg-
ments is that they are useful for obtaining "that clearness in the
concepts which is requisite for such a sure and wide synthesis as will
lead to a genuinely new addition to all previous knowledge" while
synthetic statements are the ampliative judgments which actually
extend knowledge (Bll, l4).

The number of Kantian criteria

We have just mentioned six possible criteria given by Kant to
bifurcate the class of S-P judgments into those that are analytic and
those that are synthetic. However, the sixthvexplicative description
already appears very similar to the third non-ampliative and it may
well be that Kant's discussion of the analytic-synthetic distinction
(referred to from now on as the A-S distinction) presents just one or
two criteria which Kant attempts to clarify in several differing ways.

In his article "Analyticity and Grammar," Newton Garver also
lists these six basic formulations in Kant of the A-S distinction.
Garver claims that while there is no indication that Kant regards these

formulations as alternatives no one can accept Kant's apparent view that

 

1At Bl89-l91 Kant says that the principle of contradiction is
"the highest principle of all analytic judgments."

147

all six formulations are equivalent without careful study.1 Rather

as Beck notes in his article in Kantstudien, "we can discern two
criteria for analytic judgment" and that "Kant, in apparent disregard
of their difference, uses first one and then the other as it suits his
purpose."2 Garver agrees with Beck's assessment and states that the
first, second and fifth of Kant's formulations seem to be based on the
logical form of a judgment, while the third, fourth and sixth formula-
tions seem to be based on the way in which the judgment is related to
the human beings who make the judgment. In an attempt to clarify which
formulations may actually be criteria in Kant's discussion let us con-
sider each of Kant's six descriptions of the analytic and the synthetic
in greater depth.

Kant's containment criterion

The first formulation of Kant's analytic-synthetic distinction
which we have presented has been commonly called Kant's containment
criterion for the A-S distinction. There are three chief objections
which have been raised in opposition to the containment criterion as
Kant presented it. The first objection which we shall call the vague-
ness objection, is that Kant's containment criterion is vague to the
point of being almost metaphorical. The second form objection is that
-Kant's containment criterion applies only to judgments of S-P form and
the third objection, the variability objection is a criticism of the

 

1Newton Garver, "Analyticity and Grammar," Kant Studies Today,
ed. by Lewis Beck White, p. 258.

2P. 171 cited by Garver, ibid.

148

fact that Kant's distinction applies to judgments rather than
statements or sentences.
The vagueness objection to Kant's
containment criterion
Newton Garver in his discussion of the vagueness objection
dwells on the following words of Kant-~“the predicate 8 belongs to
the subject A, as something which is (covertly)l contained in this
subject A" (810). Garver notes that either a judgment is analytic
in this manner or "8 lies outside the concept A, although it does
indeed stand in connection with it" (BlO).2 But as Garver says:
We are all perfectly familiar with the idea that some-
thing may be contained in something else, where the
alternative is that the first thing lies outside the
second. . . . The difficulty which we have in understand-
ing Kant's first explanation of analyticity is the diffi-
culty of applying this familiar concept of containment in
an unfamiliar context.3
Garver elaborates on the above by stating that concepts cannot
be literally contained in one another in the same sense in which paper
balls may be contained in or lie outside of a basket. Thus Kant's use
of "contain“ must have been a metaphorical one.
Another discussion of the vagueness objection can be found in
an article entitled "On the Distinction Between the Analytic and the
Synthetic" by John Wild and J. L. Cobitz. Lewis White Beck in comment

upon the discussion of these authors cites what they give as a rendition

 

1These parentheses are Kant's as translated by K. Smith.

2Ibi ., p. 249.

a:
0—0
0'
_a.
'0. '0.

149

of Kant's logical definition of "analytic," viz., "A proposition is
analytic if the subject in some sense includes the predicate" or "the
predicate 8 belongs to the subject A as something that is contained
covertly in the concept A."1 Beck objects that the adoption of a
spatial connotation for the word "includes" by Wild and Cobitz enables
them unjustifiably to reduce Kant's account to nonsense. Beck contends:

Take the proposition, "S includes P," which is ex_

h othesi analytic. In Euler circles, S must be the

larger, for the whole is greater than the part. But

when Kant says, “S includes P," he means that P is

included in the concept of S and not that all the memb

bers of P are included among the members of S. Kant is

using the word "includes" in an intensional sense. He

means that P is the genus of 5, its intension is less

than the intension of S, and its intension can be found

by the analysis of the intension of 5.2

Beck is quick to point out that Kant has offered in addition

to his containment criterion, the law of non-contradiction in the
manner of Leibniz as the chief underpinning of claims that judgments
are analytical. Once the analyticity of a statement is judged on the
basis of whether its contradictory is self-contradictory the intensional
sense of the "includes" in Kant's containment criterion is clear, for
as Beck points out "there is no contradiction in saying "S does not
include P," even if it is true that S does include P, provided that the
relation of inclusion is extensional.3 Self-contradiction arises only

if P is included in the concept of S, as when P is the genus of S.

 

1Lewis White Beck, Studies in the Philosophy of Kant, pp. 99-lOO.
2Ibid., p. 100.
3Ib‘ld.

150

P's intension would be less than that of S but diagrammed by use of
Euler circles the P circle would be larger.1 It is in this way that
the spatial analogy of writers such as Cobitz and Wild makes the
smaller seem to include the larger and thus makes the Kantian usage
appear paradoxical.

However, even if Kant meant "contain" in the intensional sense
of the meaning of a concept, as Beck suggests, there is still another
problem for the containment criterion. It is one that Garver mentions
in his article. Generalizing from a physical interpretation of "con-
tain," Garver states that if a crumbled piece of paper thrown toward
a waste basket gets hooked on the top edge of the basket and we are in
a quandry as to whether we can correctly say that the paper is contained
in the basket it follows that "Kant must have assumed that concepts all
have sharp boundaries," so that one can always say definitively whether
one concept is "contained in" another or not, if containment is to
serve as a definitive criterion of analyticity.2

In general the concept of containment loses its clear
dichotomous character whenever the putative container
either lacks sharp boundaries or has boundaries that
an object can straddle.3

Quite correctly Garver has pointed out a well known problem of

natural language terms, viz., the vague boundaries of the intensional

9‘

1As Gerald Massey has pointed out the P circle and the S circle
may be the same circle and thus the P circle would naturally be at
least as large.

2Garver, op. cit., p. 249.
3Ibid.

151

scope of their terms. Any reconstruction of the Kantian A-S distinction
will have to deal with the meaning of terms (if not Kantian concepts) in
such a way as to avoid if possible this notorious problem of the vague-

ness of intensions.

The form objection

The form objection is discussed extensively by Richard Robinson.
Robinson believes that the restriction of Kant's containment criterion
to judgments of S-P form was a restriction deliberately adopted by Kant.
Robinson bases his belief on the authority of two German scholars who
have assured him that the German text must mean this. In addition
Narnock has indicated to Robinson that Kant's assertion that "exists"
is not a predicate is proof of Kant's awareness that not every judgment
is of the S-P form.

Garver discusses Robinson's Opinion and concludes that if
Robinson is correct about Kant's intent in drawing the lines of his
A-S distinction then it may be that "the theorems of logic, relational
propositions, and simple existential propositions were not intended by
Kant to be considered within the scope of this distinction."1 However,
Garver does go on to state that such a suggestion, broadly conceived,
has little to recommend it and is rebutted easily. Existential state-
ments such as "God exists" or "There are aardvarks," can be regarded
as having a logical predicate even though Garver recognizes that Kant

claimed that "'being' is obviously not a real predicate" (8626). More

 

1Ibid., p. 246.

152

explicitly even, Kant himself argues that existential judgments are
all synthetic (8625-626). With regard to relational judgments and the
theorems of logic Garver adds:

Similarly, a relational judgment can be looked upon as

having a subject and a predicate, even though such an

analysis of it is not final and may not be wholly satis-

factory. Theorems of logic are perhaps more difficult to

fit into the subject-predicate mold, but it is in any case

implausible to regard them as judgments in the Kantian

sense.1

0n the other hand Garver believes that Kant's explanation of

analyticity is tailored to fit only one species of judgment within the
classification of judgments that Kant gives elsewhere. At (895), Kant
divides judgments into categorical, hypothetical and disjunctive. The
first type of judgment deals only with two concepts, the second with
two judgments and the third with several judgments in their relation
to each other (898). The containment criterion only applies to the
first type of judgment and thus many of the tautologies commonly thought
to be paradigm cases of analyticity by moderns would not be so regarded
by Kant. By way of example Garver mentions "If the Mekong is longer
than the Danube and the Yangtze is longer than the Mekong, then the
Yangtze is longer than the Danube" which would fall into the Kantian
class of hypothetical judgments. He goes on to note also that the
valid formulae of modern symbolic logic which probably cannot be

regarded as expressing judgments at all would be excluded anyway from

the range of Kant's distinction between the analytic and the synthetic.

 

1Ibid., p. 247.

153

This last point brings us well into the variability objection but
befbre discussing it we must recognize as Robinson ably points out
that had Kant merely proposed Leibniz' distinction between statements
whose denial is self-contradictory and those whose denial is not, he
would have given a criterion "which correctly and usefully divides the
whole class of judgments."1 Robinson states:

He (Kant) would then have said: "Leibniz' distinction

between those statements whose denial is self-contradictory

and the others is clear and good; but Leibniz was mistaken

in supposing it to be the distinction between necessary

and contingent statements; it is really a distinction of

its own . . . the distinction between analytic and syn-

thetic statements; and it is not the case that all

necessary statements are analytic; on the contrary,

some of them are synthetic.2

Using only the Leibniz criterion, Kant could have made all the

important points of the Critique of Pure Reason while avoiding the
disadvantages expressed in at least the vagueness and form objections
claims Robinson. It is interesting to note that once again some of the
important objections to Kant's containment criterion can be defeated by
the introduction of Leibniz' criterion of the law of contradiction into
the containment criterion formulation. Previously, Beck introduced
Leibniz' criterion into the discussion of formulation I (the containment
criterion) in order to clarify Kant's use of the term "contain" and now
Robinson resorts to the Leibnizian formulation to broaden the scope of
the containment criterion so that it will extend over the class of all

judgments rather than just those of an S-P form. It may be fair to say

 

1Richard Robinson, "Necessary Pr0positions," Mind, LXVII,
No. 267 (July l958), 29?.

2Ibid.

154

just in light of our discussion so far that Kant has not presented six
criteria for his A-S distinction or even two as Beck suggests but really
only one criterion which he states in several ways, each formulation
sharpening the cutting edge of his criterion a little bit more. This
view will be adapted presently as a working hypothesis until enough
ground is broken for our proposed reconstruction of Kant's A-S dis-

tinction.

The variability objection

Kant's A-S distinction is a division of judgments and not of
propositions, statements or sentences. However, a given judgment may
be considered at one time analytic and at another synthetic depending
on which statement is being scrutinized in cases where several sen-
tences express the same judgment. To express the judgment that a given
day is cloudy, we might say "This day is cloudy" or "This cloudy day is
cloudy." The first statement indicates a synthetic judgment while the
second indicates an analytic judgment. The meaning of terms in natural
languages also change and an analytic sentence today may again not be
analytic tomorrow. Another aspect of variability is rightly noted by
Jonathan Bennett. As he says one and the same sentence may express more
than one judgment. By way of example he mentions the sentence, "the
judgment that what a man voluntarily does is always what he wants to do"

does not uniquely refer to a single judgment."1 One of the judgments to

 

lBennett, op. cit., p. 6.

155

which it refers can be said to be true by means of the meanings attached
to the words in it and the other judgment can be said equally prOperly
to say something which is simply false. As Bennett notes:

Kant would no doubt reply that what he calls "analytic“

are not sentences but judgments, and that a judgment is

indeed either always analytic or always synthetic. . . .

But . . . if "judgment" is to be used like that, then

expressions of the form "The judgment that . . ." will

refer unambiguously to a single judgment only if the

sentence in the blank admits of only one normal

construction.‘

It would seem in light of the variability objection that the A—S
distinction as pr0posed by Kant will fail its purpose as long as it is
thought to be applicable to judgments instead of the class of individual
statements which express judgments.~ This then is another point that we
shall consider in framing our reconstructed A-S distinction.

In addition to the above objection, Bennett discusses what we
call the oversight objection. 'Analytic judgments are understood as
those which are true by virtue of concepts or whatever. Bennett asks
then if on Kant's account self-contradictory judgments are to be con-
sidered synthetic. In answer he claims:

If we take Kant literally, they are: in its normal
meaning, the sentence "All squares are circular" is
taken to "add to the concept of the subject a predicate
which has not been in any wise thought in it, and which
no analysis could possibly extract from it," which is
Kant's formula for a synthetic judgment.2
Bennett suggests that while this Kantian oversight is a minor

one, which can be easily remedied, it indicates the strong leaning of

 

 

156

Kant to state his analytic-synthetic distinction in psychological terms.
He believes that Kant wanted a classification of judgments which might
be judged or thought to be true and Kant tends to 'assume that no one
could think to be true something which was in fact false, by virtue of
the concepts involved.‘
Kant's second formulation of his A-S
distinction
According to Kant's second identity criterion of his A-S dis-

tinction, "analytic judgments (affirmative) are therefore those in which
the connection of the predicate with the subject is thought through
identity" (BlO). Garver once again concentrates on Kant's odd language.
While there is no ordinary idiom according to which we speak of thinking
something through something else, Garver thinks that Kant's odd phrase
“makes it clear that an analytic proposition is not the same as an
identical pr0position, although there is a very intimate connection
between the two."1 Garver quotes from Kant's essay on the progress of
metaphysics since the time of Leibniz and Wolff in order to clarify
Kant's position:

For example, that every body is extended. If one wished

to call such a judgment "identical," one would invite

confusion; for judgments of that sort contribute nothing

to the elucidation of concepts, which must be the aim of

all judgments and hence are said to be empty. For example,

that every body is a bodily (in another word, material)

substance. Analytic judgments are indeed based on identity

and can be resolved into it; but they cannot be identical
since they require analysis and thereby contribute to the

 

lGarver, op. cit., p. 250.

157

clarification of concepts which would not be done at all
if they were identical idem per idem.1

Despite the above quotation, Garver still puzzles as to exactly
how analytical judgments are to be resolved into identical propositions.
He suggests, as Parkinson2 and Beck3 before him, that Kant may mean that
an analytical judgment is one which depends upon our "acknowledging that
the concept of a body is identical with the conceptual combination of a
number of elements, one of which is the concept of being extended.“ In
the symbolic form proposed by Beck and Parkinson the form of an analytic
judgment would be--All A = BX are 8 where X is a concept or a set of
concepts which joins with B in a conceptual complex identical with A.
Garver thinks that the foregoing symbolic representation of analytical
judgments is interesting in that it illustrates how analytic judgments
depend upon being thought through identiy since in the symbolic formu-
lation "the subject concept must be conceived as identical with a
conceptual complex of which the predicate is one component."5 The
second point of interest to Garver is that the symbolic formulation

makes explicit the connection between the first containment criterion

 

1Immanual Kant, "Preisschrift uber die Fortschritte der
Metaphysik," Kant's gesammelts Schriften, XX, 253—332, cited by
Garver, p. 2582

2G. H. R. Parkinson, "Necessary Propositions," Mind, LXIX
(l960), cited by Garver, p. 251.

3L. W. Beck, Studies. PP. 74ff, cited by Garver, p. 251.
I“Garver, op. cit., p. 25l.
5Ib‘ld.

158

and the second identity formulation. 'The view that these two
formulations go hand in hand is held also by Parkinson and it serves
to confirm our working hypothesis that Kant really presented only one
criterion for his A-S distinction. However, there is still an objection
to this second formulation which bears discussion. We will call it the
epistemological objection.
The epistemological objection to Kant's
identity,criterion
Garver raises the epistemological objection in the following

manner:

If I conceive A as identical with BX, the judgment that

all A and B is analytic. But how am I to know whether

I rightly conceive A as identical with BX‘- that is,

whether A is identical with BX?1

The above passage which constitutes the epistemological objec-

tion is the same concern again that terms (concepts) have vague meanings
that was expressed in the vagueness objection with one major difference.
The epistemological objection has more bite than the vagueness objection
for it raises a difficulty even for concepts which are clear cut and for
terms whose meaning may be completely unambiguous. A concept A may be
clearly understood to be identical with a conceptual complex 8X by
everyone but Kant or in this case Garver or whoever might be assessing
a given judgment in order to determine its analyticity. This objection

thus points out an epistemological consideration which may enter discus-

sion of Kant's second formulation.

 

1Ib‘id., p. 252.

159

Kant‘s third formulation the non-ampliative
criterion

According to Kant's third formulation, a judgment is analytic
depending on how we think it. In analytic judgments the subject is
divided into constituent concepts which are always conceived (thought)
as existing within it, although confusedly (Bll). It follows from this
phrasing, as Garver also notes, that the analyticity of a judgment may
vary from person to person depending upon their perhaps unique inten-
tions or perspective. The perspective objection then which might be
raised against this third Kantian formulation is the same as the epis-
temological objection. 'Again any reconstruction of Kant's A-S dis-
tinction must be one which will avoid the psychological trappings of
Kant‘s third formulation which beclouds any discussion of analyticity
almost before it starts.

Kant's fourth formulation the empirical
independence-criterion

Garver finds the same psychological overtones in Kant's fourth
formulation which we call the "empirical independence" criterion or
formulation. Garver understands Kant to be saying that in a synthetic
judgment such as 'all bodies have weight' one must go beyond the subject
concept, body, and resort to empirical evidence "to the effect that
there is a constant connection between the cases where the concept
'body' is applicable and the applicability of the concept weight or
of some scale of weights.“1 But it is difficult to determine the

 

1Ibid., p. 254.

160

difference that there is in the way of knowing the proposition that
"all bodies have weight" and the analytical judgment that "all bodies
are extended." Garver notes that the common view of both British
idealists and contemporary American philosophers is that there is no
difference. "Weight" might be part of one person's concept of 'body'
and 'weight' might not be part of another person's concept of "body."
As Garver says, "Nothing that Kant says makes it clear why these alter-
native perspectives are not legitimate." Analyticity is again relative
to varying personal perspectives as maintained by both the epistemolog-
ical and perspective objections.

Aside from this last objection there is one which applies
perhaps more sweepingly against the empirical independence criterion

which we shall call the ambiguity objection.

The ambjgpityobjection

.Already in the first three explanations of the A-S distinction
Kant's language is at best vague but his fourth formulation provides
almost no clue as to what point he is trying to make. There are two
possible interpretations applicable to his words but on close examina-
tion neither of these interpretations (l) or (2) is acceptable.

When Kant says in his fourth characterization that it is

"absurd to fpprDan analytic judgment on experience," does he mean
(1), that it is absurd that the source of an analytical statement be
empirical, i.e., that an analytical judgment cannot come to mind through
experience, or does Kant mean that though experience may inspire or pro-

vide occasion for one to think of an analytic judgment, one cannot know

161

that such a judgment once it is entertained is true by looking to
experience or empirical fact. Surely it is difficult to adopt inter-
pretation (2) for it would seem from Kant's words "there is no need to
appeal to the testimony of experience in its support," that 'cannot'
is too strong a word to use in connection with establishing the truth
of an analytic statement on empirical testimony. Rather Kant's words
lead one to conclude not that one cannot check the truth of analytic
statements by referring to experience but rather that one need not
resort to experience to do so. It would seem that one could always
check these statements by referring to experience but why bother when,
as already noted, Kant says "I have only to extract from it" (a concept
like body) "in accordance with the principle of contradiction the
required predicate" (having extension . . .) to become aware of the
necessary truth of the judgment" (812).
On the other hand it seems equally unlikely that interpretation

(1) correctly applies to Kant's words for in discussing his usual
example of an analytic statement "A body is extended" he says:

I can apprehend the concept of body analytically through

the characters of extension, impenetrability, figure,

etc. all of which are thought in the concept. Now,

however, looking back on the experience from which I

have derived this concept of body. . . .1

In view of the above passage it is doubtful that Kant could have

meant that analytic judgments could not originate or be derived from

experience even though at other places he talks of not going outside

 

1Kant, op. cit., p. 50. (Edited by Smith.)

162

one's concept in framing, i.e., originating or forming, an analytic
judgment. If a judgment is framed from a concept and the concept is
derived from experience it surely seems that analytic statements such
as "a body is extended“ can be said to originate at least indirectly
in experience. This of course is not to say that analytic judgments
‘mpgp originate or be formed from experience, for this would not be true
of some analytic judgments in the area of mathematics.
Objection to Kant's fifth Leibnizean
fermulation

We have already noted that several writers, Beck, Robinson, and
Garver, have had recourse to Kant's fifth formulation or description
when defending his first (the containment) criterion against criticisms
and thus it appears that the fifth formulation would be an integral
aspect of any reconstruction of Kant's A-S distinction. However, there
is a problem1 reconciling the Leibniz non-contradiction criterion, as
interpreted by moderns, with its interpretation by Kant. Garver notes
what Beck points out in his article "Kant's Theory of Definition," viz.,
that Kant regards analytic judgments as more basic than definitions
and prior to them. Rather than the formal contradiction which arises
logically from definitions and general logical laws in the Fregean
interpretation of Leibniz' criterion, Kant seems to be thinking of a
psychological conceptual contradiction of some sort. It may be, how-

ever, as Garver suggests, that Kant's empirical independence criterion

 

1We shall refer to this problem in the future as the psycho-
logistic objection.

163

can be encompassed by Frege's interpretation. Though logical proof
is a matter of form, it is also a way of knowing, i.e., of ascertaining
the cognitive acceptability of a meaningful string of symbols. As
Frege says:

If, however, it is impossible to give the proof without

making use of truths which are not of a general logical

nature, but belong to the sphere of some special, science,

then the proposition is a synthetic one.1

The Fregean interpretation, if acceptable, actually assimilates

Kant's first, second, fourth and fifth formulations, though Frege's
criterion violates the spirit of Kant's second identity formulation
which was detrimentally psychological. Aside from this, it cannot be
expected that a rational reconstruction should capture every character-
istic of Kant's A-S distinction. This is especially true if the
ommission of some characteristics such as those of Kant's identity
formulation increase the intelligibility of the distinction. Neverthe-
less Garver admonishes that to adopt Frege's criterion is to abandon
more of Kant's intent than merely formulation two. The Fregean analysis
would require the acceptance of definition as basic. It also applies
paradigmatically to logical formulae and tautologies while Kant sought
to disregard tautologies as examples of analytic truths.2 Frege's
account also precludes any assessment of arithmetical expressions as

synthetic a priori and such an assessment was an important part of

Kant's view.

 

1G. Frege, The Foundations of Arithmetic, p. 4.

2Garver, op. cit., p. 26].

164

Kant's last formulation of the A-S
distinction the explication
criterion
It is difficult to regard Kant's last discussion of the A-S

distinction as any kind of formu1ation or near criterion at all. Rather
Kant's discussion at (Bl3) seems more a description of the role of
analytic judgments. It would seem that Kant's discussion of the role
that these judgments play is only an elaborative postscript to his
criterion for determining which judgments are analytic. Once "analytic“
is defined or a criterion is given for deciding which judgments are
analytic then a description of the role these judgments play in dis-
course or thought is useful. However, befbre we dismiss this sixth
formulation which is similar to the third and subject therefore to
similar objections, it is important to consider another objectionl
raised by Garver. Attending to Kant's words at (A8) "the concept which
I already have is . . . made intelligible to me,“ Garver asks the
obvious question, "How can a concept which Kant already has be made
intelligible to him?" As Garver says:

If I have a firm grasp on the concept body--that is, if

I know all the rules of language pertaining to the word

'body'--, then I am in a position to know that the judg-

ment all bodies are extended is analytic. . . . If I have

the concept, it must already be intelligible to me, and

cannot be made intelligible by the analytic judgment

. if I have such a vague and uncertain understanding

of the concept, I can hardly be in a position to warrant
the proffered judgment as truly analytic.2

 

1We shall call this objection the definition objection in later
parts of this chapter.

2Garver, op. cit., p. 260.

165

Garver surmises that Kant may have seen the absurdity of his
explication claim when he dropped the (A8) passage from his second
edition. This definition objection is important, for it indicates
that Kant's view that analyticity is prior to definitions may be
untenable. Having a concept already can only sensibly be interpreted
to mean that the person P who has the concept must have a definition
(public or private/implicit or explicit) for the concept before he can
assert that judgment J1 in which the concept serves as a subject is
analytic. If the definition is a public one then Jl will be analytic
for others as well as person P. If the definition is private then Jl
nay be analytic only for person P. If the definition is implicit,
person P may be assessing Jl as analytic without attending to the
definition which he has given to the subject concept of J1. If the
definition is explicit then person P will assess J1 as analytic by
consciously referring to his definition of the subject concept. If
the definition objection points to the above as it does appear to, then
a Frege type explication of analyticity may be as faithful to Kant's
criterion for the A-S distinction as any consistent and correct
formulation can be.

Garver goes on to maintain that the definition objection can be
avoided if the explicative function of analytic judgments is pertinent
not to any person P who might know that J1 is analytic but rather to a
listener who, unlike the speaker P, does not "have the subject concept
already," i.e., does not know the definition of the subject term. The

speaker P might utter an analytic statement in order to clarify for the

166

listener the subject concept of J1. In this way analytic judgments
would serve an explicative function. Garver supposes that when Kant
spoke of analytical judgments as explicative, he must have had a
scientific context in mind. Garver points out that "analytic preposi-
tions are important and necessary in such contexts because they enable
scientists to explain to one another how they understand the concepts
and terms which they use."1 In this chapter we are chiefly concerned
with analytic statements in the context of science and the role of
analytic statements described by Kant's sixth formulation as interpreted
by Garver is especially plausible in this context. Therefore, in our
reconstruction of Kant's A-S distinction we shall endeavor to fOrmulate
a criterion for analyticity which will be compatible with the explica-
tive function of analytical statements especially within scientific

discourse.

Directions of a reconstruction

Review of the foregoing major objections to Kant's distinction
between analytic and synthetic judgments reveals much about the direc-
tion that any proposed explication of the distinction must take. First
it is clear from the variability objection that the distinction must be
one made among sentences when they are used as statements in various
contexts and the distinction cannot be one which divides judgments.
In discussion from this point forward we shall not speak of analytic

or synthetic propositions as many writers do since the term 'proposition'

 

1Ibid., p. 265.

167

is a catchall word which sometimes can mean things as diverse as states
of affairs and the statements made by Speakers in describing states of
affairs. Rather this paper will discuss analyticity as a property of
certain statements and by statements we shall mean declarative sentences
not considered as sign types but as spoken or written sign tokens which
are always bound to a particular context of liguistic activity.1

By considering analyticity to be a property of individual
concrete linguistic sign tokens,2 i.e., particular statements, there
can be no problem of one judgment being both analytic and synthetic on
differing occasions depending on the particular sentence that is chosen
to express it by way of declarative statement on varying occasions. No
particular statement or sentence token will ever be said to be sometimes
analytic and sometimes synthetic when used in varying contexts because
sentence tokens cannot even be used in varying contexts. A particular
statement qua sentence token can be written or spoken only once. Thus
the sentence type "A planet is a heavenly body" will yield a different
particular statement (token) every time it is uttered or written. It is
possible, therefore, that the sentence type "A planet is a heavenly body"
is at one time analytic and at another time synthetic but no particular
statement or sentence token ST1 "A planet is a heavenly body" spoken
or written on a given occasion is sometimes analytic or synthetic for

this statement ST1 only occurs once. If the exact same string of

 

1For discussion of this distinction see Henry S. Leonard
Principles of Right Reason. PP. lSl-l7l.

zLeonard considers these as statement occurrences, but statement
«occurrences are defined by him similarly to sign tokens.

168

symbols as in ST1 is spoken or written on another occasion it will not
constitute a case of repeating sentence token ST1 but rather it will
constitute a different unique sentence token 5T1. Our first point
then is that analyticity as a property of a statement token will be
one which always excludes syntheticity as a property of a given sen-
tence token, e.g., statement STI. The domain of discourse then for
our pr0posed explication will be the class of particular statements

or sentence tokens, written or spoken.
A Distinction Reconstructed

The meaning criterion--our proposed
reconstruction

 

The objections to Kant's view already discussed were chiefly
aimed at his seemingly metaphorical containment criterion of analyt-
icity. On the assumption that Kant gave more than one criterion for
analyticity, it might seem reasonable prima facie to just discard the
containment criterion as Richard Robinson has suggested and adopt the
Leibniz' criterion that statements are analytic when their denial
involves self-contradiction. However the previous discussion of the
interdependence of formulations I and II reveals an interesting and
important aspect of Kant's analytic-synthetic distinction which must
be reflected in any adequate reconstruction of his distinction.

Beck offered reply to the containment criterion objections of
Wild and Cobitz by proposing an intelligible interpretation of "includes"
as this term is found in Kant's statement of the containment criterion.
Beck suggested that Kant wished to take account of the intensional mean-

ing of terms (concepts in Kantian terminology). This Kantian intention

169

may have been noted also by some of the many writers who have adopted
a modern criterion of analyticity not to be found explicitly anywhere
in Kant. It is, however, implicitly suggested by Kant's containment
criterion. This revised criterion which we shall refer to as the
meaning criterion allows that a statement is analytic iff it is true
or false by reference to the meanings of the terms within it.1

The containment criterion, a precursor

of the meafiihg_criterion

If we look more closely at Kant's statement and use of his
containment criterion of analyticity it will become clear not only
that Kant's intention in drawing the distinction the way he does is
accurately reflected in the more modern meaning criterion but also that
the modern meaning criterion can avoid the pitfalls of psychologism
which so thoroughly pervade Kant's discussion of his containment
criterion.

By recourse to his containment criterion, Kant claimed that
the statements of mathematics such as "7-+5==12" are synthetic. Many
modern authors, in particular Ayer, have held just the opposite view
that such mathematical statements are analytic.

These writers notwithstanding, there can be some understanding
of why or how Kant could claim such statements to be synthetic if one

notes the role of the containment criterion as Kant used it. This role

 

1Previous meaning criteria have been proposed which are usually
formulated as--A statement is analytic iff it is true in virtue of the
meanings of its terms. Our criterion is a revision of these previous
ones.

 

170

is illustrated in Beck's answer to the critical question of Wild and
Cobitz "In what sense is the subject term '7-+5' in the proposition
7-+5==l2 logically wider than the predicate term?"

The answer is, in no sense. The judgment is, according

to Kant, synthetical though necessary. . . . Its inten-

sion (that of the predicate term) is different from that

of the subject, as they (Wild and Cobitz) admit, though

12 is not (as might be expected from my foregoing argu-

ment) the genus of which 7-+5 is a species. To define

a synthetic judgment it is not necessary to stipulate

that P shall include S; it suffices merely to stipulate

that S does not include P. Extensionally, 74-5 and l2

are equivalent, but the intensions of the two terms are

different, and that of one is not included in that of

the other.1

If it was Kant's aim via his containment criterion to make

the point which Beck describes then it would seem that adoption of the
modern meaning criterion which we will advocate herein will compel one
to agree with Kant that mathematical statements such as "7+-5==12" are
synthetic. This follows from the previous assessment of the meaning
criterion, that it reflects Kant's aim in suggesting his containment
criterion. However it seemed very much the case that when Frege argued
that such statements in mathematics are analytical he may have been
using a meaning-type criterion to argue for this conclusion. It is not
obvious, though Beck may think so, that the intensions of "7-+5“ and
“l2" are different if for no other reason than that the complete mean-
ings of the symbols "7," "+," "5," "=" (which Beck omits) and "l2" can
be given by definitions within a mathematical system such that the truth

of the statements “7-+5==12" could be determined completely by recourse

 

1Beck, op. cit., p. lOl.

 

to the meanir

toes this con

containment a
undesirable d
arguing that

that the form
aean only the
but rather a i
something more
self-evidence

an effort to c
actually thoug
ter.-cept, Thus
analytic to om

thinking about

e;istemologica‘
has been great‘

(7 *
.dnL'S adOptio,

 

 

 

171

to the meaning of "7," "+." “5," "=," and "l2" as given in the system.
Does this conclusion then indicate a difference of purpose between the
containment and the meaning criteria? We think not, at least not any
undesirable difference. 'It seems that Kant, if Beck is correct, was
arguing that the meanings of the subject and predicate terms are such
that the former include the latter and that by “meaning" Kant did not
mean only the extensional brand of meaning employed by some logicians
but rather a concept of intensional meaning. However, Kant also meant
something more. We suspect that Kant, still influenced someWhat by a
self-evidence view of necessity, embraced the containment criterion in
an effort to claim that a judgment (statement) was analytic if a person
actually thought of the predicate concept as included in the subject
concept. Thus it would seem that in Kant's view a judgment might be
analytic to one person but not to another depending on what each was
thinking about the concepts involved. This brings Kant from a public
epistemological endeavor to a private psychological one for which he
has been greatly criticized. Robinson notes much the same emphasis in
Kant's adeption of the containment criterion:

1 have sometimes thought that perhaps Kant changed from

the contradiction-criterion to the containment-criterion

because he felt unequal to the task of showing that

mathematical statements are synthetic by the contradiction-

criterion, but able to show that they are synthetic by the

other criterion . . . (he quotes from Kant) “At first one

would think that the statement '7-+5==l2' was a merely ana-

lytic statement, following from the concept of a sum of

seven and five in accordance with the Law of Contradiction.

But when one looks more closely one finds that the concept

of the sum of 7-+5 contains nothing more than the union of

the two numbers into one, without its being in any way

thought which this number is that includes the two" (Bl5).
This passage has suggested to me the guess that Kant

adapted the
show in (1915
is synthetic
l

The abo
that the nether:
a-school boy W

be that ”79 + 95

'ad gone to a 8<

  
  
  

the inconsisten:
anal tic but the
conclude that a'
if statements wt

"étute relative

 

 

172

adopted the containment-criterion because he could not

show in detail by Leibniz' criterion that mathematics

is synthetic.1

The above quotation from Robinson indicates how it could be

that the mathematical statement "7-+5==l2" might be analytic to every
ex-school boy who had memorized his addition table but how it might not
be that “79i-95==l74" was analytic to the same ex-school boy unless he
had gone to a school which used very long additional tables! To avoid
the inconsistency of stating that the simple addition statements are
analytic but that the more complex are not, Kant may have decided to
conclude that all mathematical statements are synthetic. The number
of statements whose answers are memorized in early childhood are so
minute relative to all the possible addition statements whose answers
we could not possibly have in our mind but would have to compute that
there was hardly any point for expecting them as a special class of
analytic statements. Frege surmises the same in saying:

Kant, obviously, was thinking only of small numbers. So

that for large numbers the formulae would be provable though

for small numbers they are immediately self-evident through

intuition. Yet it is awkward to make a fundamental distinc-

tion between small and large numbers, especially as it would

scarcely be possible to draw any sharp boundary between them.

If the numerical formulae were provable from say, 10 on, we

should ask with justice "Why not from 5 on? or from 2 on?

or from l on?2

The meaning criterion proposed herein differs from the con-

tainment criterion by eliminating its psychological overtones when

 

1Robinson, op. cit., p. 298.
2Frege, op. cit., pp. 6-7.

stated with ins
of the terms" h
*neneanin CH

a-‘steno o ' 3‘

Severa]

 

a; the inserti0
:he need for a

terns constitut
:iain that the

man who hear
Only in princip'
:y reference to
as English) or i
arithmetical sys
inediately or

Me or false bl
ii! that it is 5
Case in mathemai
E'aghatically wi
file or false.

‘l‘Ch when appl“

 

3530!) P 355'

“‘n o
the, i.e., hal

"mt has been el

 

173

stated with insertion of the phrase "by reference to the meaning
of the terms" which shall be unpacked soon in our discussion.
The meaning criterion avoids the form,
epistemology and perspective objections

Several phrases in this proposed meaning criterion are crucial.
By the insertion of the phrase "by reference," we intend to eliminate
the need for a person to know or have in his mind the meaning of the
terms constitutive of a particular statement in order for him eVer to
claim that the statement is analytic. It is only required that the
person who hears or reads such a statement be able to determine not
only in principle but pragmatically that the statement is true or false
by reference to a dictionary (in the case of a natural language such
as English) or by reference to the definitions of an axiomatic or
arithmetical system. Therefore, it is not required that a person know
immediately or in some self-evident way that a sentence is analytically
true or false but only that it is pragmatically possible for him to find
out that it is so even through involved demonstration as might be the
case in mathematics or logic. Our proposed meaning criterion states
pragmatically when any statement has the property of being analytically
true or false. Our new meaning criterion is an epistemological one
which when applied serves as the epistemological justification for any
person P to assert that a statement is or is not analytically true or
false, i.e., has or has not the property of being analytical. This

point has been a critical one in some work on analyticity in logic.

During
Carnap sought ‘
‘emeted calcu
‘r. his article

It was nati
essential 1
evidence we
sentence m
minds. . .
and transfc
nunoer thec
We neithe
This meant
sentences c

Bohnert
mmeet this as
extension of th

3T2cesses .2 Ho

('fither thl 5 pr

yum}

“ed 55 hun

lit

 

 

174

During his earlier period of purely syntactical work, Rudolf

Carnap sought to characterize analyticity in the context of an unin-
terpreted calculus or language. But as Professor Bohnert describes
in his article "Carnap on Definition and Analyticity":

It was natural in this context to take provability as the

essential trait of an analytic sentence. While self-

evidence was to be discarded it seemed that an analytic

sentence ought to be provable by means available to human

minds. . . . Gddel showed that languages with formation

and transformation rules adequate to the construction of

number theory contained "purely logical" sentences which

were neither provable or refutable by finite processes.

This meant that there would be sentences true but unprov-

able if one maintained the law of excluded middle for the

sentences of such a language.1

Bohnert goes on to describe historically that Carnap attempted

to meet this difficulty in his work Logical Syntax of Languagg_by an
extension of the concept of provability which would admit transfinite
processes.2 ‘However, this "gave rise to doubts among intuitionists
whether this provability concept does not go beyond what can be honestly
accepted as humanly provable."3 This aspect of provability or as some
would call it the pragmatics“ of an analyticity criterion is provided
for by our criterion as already stated and we shall find this to be the

case with writers other than Carnap, e.g., Frege and Max Black.

 

1Herbert Bohnert, "Carnap on Definition and Analyticity,"
The Philosophy of Rudolf Carnap, ed. by Schilpp, p. 4l2.

2Allowing deductions based on infinite classes of premises.
3Ibid.., p. 413.

“By "pragmatics" we mean the employment of a criterion, be the
criterion semantic, syntactic or in anotherssense pragmatic.

 

Cr' teri 0n

 

There a
that while a me
the Leibniz CH
their negation
criterion is re
eatic statement

This criter
definition I
provide 9P6
analytic or
Whether

:antradicti on El

is really a moc

is Frege clear‘.

terion of analsI
l
truth is one irl

I

Eiflnition "

It I
er. only the ‘

i""i'aticity

 

13“”195 of the

 

175

The meaning_criterion and the Leibniz‘
criterion

There are some writers such as Hospers who have maintained

that while a meaning criterion for analyticity is a pragmatic one,
the Leibniz criterion of non-contradiction is a criterion which states
only a semantical property which analytic statements have, i.e., that
their negation involves self-contradiction. However, even this Leibniz
criterion is regarded by some philosophers such as Hamlyn as the prag-
matic statement of a semantical decision procedure. He says:

This criterion can scarcely be said to suffice as a

definition of an analytic statement, although it may

provide grounds for saying whether a judgment is

analytic or not.1

Whether one considers the denial of a statement resulting in

contradiction a semantical property or a pragmatic decision procedure
is really a moot point in this present discussion since Liebniz' cri-
terion of analyticity is not the one being directly presented herein.
As Frege clearly pointed out in the field of mathematics, an analytic
truth is one in whose proof one finds only "general logical laws and
definition." Frege showed that proofs in arithmetic were not possible
when only the law of non-contradiction was adopted as a criterion for
analyticity. One also had to refer, as our criterion states, to the
meanings of the terms composing the statement. By the same token,

however, it would not seem that one could justify the claim "Statement

St is analytic" if one's criterion is not a combination of a meaning

 

1D. W. Hamlyn, "Analytic and Synthetic Statements," Encyclopedia
of Philosophy, 1, p. l06.

criterion and i
opposed to be
contradicti on <
ncted that Beci
Kant's containn
introduction 01

Our mee
non-contradicti
‘nvolved in dis
actually have i
ceronission of
analytic statem
sosed should pe
Sitters of math
System of many
3C0sited as a th
36 ascribed to
”ECOUf‘se t0 the

5:.
We can find I

 

0n the I
eat (ina ver.

   
 

we 13” 0f non

  

176

criterion and a non-contradiction criterion.1 This is just what we
supposed to be Kant's criterion--a combination of the containment and
contradiction criterion rather than two separate criteria. We also
noted that Beck was able to give an intelligible interpretation to
Kant's containment criterion, thereby countering its opponents, by
introduction of the law of non-contradiction.

Our meaning criterion does not explicitly mention the law of
non-contradiction or for that matter the law of identity which is also
involved in discussions of analyticity but the meaning criterion does
actually have incorporated within it the Leibniz criterion. However.
our omission of explicit mention of Leibniz proposed characteristic of
analytic statements was purposeful. The meaning criterion here pro-
posed should pertain not only to natural languages and the axiomatic
systems of mathematics, traditional logic and science, but also to
systems of many-valued logic where the law of non-contradiction is not
adopted as a thesis of the system. In such systems, if analyticity can
be ascribed to any statements at all, it will have to be exclusive of
recourse to the law of non-contradiction. In such a many-valued system
if we can find analytic statements we will probably be able to do so by
reference to the meanings of the terms within these statements.

On the other hand, in all the other fields of systematic state-
ment (in a very loose sense, natural language is included here) where

the law of non-contradiction does hold or is adopted as true, it can

 

1See page 149 of this chapter.

be regarded as
tart of their 1
tierefore the I
this law of nm
criterion phra:
Sin." Since ti
Sriterion, we

triy to judgme
390"! 0f mean
Wing which i
an attempt to

Midi dlSCus

W
It may
3 the Phrase
Sm!" which ha
this clarifica
the tenn "Map
”any t
that is rTreant
acceiltd by U

ht

\

1H

\
w

as,

 

2

177

be regarded as another definition within the system that gives terms1
part of their meaning,2 e.g., non-contradiction = df - (A‘-- A) and
therefore the meaning criterion captures the effect of reference to
this law of non-contradiction, for it is incorporated within the
criterion phrase "by reference to the meaning of the terms of statement
STn." Since the Leibniz criterion is implicitly included in our meaning
criterion, we avoid the form objection that Kant's distinction pertains
only to judgments (statements) of S-P form. We shall present our own
theory of meaning which we shall call the metalinguistic theory of
meaning which we shall call the metalinguistic theory of meaning in

an attempt to avoid several other objections to Kant's presentation

already discussed.

The meaning_of ipeaning" that is meant

It may be well at this point to clarify exactly what is meant
by the phrase "by reference to the meaning of the terms of the statement
Stn," which has been under discussion thus far. It is necessary that
this clarification should be made since there is much controversy about
the term "meaning" itself.

Many theories have been advanced in efforts to explicate exactly
what is meant by "meaning" and few if any thus far have generally been
accepted by the majority of philosophers. In keeping with the aims of

this paper we shall be interested only in discussing the term "meaning"

1We use "term" here as we would "word" and thus we consider
words such as "and," "is" etc. to be terms.

2Frege, op. cit., p. 7.

as it is applii
sentences. We
which ever shoa
better choice i
brevity we sha'
later. Howevei
theory of mean'
It has
me" a denote
aid intensiona‘
if one adepts
liiects to Whi
i“ the denotat.
”Harem-a] the
ZhEOl‘y are '1 0H
itwil] SU-ffic
”leaning, vi
w or "is" \
w HWI
:‘emy by Irv

'br

 

 

178

as it is applied to terms and not as it is sometimes applied to
sentences. We have reason to suspect that "meaning" is not a term
which ever should be applied to sentences, "meaningful" being a much
better choice of words in reference to sentences but for sake of
brevity we shall not present reasons for this view in the present
paper. However, it will of course be important to remember that our
theory of meaning does presuppose that the above view is correct.

It has been widely held that terms or words have meaning in
either a denotative sense, a connotative sense, or both. Extensional
and intensional meaning are respective synonyms of the former pair.

If one adopts the view that the meaning of a term consists of the
objects to which the term may be applied then he is considering meaning
in the denotative or extensional sense. Such a view has been called the
referential theory of meaning. The serious problems related to this
theory are now so well known that it is needless to review them here.
It will suffice to recall only the most obvious objection to this theory
of meaning, viz., that there are many words such as "centaur" or "uni-
corn" or "is" which have no denotation but which do have meaning. A
further limitation of any extensional theory of meaning pointed out
clearly by Irving Copi is that it is not extension which determines
intension but rather intension which determines extension. In the
words of Copi:

Any given object has many, many properties and is there-

fore included in the extensions of many, many different

terms. Hence the example mentioned in the denotative
definition of any one term will be just as appropriately

mentioned
terns.l

Copi c
"informative 5
narrower than
the emotive as
the different
sense and he d
iSTement by t
55'“? set of pr
if the item's
tion and so th
tation 0r COnv

It is
reed not be th
50“? philosomi

lion allusmnS

aCOM/9mma1
r
w.

Mystic cor

 

 

 

179

mentioned in the denotative definitions of many other
terms.1

Copi claims that the intension of a term is a part of its
"informative significance," (this is the view of logicians which is
narrower than those who believe that intension or connotation includes
the emotive as well as the descriptive significance of a term).2 Of
the different senses of "connotation," Copi prefers the conventional
sense and he describes conventional connotation as the outcome of
agreement by those who attach the same meaning to a term, to use the
same set of properties in deciding for any object whether it is part
of the term‘s extension or not.- "This agreement establishes a conven-
tion and so this meaning of a term is known as its conventional conno-
tation or conventional intension."3

It is important to note that the agreement to which Copi refers
need not be the outcome of a formal meeting or assembly of some sort.
Some philosophers, Russell for one in Analysis of Mind, have seized
upon allusions to agreement among linguistic performers to argue against
a conventional view of language. Not considering the possibility that
there can be informal conventions which come about gradually in the
history of a people, Russell stresses the implausibility of thinking
that there ever were large formal councils or meetings to determine

linguistic conventions and he then stresses the implausibility of

 

1Irving Copi, Introduction to Logic, p. ll4.
2Ibid., p. 108.
31b1do’ p0 1090

180

regarding natural language as the outgrowth of any agreement or
conventions at all! He says:

It is natural to think of the meaning of a word as

something conventional, either from the point of view

of the individual or from that of the community. . . .

If we trace any Indo-European language back far enough,

we arrive h pathetically (at any rate according to some

authorities at the stage when language consisted only

of the roots out of which subsequent words have grown.

. . . We can hardly suppose a parliament of hitherto

speechless elders meeting together and agreeing to call

a cow a "cow"1 and a wolf a "wolf."2

It would seem that those such as Copi who speak of meaning as

a product of convention never envisioned a parliament called for the
purpose of forming a language any more than Russell can envision it.
What they probably do mean is that when the first person began to make
the same sound repeatedly while pointing to a rock or whatever, his
fellows tacitly agreed with his usage of this sound in succeeding
circumstances when they wished the original speaker or speakers to
notice a rock. '1 'al ‘of-us made up our own word for an object and
persisted to use it despite the differing words used for it by everyone
else there would be little chance of linguistic communication. The
question is not one of the origin of particular sounds coming to be
used by individuals in the sense of a private language but rather of
the informal convention via correction and compliance in inter-
individual communication which must be the basis of any public

language.

 

1We have inserted the use-mention quotes.

2Bertrand Russell, Analysis of Mind, p. 190.

181

Aside from objections against the‘view expressed by Copi that
language is conventional at least in an informal sense, other writers
such as Arthur Pap have disagreed with Copi's contention that exten-
sional meaning is determined by intensional meaning in the case of
simple ideas such as the idea of red or of a standard meter.1 Using
the example sentence, “there are red surfaces" Pap claims:

Such indubitable existential statements would be analytic
in terms of being true by ostensive definition just as
much as statements analytic by verbal definition. Like
"red" if the expression “1 meter" is to have a meaning

at all, one rod in the universe must have the predicated
property by ostensive definition.2

Pap goes on to say that what he is really doing is nothing more
than drawing an obvious inference from what he views as the well-founded
positivistic thesis that all connotative meaning which is ultimately
intelligible must merge into an ostensive definition.3 As Pap notes it
would be unreasonable to postulate that every significant term should
be capable of both ostensive and verbal definition, but it is reasonable
in the case of terms which are not oStensively defined, e.g., many
scientific terms, to expect that these terms will be "capable of being

related by a series of verbal definitions to terms which are ostensively

definable.“

 

1Arthur Pap, "Indubitable Existential Statements," Mind,
l946, p. 238.

182

By verbal definition Pap means those strings of symbols (words)
that give connotative meaning1 and by ostensive definition Pap means the
defining of terms in which meaning may be exhibited by pointing to one
or several instances to which they are applicable. Ostensive defini-
tions in Pap's discussion give a term denotative meaning. As Pap
explains:

In common discourse denotative meaning is as a rule
genetically prior to connotative meaning. Thus the

first and rudimentary method by which the meaning of
words is usually explained to children is the method

of ostensive definition. The intelligibility of abstract
verbal definitions, exhibiting the connotation of a word,
presupposes some intellectual maturity not possessed by
infants.2

The foregoing quotation brings to focus an important point that
we wish to make about the meaning of "meaning." Contrary to Pap's con-
clusion, it is our belief along with some other writers such as Dewey
that ostensive definition is not an instance or type of definition at
all. It is not the meaning of words which is usually explained to
children by the activity of pointing to the objects which they denote
but rather it is the use of words which is explained, or as we prefer
to say, given to children by what has been traditionally called osten-
sive definition. 'It is our view that the meaning is given by definition
and ostensive pointing does not give the meaning of a word but only its
use. Therefore, we do not regard ostensive definition as a type of

definition at all. This claim needs more elaboration since there is

 

1Ibid., p. 239.
2Ibid., p. 243.

183

a philosophical view which identifies “meaning"-and-9use9 by claiming
that the meaning of a word is its use. It was Wittgenstein's directive1
"If you want to find the meaning of x look to the use of X," however,
“looking to the use of X to find the meaning of X" is not the same
thing as saying that "the meaning of x is the use of X" or vice versa.
"Meaning" in our view is a metalinguistic term used to relate
words or word phrases. Therefore, we maintain that when a speaker
says, "'X' means Y" his utterance is more correctly and precisely
written “'X' means 'Y.'" The word "means" functions in the sentence
"'X' means 'Y'" as a metalinguistic term relating the two verbal expres-
sions 'X' and 'Y' which normally occur on the material2 level of a
language, e.g., English, to talk about things. Viewed in this way the
meaning of terms is expressed or given only in definitional statements
and definitional statements are always verbal entities that are used to
talk about other verbal entities. The act of pointing to objects while
trying to teach the names of those objects to a child (usually called
ostensive definition in the literature) is not a definition at all since
pointing is not a verbal statement or utterance but a gesture which
attempts to relate not strictly verbal entities but rather one verbal
entity a common noun and an object. Stephen Pepper describes the so-

called ostensive definition as:

 

1Some writers prior to Wittgenstein made this same suggestion.

2Sellars uses the term "material" in this regard. We shall not
follow his convention in succeeding pages but we shall refer to this
level of language as the object level.

184

Here 5 is the symbol defined, and O is some empirical

fact. The relation of indication between S and O is

presumably a factual reference of some sort. Probably

the relation of indication is at the least an operation.

It is a set of directions for the performance of certain

acts which being carried through reach a certain result.

The symbol S is at one end of the operation and the object

O at the other, and the one "indicates" and is said to

"mean" the other in virtue of the operation.1

Dewey criticizes Pepper for still calling the old (ostensive)
indicating operation as described in the preceding quotation, a defi-
nition for in doing so Pepper shows the "old mixture of word and
object."2 Dewey reflects that John Stuart Mill remarked a hundred
years ago that however unambiguously one can make known who the
particular man is to whom a name belongs by pointing to him such
pointing has not been esteemed one of the modes of definition.3
As already expressed, we agree with philosophers such as Dewey

and Mill in maintaining that ostensive pointing of the sort that is done
when one teaches a young child his first words is not a form of defini-
tion nor does it convey to a child any meaning. However, ostensive
pointing does teach a child the use of his first words. Once the child
matures to the extent that he knows the use of many words, it becomes
possible to teach him the meanings of further new words by means of
what Pap calls verbal definition. When definitions are employed there

is no pointing but only a verbal utterance or written expression that

 

7*

1Stephen Pepper, "The Descriptive Definition," Journal of
Philosophy, XLIII (1946), 30.

2John Dewey and A. F. Bentley, "Definition," Journal of
Philosophy, XLIV (1947), 300.

3Ibid., p. 304.

185

metalinguistically informs the child that the new word or word phrase
has or ought to have the same use as another word or phrase whose use

he has already learned ostensively.1 Definitional statements, which
often employ the term "means," relate words or terms by expressing the
fact that the definiendum and definiens have and should have the same
use and the definiens whose use is known is said to be the meaning of
the definiendum whose use is unknown. 'A meaning (we do not like using
"meaning“ as a subject term but English does not seem to afford a way
around it here) is always a linguistic word or expression that has a
known use. The meaningz of a definiendum is another verbal expression--
the definiens. The meaning of the definiendum is not the use of the
definiens but the definiens itself--a verbal entity! Of course it can
be said that construed in this manner the meaning of the definiendum
depends on use-—the known use of the verbal expression or word that is
the definiens. Thus Wittgenstein's imperative that to find the meaning
of a term one must look to the use of X is somewhat in agreement with
the view here proposed, except that the meaning of 'X' is the expression
'Y' whose use happens to be known and the use of 'Y' is stated to be the
same for future discourse as the use of 'X' by the definitional

statement.

 

1This is not to say that the use of a word is always learned
ostensively. It is well known that children will learn the use of
terms by hearing them used in a given context. They often use the
term correctly in later situations though they do not know the meaning
of the term or what they are saying.

2Usually dictionary entries include several definiens, i.e.,
several senses of a term. When we speak of the meaning of a term we
mean only one of these definiens taken at a time for consideration.

 

186

The metalingyistic theory of meaning
ci rcumverfis criti cisW

Adoption of this metalinguistic theory of meaning curcumvents
many of the common problems usually associated with other meaning
theories. For example as Alston notes:

First, whatever sort of entity is identified with mean-
ing, there will be meaningful expressions corresponding
to which no such entity can be found. Not all words
refer to something and not all have connotation, e.g.,
"is" "although."1

On the present proposed theory in which the sort of entity
identified with meaning is another verbal word or expression which can
easily be found in any dictionary there will not be any meaningful terms
in English for example to which no such entity that we propose as a
meaning can be found to correspond.

Alston focuses attention on the difficult problem that arises
in the meaning theories which treat meaning statements (definitions)
like statements that employ 'is‘ to establish an identity relation
among entities. In illustration he mentions sentences such as:

l. The capital of France is Paris.
2. The conductor of the Philadelphia Orchestra is Eugene Ormandy.
3. The wife of Henry Luce is Clare Booth.2

In the above sentences Alston claims "that we can always "make

explicit what we are saying about a country when we give its capital by

 

1W. P. Alston, "The Quest for Meanings," Mind, LXXII (1963),
p. 79.

zlbi ., p. 80.

187

saying what sort of thing a capital is." This will involve specifying
(l) the general category to which it belongs (city) and (2) the sort of
relation a member of this category must have if it is to be the capital
of that country."1

However, in the case of a definitional statement such as "The
meaning of 'procrastinate' is " it is not generally true
that what follows 'is' in the blank designates anything at all. There-
fore, it is impossible to determine any category to which the meaning
of 'procrastinate' belongs-or how it might be related to 'procrastinate'
in order to be the meaning of the-word.2 ‘Alston states, e.g., that
"The meaning of"auspicious" s favorable" and "The meaning of 'grad-
ually' is bit by bit," but neither meaning statement consists of the
identification of some entity which is related to these words as its
meaning.3 Further in cases where what follows 'is' is a referring
expression "we cannot construe the meaning statement as an assertion
that the entity referred to "by the definiens" is identical with the
meaning of the expression in question," the definiendum. Taking the
example of 'courage' as a definiendum, Alston stresses that though
'courage' refers to a certain trait of character as does 'steadfastness
in the face of danger,‘ as typically found without quotes as a defin-
. iens 'steadfastness in the face of danger' does not refer to that to

which "the meaning of 'courage'" refers. On the present theory this

 

1Ibid., p. 80.
2Ibid., PP. 80-81.
8Ibid., p. 81.

188

problem does not arise: ‘Alston'pointS'ou -the problem more clearly by
juxtaposing'the‘following'sentences-while°using'the‘rule'of'substitution.

l. The meaning of 'courage‘ is steadfastness in the face
of danger.

2. Steadfastness in the face of danger is all too rare in
these times.

3. The meaning of 'courage' is all too rare in these times.

In our theory the identity established by sentence (1) does not
arise. Rather we would write sentence (1) as "The meaning of 'courage'
is 'steadfastness in the face of danger.'" In terms of the identity
established by the revised sentence (la) of our theory, sentence (2)
becomes by substitution of the definiens—-"'Steadfastness in the face
of danger' is all too rare in these times" which reads just as strangely
as sentence (3) in which Alston has substituted the definiendum--"The
meaning of 'courage"is all too rare in these times.

' Another problem, involved in more usual theories of meaning,
is that none of the predicates which can be attached to the meaning of
Fcourage,l i.e., 'easy to grasp,‘ 'rather vague,‘ can sensibly be
coupled with the definiens 'steadfastness in the face of danger.‘1
Thus he claims, "the trait and the meaning can hardly be identical
though the meaning of the word 'courage' is steadfastness in the face
of danger. But again on our present theory the predicates 'easy to
grasp,‘ 'rather vague' which might be attached to the meaning of 'cour-

age' could also be attached to its definiens 'steadfastness in the face

 

1This is when the definiens is written without quotation marks
as in usual theories of meaning.

189

of danger' the linguistic phrase that we are here claiming is the

meaning of the term 'courage.'

Objections to the metalinguistic theony
of’meafiing

The view of meaning presented in our discussion has not been
commonly proposed and has therefore not received much critical attention
in philosophical literature. However, the philosopher Max Black has
commented on this kind of theory in Models and Metaphors. He analyzes
what he calls meaning formulas. In Black's discussion a meaning formula
is “used to stand for any statement of the form, '5 meant (A) by (X),'
in which '5' is replaced by a reference to some person, 'X' by a speci-
fication of some gesture or utterance produced by him on a particular
occasion, and by"A' any expression that stands in the place of 'A';
the accusative of the meaning formula."1 He also counts shorter ver-
sions of the above such as "S meant A" as meaning formulas. With
regard to these meaning formulas Black says:

When a meaning formula is printed, it is customary to use
quotation marks at each end of the accusative, so that we
have "He meant 'Dim the lights'" rather than "He meant
Dim the lights." Now since a common use of quotation marks
is for mentioning the word or expression that appears between
the marks, we might suppose that the accusative of a meaning
formula mentions a verbal expression but this is easily
shown to be wrong.2

Black argues that it is wrong to suppose that the accusative

of a meaning formula is a verbal expression since such formulas would

 

1Max Black, Models and Metaphors, p. 17.
2Ib'ld., p. 19.

190

be exhibiting the mention of an expression in the accusative. If this
were so then it would be apprOpriate to make the mention of the expres-
sion explicit by inserting the words "the expression" at the appropriate
place in the sentence, e.g., the mention of an expression in the state-
ment, “'Dim the lights' consists of three words," is made explicit by
transforming the statement into "the expression, 'Dim the lights,'
consists of three words.“ However, if one inserts the words 'the
expression' into a meaning formula Black contends that the results are
nonsensical. The nonsensical result would be "He meant the expression
'Dim the lights.”1 Black does admit though that if the original mean-
ing formula is replaced by the longer formula, "He meant what would be
meant by saying 'Dim the lights,'" the result does not seem to be
nonsensical since the longer formula clearly is a statement about an
expression and could serve the same purpose as the original explanation.2
Nevertheless Black rejects this alternative treatment by saying:

. . . There might be some inclination to say that the

latter was after all "really" about words. But this

would be a mistake. The statement "This is red" is also

replaceable by, and in some way equivalent to, the longer

statement "This has the color usually called 'red.‘ "But

while the latter statement clearly mentions the word "red,"

it by no means follows that the original statement does.

And the same can be said of our original example.3

In answer to Black's objection it must first be noted that his

analyzandum is not the same as the one discussed herein. Meaning

 

1Ibid., p. 19.
2Ibi ., p. 20.

 

0..

0—.
U"
.5.
D.

191

formulas such as “He meant 'dim the lights,'" are in no way comparable
to what we have referred to in this paper as meaning statements, i.e.,
definitions. Black himself makes it clear at the outset of his discus-
sion that he does not intend to give an account of statements such as
"'Loch' means a narrow lake." 'But such a statement is precisely the
kind of statement that we are solely concerned with in the present view
of meaning. Black states that a dictionary entry such as "'Loch' means
a narrow lake" will not be referred to as a meaning fermula,“ since it
does not refer to a particular speaker and a particular occasion.

In this paper we have chosen not to discuss statements similar
to Black's meaning formulas simply because they do refer to a particular
speaker and a particular occasion. ‘A theory which would treat such
statements adequately would be a theory about the meaningfulness of
statement tokens, not a theory about the meaning of terms.1 As stated
earlier, it is the present view that with regard to whole sentences,
one cannot correctly pose the question "What is the meaning of that
statement STn?" Rather it is only appropriate to ask either the ques-
tion "What is the meaning of this word ‘W'?" or the question "Is this
statement Stn meaningful?" Whether a statement STn is meaningful or
not will depend on many things--the meaning of the terms in it, the
context in which it is spoken, the syntactical and grammatical rules
of the language it was written in to name just a few. Therefore, we

cannot at all accept Black's meaning formulas as similar to the meaning

 

1Our theory does not deal with what people mean (intend) when
they utter words in sentences.

192

statements of our metalinguistic theory of meaning. In the sentence

"He meant 'X'" the term 'meant' is expressing some sort of relationship
between a speaker and a word or expression. But in our view 'means' or
'meant' expresses a relationship between linguistic expressions whose
uses may or may not be known to speakers and writers. When Black uses
the cognate of 'mean,‘ i.e., 'meant,‘ he does not make explicit the
sense of 'mean' that he intends.‘ Alston in Philosophy of Langppngnotes
that linguistic meaning as we describe in our theory is only one of the
uses of the term 'mean.‘ On the other hand Alston says:

There are many other uses of 'mean,‘ some of which might
be confused with our sense

1. That is no mean accomplishment (insignificant)

2. He was mean to me (cruel)

3. I mean to help him if I can (intend)

4. The passage of this bill will mean the end of second
class citizenship for vast areas of our population
(result in)

5. Once again life has meaning for me (significance)

6. What is the meaning of this (explanation)

7. He just lost his job. That means that he will have

to start writing letters of application all over

again (implies).l

In these cases we are talking about people, actions, events or
situations rather than about words, phrases or sentences. In Black's
meaning formulas it seems that 'meant' is being applied to a linguistic
expression in the accusative but it also seems to be applied to a person--
"What did S mean when he said 'Dim the lights.‘ Such a question might

arise not only and perhaps even rarely when the meaning of the phrase

'Dim the lights' is unknown to a hearer or reader. The meaning of the

 

1W. P. Alston, Philosophy of Language, p. 10.

193

phrase might be perfectly clear but as a sentence used by 5 it might
misfire. It is possible that S may say 'Dim the lights' in a room

that is already quite dim. Then meaning formula statements might be
given such as "S meant 'make the room still darker' by 'Dim the lights'"
or "S meant sarcastically 'that this room certainly is dim' by 'Dim the
1ights.'"

It appears then that Black's objection does not apply to the
meaning statements of our theory and it is difficult to assess his
objection even in the context of his own discussion since he seems to
be analyzing 'meant' as applied to an "uneasy mixture"1 of persons and
linguistic expressions. It may simply be that he is taking statements
or sentence tokens as the smallest unit of his analysis while our theory
is confined to individual words or word phrases that are not sentential.
The possible ambiguity of Black's discussion of 'meant' is even more
apparent by his reference to the statements "This is red" and "This
has the color usually called 'red.'" In the passage formerly quoted,
he seems to be objecting that the longer transformation of a meaning
formula (which seems to indicate that a meaning formula is really about
expressions) does not really indicate that a meaning formula is about
words since "This is red“ is not about words while its longer transfor-
mation "This has the color usually called 'red,'" seems to indicate that
"This is red" is in part about words. But it certainly is not obvious

how the statement "This is red" is in any way analogous to Black's

1We here refer to a quotation from Dewey given on p. 184.

194

example meaning formula such that what is true of the former can
correctly be said to be true of the latter. In any case "This is red"
is far removed once again from those examples of meaning statements

that we are now discussing.

Meaningpstatements

The present theory of meaning has been called the metalinguistic
theory of meaning. It is so-called because the paradigm example of a
meaning statement in this theory is "'X' means 'Y'" which is a first-
level meta-statement1 about linguistic expressions. The term "means"
in the meaning statement is a metalinguistic term relating the definiens
expression to the definiendum expression in the manner previously
described. Such a meaning statement is a definition. If we now return
to our meaning criterion of analyticity it is clear that a statement Stn
is analytic iff its truth or falsity can be determined by reference to
the meaning of the terms of which STn is composed. Thus an analytic
statement will be a statement which is deducible from the definition
statements of its terms.

Let us take again as an example of an analytical statement ST,
"All planets move round the sun," a token which might occur in a first-
order extensional language E which contains the usual connectives. and

primitive descriptive predicates P, H and M to stand for "planet,"

‘

1We shall distinguish meta-statements about linguistic expres-
sions in the object language from meta-statements further removed from
the object language according to various levels. A meta-statement
about a first-level meta—statement would be a second-level meta-
statement in our view.

195

"heavenly body" and “moves round the sun;f 'The meaning of fplanetf

in a hypothetical dictionary might be "a heavenly body that moves round
the sun." However this dictionary entry cannot be expressed in E
according to our theory but only in a metalinguistic language E' which
contains the same connectives, predicates and individual constants and
variables as E with the addition of intensional verbal expressions such
as “is necessary" and "is to be necessary." Our formulation of this
dictionary entry will differ from more usual ones in yet another way

as well.

Some writers such as F. Waismann who write definitions as object
language expressions have preferred to write meaning statements such as
"'Planet' means a heavenly body which moves round the sun" as D1 '(x)
(Ponx - Mx)’ :1 however, we have chosen to reflect the term "means" in
meaning statements by the equivalence symbol '2' instead of the 'a'
symbol. Thus our definition is compatible with Beck's symbolization
for analytic statements in terms of Kant's second identity formulation
of the A-S distinction.' In our present example Beck's symbolization
would be P==Mx where X is a collection of the meaning terms of P one of
which is H. MX together constitute the whole collection of these terms.
The predicate of STl-"moves around the sun" is thought through identity
with the subject term "planet." We thus write DI, the meaning statement
of "planet" as "'(x) (szst ~Mx)‘. . . ."2 In order to complete D1

above let us examine metalinguistic statements in more detail.

 

1This is probably because H and M may not be the only applica-
tion criteria for P. We are assuming that H,M is a complete list of
these criteria.

2Again we assume that X consists here only of H.

196

To adequately characterize definitional meaning statements we

shall borrow an analysis of entailment statements proposed by P. F.

Strawson in his article "Necessary Pr0positions and Entailment State-

ments."1

Strawson deScribes statements which he calls intensional ..

contingent statements and he makes the following six points concerning

them:

1.
2.

Statements descriptive of the use of expressions'are ‘
contingent (not either necessary or self-contradictory)
Entailment--statements (i. e. statements in which the-
main verb is 'entails' ) are "intensional contingent
statements in which the main verb is the verb "to be"
and the complement of that verb is some expression
such as ”necessary," "impossible," "consistent,“ used
in the sense "logically necessary". . . . Such words

as "entails," and, when used in this way, "impossible, "
"consistent," "necessary," etc. , will be referred to as
"intensional words." “

In making contingent intensional statements, we use
the main intensional words, and mention the other
expressions they contain.

Some expressions mentioned in some contingent inten-
sional statements are themselves intensional expresa'
sions.

Every contingent intensional statement of the form

"'p‘ entails 1'q"" is logically equivalent to another
contingent intensional statement of the form" 'p:1q'

is necessary" or “'p onot-—q' is impossible.

To avoid the paradoxes arising from C. I. Lewis'
analysis of entailment--Strawson claims that no
necessary statement and no negation of a necessary
statement can significantly3 be said to entail or be
entailed by any statement.

Strawson developed this account of intensional contingent state-

me"ts in an effort to adequately characterize metalinguistic entailment

\_

Sta temen

1P. F. Strawson, "Necessary Propositions and Entailment-
ts," Mind, LVII (1948),

2We shall regard."means" as this kind of intensional word also.

31bid., pp. 185-186.

197

statements of the kind '"__? entails fi__,"' after rejecting the account
of entailment statements given by C. 1. Lewis. We also reject the
Lewis' account of descriptive entailment statements (statements which
report that '"p" entails I'q"' within some deductive system). However,
Strawson's theory of IC statements is acceptable1 and can be expanded,
for Strawson's account is not only a suitable analysis of descriptive
entailment statements but it also seems to be an excellent character—
ization of lexical definitions, i.e., meaning statements which we shall
consider quasi-reportive legislative entailment statements. A meaning
statement 'x' means 'y' can easily be translated into a statement such
as ‘p' §h311_entail 'q." This latter statement after instantiation
would read, "'b being or having xl shall entail 'b being or having y.'"
Each meaning statement can be translated into such a quasi-reportive
legislative entailment statement.

We will therefore adopt the first five points of Strawson's
theory as a characterization not only of‘a metalinguistic descriptive
statement that "p" entails "q“ but also of a metalinguistic quasi-
reportive legislative statement that "p" and "q“ are to be considered
as related in all instances by entailment.

A definition says not only that it happens to be the case that
P means H 8. M, i.e., that '(x) (PXEHx - Mx)‘ but also that '(x)
(szHx - Mx)‘ is to be considered true for all future cases, i.e.,

"(x) (PxEHx -.Mx)" is to be necessary. Thus the metalinguistic

 

1We do not accept no. 6.

198

definition statement "'planet' means 'heavenly body that moves around
the sun'" corresponds to the intensional contingent statement '"(x)
(szHx - Mx)?‘ is to be necessary.‘ We shall refer to this last state-
ment as 1C1. As a metalinguistic statement, IC1 is true and yet pre-
scriptive. IC1 itself is not necessarily true and it correSponds to
the metalinguistic definitional statement in a natural language, e.g.,
English which establishes, presents, or legislates a connection which
is to be necessary between object-level expressions of the natural
language. In this view then lexical definitions can be both legislative
and informative, i.e., true‘or‘false.’l It is also the case that a quasi-
reportive legislative statement which exhorts that certain words are to
be related necessarily is a necessary and sufficient condition for say-
ing truly that they are so related. Thus "'(x) (PXEHX - Mx)‘ _i_§_ to be
necessary" entails"'(x) (PXEHX' Mx)l is necessary" and vice versa.
Words do not mean each other in our theory until we have given them a
verbal definition which establishes that they are not only accidentally
used in the same ways by some linguistic performers but that they ought
to be used in the same ways. This of course does not mean that there
are or will be no renegade linguistic performers for rules are broken.
It must also be remembered that not all entailment statements whether
strictly descriptive or quasi-reportive legislative IC statements are
definitional IC statements. As we shall see in Chapter V a necessary
relation can be established conceptually among things, properties or

events as well as among words.

 

1This will soon be explained in greater detail.

199

Analyticity

 

The entailment relationship between IC1 and our analytic
statement ST1 can be expressed by another purely descriptive meta-
linguistic entailment statement MST11 "IC1 entails ST," which
corresponds to "'(IC,:ST1)' _i_§_ necessary." In carrying out the
deduction of analytic statements from definitions their necessity
(that of the analytic statements) referred to by Kant in formulation
five (the Leibniz' criterion) is evident. The premise "'(ICI:JST1)'
is necessary" together with the premise IC1 entails STl. If these
premises are true STl must be true. We say therefore that ST1 must be
true in relation to these premises. If we construe "true necessarily"
not as "true" followed by a modal qualifier but as "true? followed by
a 2-place relational predicate which obtains between true premises and
the statements they entail we can say that ST1 an analytic statement
is true necessarily. An analytic statement cannot be false relatiVe
to the premises which entail it. However an analytic statement can
be falsified whenever these premises are altered in appropriate ways.
In support of our views on analyticity, we shall examine in Chapter IV
the various senses of "necessity" to determine if indeed there ever are
any statements which are absolutely necessary, i.e., in no need of
demonstration and incapable of becoming false. We claim that analytic
statements are not of this kind and in Chapter IV we shall claim the

same for a priori statements.

 

1This is our notation for meta-statements, one level removed
from the material or object level of language.

200

Definition

The theory of meaning being presented is confined to definitions
which have been referred to as meaning statements. The scope of our
analysis of meaning is deliberately narrow so that the confusion that
arises from talking about "meaning" in other contexts can be avoided.
Our metalinguistic theory of meaning which indirectly is part of our
revised criterion for the analytic-synthetic distinction is concerned
only with the meaning of individual words or word phrases since sameness
of meaning among the terms of analytic statements is the sine qua non of
their necessary character. However, narrow as our theory of meaning is,
there still are many kinds of definition, i.e., meaning statements, even
within the scape of our theory which will need at least brief discussion.

Richard Robinson in his book Definition gives two major classifi-

 

cations of definitions--real definitions and nominal definitions. The
purpose of real definitions, he says, is to relate things with things
while the purpose of nominal definitions is to relate words with words
or words with things. We will not approach the classification of
definitions in this manner since it has already been stated in this
paper that definitions should be regarded as meaning statements which
always relate words with other words. It may happen that a definition
or a meaning statement which relates two symbols may indirectly relate
the things to which the symbols refer (if they refer) but this secondary
relation cannot be considered the chief function or main character of a
definition statement. As an example of real, or as he Robinson calls
them thing-thing definitions, Robinson discusses the work of Plato.

As Robinson claims:

201

. .Socrates and Plato were obviously thinking only of
the definition of things and not at all of the definition
of words. The search for the definition of piety in
Plato's Euthyphro is certainly an inquiry about the thing
piety, not about the word 'piety' . . . the same is cer-
tainly true of every other place in Plato's writings
where there is discussion of a question having the fOrm
"What is x?" that is to say, a request for a definition.

Contrary to the last sentence of the above passage, it is our
contention that the question "What is it?" is not a request for a
definition but rather for an identification. It is the vagueness
of the term "is" which allows for Robinson's conclusion that a search
for identification is a search for meaning.

Statements about identity and statements about meaning are
similar. Both kinds of statement explicitly express a relation legis-
lated as necessary, the former between things and the latter between
words. The emotion love is identified with the emotions joy, exhil-
aration, attachment by the identification statement "Love is joy,
exhilaration, attachment, etc." When this identification becomes a
necessary one intensionally, then we can write the statement "The word
'love' is identified with the words 'joy,' 'exhilaration,' 'attachment'
etc." and we can frame the particular kind of IC statement which is a
definition, i.e., "Love" means "joy, exhilaration, attachment etc."
Statements that express an identification between things can result

in statements that express an identification between the words that

refer to those things if the things connected become identified with

 

1Richard Robinson, Definition, p. 149.

202

one another as an identity.1 In this case the speaker or writer intends
that one thing be necessarily connected, i.e., identified necessarily,
with the other. While some blue things are flower things and some
flower things are blue things unless a speaker or writer intends that
all flower things and all blue things are necessarily connected, i.e.,
identical, there will result no true statements such as "blue" means
“flower." The statement "Flower things and blue things are necessarily
connected" is itself a contingent statement which expresses the fact
that there is a partial identity between flower things and blue things.
Such a statement is contingent because it is not necessarily true that
flower things be identified with blue things all the time or ever until
the speaker or writer declares that he intends that this should be so.
Once the speaker uses a statement which necessarily relates things via
identity, it is then possible for that speaker to utter or write meaning
statements or definitions involving the words that refer to those things
and these meaning statements express an identity between these words.

It is important to note, however, that this genetic relation between
thing-identity statements and word-identity statements does not always
occur. The thing-identification statement "That flower is red" does not
express an intensional necessary identity between the two things, i.e.,
the flower and red thing such that being the first thing, the flower,

is always correlated with being the second, the red thing. Therefore

no word-identity statement can arise from it such that we would say

 

1One thing becomes a mark of the other.

203

"flower" means “red." Likewise there are many word-identity statements
that do not arise from any thing-identity statement. Many definitions
express a relationship between words which themselves do not refer to
any things such as 'and' or 'although.‘ These definitions are licenced
differently.

Our conclusion then is that there are no thing-thing definitions
as Robinson supposes, but rather identity statements which express
necessary relations between things and a special subset of these iden-
tity statements are definitions which express a necessary relation

between linguistic things, i.e., words.

Types of linguistic meaning statements

It should not be construed from the above discussion and the
theory of meaning here advocated, that all definitions are nominal, if
by "nominal" the reader means "stipulative." These two terms have often
been used synonymously and in contrast with the terms "real" and “lexi-
cal." But as we have just seen in Robinson's discussion "real defini-
tion" has quite a different referent from "lexical definition." The
first refers to what he calls thing-thing definitions while the second
refers to what he calls word-thing definitions.‘ Word-thing definitions
which he claims are all nominal can be lexical definitions or stipula-
tive definitions. Other writers such as Irving Copi who believe as we

do that there are only word-word definitions refuse to use the old and

 

1I id., p. 19.

204

confusing terms "nominal" and "real" in application to definitions and
we shall here follow suit. Let us consider the two major classifica-
tions of definitions which most writers seem to recognize--lexical and

stipulative definitions.

Lexical definition

Dictionary definitions are often referred to as lexical defini-
tions. Irving Copi states that when the term being defined is not new
but already has an established usage, its definition is lexical.1
Lexical definitions report the meaning of terms and so can be true or
false. At times when we might doubt the definition that someone had
given us of a commonly used term in English we might consult Webster's
Dictionary and find that the definition statement given by the person
was indeed false. At other times we might note a change in the wide-
spread usage of a dictionary term and find that older dictionaries
contained a false definition of that term.

It has sometimes been maintained that no definitions are true
or false, i.e., that no definitions are descriptive or reportive but
only normative. In this view dictionary definitions express only how
words should be used or what they ought to mean but not what they do
mean. To this claim there is overwhelming reply in philosophical
literature. For one, Robinson notes:

Though a rule is neither true nor false, a statement
that certain people observe a certain rule is either

 

1Copi, op. cit., p. 102.

205

true or false, and a lexical definition is such a
statement.1

Robinson is also careful to note, however, that dictionaries
“tend to be histories not of all the usages prevailing at a given time
and place but of those of the preferred group of persons.2 Copi main-
tains that the word "usage" refers to a statistical matter. In his own
words:
The need tor lexical statistics cannot be evaded by
reference to "correct" usage for that too is a matter

of degree, being measured by the number of "first-rate"
authors whose usages of a given term are in agreement.3

Lexical definitions according to the
metalinguistic theory of meaning
In the meaning theory proposed a definition relates terms in
the following ways:
1. The use of, or a meaning of the definiendum is part
of the public language and known to the speaker but
not to the hearer while the use of or a meaning of
the definiens is known to both.
2. The use of, or a meaning of, the definiendum is not
part of the public language or known to the speaker
or hearer while the use of or a meaning of the defin-
iens is known to the speaker or to both speaker and
hearer.
A lexical definition is one which relates terms in sense (1) and
is thus covered by our theory. The relation of words described in (2)
is one expressed by stipulative definition which is also covered by our

metalinguistic theory of meaning.

 

1Robinson, op. cit., p. 39.
2Ibid., p. 37.

3Copi, op. cit., p. 103.

206

Inspection of (1) indicates more clearly how it is that lexical
definitions convey information to a bearer and can therefore be said to
be true or false. The hearer knows at least how to use the definiens
verbal expression. Perhaps he knows not only how to use the definiens
expression but also that another verbal expression found in the diction-
ary, the meaning of the definiens, has the same use as it, the definiens,
has. The lexical definition token spoken or written by the hearer's
companion (the speaker or writer) presents a meaning and perhaps a use
theretofore unknown to the hearer. Once the definition token statement
MSTn or as we have previously been calling it, 1C1, is uttered, the
bearer is informed about the meaning or use of the definiendum. However,
the hearer is not only informed that many speakers use a term in a cer-
tain way or that the term has a certain meaning (another verbal expres-
sion in our theory) but also that this use of the definiens or the
meaning of the definiendum which is the definiens must be adopted as
correct. Why? The definiens is already accepted by most speakers or
the preferred writers as the meaning of the definiendum and it is
intended that the hearer communicate by means of this public language
already adopted. Lexical definitions are once again the contingent
intensional statements which legislate that two verbal expressions the
definiendum and the definiens must be considered equivalent, i.e., as
an identity that is to be necessary. But as we have already said
lexical definitions are also informative and descriptive. It appears
thus that these very unique identity statements elliptically combine

both a descriptive and prescriptive element. The speaker is really

207

saying via his utterance of the lexical definition that you the bearer
must mean 'X' by 'Y' in order to communicate with me because the rest
of us who use a particular language already use 'X' as though it means
'Y.' The prescription and the descriptive reason for the prescription
are both included within these statements. There is in addition to
this informative aspect of definitions another which is that we are
informed by the speaker of his intention to use a term 'Y' as part of
a public language and that he wishes the hearer to communicate with
him on the level of the public language. It must also be noted that
though the definition statements themselves establish certain identi-
fications as identities yielding necessary connections, the definition
statements are contingent and not necessary. Lexical definitions which
are assertions of an empirical fact about speakers are not themselves
necessary. What is true necessarily is the analytic statements that
follow from definitions. These analytic statements which arise from
definitions and are true by reference to them are thus indirectly both
discovered and justified by empirical investigation. In this same
indirect way analytic statements can be falsified as when a definition
is or has become false. But the analytic statement does not chiefly
function as a conveyor of empirical facts about linguistic entities.
It may be, as Norman Malcolm suggests, that analytic statements serve
a role in our calculations within a language facilitating inferences.
We agree with him that a necessary proposition is not really a propo-

sition describing how people use words.1 However, whereas Malcolm

 

1Norman Malcolm. "Are Necessary Propositions Really Verbal?"
Mind, XLIX (1940), 199.

208

concludes that analytic propositions are not empirical and not even
statements we conclude that they are statements for they have a truth
value, truth and that they may be said to be empirical in many of the
senses that could be given to the statement "A is an empirical propo-
sition." We have noted that analytic statements which may originate
from our knowledge of definitions can be said to have an empirical
origin. We have also noted that analytic statements may be indirectly
falsified. They are justified, i.e., regarded as true necessarily via
the truth of a definition which can be called an empirical statement in
any of the usual senses of the phrase. Further, the words which occur
in analytic statements may refer to empirical or observable entities
thus analytic statements may be about empirical entities. Thus in what
sense can analytic statements be regarded an non-empirical or as non-
statements? The whole appellation "empirical proposition" in this
context seems vacuous or if meaningful at all, confusing. It is for
this reason that we have sought to avoid including Kant's fourth formu-
lation of his A-S distinction in our reconstruction in any but the
weakest sense of "empirical independence" viz., that analytical state-
ments need not be verified by empirical investigation other than that
of checking a dictionary. Since checking a dictionary can be thought
of as a form of empirical investigation the whole discussion of the
empirical aspects of analytic statements is unenlightening and detri-
mental to effective discussion. Even writers such as Gover Maxwell
who strongly assert the prescriptive aspect of definitions admits:
Facts are involved in analytic sentences in the following

way. When we give reasons for adopting rules or stipula-
tions [definitions in our terms] which render certain

209

sentences analytic we usually appeal to facts. We point

to the linguistic fact that most people have a certain

set of habits involving the word 'bachelor."
Stipulative definition

We have described cases of stipulative definition already in (2)

as being those in which the use of or the meaning of the definiendum is
not known to the speaker or hearer while the use of or a meaning of the
definiens is known to the speaker or to the speaker and hearer. The
chief difference between stipulative and lexical definitions is that
the former introduces a verbal expression whose use or meaning is
unknown to both speaker or hearer while both may be familiar with the
definiens, whereas the latter introduces only to the hearer a definien-
dum whose use is already known to the speaker via a definiens known
again to both. Irving Copi suggests that the definiendum need not be
totally novel to the speaker and hearer in the giving of a stipulative
definition but only that the definiendum be new in the context in which
the defining takes place.2 Given that the introduced expression, the
definiendum, is new, it would seem that stipulative definitions unlike
lexical definitions are purely prescriptive. As Copi claims:

Since a symbol defined by a stipulative definition had

no prior meaning the definition cannot be regarded as a

statement or report that the definiendum and the defin-

iens have the same meaning"3 (this phrasing is contrary

of course to that compatible with our Metalinguistic
Theory).

 

1Grover Maxwell, "The Necessary and the Contingent," Minnesota
Studies in the Philosophy of Science, III (1962), 399.

2Copi, 0p. cit., p. 100.
31b1d-9 p. 101.

210

Despite the above claim even if it were agreed that stipulative
definitions do not convey information about past usages, it must be said
that they do nevertheless convey information to the hearer about what
ought to constitute a future usage as well as the intentions in this
respect of the speaker. As Robinson suggests:

There is still another way in which truth or falsehood
enters into stipulative definition. In stipulating a
meaning for a word a writer demands that his reader shall
understand the word, in that sense whenever it occurs in
that work. The writer thereby lays upon himself the duty
of using the word only in that sense, and tacitly promises
to do so, and tacitly prophesizes that he will do so. But
sometimes a writer does not use the word only in the sense
he has stipulated. Then his stipulation implied a false
promise and a false prediction.‘

In addition to the above description of a possible factual
element in stipulative definition Robinson also describes another:

But though stipulation as such cannot be true or false,
it usually implies an element of assertion, and conse-
quently of truth or falsehood, in the following way. The
assumption that the word and the thing both exist in some
way. The assumption that the word exists cannot be false
in stipulative definition; for in uttering the definition
we utter the word and thus make it exist. But it can be
false in lexical definition, for the existence there
implied is utterance by other persons. . . . Thus the
geometers who defined 'horned angles' as those made by
the intersection of curves as opposed to straight lines
were implying and believing in the possibility of such a
thing, whereas the angle between any two curves can only
be the angle between the two straight lines that are
tangent to the curves at the point of intersection.2

The point Robinson makes in the above passage can also be made

in terms of the metalinguistic theory of meaning. As discussed

-_

‘Robinson, op. cit., p. 64.
2Ibid.

211

previously, definition statements (stipulative ones here) in the
metalinguistic theory of meaning establish a necessary identification
which expresses a relationship only between verbal entities, not between
words and things. However, we pointed out, too, that a definitional
identity statement often arises from non-definitional identity state-
ments expressing a relationship between things. ”In the example above,
given by Robinson there may be an identity statement which serves to
pick out various elements of the given as constitutive of but one
phenomenon or event, e.g., the intersection of curves. From this
identity establishing statement among elements, which ferm one phenom-
enon, e.g., a certain angle, may arise the stipulative definition that
the unknown phrase 'horned angles' shall mean the known phrase 'the
intersection of curves.‘ Indirectly the stipulative definition may be
considered false if the existence statement that there is something, the
intersection of curved lines, is not true. Once again then we see the
vagueness of claims that a certain definition statement is or is not
factual, is or is not true or false. As is also the case with lexical
definitions stipulative ones are contingent intensional statements which

establish a necessary identification between verbal entities.

Synonymy

There are some types of supposed stipulative definition which
are not characterized by (2) on page 205. In such definitions it may
happen that the use of the definiens is not known to speaker or hearer
just as the use of the definiendum is unknown to both. These defini-

tional statements relate symbols which are mere marks. Pepper classifies

212

such stipulative definitions under the heading equational nominal
definitions.‘ However Pepper claims that even in the category of
equational nominal definitions, the symbols that are the definiens

are "generally assumed to be symbols having meaning in their own
right."2 But for the moment let us consider definitional statements
where a symbol of unknown meaning is said to mean another symbol of
unknown use or meaning. Can such a statement really be considered a
definition? Certainly not on the present theory of meaning advocated
herein. Rather such a statement would be a mere rule of synonymy. By
"rule of synonymy" we mean a statement which tells a reader or hearer
that two symbols have the same meaning, such that one can always be
substituted for another salva veritate but it does not inform the reader
or hearer what the common meaning is for the use or meaning of neither
symbol may be known. The stipulative definitions which might be con-
sidered the exception to the rule of our theory of meaning are therefore
really not exceptions or counter instances for they are not definitions
but rather rules of synonymy. It is important to note that such defini-
tional statements are not disallowed as definitions on the basis of our
meaning of "meaning" and "definition" which would involve a circularity
but rather on the basis of a separate characterization of ”rule of
synonymy" which seems to capture these few stipulative definitions
within its extension. To say therefore that "'X' and 'Y' are synony-

mous" is not to give a definition or the meaning for either directly.

 

‘Pepper, op. cit., p. 30.
2Ibid.

213

The synonymy statement may be written as "'X' ='Y'" but the truth of
"'X'=='Y'" depends on two definitions, e.g., "'X'==df 'Z'" and
"'Y'==df 'Z.'“ Thus by these definitions it is true that "'X'=='Y.'"

In his interesting article on synonymy, Benson Mates cites
several circumlocutions for expressing synonymy, e.g., we might say
"to say A is only to say 8“ or "A; in other words, 8" or “when I say
A, I only mean B."‘ In none of these expressions do we find the term
"means"2 or any suggestion that knowing the meaning of any terms is
expected. Within the framework of our metalinguistic theory of meaning
synonymy statements are metalinguistic expressions which convey the
information that two linguistic entities have the same meaning such
that if we looked them each up in our dictionary we would have in some
way the same verbal entry written next to each. These statements of
synonymy however are not used to give the meaning of either of the
expressions they relate for they would then be definitions and not
statements of synonymy. Failure to recognize this difference between
definitions and synonymy statements can lead to serious confusions such
as are found in comments by Max Black on analytic statements.

Max Black describes analytic sentences in natural languages as
necessary statements which are certified by checking against a corre-

sponding rule of language.3 Using an example analytic statement

 

‘Benson Mates, "Synonymy," Semantics and the Philosophy of
Language, ed. by Linsky, p. 118.

2Here we mean "means" in the metalinguistic definitional sense
as we have already defined it. "Mean" used in the sentence above seems
to be used by Mates as one would use "intend"--see page 192.

3Max Black, "Necessary Statements and Rules," Philosophical
Review, 1958, PP. 313-341.

214

"Monday is the day before Tuesday," Black conjures the linguistic rule
that "Monday" may be replaced by "the day before Tuesday" and vice
versa. Such a rule certifies the example statement. Thus in order

to "certify" analytic statements one must look for rules of synonymy
which are in force in the English language. However, is it not plau-
sible to ask why two terms are synonymous? Finding a synonymy rule may
explain how it happens that a statement is analytic, e.g., we find
within it a term joined with its synonym but why stop explanations of
analyticity at the level of synonymy, when synonymy itself is in need
of explanation? One word can be substituted for another salva veritate
simply because they have the same meaning. If one can have recourse to
the meanings of the terms he uses he can then "certify" Black's rules
of synonymy themselves. How would one know he has seized upon the
correct synonymy rule unless he can check the meanings of each term to
see in fact if the terms are synonymous--substitutable salva veritate.
Synonymy statements then are true or false depending on the meaning of
their terms as given, legislated or conveyed by definitions and defini-
tions which are verbal and establish one linguistic entity as the
meaning of another are true or false and depend ultimately on usage

which is not learned by definition at all but by observation.

The reconstruction of the A-S distinction

In view of the foregoing discussion, the virtues of our proposed
reconstructed criterion of the analytic-synthetic distinction and our
proposed meaning theory should be clear. 8y pr0posing that a statement

is analytic iff its truth or falsity can be determined by reference to

215

the meaning of the terms which constitute the statement, one can avoid

all the objections levelled against the distinction as drawn by Kant.

Re: the vagueness objection-~Our reconstructed criterion is not
metaphorical or vague.

the form objection--Our reconstructed criterion divides
exclusively all declarative statements
not just those of S-P form.

the variability objection--Our reconstructed criterion applies only
to particular statement tokens which occur
only once and are at that time either
analytic or not. No sentence or judgment
therefore is at one time analytic and at
another synthetic.

the oversight objection-~Our reconstructed criterion provides for .
statements whose falsity can be determined
by reference to the meaning of the terms
in the statement.

the epistemological objection--Our reconstructed criterion prescribes
that we refer to the meanings of terms
within particular statements in order
to determine whether the statement is
analytic. This meaning is, according to
our proposed.metalinguistic theory of
meaning a verbal expression found in a
dictionary. Thus any person making an
assessment of a statement need not depend
upon his own understanding of terms within
the statement.

the perspective objection--Our reconstruction circumvents this crit-
icism aimed against Kant's psychologism in
the same way as above (the epistemology
objection).

the ambiguity objection--Our formulation makes no claim about any
characteristic of empirical indifference
with respect to analytic statements. They
may originate from empirical considera-
tions. They also may be indirectly falsi-
”fied by experience (as already explained).
If we Wish they may be presently justified
by empirical data though we need not
resort to empirical investigation in
order to justify them. The terms within

216

analytic statements may refer to observa-
tional entities and therefore analytic
statements may indirectly pertain to the
world of empirical fact.
the psychologistic objection--Our reconstruction as that of Frege

captures Kant's fifth formulation and
makes it more stringent by requiring that
definition is basic to analysis. In this
way the psychological relativity of Kant's
view is avoided.

the definitions objection--Our reformulation allows for the fact that
analytic statements can serve in explica-
tion of concepts which we already have
(as Kant might say) in the sense that they
are in a dictionary not presently open to
us.

Aside from avoiding the many difficulties of the A-S distinction
as Kant discusses it, our formulation captures in the ways already
explained his first, fifth and sixth formulations as well as the spirit
of the second and fourth. Analytical statements are necessary, entailed
by intensional contingent definition statements (which may include the
law of non-contradiction) which establish identities between terms as
necessary. As such the predicate term may be contained in the subject
term as part of its meaning. Given these characteristics analytic

statements need not be verified or falsified by empirical investigation.

Criticism of the A-S distinction

The distinction between analytic and synthetic propositions has
come under severe attack in recent literature by W. V. O. Quine, whose
work suggests that such a distinction is untenable. In his now classic
article "Two Dogmas of Empiricism" Quine offers two major objections to
the analytic-synthetic distinction. The first objection rests on three

assumptions Quine appears to hold:

217

l. analyticity = df a characteristic of sentences which
can never be given up as false

2. "analytic" and "a priori" are synonymous
3. sentences are enduring entities.

If one holds the above three assumptions then one cannot
escape Quine's first critical conclusion that there is no1 clear
dividing line between analytic and synthetic statements at least if
one admits any statement can be given up if the epistemic situation
demands it. But the mark of analytic statements assumed by Quine in
#l is the least desirable mark out of the six which we discussed in
Chapter III. Ne there referred to Quine's assumed criterion of
analyticity as the empirical independence criterion and described the
ambiguity of the phrase "empirical independence" as used by Kant and
his successors. In Kant's text, there is evidence to suggest that by
this phrase he did not mean either that empirically independent prop-
ositions could not indirectly originate in experience nor that they
could not be verified by empirical investigation. Nevertheless this
problematic criterion is assumed by Quine though Kant offers several
other criteria of perhaps greater merit.

Quine may assume #1 because he assumes #2 that analytic and
a priori statements are the same kind of statement. Historically "a
priori" has been conmonly characterized as referring to judgments
(sentences) not subject to falsification. However as we will show

in Chapter IV, this criterion is not even a good one for the a

 

1In fact according to Quine the notion of analyticity should be
discarded.

218

priori-a posteriori distinction; and we will also show that in modern
times the terms "analytic" and "a priori" have become identified as
synonymous terms by assumptions on one side (e.g., the work of Ayer
among others) and default on the other (with the notable exceptions
of L. w. Beck and N. R. Hanson).

The third of Quine's assumptions, that analyticity is a
characteristic of sentences as enduring entities, leads him to the
conclusion that sometimes a given entity, a sentence, can be analytic
while at other times it can be synthetic and that once again the A-S
distinction cannot be clearly drawn. Quine claims that at one time a
sentence will seem conventional and at another time it will not seem
nerely conventional. Precisely due to the apparent and clear truth of
this claim we have sought in Chapter III to explicate "analyticity" with
respect to only concrete particular statements which are always sentence
tokens never enduring signs unvarying from context to context. A state-
ment is fleeting and transient. It occurs only once and whatever it is
on the occasion it occurs, analytic or synthetic, that is what it is.

It seems odd that Quine or anyone should have ever talked about sen-
tences simply because he recognizes they change their analytic or
synthetic status over time.

Earlier in this chapter, mindel of Quine's logical but unsound
conclusions as well as the historical objections to Kant's theory, we
deliberately avoided discussing sentences and chose only to discuss
discrete non-enduring sign tokens--concrete statements. At one point a

given statement is definitionally true. At another point a numerically

219

different statement which is expressed by the same sentence as the
first is not definitionally true but as Quine might say discursive.

We have also formulated a metalinguistic theory of meaning to
avoid Quine's other objections to the analytic-synthetic distinction.
In our theory, meanings are not regarded as non-physical quasi-
metaphysical shadows behind words but as simply other words, strings
of symbols physically existent on the pages of our dictionaries.

In efforts to embrace a nominalism which does not allow for
the existence of meanings as obscure entities Quine finds attempts to
explain analyticity in terms of meaning, definition or synonymy
inadequate.

If one were to suggest as does Benson Mates, that synonymy
is the interchangeability of linguistic expressions (except within
words) which is salva veritate, there would be no way to claim as one
would want to claim that the statement "All creatures with a heart are
creatures with a kidney" is synthetic while "All bachelors are unmarried
men" is analytic. The interchangeability of the terms in the first
statement is salva veritate just as is that of the terms in the second.
It seems here that Quine's second objection is much the same as our own
with respect to Black's discussion of certifying rules of synonymy.

How are we to know even in terms of interchangeability salva veritate
which rules (statements) of synonymy are true or false? It certainly

appears that the synonymy statement "'Creatures with a heart' is

synonymous with 'creatures with a kidney,'" is false. In terms of

our present theory we would find out a posteriori that such a synonymy

220

statement is in fact false as soon as we consult our dictionary.
"Creature with a heart" and "creature with a kidney" are both co-
extensional. But if we look in our dictionary we find no definition
that "creature with a heart 'means' creature with a kidney."1 Thus
these two phrases fail to be united by an IC meaning identity statement
and we cannot write any true synonymy rule for them. According then to
our view of analyticity and synonymy the two phrases "creature with a
heart" and "creature with a kidney" are not synonymous and therefbre
there is no true metalinguistic statement that "creature with a heart"
and "creature with a kidney" have the same meaning. Since the two
phrases have not the same meaning there is no question according to
our revised meaning criterion but that Quine's example sentence "All
creatures with a heart are creatures with a kidney" is synthetic.
Quine's puzzlement concerning synonymy is reflected in the following
passage:

Some say that statements of the second class reduce to

those of the first class the logical truths by definition.

"Bachelor" for example is defined as "unmarried man."

But certainly the definition which is the lexicographers

report of an observed synonymy cannot be taken as the

ground of the SVUOHYMY-

Quine's second objection might bear some force if the above

passage were a correct view of definition. But as we have tried to

 

1The definitions of "heart" and "kidney" are very different!

2H. V. Quine, "Two Dogmas of Empiricism,” From A Logjcal Point
of View, p. 24.

221

show already it is not. Definitions are based on similar uses between
terms. But definitions via their report of similar uses establish the
meaning of a term (explained in our theory in terms of IC statements).
When we have a sufficient cross reference repertoire of definitions
(IC statements) we can then speak about the definienda which have
similar written definientia, i.e., we would write or utter statements
of synonymy. In contrast with Quine, we claim that a statement of
synonymy does not refer to the sameness of use of two terms but rather
to the sameness of meaning of two terms. Quine's own example sentence
indicates the importance of not confusing the uses of a term with
the meaning of a term. The two phrases "creature with a heart" and
"creature with a kidney" might well be put to the same use, i.e., to
refer to animals but they hardly have the same meaning. It is indeed
a strange idiom which Quine uses when he says that definitions are
reports of a lexicographer of observed synonymies. On our theory
"synonymy" does not refer to some elusive thing "out there" that a
lexicographer might observe. Rather just as "meaning," "synonymy"
is a metalinguistic term which refers to a relation between linguistic
expressions.
Our characterization of definitions as IC statements also

circumvents Quine's pessimism about semantical rules. He says:

Not every true statement which says that the statements

of some Class are true can count as a semantical rule--

otherwise all truths would be 'analytic' in the sense

of being true according to semantical rules. Semantical

rules are distinguishable apparently only by the fact of

appearing on a page under the heading "Semantical Rules"
and this heading is itself then meaningless.

 

1Ibid., p. 33.

222

In reply R. M. Martin states:

Both Quine and White seem to have had difficulties under-
standing what Semantical Rules are. . . . Now it seems
fairly evident that in Carnap's formulations of semantics,
semantical rules are to be regarded either as (l) defini-
tions in the metalanguage, or as (2) semantical axioms in
the metalanguage, . . . in either case we know quite well
what the semantical rules are. To object to them in the
sense (I) is to object to semantical truth definitions of
the Tarski kind. And to object to them in the sense (2)
would seem tantamount to objecting to the very kind of
general semantics Quine is . . .demanding!

Our view of 10 statements adapted from Strawson characterizes
definitions as unique in form and evidence of their form enables us to
write such statements under the heading semantical rule. In sum Alan
Gewirth writes the following reply to Quine's first criticism:

But if, following Lewis, we say that these difficulties
prove that we must refer to "intensional meaning“ since
extensional meaning does not meet the salva veritate
test, . . he (Quine) intimates the need for a Trprag-
matic" as against a merely "semantic" explication of
synonymy and analyticity; but on grounds of unclarity
he rules out, as constituents of such explication, any
reference to such intensional or mentalistic factors

as "meaning," “sense," “understanding," "definition,"
"necessary," by which one might naturally seek such
explication. Paradoxically however he uses those very
terms and others like them to refute proposed explica-
tions of synonymy which themselves use such terms.

Rather than reject the use of terms such as "meaning," "defini-
tion" and "synonymy" we have sought to clarify them and then use them
in reconstructing Kant's A—S distinction. In order to avoid Quine's

criticism we have closely reviewed the problems of Kant's fourth

 

1R. M. Martin, "0n Analytic," Philosophical Studies, LXI (1952),
45.

2A. Gewirth, "The Distinction Between Analytic and Synthetic
Truths," The Journal of Philosophy, L (l953), All.

223

formulation-~the empirical independence criticism--and blunted much of
its force. Many who have not done so have erroneously and superficially
characterized analytic statements as those which are immune to falsifi-
cation. We strongly disagree with this claim as does Quine (page 31)
and in the next chapter we shall examine its wellspring, for such claims
do not arise solely from Kant's fourth formulation but also from a fusion
of the terms "analytic" and "a priori."

We will discuss this identification of "analytic" and "a priori"
in Chapter IV. But before we begin our discussion there of Kant's a
priori and a posteriori distinction let us recap our explicative prog-
ress thus far. We have seen already that analytic statements are true
necessarily in relation to the IC definition statements which entail
them. We have also seen that analytic statements may become false when
the IC definition statements that entail them become false. Words change
their meaning gradually but there comes a point when the meaning of a
word has changed so much that the legislation that the word "X“ should
mean "Y" is false; everyone uses "X" in the same manner as “Z" and not
"Y.“ It is also the case that we can empirically verify analytic state-
ments if we want to do so. We can take a survey to see if in fact all
bachelors really are unmarried men. Indeed the phrase "empirically
verify" can be construed to include the checking of a dictionary to see
if "bachelors" means "unmarried male." It is also the case that ana-
lytic statements may contain factual content and apply to the empirical
world in as much as the terms within them apply to the empirical world.

Therefore, we maintain that analytic statements though true necessarily

224

may be in many senses empirical. Does this mean that they are not
necessary? No, for we have explained how it is that they are true
necessarily; how they cannot be false in a relative sense. However,
there are other senses of "necessary" and it may be that analytic
statements are not necessary in any of these senses given their empir-
ical aspects. It may also be suggested by traditionally minded critics
that a priori principles are necessary in some absolute sense of "neces-
sary.“ They might also argue that a priori truths are independent of
experience despite our previous novel object objection to Hume's a
priori argument in Chapter I. Let us move on now to consider what can
be meant by a priori, i.e., whether a priori truths are the same as
analytic, whether a priori truths are necessary in an absolute sense
or necessary at all in any sense and whether a priori truths can be

independent of experience.

CHAPTER IV

THE A PRIORI--A POSTERIORI DISTINCTION

Kant's second distinction

- In the Introduction of the Critique of Pure Reason (2nd ed.),
Kant begins his discussion by explaining the difference between a priori
and a posteriori knowledge. A priori knowledge is independent of expe-
rience and even of all impressions of the senses. Kant notes at 82
that, though it has been cutomary to say, even of much knowledge derived
from empirical sources, that it could have been known a priori; this is
not his intended use of the term “a priori." According to customary
usage, a mystified neighbor might say that Mr. Smith could have known

a priori that his house would fall down based on the past experience
that mankind has had with respect to old houses with cracked foundations
and sagging beams of rotted wood or other similar defects. But when
Kant says that Mr. Smith could have known something a priori, he does
not mean that Mr. Smith could have known that his house would fall
independently of this or that experience of its actually doing so but
rather that Mr. Smith knows something absolutely independently of all
experience. In contrast to a priori knowledge, empirical knowledge is
possible only through experience; and empirical knowledge is therefore

a posteriori.

225

226

Kant goes on to give a criterion by which-one can distinguish

a priori from a posteriori knowledge with certainty. In establishing
this criterion, Kant gives two marks whereby a judgment can be known
to be a priori. He says:

First, then, if we have a proposition which in being

thought is thought as necessary, it is an a riori-

judgment; and if, besides it is not derived grom any

proposition except one which also has the validity of

a necessary judgment, it is an absolutely a riori

judgment. Secondly, experience never confers on its

judgments true or strict, but only assumed and compar-

ative universality through induction. . . . If then, a

judgment is thought with strict universality, that is,

in such manner that no exception is allowed as possible,
it is not derived from experience, but is valid abso-

lutely a priori (B4).

These two criteria, necessity and strict universality, were
thought to be inseparable by Kant; but he claimed nevertheless that
it was advisable to use them separately as criteria for assessing a
priori judgments since the necessity of some judgments is more easily
shown than their universality while, in other judgments, their univer-
sality is more apparent than is their necessity. Interestingly enough,
it is in not many lines later that Kant seems to slip into yet a third
mark of a priori judgments or principles. At BS he claims that "it is
possible to show that pure a priori principles are indispensable for
the possibility of experience and so prove their existence a priori."
If a principle is indiSpensable for the possibility of experience, its
existence (to use Kant's wording) can be proved a priori; it is an a
priori judgment. Thus if it appears that some judgment seems neither
necessary nor universal, it may nevertheless be both if one can show

that this judgment is a principle indispensable for the possibility of

227

experience. A good example of such an elusive a priori principle is
the general law of causality. As we have already seen, Hume for one
in his Treatise questioned the necessity and strict universal character
of this judgment--that every event has a cause. But Kant sought to
prove that the Law of Causality was necessary and universal and there-
fore a priori in the "Second Analogy" by demonstrating that the general
law of causality is indeed a necessary presupposition of experience as
we know it, i.e., as an experience which partly consists of events.
Such principles if not a priori in origin could hardly be regarded as
first principles from which experimental judgments can derive their
certainty since these principles themselves would be derived from
experience. Kant gives then in these last remarks a third criterion
for a priori judgments, viz., that they are not derivable from expe-
rience. But already in his text we note a contradiction concerning
this third criterion of a priori judgments. As an example of a pure
a priori judgment (BS), Kant mentions the proposition “every alteration
must have a cause"--the Law of Causality once again. He claims:

. . the very concept of a cause so manifestly contains

the concept of a necessity of connection with an effect

and of the strict universality of the rule that the con-

cept would be altogether lost if we attempted to derive

it, as Hume has done, from a repeated association of that

which happens with that which precedes, and from a custom

of connecting images, a custom originating in this

repeated association and constituting therefore a merely

subjective necessity.

At once it is obvious that this passage contradicts his words

of a few lines previous when he says:

228

A riori modes of knowledge are entitled pure when there
15 no a ixture of anything empirical. ~Thus, for instance,
the pr0position, "every alteration has its cause,“ while

an a riori proposition, is not a pure proposition,
because alteration is a concept which can be derived
only from experience (BB).

What has been given as an identifying mark of first principles
in the fbrmer passage has been reclaimed again by this latter passage.
There are pure a priori principles which are completely underived (in
some unexplained sense of "derived") from experience but there are also
other non-pure apriori principles which are not completely underived
from experience in that their concepts are derived from experience.
Apparently, judging from the above contradictory passages, Kant has not
decided whether the general law of causality is a pugg_a priori judgment
or not. Aside from this contradiction, there is something additionally
curious about the former passage in that Kant explains in it why the
general law of causality is'a pure a priori judgment very much in terms
that one would expect him to use in explaining why a judgment is analytic.
Yet of course we know from other places in his text that he distinctly
denies that the law of causality is an analytic judgment. Nevertheless,
in the former curious passage, Kant speaks about the concept of cause
manifestly containing the concept of a necessity of connection with an
effect; and containment of a predicate concept by a subject concept is
one of Kant's criteria for analytic judgments. This description overlap
between the present a priori-a posteriori distinction, which we shall
refer to as the P-P distinction, and Kant's A-S distinction, previously
discussed, raises a critical problem for Kant's whole treatment of these

subjects. It will be remembered that Kant's treatment allows for the

229

possibility that some judgments are both*synthetic and a priori. Yet
his A-S distinction is usually considered a mutually exclusive division
of all judgments at least in the class of S-P judgments (on our view
herein it would be of the class of all declarative token statements).
But if a judgment is analytic and thereby in no way synthetic, how is
it possible that both analytic judgments and synthetic judgments may

be a priori? Would not a priori analytic judgments and a priori
synthetic judgments have something in common, some common characteris-
tic which would make them both a priori?' If so, it would seem that this
characteristic, which would be common to both analytic a priori judg-
ments (statements) and synthetic a priori judgments (statements), should
not be any characteristic proposed in Kant's A-S distinction or in our
reconstructed criterion for this distinction. For if the/a character-
istic which renders a statement a priori were the same as the/a charac-
teristic which renders a statement analytic, then synthetic a priori
statements would be closer in formal status to analytic a priori state-
ments than they would be to other synthetic statements which were a
posteriori.1 By the same token, if the/a characteristic which renders
a statement a priori were the same characteristic as the/a characteris-
tic which renders a statement synthetic, then analytic a priori state-
ments would be closer in logical or epistemological status to synthetic
a priori statements than perhaps to any other analytic statements that
might possibly be a posteriori. The claim we are making then is simply

 

1There would be no synthetic a priori statements.

230

this: If Kant‘s AaS and PaP distinctions are going to cut across each
other with respect to the class of statement tokens, then these two
distinctions must have independent characterizations. If they do not,
as illustrated above, one or both distinctions will fail to divide the
class of statement tokens into mutually exclusive categories and both

distinctions may be defunct.

Are there really two Kantian distinctions?

The three criteria for a priori judgments given by Kant, viz.,
necessity, strict universality, and independence from experience (the
empirical indifference criterion again), are the same as1 some of the
criteria for analyticity, e.g., necessity and empirical indifference
(Kant's 4th and 5th formulations). If Kant's view of these distinctions
is to be workable, then ideally it should not be the case that the cri-
teria for a priori judgments should be the same as any of the criteria
fer analytic judgments if some a priori judgments are to be considered
synthetic. We will thus examine Kant's criteria for the P-P distinction
in order to determine if there is a P-P distinction to be drawn at all

beyond the A-S distinction already discussed.

The a priori criteria
The first criterion, strict universality, and the second crite-
rion, necessity of Kant's P-P distinction, are interdependent, as

Robinson suggests,2 in that the universality of a judgment (statement)

 

1We shall see by the end of this chapter if this sameness is
but a superficial feature of both sets of criteria or if it runs deeper.

2Richard Robinson, op. cit., p. 293.

231

depends upon its necessity. However, if necessity is viewed as the
chief mark of a priori judgments, it may well be that Kant has not

made a case for two separate distinctions, A-S and P-P, since necessity
is also a characteristic of analytic judgments. It would seem then that
the saving of Kant's P-P distinction and any reconstruction of it will
depend upon an analysis of the term "necessity." If there are several
senses in which a statement can be necessary, then it will be possible
for analytic statements and a priori statements to be necessary in dif-
ferent ways. If there are at least two distinct senses of "necessary,"
one which can apply to analytic statements and one which can apply to

a priori statements, then there may be a genuine P-P distinction in

addition to Kant's A-S distinction.

Necessity

There are at least two major Classifications of necessary statements
which were discussed as far back as Aristotle in his " Posterior Analytics."
Aristotle claimed that the truth obtained by demonstrative knowledge is
necessary. However, necessary conclusions are relatively necessary
truths since their truth follows demonstratively from necessary premises.
Concerning absolute necessity (necessary premises), Aristotle states:

Assuming then that my thesis as to the nature of scien-
tific knowing is correct, the premisses of demonstrated
knowledge must be true, primary, immediate, better known
than and prior to the conclusion, which is further related
to them as effect to cause. . . . The premisses must be
true. . . . The premisses must be primary and indemonstra-
ble. . . . In saying that the premisses of demonstrated
knowledge must be primary, I mean that they must be the
'appropriate' basic truths, for I identify primary

premiss and basic truth. A 'basic truth' in*a

232

demonstration is an immediate pr0position. An immediate
proposition is one which has no other proposition prior
to it.

In an effort to explain exactly what the premises of demonstra-
tion are, Aristotle distinguishes three types of attributes. There is
a kind of attribute which is (I) true in every instance of its subject,
(11) an essential attribute, and (III) a "commensurate and universal"
attribute. Of type (I), Aristotle says:

I call 'true in every instance' what is truly predictable
of all instances-~not of one tO‘the exclusion of others--
and at all times, not at this or that time only; e.g.
then if it be true to say 'this is a man,’ 'this is an
animal' is also true, and if the one be true now the

‘ other is true now.2

0f type (II) Aristotle says:

Essential attributes are (1) such as belong to their
subject as elements in its essential nature (e.g.

line thus belongs to triangle, point to line; for

the very being or 'substance' of triangle and line

is composed of these elements, which are contained

in the formulae defining triangle and line; (2) such
that, while they belong to certain subjects, the subjects
to which they belong are contained in the attributes own
defining fermula. Thus straight and curved belong to
line, odd and even, prime and compound, square and
oblong, to number; and also the formula defining any

one of these attributes contains its subject--e.g.

line or number as the case may be.3

Attributes which are not related in either the first (I) or the
second (11) way are called accidents or "coincident," "e.g. musical or

white is a 'coincident' of animal.““ Attributes which are of yet a

 

‘Aristotle, "Posterior Analytics," The Basic Works of Aristotle,
ed. by Richard McKeon, p. ll2.

211:111., p. 115.
’Ibid., p. 115.
I id.

 

8

U’

233

third kind'(III), "comnensurately universal,“ are those which belong
to every instance of its subject and to every instance essentially
"from which it clearly follows that all commensurate universals in-
here necessarily in their subjects."1 Aristotle states:

An attribute belongs commensurately and universally to
a subject when it can be shown to belong to any random
instance of that subject and when the subject is the
first thing to which it can be shown to belong. Thus,
e.g., (l) the equality of its angles to two right angles
is not a conmensurately universal attribute of figure.
For though it is possible to show that a figure has its
angles equal to two right angles, this attribute cannot
be demonstrated of any figure selected at haphazard, nor
in demonstrating does one take a figure at randome—a
square is a figure but its angles are not equal to two
right angles. 0n the other hand, any isosceles triangle
has its angles equal to two right angles, yet isosceles
triangle is not the primary subjectof this attribute
but triangle is prior.2

After presenting his distinction of the three kinds of attri-
butes, Aristotle concludes that basic premises, necessary truths, in an
immediate (absolute) sense must express essential "connexions," i.e.,
they are premises which mention essential attributes. As he says:

Now attributes attaching essentially to their subjects
attach necessarily to them: for essential attributes
are either elements in the essential nature of their
subjects, or contain their subjects as elements in their
own essential nature. . . . It follows from this that

premisses of the demonstrative syllogism must be con-
nexions essential in the sense explained.3

We see in this Aristotelian account of demonstrative reasoning

a distinction between premises that are immediately necessary (absolute

 

1Ibid., p. 117.

 

234

necessity) and other propositions, conclusions, of demonstrative
reasoning which are necessary only because they are deduced from
necessary premises. The latter which are only relatively necessary
must be demonstrated while the truth of the former is independent of
demonstration.

In Probability_and Induction, William Kneale also refers to
Aristotle's discussion of absolute and relative necessity in the
"Posterior Analytics." Kneale notes that the premises used in demon-
strations are themselves independent of demonstration. Their necessity
is thus absolute rather than relative to any other premises. These
primary premises must be grasped by intellectual intuition, and Kneale
states that this intellectual intuition is not the uncovering of some-
thing innate in the mind but a kind of induction which "exhibits the
universal as implicit in the clearly known particular."1 It appears
then that this intuition would be impossible without experience. As
Hamlyn writes:

We come to see the truth of the axioms of particular
sciences through a form of induction that may be called
intuitive; we see their truth in particular instances.2

Though Aristotle has distinguished between truths which are
relatively necessary and truths which are absolutely necessary, it is
clear in view of the above comments that both kinds of necessary truths

may be empirical in several senses of the term "empirical.“ It was only

 

1Kneale, Probability, p. 31.

2D. W. Hamlyn, "Contingent and Necessary Statements" (already
cited in Chapter III along with another Hamlyn article), p. l99.

235

in seventeenth and eighteenth century empiricism that the realm of
necessary truth was restricted to relations between ideas and contingent
truths (those regarded as accidental by Aristotle) became identified
with truths that state facts about the empirical world.1 In light of
this change, let us now examine various views concerning the meaning of

"necessity."

Kinds of necessity

In his very interesting article "Necessary Propositions,"
Richard Robinson maintains that neither the origin nor the meaning
of the phrase ”necessary truth" is clear even though philosophers often
say that a proposition is a necessary truth if it is impossible that it
should not be true, i.e., if there is no possible alternative. However,
the layman or philosopher upon some reflection might come up with as
many as five different meanings for "necessary proposition."

First, according to Robinson, a thoughtful person might venture
that a necessary proposition was a proposition which it is necessary for
at least some of us to believe.

A proposition could be necessary for us to believe
because it seemed to us obviously true, or because
something compelled us to believe it, or for other
causes. Or it could be necessary for us to believe
if a certain purpose was to be fulfilled . . . in the

sense that it is necessary for you to believe this
proposition if you are to be saved.2

 

1Ibid., p. 199.
2Robinson, op. cit., p. 290.

236

A second sense, offered by Robinson, of “necessary proposition"
is one discussed by Aristotle in his "Prior Analytics" (i, 8-l2) which
is the modal proposition of the apodeictic kind. In English these
propositions would be of the form "S must be P" or "5 cannot be P“
or "S is necessarily P."1

A third more familiar view of necessary propositions is the one
of Leibniz that "A truth is necessary when the opposite implies contra-
diction, and when it is not necessary it is called contingent."2 In
this third sense Robinson thinks a definition of "necessary preposition"
can be formed which is that "a pr0position is necessary if either itself
or its contradictory is'self-contradictory."3 He says:

I will call this the analytic kind of necessary proposi-
tion, because Leibniz remarked that 'when a truth is
necessary, its reason can be found by analysis resolving
it into more simple ideas and truths, until we come to
those which are primary.“

A closely related view to that above is a widespread belief
especially among logicians that a necessary proposition is one which
is true in all possible worlds, that lS‘Of course logically possible
worlds. Lord Russell quotes Baldwin's Dictionary in discussing this

brand of necessity in_Mysticism and Logic, "Necessary. That is neces-

sary which not only is true, but would be true under all circumstances."5

 

1Ibid., p. 290.

bid.

sIbid-‘g p. 29] o

N
0-.

“Ibid., p. 291, quoted in part from the Monadology, p. 33.

5Bertrand Russell, Mysticism, p. 181, cited from Baldwin's
Dictionary. '

237

Russell maintains that*propositions can only be said to be true
or false and that it is only propositional functions which can be said
to be true in all circumstances.‘ But nevertheless he says:

A proposition is necessary when it is a value of a prop-

ositional function which is true under all circumstances,

i.e., for all values of its argument or arguments . . .

by specifying the constituent which is to be regarded as

argument. . . .2
Similarly, Wittgenstein in the Tractatus tried to describe logical truth
as "a proposition which is true for all truth-values of its constitutent
propositions and true under all truth conditions as revealed by the use
of truth tables."3 But as Hamlyn notes, there cannot be a general use
for necessity described in this manner since not all logical truths are
truth functional. Hamlyn states:

The problem here is equivalent to the decision problem

for logical calculi. What means can be provided for

determining all logical truths? It is now known, thanks

to Alonzo Church, that this problem cannot be solved

generally for all calculi. Hence, there cannot be a

single criterion for logical truth. The one that

Wittgenstein offered for truth~functions was roughly

equivalent to one of Leibniz' criteria for truths of

reason - that they are true in all possible worlds.“

A fourth and last sense of necessary proposition is one which
Robinson calls the universal sense of necessary proposition. This

would be any proposition which asserts a universal connection with

 

1Ibid., p. 182.
2Ibidu Pp. 182-183.

3Hamlyn, op. cit., p. 201.
l‘Ib‘id.

Inf-0V. fl

ZhIA

 

238

unrestricted generality, e;g., "All A'is B" when this is understood
to be unrestrictedly general.1

Robinson claims that the several senses of "necessary proposi-
tion" described above are independent of each other (he does not con-
sider Russell's view which is distantly a Leibnizian theory of necessary
proposition). 0f the four main senses noted by Robinson, a universal-
necessary proposition may be an Aristotelian type of necessary proposi-
tion or a compulsory-belief necessary proposition. A necessary
pr0position in Leibniz sense might be expressed modally or it might
be a compulsory-belief proposition.‘ However, Robinson claims that
Kant's necessary proposition in the-Critigue is in part all of these
types of necessity.2 The Aristotelian modal view is found in those
Kantian examples where he uses the term "must."

Kant's necessary proposition is also partly the fourth sense
of necessity, that of universality, when he speaks (B3) of necessity
and universality as two marks of the a priori. Robinson thinks that
when Kant says necessity and universality belong together inseparably
he does not realize that this is so because they are one and the same
thing.3 As Robinson states:

The unrestricted universal "All S is P" has necessity
in that it entails that, if anything were 5, it would

necessarily be P also. . . . If absolutely all S's are
P, then any particular 5 must be P. There is no other

 

1Robinson, 0p. cit., p. 291.
21b1d0 9 p. 293.
3Ibid.

239

necessity about many of Kant's examples than just their

universality, which necessitates something about every

particular falling-under-thesubject-term.1

When Kant speaks about analytic judgments being necessary, he

is thinking of necessary propositions of course in the Leibniz sense;
but as Robinson observes, when Kant proceeded from Leibniz' "necessary
truths" to his own different kind of a priori necessary propositions,
he failed to notice that his "necessary propositions" were no longer
guaranteed to be true.‘ Kant assumed that universality entailed neces-
sity (a nonéLeibniz view of necessary propositions), and that necessity
entails truth, but necessity does not entail truth Robinson stresses in
the case of some universal necessary statements such as those of science
which may in fact be falsified.2'~0nly the Leibniz kind of necessary
pr0positions, or what Kant would have called analytic propositions, are
those whose truth is guaranteed by their necessity. At the same time,
Kant exhibits in his writing the first view of necessary proposition,
i.e., that it is one which we are compelled to believe. As Robinson
states:

If there is a statement to the effect that any S must be

. P, and I feel myself obliged to believe it, that shows
'that it is a necessary statement. For Kant, it goes
without saying, if it is a necessary statement it is a

true statement. Why was Kant confident that necessary
propositions were true pr0positions? . . . Leibniz had

 

1Ibid., p. 293.

2In Chapter I we have seen that some philosophers characterize
the laws of science as unrestricted universals. They also think that
as such law statements are in some unexplained sense necessary. Yet
we know that the laws of science are not always true; some have been
falsified and rejected over the years.

240

spoken only of necessary truths, andgiven them a
definition which made them indeed necessarily true.
Kant spoke of necessary propositions and tacitly
repudiated Leibniz's definition without giving
another; but he went on assuming that they were
all necessarily true.1
It is this muddle in Kant with which we will contend after
further consideration of the term "necessary."
The kinds of necessary proposition
which are relativelyynecessary
After review of the various senses of necessary propositions,
one aspect of them is particularly striking. None of these kinds of
necessary propositions are necessary in and of themselves (absolutely).
-In the first compulsion sense of necessary proposition discussed
by Robinson, something may have compelled us to believe a proposition
(perhaps another belief?); or it could be a certain purpose that compels
us to believe a given proposition. ~In either case, the proposition
which we must believe is necessary only in relation to something else,
i.e., another belief or purpose.
~In the Leibniz sense of necessary proposition, the relative
aspect of the necessity which characterizes them is most clearly seen.
These propositions are necessarily true relative to the meaning of the
terms of which they are constituted usually together with the laws of
non-contradiction and/or identity. As already discussed, analytic

statements are true necessarily but only relatively to the intensional

contingent definition statements from which they can be deduced.

 

1Ib'id., p. 295.

241

Robinson claims that statements of the type "All S is P" have
necessity in that they entail that if anything were S, it would neces-
sarily be-P also. Aside from the awkwardphrasing "have necessity,"
is it not the casethat the particular statement "If this is S then
it is also P" is the statement which is true necessarily relative to
"All S is P."

0f the view of necessity advanced by Russell, that "a pr0posi-
tion is necessary when it is a value of*a propositional function which
is true under all circumstances," we can once again say that no neces-
sary proposition is necessary simpliciter (absolutely) but rather only
necessary (if necessary at all) in relation to a propositional function
which is true under all circumstances. There are those who say that if
all the values of'a propositional function are true, then any one of
these values is just true and not necessarily true. However, one may
surmise that those who hold a Russellian view may be basing their claim
of necessary truth on the following kind of move. All the values of a
given propositional function are true. X'is a value of this same
propositional function. Therefore X must be true. The notion of
necessity is dragged in at the metalinguistic level as a relational
predicate said to obtain between the premises and conclusion of a
metalinguistic deductive argument.

Up to this point, we have omitted discussion of Aristotle's
apodeictic sense of necessary proposition since it is very likely that
it is parasitic on some other sense of necessary proposition. If we

write "5 must be P," it is probably because we regard the statement

242

“S is P" as necessary in one of the other senses mentioned herein.
Thus we shall not consider this an independent view of necessity.1
It appears in view of the foregoing survey that if these
five/four2 senses of "necessity" constitute a complete list of the
senses of “necessity," then there are no necessary statements sip?
pliciter but only statements necessary relative to other statements
which are assumed to be true. As D. W. Hamlyn states in his article
"Contingent and Necessary Statements“:
To maintain this categorically would be to maintain
that necessary truth, as such, is relative. A
conclusion of an argument is necessary if the
premises are true.3
All necessary statements are necessary relative to other true
statements which entail them. The true premise statements themselves
are all contingent though some such as "'All S are P' is unrestricted"
are intensional contingent statements which establish or legislate a
necessary connection between 5'5 and P's. Necessity arises therefore
from entailment, and the necessity which characterizes any necessary
statement is fundamentally a type of logical necessity. This rela-

tionship between necessity and entailment is thoroughly discussed by

William Kneale.

 

1We shall soon see that there is no absolute sense of necessity
to which this view might refer.

2We mean here the four main senses discussed by Robinson and
the fifth somewhat related view of Russell.

3Hamlyn, op,_git,, p. 198.

243

Relative'necessity
William Kneale, in his wellaknown'article'"Truths of Logic,'

discusses two traditional views about what sort of truths constitute
the science of logic. According to one view expressed by Aristotle
in the "Prior Analytics," the truths of logic are principles of entail-
ment which, qua schemata, tell us what propositions follow from what
others. Formal principles of entailment are stated as general prin-
ciples or schemata with arbitrarily chosen letters to mark gaps which
would have to be filled if a genuine argument were to be produced.
Such an entailment schema might be:

"If p then q

But p

“Therefore q [pjpfl";'
and such a schema of entailment is distinct from a particular truth of
entailment that is an instance of it. The former would be a truth of
pure logic while the latter would be a truth of applied logic. When
examining the validity of individual arguments, Kneale maintains that
principles of entailment may be employed either as premises in the
presentation of the argument or, as is more generally the case, as an
implicit rule of inference.

The alternative account of what are considered the truths of

logic refers to certain propositions such as the principle of non-
contradiction and the principle of excluded middle which do not them-

selves look like entailments. According to this second account of

 

1William Kneale, "Truths of Logic," Aristotelian Society
Proceedings, XLVI (l945-l946), 209.

244

logic, its truths are principles of necessary truth or falsity, e.g.,
that the disjunction of any proposition and its negation is necessarily
true.

Both views concerning truths of logic can be synthesized into
one basic view, Kneale claims, if one but realizes that the metalin-
guistic entailments described by the first view "can easily be presented
as second-order propositions of necessary truth [315]" in the manner of
the law of nonécontradiction as described by the second theory of
logical truths. One might say as Kneale does:

If one proposition entails another we may say that the
hypothetical pr0position which has the first as protasis
and the second as apodosis is necessarily true.1

Due to the greater comprehensiveness of this second account of
the truths of logic, it has been the more widely accepted view. How-
ever; Kneale stresses that a confusion between "formal truisms" and
"logical truth" often enters into discussions of the second view.
Kneale regards object-level propositions, for example, the formula
"P or not P" as formal truisms. But a metalinguistic proposition
expressed in the formula "It is logically necessary that P or not P"
is a logical truth or truth of logic. In the past, Kneale maintains
that the confusion was strengthened by an unsatisfactory theory of
modality according to which phrases such as "it is necessary that" were
"supposed to refer only to the speaker's ground for asserting what

followed.“ In more modern times philosophers have often used the

 

1Ibid., p. 2l2.

245

phrase "it is a logical truth that p or not p" when they mean only

that the truth of the proposition is determined by its form. However,
this use of "logical" “can easily be taken to imply that the science
of logic consists of such truths as thatyp or not p.”1 Such a view is
found according to Kneale in the work of Frege,Whitehead, and Russell,
who supposed the science of logic to consist of formal truisms.

With this much apparatus, we can now turn to the crux of our
concerns in our discussion of necessity. Kneale has presented three
kinds of formulae as necessarily true: (l) formal truisms, (2) logical
truths which claim that such and such a truism is necessarily trUe or a
'logical truth, and (3)‘entailmentstatements2 which can be’ assimilated
to formulas of type (2). 'Though it might appear on Kneale's account
of formulae of types (1), (2) and (3) that theremay be prepositions
which are necessary in an absolute sense, we maintain as does Kneale
that there are not. Kneale refers to C. I. Lewis' view of strict
-implication'which we have previously rejected as an adequate accoUnt
of the entailment relation. Lewis brings the notion of modality within
his calculus by saying that p strictly implies q" is equivalent to "It
is necessary that p.3 q." The commonly cited disadvantages of Lewis'

theory* of entailment are regarded as points of advantage by Kneale in

‘—

1Ibid.,p. 213.
2Ibid., p. 2l4.

3This is in reverse order of presentation in previous
discussion.

“Though we have rejected the Lewis theory, it points toward
our own conclusion about necessity in this instance.

246

illustrating the relationship between entailment and necessity.
According to the Lewis view, a necessarily false proposition strictly
implies every proposition; and a necessarily true proposition is
strictly implied by every proposition. Kneale regards this supposed
predicament as a strength of Lewis' view in the following way:

For it can be shown that accordin to ordinary notions

of entailment a formal absurdity ‘i.e., the negation of

a formal truism) entails every proposition and a formal

truism is entailed by every proposition. These assertions

seem strange only because we do not commonly think of

absurdities or truisms in connection with entailment.

. In some recent works on logic it is sometimes said

that the propositions I call truisms are consequences of

the null class of premises. . . The special interest

of this result is that it enables us to understand the

relation between principles of entailment and those other

principles of logic which do not seem to have the form of

entailments. We have seen that entailment can be defined

in terms of necessity, but it now appears that necessity

*can be defined in terms of entailment.1

If the view expressed in the above passage is accepted, it would

appear that the necessity of at least logical truisms can be said to be
relative in that they are entailed or logically follow from the null
class of premises. This, however, is a way of arguing for the relative
necessity of these propositions which may be hard to accept. A better
way of stating the point might be that formal truisms are true neces-
sarily because of the semantical rules of the system in which they
appear. "As Carnap has pointed out (p. 30),2 the truth-value tables

in which we try to explain to ourselves our usage of such words as "and"

 

1Ibid.. p. 217.

 

aneale is referring to Carnap's Introduction to Semantics.

247

are really formulations of semantical rules."1 If we combine this last
thought of Kneale's thoughts on entailment and necessity, we find the

following explanation of the connection between necessity and entailment

 

previously expressed. Kneale writes, quoting from the Tractatus:

Those possible combinations of truth - values of its
arguments for which a truth - function is true may be
called its truth - grounds (5.101) and for a finite number
of arguments these can be indicated most conveniently in
a truth - value table of the form which now has become
familiar (4.31). One sentence entails another if the
truth-grounds of the first are included among the truth-
grounds of the second (5.12), and this is the only kind
of necessitation, because all elementary sentences are
independentof each other (5.134). It folldws that if
a sentence is true in all possible circumstances it is
entailed by every sentence, and just because it excludes
no possibility it says nothing (5.142).. . . Similarly
a self4contradictory sentence, which is not true in any
possible circumstances, entails every sentence.2

Expressed in these terms, Kneale's assertion of the connection
of necessity with entailment and entailment with necessity is perhaps
more acceptable.* Metalinguistic propositions such as "it is necessary
that 'p v ap' is true" might well be said to hold necessarily themselves
relative to the necessary truth of the truisnlwhich they mention. How-
ever, this latter contention might well be challenged by some logicians
such as Quine. In his article "Truth by Convention" Quine points out
the fact that some truths of logic such as the law of non-contradiction
(it is not obvious whether it is meant as a truism or a metalinguistic
formula) antecede the adoption of conventions (semantical rules) since

the very application of these conventions presupposes the adoption of

 

1Ibid., p. 233.

2Ib'id., p. 220.

 

248

logical truths such as the law of non-contradiction. In this view,
it would seem that a statement of the law of non-contradiction, if
necessarily true, is not necessarily true relative to anything but is
perhaps pragmatically adopted. This conclusion, however, need not
invalidate the view that all necessity claims are relative. Rather
we maintain that logical truths, even those referred to by Quine, are
themselves contingent assertions1 which legislate that certain proposi-
tions are to be considered as necessarily true. For example, Hospers
'writeS'the law of non-contradiction as "not both A and not A." He
then-says:

Sometimes, however, they are fermulated as truths about

propositions, in which case they (he's referring to the
laws of thought here) are all tautologies:

Law of Identity: If p, then p p :1 p

Law of Non-contradiction: Not both p -(p --p)
and not p

Law of Excluded Middle: Either p or pv-p
not-p

There is an advantage in this formulation in that the
three laws cannot be used as rules of inference in the
logical deduction of propositions. -But these are never-
theless only special cases-of our first formulation:
unless it were true that A is A, one could have no basis
for asserting that p is p (or have p implies p)."2

In view of the above passage, the statement of the law of non-
contradiction 'not both A and not A' is itself contingent but legisla-

tive.3 The logical tautologies such as '-(p--p)' are as Quine maintains

 

1Here we reject the Kneale solution at *.
2John Hospers, op. cit., p. 210.

3In terms of our IC theory in Chapter III we would write "'Not
both A and not A' is to be necessary

249

not mere truths of convention but necessarily true relative to the law
'Not both A and not A' which j§.a truth by convention.

Another candidate for a truth which may be necessary in an
absolute sense has been discussed extensively by Kneale himself as an
exception to his entailment-necessity views. Exceptional necessary
truths of this type are not formal entailment statements. An example
is "It is necessarily false that a thing is both red and green." Kneale
thinks that this statement reflects an "objective necessity independent
of all linguistic conventions."1 However, it might be countered to this
contention that if experience imposes belief in such a principle upon us,
it does so with no greater force than it does any other description
which is true repeatedly with no known violation. In an approaching
age where psychedelic phenomena may be the norm, it may well be that
something will appear to be red and green at the same time; or it may
be that principles of identity and individuation may change in such a
way that something identified as one thing might be regarded as both
red and green simultaneously as we now regard one thing as both short
and fat at the same time. Thus these non-formal principles have much
the same status as a highly confirmed hypothesis, and statements which
express these principles can be regarded as themselves contingent though
they may establish and report that it is to be considered necessary that
certain relationships pertain, in our example that no one thing can be

wholly two colors at the same time.

 

1Kneale, "Truths," p. 232.

250

From the preceding discussion, it can be concluded that those
propositions commonly regarded as necessary are necessary relative to
the semantical rules and definitions within the systems in which they
occur. Those other propositions which appear to be necessary in an
absolute sense are really not necessary at all but are more adequately
viewed as intensional contingent statements presented by Strawson which,
we have adapted as well as adopted herein. This conclusion leads us
back at this point to our main line of work--the explication of a
criterion for Kant's P-P distinction.

Before resuming our main discussion, we must pause to clarify
our position in light of Kneale's article with respect to the two
symbols "3" and "+3" According to Kneale, it would seem that Lewis'
characterization is to be adopted for all entailment relations (this
would include logical deductions). This, however, is contrary to much
of contemporary logic which seems to recognize a model interpretation
of the binary English connective "that . . . entails that . . ." but
yet seems to also recognize a non-modal sense of entails" such that
"a sentence A entails a sentence B iff it follows from the meanings of
A and B that B would be true if A should be true."' Of this second
interpretation of "entails" G. J. Massey says:

. if A and B are sentences of a P- language, then A
entails B if A implies B. . . . It can also be readily
shown that implication is a sufficient but not a neces-

sary condition of entailment among sentences of a P-
language. 2 .

1Gerald Massey, Understanding Symbolic Logic, pp. 79-80
(1 of 2) galley.

2Ibid., p. 80 (l of 2) galley capy.

251

In succeeding pages, we shall use "entails" in the non-modal
sense as it is defined by Massey. In the next section of this chapter,
we shall give a definition of "entails" more precisely tailored to our
discussion of presuppositions. However, "entails" should always be
understood as a 2-place metalinguistic predicate. If A entails B and

A is true, B is true necessarily, relative to A.

The necessity of a priori statements

We have already noted that necessity has been offered by Kant
as a mark not only of analytic judgments but also of a priori judgments.
The manner in which analytic statement tokens are necessary in our view
has already been described. The necessity of analytic statements is
relative to the definitions of the terms within them from which they
can be derived in accordance with certain logical laws.

If Kant's view that there are a priori statements which are not
analytic but synthetic is to be accepted, we must now discover in what
sense a priori statements are necessary which is different from the
sense in which analytic statements are necessary. Since all concepts
of necessity are relative, it must be the case that a priori statements
are necessary relative to something other than definitions or semantical

rules.

Presuppositions

We have already noted a reference in Kant's discussion of a
priori truths to the claim that they are necessary presuppositions of
experience. In the case of his paradigm example of an a priori truth,

the general law of causality, both the usual marks of a prioricity,

252

i.e., strict universality and necessity, are challenged by Hume. Kant
seeks to demonstrate in the "Second Analogy" that the general law of
causality is a necessary presupposition of experience and therefore
necessary, strictly universal, and of course an a priori truth. Before
attempting a reconstruction of "a priori," we think it a fruitful tact
to follow Kant's lead and discuss presuppositional truths in order to
determine if necessity and strict universality are adequate marks of

a priori truths qua presuppositional truths.

Little has been written on the subject of presuppositional
truths or presuppositions1 even though there has been intense interest
during the past 25 years in the related topic of contextual implication.
Contextual implication has been described as a "concept that applies to
those conditions that must be satisfied before an utterance can count
as normal in the circumstances in which it is made."2 Discussion of
contextual implication is mainly connected with develOpments in moral
philosophy, especially with the work of G. E. Moore. But the notion
of presupposing has arisen from developments in logical theory, partic-
ularly in the work of P. F. Strawson whose view of presuppositions was
developed in opposition to Russell's theory of descriptions. We shall
refer to Strawson's view in coming pages; however, we agree with writers
such as Wilfred Sellars, Avrum Stroll, David Rynin, and Roger Hancock

that Strawson's theory is an undesirable explication of "presupposing."

 

'We shall always mean what is presupposed and not the activity
of presupposing.

2Avrum Stroll, "Presupposing," Encyclopedia of Philosophy,
lst ed. (1967), p. 447.

253

As Sellars points out, Strawson's theory is more an account of when
statements are correctly asserted, i.e., contextual implication, than
it is an account of presupposing.

A more encompassing treatment of presuppositions is contained
in an older work, An Essay on Metaphysics by R. G. Collingwood. Though
Collingwood's treatment is somewhat confused and unclear, it has been
central in less recent discussions of presuppositions; and it covers
to some degree the major topics that are involved in any discussion of
presuppositions. We shall explore some of the lines of Collingswood's
view contrary to pronouncements by Alan Donagan, one of Collingwood's
major critics, who says:

'Collingwood's projected reformation was therefore too
riddled with confusions and contradictions to stay
afloat: and no considerable parts of its wreckage
can be salvaged.1

Despite this criticism, we believe that some worthwhile parts
may be either salvaged from Collingwood's discussion or suggested by
his mistakes. As D. Rynin contends:

No man willfully utters stupidities, involves himself

in obvious inconsistencies or commits his reputation to
patent errors. Collingwood could of course be as poor

a thinker as Donagan makes him outto be on the subject
under discussion (presuppositions). . . . In any case,

I wish to read Collingwood on the assumption that he is
not an intellectual dolt, and the least I can do is to
seek out, if possible, some interpretation of his thought
that leaves it not utterly ridiculous and obviously mis-

taken to see whether what he seems to be saying may not
be worth saying and even listening to.2

 

'1A1an Donagan, The Later Philosgphy of R. G. Collingwood, p. 279.

‘ 1 2David Rynin, "Donagan on Collingwood: Absolute Presup ositions,
Truth and Metaphysics," Review of Metaphysics, XVIII (1964-1965 ,
pp. 307-308.

254

Let us begin our quest of such an interpretation by considering
Collingwood's-initial'pronouncement'concerning'presuppositions.
Collingwood writes:

Whenever anybody states a thought in words, there are
a great many more thoughts in his mind than are
expressed in his statement. 'Amon 'these are some
which stand in'a peculiar relation to the thought he
has stated. 'They are not merelyits context, they
are itspresuppositions.1

The peculiar relation to which Collingwood refers in the above
passage is described by him at the end of his book as:

The relation here described as "presupposing" or
'being derived from' might,I take it, be understood

. as a logical relation, where 'b presupposes a'

means that a state of things a exists contemporaneously
with a state of things b, and a is an indispensable
condition of b.2

The former passage speaks about a relation obtaining between
the thoughts expressed by a statement and unexpressed thoughts; but if
this is any kind of logical relation, it must be regarded as holding
only between statements or what they express when stated but not between
expressed and unexpressed thoughts. ‘We shall thus view these thoughts
in the mind, referred to by Collingwood, as mentioned linguistically in
metalinguistic statements such as "My object-level statement ST2 is
logically related to the claim CM2 that . . . , which was in my mind
but not stated." In keeping with this linguistic approach, we might

rewrite Collingwood's latter passage as:

 

1R. G. Collingwood, An Essay on Metaphysics, p. 21.
2Ibid., p. 291.

 

255

A statement 8 which describes a state of things b
presupposes A, a statement which describes a state of
things a, when A is true if'B is true, i.e., when a is
a necessary (indispensable) condition of b. It is also
the case that the truth of'a statement A'is a necessary
condition for the truth of°B (present author).

In reading these two passages, two claims already can be made
about presuppositions. 'First, presuppositions are not stated. They
do not enter our object-level discourse. Secondly the truth of a
presupposition as a necessary condition for the truth of various
statements STZ, . . . , ST; is logically prior to those statements.
We may therefore write that if STZ presupposes CMz, then ST2 entails
CMZ. .

'It might seem that logical priority is a better characterization
of the antecedent ST2 than it is of the consequent CMZ. However, the
truth of ST, is only a sufficient condition for the truth of CM2 while
the truth Of'CMé'IS a necessary condition for the truth of 5T2. As a
sufficient condition for the truth of CMZ, the truth of ST2 may be
temporally or epistemologically prior to CMZ; but it is not logically
'prior; for the truth of some other antecedent, let us say 5T3, might
also be a sufficient condition for the truth of CMZ; and we might
mention CM2 in a metalinguistic discourse, saying that it is a pre-
supposition of ST, without considering the sufficient but non-necessary
condition ST2 as true. 0n the other hand, if the truth of CM2 is a

necessary condition for the truth of ST2 whenever 8T2.1§_mentioned as

 

1We useSTz as the name of a statement in this chapter since we
have already used ST1 in Chapter III.

256

true in a metalinguistic discourse CMZ, its presupposition must be
considered true; and thusCM2 is logically prior to 5T2.l

It is important that logical priority not be confused either
with temporal priority or epistemological priority. It is fairly easy
to note that if'CM, isa presupposition of 5T2, CM2 may be both logi-
cally prior to Slzand temporally posterior to 5T2. CM2 may be so
temporally posterior to ST: that no one may ever mention CMz though
they constantly state 5T2. It would seem, in fact, that since pre-
suppositions are always unstated, the statements which presuppose them
are always temporally prior to them. By the same token a presupposition
CMZ, qua unstated claim, may or may not be in our conscious awareness
when we state 5T2, a statement which presupposes it. We may discover
CM2 only when someone else points out the fact that our assertion STz
presupposes it. ’In this situation, CM2 would be logically prior to ST2
but epistemologically posterior to 5T2. With respect to epistemological
justification of CM2 as opposed to discovery of CMZ, it may also be the
case that CM2 is epistemologically posterior while logically prior to

‘ST Suppose we have always considered ST2 a true statement and ST2

2.
presupposes CMZ. In thiscontext,ST2 entails CM2 and CM2 is pre-
supposed. However, evidence may later indicate that, in fact, ST2 is
not true; and the rules of implication indicate then that cnz. as

merely a necessary condition for the truth of 5T2, may be true or false.

 

"This'assumes that-6M5 is also sufficient as well as necessary
or that all the other necessary CM's are mentioned.

257

CM2 which is presupposed'by'ST2 may not be true; its justification is
pending; and'therefore,'CMé'is logically prior but epistemologically
posterior to ST2 in the sense of epistemological justification.

In contrast to our characterization of presuppositions as
logically prior to the statements that presuppose them in the sense
that the truth of'a presupposition is a necessary condition for the
truth of these statements, P. F. Strawson claims that a statement S
presupposes a statement S"i 'th 'sense that the truth of S' is a
precondition of the truthaor falsity of S.1 In the sense given
"presuppose“ by Strawson, S does not entail S' as we claim in our
theory. As Strawson says:

It is selfacontradictory to conjoin S with the denial of
S' if S' is a necessary condition of the truth simply of
S. It is a different kind of logical absurdity to conjoin
S with the denial of S' if S' is a necessary condition of
the truth or falsit 'of S. The relation between S and S'
in the first case is that S entails S'. We need a differ-
ent name for the relation between S and S' in the second
case; let us say as above, that S presupposes S'.2
Objections to Strawson's theory

Several thinkers have raised objections both to Strawson's
definition of "presuppose" and to his motivation for formulating it
as he does. Let us consider objections to his definition.

Roger Hancock notes that the relation of presupposing as

Strawson defines it cannot be one which obtains between statements if

 

'P. F. Strawson, Introduction to Logical Theory, p. 175.
21bid.

258

statements are true or false by definition.1' A false presupposition
is presupposed by a nonastatement since Strawson claims that a true
presupposition is a necessary condition for either the truth or
falsity of'a statement. If this necessary condition is not satisfied,
i.e., if the presupposition is false and not true, then the so-called
statement that presupposes it is not true or false and so by definition
not a statement at all.2 a
Avrum Stroll notes another objection to Strawson's view raised
by David Rynin.' Stroll writes:
. . when "necessary condition" and "truth or falsity
‘of the statement that" are interpreted in the ordinary,
truth- functional way, the definition has the paradoxical
‘ consequence that all presupposed statements are true.
Rynin' s demonstration is that (S o S' ) and (- 5.3 S' ),
but (S v-S) therefore 5'.3
Stroll also discusses an objection which deals more specifically

with the goal Strawson seeks to achieve via his definition of presuppo-
sition. ‘Strawson rejects Russell's theory of descriptions. It will be
remembered that according to Russell a sentence such as "The king of
France is bald" contains in part an assertion that the king of France

exists. Strawson argues that this existence claim is not an explicit

part of the sentence "The king of France is bald" but is rather a

 

1Bas C. Van Fraassen discusses presuppositions when this
principle of bivalence is given up in his article "Presupposition
Implication, and Self Reference," Journal of Philosophy, LXV (1968),
136-151.

2Roger Hancock, "Presuppositions," Philosophical Quarterly, X
(1960), 73.

3Stroll, op. cit., p. 448.

259

presupposition of such a sentence when a speaker uses it to make a
statement in normal circumstances. Stroll raises the objection that
"if 'The king of France no longer exists' is used to make a true
statement then by Strawson's criterion one who employs it thereby
presupposes the existence of the king of France. If we disregard
this last criticism for a moment, there surely is some sympathy for ‘51

Strawson's wish to maintain that we cannot say whether the sentence

 

"All John's children are sleeping" is true or false unless it is true £
that John has at least one child. However, as Sellars points out,

there is a big difference between its being false that 'All John's

children are sleeping' when in fact John has no children and its being

correct to say "'All John's children are sleeping' is false" when John

has no children. Sellars comments:

. . . if John were to say "The table is large," it would
be'a mistake for Smith to say "No, that's false" unless
' he believes that the uniqueness condition is satisfied.
‘-If-he doesn't he should say rather, “There's no table
over there," or, "There are seven tables over there,
which one do you mean?" Strawson infers from this that
the utterance is neither true nor false unless the unique-
ness condition is satisfied. . . . But even though both
the original utterance and the reply presupposes that the
uniqueness condition is satisfied, the utterance is never-
‘theless false if the uniqueness condition is not satisfied.
‘The utterance is false if the uniqueness condition is not
satisfied, even though it is not correct to say that it is
false unless one believes that the condition is satisfied.1

In agreement with Sellar's conclusion, we maintain that a

presupposition is an unstated claim whose truth is a necessary condition

 

1Wilfred Sellars, "Presupposing," Philosophical Review, LXIII
(1954), 207-208.

260

for only the truth of a statement which presupposes it. 'We maintain
this not only in light of the above objections to Strawson's view but
also in keeping with the ordinary sense of "presuppose“ in English
which we shall soon discuss.- This ordinary use of "presuppose" (one
of the senses of "presuppose“ in an American dictionary) is a stronger
meaning than that of terms such as “assume" or "suppose." But let us [H‘
begin the next section by discussing the weaker sense of "presuppose" E

which identifies it with "assume" and "suppose." ‘

The weaker sense of fipresuppose"

Collingwood in his discussion of presuppositions goes on to
define a proposition as that which is stated, i.e., that which can be
true or false, and the stating of propositions is propounding them.

Thus in keeping with Collingwood's somewhat dated terminology, only
propounded propositions are statements. It should be noted here that
Collingwood is implicitly stating, via his definition of proposition,
that statements are those linguistic entities which may be either true
or false. Thus far in our discussion, we have been talking about state-
ments which presuppose unstated claims. But Collingwood now goes on to
maintain that every question involves a presupposition; and that "to say
that a question does not arise" is the ordinary English way of saying
that it involves a presupposition which is not in fact being made.1

For example, we gpplg_say that the question "Is this clock accurate"

presupposes that the clock 1§_sometimes not accurate; and if such a

 

1R. G. Collingwood, op. cit., p. 26.

261

presupposition is not made, the question of accuracy does not arise.
But what claim is Collingwood really making here? When such a question
arises, it indicates more clearly perhaps that we no longer presuppose
that the clock is accurate which is not to say that we do presuppose
that the clock is not accurate. ‘Perhaps it can be said that when a
question of this sort arises we presuppose that it mpy_be the case that
the clock is not accurate. 'But if we presuppose that the clock may not
be accurate, it could still be said that we presuppose also that the
clock may be accurate. What kind of presupposing is this?
The dictionary definition of "presuppose" is:
l. to suppose or assume beforehand; take for granted.

2. to require or imply as a preceding condition as,
‘a healthy body presupposes healthful living.1

As a synonym, the dictionary gives "presume." The definition of
"presume“ is the fellowing:
l. to take upon oneself without permission or authority.
2. to take for granted; accept as true until proof to
the contrary is furnished; suppose, presuppose . . .
'“presuppose" is the broadest term here, sometimes
suggesting a taking something for granted unwarrant-
edly . . . and in another sense, implying that some-
thing is required as a preceding condition.2
The above definition confirms the uneasiness expressed pre-
viously about Collingwood's claim that every question presupposes some
statement. In the case of the question "Is this clock accurate?" we

indeed are not taking for granted that the clock is not or may not be

 

1Webster's New World Dictionary of the American Language, ed. by

J. H. Friend, p. 11551
2Ibid., p. 1154.

 

262

accurate (dictionary sense (1) of 'presuppose'); nor would the possible
inaccuracy of the clock be a necessary precondition for asking such a
question (sense (2)of 'presuppose'). Thus, it appears that Collingwood
is at the outset of his discussion using"presuppose' in some unex-
plained and weak sense which is neither of the usual senses of the term
as found in our English language dictionary. Rather he seems to be ‘1‘

using 'presuppose' loosely as a synonym for 'surmise' which means

 

according to Webster's dictionary: ". . . an idea or opinion formed )
from evidence that is neflther positive nor conclusive; conjecture;
guess."1
Faced with two timepieces with different readings, the question
"Is this clock accurate?" reflects not any particular presupposition but
only a guess that of the two timepieces it may be the clock which is
inaccurate. The word "guess" is indeed a weak synonym for "presuppose“;
however, it may not really be weaker than "assume," a synonym mentioned
in dictionary sense (1) which did not seem prima facie to capture Col-
lingwood's use of "presuppose" with regard to questions (as we claimed
above in our clock example).
In his article "Presuming," Roland Hall says of "assume":
- Elsewhere I have argued that the latter in its positing
uses (except the prothetic uses) indicates, when used
protopersonally, the possession on the part of the
speaker of slight grounds for making the assertion which
it introduces or which flank it, and rules out knowledge

of the truth of that assertion while leaving the speaker's
commitment to its truth-intact.2

 

11bid., p. 1467.

~ 2Roland Hall, "Presuming," The Philosophical Quarterly, XI
(1961), lo.

263

Hall's account of the meaning of "assume" is quite compatible
with the definition of “surmise“; and we conclude, therefbre, that the
weak sense in which Collingwood is using "presuppose" with respect to
questions’is'probably'dictionary'sense'(l)of‘”presuppose."

'In view of this initial confusion between the weak and strong
senses of "presuppose" in Collingwood's discussion (he explicitly adopts
sense (2) in his formal definition of "presuppose" but then uses the
term in sense (1) when discussing questions), it may be well now to
consider other relations that may hold between statements such as
supposing and assuming. But before undertaking this discussion, let
us recap what can be said about presuppositions thus far. A presup-
position is:

l. logically prior to the statement(s) that presuppose(s) it.
2. 'not necessarily or even usually epistemologically in either
' *a discovery or justificatory sense of the word “prior" to

'the statement that presupposes it.

3. an unstated claim CMn which does not occur within the context
of the discourse in which STn the statement which presupposes

it occurs.

4. in one sense, dictionary (1), much like an implicit or hidden
assumption which we take for granted as true.

5. in another sense (2) not only an entailed hidden claim that is
true but one that must be true since its truth is a necessary
condition for the truth of STn, the statement which presupposes
it.

6. best understood in sense (2) formally adopted by Collingwood,
since sense (1) is very general and can serve as the meaning
of many related terms such as supposition, or assumption.

 

264

Assumptions

Collingwood goes on to claim that when a person makes an
assumption, he is making a supposition about which he is aware "that
he might, if he chose, make not that but another."1 Thus for Colling-
wood all assumptions are suppositions, but all suppositions are not
assumptions, for some suppositions are made when the person making them
is unaware that there are other alternative suppositions which he could
have made. It may be well to note here that Collingwood is introducing
a psychological factor in discussing assumptions. A person may think
that there are alternatives to his supposition even when there are not
and therefore consciously think that his supposition is an assumption.
‘-0n'the-other'hand 'in-cases wher 'indeed there are alternative supposi—
tions that could be adopted, a person may not know that there are and
thus not regard the adoption of a particular supposition as an act of
assuming.' Collingwood sums the matter up well by claiming that "To
assume is to suppose by an act of free-choice."2 When one says "let
us assume x=rlO," there is an implication of free choice attaching to
the term "assume."

0f suppositions, Collingwood claims that their logical efficacy
does not depend upon the truth of what is supposed or even on its being

thought true but only on its being supposed. As he says:

 

-'Collingwood, op 'cit., p. 27.
2Ibid.

2. nlllie

265

It (that the logical efficacy is constant) is a matter
of'common‘knowledge'in'the'conduct'of'scientific'thinking;
where it is possible and profitable to argue from supposi-
tions which we know to be false, or which we believe to be
false, or concerning which we have neither knowledge nor
belief as to whether they are false or true.1

It should be noted in the passage above that Collingwood is
using "supposition" as if it is equivalent to "assumption" though he
has already said that the two terms are not equivalent. If an assump-

tion is implicit, we probably are not even aware that we are assuming I,

 

it. Thus we are not aware that our implicit assumption has alternatives )
since we are not even aware of our implicit assumption in the first
place. It is hidden! It seems then in terms of Collingwood's own
criterion concerning the awareness of alternatives that he has not made
a distinction between suppositions and assumptions but rather a distinc-
tion between explicit or implicit assumptions or suppositions. There-
‘fore, from this point on, we shall use "supposition" and "assumption"

as synonyms and dISCUSS'thOSE that are explicit and those that are
implicit. Assumptions, even implicit ones, may have alternatives
whether we are aware of them or not; but as we have said already, pre-
suppositions as necessary preconditions do not. Let us move on now to
consider the specific differences between assumptions (suppositions) and
presuppositions. We shall start by discussing what is meant by our use

of the word "alternative."

 

1Ibi ., p. 28.

266

Alternatives

Collingwood distinguishes between suppositions that are taken
for granted, sometimes unwarrantedly, and suppositions1 which are not
taken for granted but consciously adapted as true when it is realized
that there are alternative suppositions which could have been adopted
equally as well.' This latter underlined phrase can be interpreted in
'two ways. 'First, (I) we might say that an assumption A1 has viable
alternative assumptions A; or A3 or A, or A, or A, , any one of which
we could have adopted as true instead of A1 in order to derive a
particular statement ST2 which we regard as true. 0n the other hand
(II), it might be the case that we have no great attachment for 5T2;
and we could have adopted A, or A, or A“ or A5 or A6 instead of A1,
any one of which would not yield'ST2 deductively but rather 51,, ST“,
and STS, respectively. When mathematicians adapt an assumption, e.g.,
A1, they are usually aware that they could have adopted instead another
assumption A: or A3 or A“ in sense (II), i.e., they adopt an assumption
in order to follow wherever it might lead perhaps to S3 5“ SS, respec-
tively, or $6.

In this paper, we wish to confine our discussion of assumptions
to any object-level discourse in which a particular true statement ST2
does occur and therefore, when we speak of alternative assumptions to
A1, we will mean view (I) that there are other assumptions which when

adopted as true entail‘ST2 as well as A1 entails ST2 when adopted as

 

1We identified these with assumptions.

 

267

true. 'Therefore,'any'assumption'Ai,..., At individually is sufficient
for the truth of ST,; but none are individually necessary for the truth
of STz. None are individually a required condition for the truth of 5T2.
Thus if we look for the presupposition of 5T2, it would pgpppp§_be the
unstated claim CM: which would be the disjunction of all the sufficient
COHthTOHS'fOT'STé'Tn cases where it is possible to list them, when they '7‘

are finite in number.‘;;fithis disjunction of sufficient conditions

 

constitutes a necessary condition for the truth of 5T2, as it often )
does, and no disjunct is stated on the same level of discourse as 5T2,
then this disjunction is a presupposition of 5T2.1 However, if all the
assumptions or any of them are stated within the same level of discourse,
then our disjunction is not a presupposition but perhaps a presumption
(see def., p. 261). We see here the connection between the phrase
"hidden assumption" and "presupposition." Hidden claims without alter-
natives are presuppositions; whereas, hidden claims with alternatives
are merely hidden assumptions. 'Hidden assumptions have alternatives
even when we are not aware of them. 'But presuppositions do not no
matter what we think. The truth of the antecedent is sufficient for

the truth of the consequent; and the truth of the consequent is neces-
sary for the truth of the antecedent; therefore, the consequent is a

necessary precondition, a presupposition, of the antecedent.

 

" 1This excludes 5T2 itself as one of the A1,...,A5. Though ST2
is sufficient to entail itself, 5T2 is always a stated Ax. But when we
rule ST2 out as one of the A's we also leave open the possibility that
an exhaustive disjunction of sufficient conditions is not automatically
‘ a necessary condition. But this possibility is also the possibility
that there are some statements which presuppose nothing but themselves
which is doubtful.

268

To summarize then our additional conclusions about
"presuppositions" up to this point we can say:
7. Suppositions are assumptions, and assumptions are suppositions.

8. Suppositions, i.e., assumptions may be explicit or implicit
(hidden).

9. When assumptions are implicit, we are unaware that they have .
'alternatives or that there are alternatives we have not taken 15‘
into account. 'If a claim has alternatives of which we are
aware or are not aware, a claim is an assumption. But if an .
unstated claim has no alternatives, then it is a presupposition. .5,
If the truth of a stated claim is necessary for the truth of a "
statement which entails it, then this stated claim is not a 1
presupposition but a presumption.

 

10. ‘A presupposition in sense (1) is the same as an implicit
assumption or a hidden assumption.

ll. All assumptions have alternatives which we can adopt in their
place either in view (I) or (II); thus if a "hidden assu tion"
can be established as lacking any alternative in sense (I , it
is a necessary precondition and, therefore, not an assumption
'at all but a presupposition in the dictionary sense (2).
‘Before'going on, let us look at the two dictionary senses (l)
and (2) of “presupposition" more closely. Collingwood makes a final
distinction concerning presuppositions, one between relative and
absolute presuppositions. But careful reading will indicate that this
distinction is none other than the distinction between the dictionary
sense (1) and (2) of "presuppose."

Collingwood declares that by a relative presupposition he means
one which "stands relatively to one question as its presupposition and

relatively to another question as its answer."1 In explanation of this

rather enlightening distinction, Collingwood describes the situation

 

1Collingwood, op. cit., p. 29.

269

where one asks “What is the distance between these two points?" while
measuring the distance with an old measuring tape.' This question pre-
supposes that the old tape has not stretched and that it is still an
accurate measuring device.' Such a presupposition is relative because
"the accuracy of the measuring tape, which while I-am using it is a
presupposition of the questions I ask, is one of the two possible ‘5‘

answers, the affirmative answer, to the question I ask while I'am thus

 

checking it."1 If, as Collingwood describes, the accuracy of the tape “
is the presupposition which gives rise to the questions of "How long
or what distance does the tape indicate?" then it is certainly not the
case that the presupposition itself is being questioned. The other
"checking question“ to which he alludes, "Is this tape accurate?"
arises not from a presupposition at all but from a supposition, "This
tape may have stretched" or the alternative supposition that this tape
may have been manufactured faulty. 'If eithe 'of these suppositions is
entertained, then the presupposition that the tape is accurate is no
longer presupposed. It appears from the rest of his discussion that by
relative presupposition Collingwood means something which, though taken
for granted, can nevertheless be questioned; for he says:

To question a presupposition is to demand that it should

be 'verified'; that is, to demand that a question should

be asked to which the affirmative answer would be that

presupposition itself, now in the form of a proposition.

To verify the presupposition that my measuring-tape is

accurate is to ask a question admitting of the alterna-

tive answers ”the tape is accurate," or "the tape is not

accurate.“ Hence to speak of verifying a presupposition
'involves supposing that it is a relative presupposition.2

 

lI id., p. 30.
‘ZIbid.

 

270

The above passage clarifies Collingwood's position, but there
appears now to be something amiss with the position itself. Once we
question a presupposition, can we really still regard it as a pre-
supposition?' There seems to be afoot an implicit contradiction in
terms. Presupposing is not questioning, and questioning is not pre-
supposing. Questioning the truth of'a claim is what we do when we no
longer presuppose it. We no longer take it for granted but suppose that
some other claim, e.g., CMS, may be true.~ However, these last remarks
arise from an understanding of "presuppose" in sense (1) which is really
the definition of assumption or supposition. It is more clearly the
case then, on our reconstructed view, that assumptions and suppositions
when questioned are no longer assumed or supposed; and this contradic-
tion brings us back to the confusion in Collingwood with respect to
suppositions (though it appears that in his passage he means specif-
ically suppositions that are assumptions). He says that the logical
usefulness (efficacy) of*a supposition does not depend upon whether the
supposition is thought true or not.1‘ In fact, we may well know it to be
false." However, even if we know a statement or claim is false, if we
adopt it as an assumption in our deduction, we are treating it as if we
regard it as true. Even in reductio proofs, where we wish to demon-

strate that an initial premise is in fact false, in the act of assuming

 

1It will be recalled that suppositions in Collingwood's View
are entertained when one is unaware of their alternatives. Assumptions
are entertained when we are aware of their alternatives. We have chosen
to disregard :this awareness factor and regard suppositions as assump-
tions if they.in fact have alternatives whether we are aware of them .
or not.

 

271

it, we adopt it as true and treat it as if it were true in order to
show that when treated as true it leads to an absurdity; and therefore,
it must be false. The same mistake is made by Collingwood in regard to
presuppositions understood in sense (2). If a statement STx occurs in
our discourse and we investigate which claim CMx it presupposes, we may
find that STx entails CMX. But if STx does entail CMx and we consider
STx true, then we cannot question the truth of CMx. CMx must be true
if STxis true when STX entails CMX. To question the truth of CMX when

 

 

the underlined conditions obtain is to question the fact that it is
presupposed. We may perhaps question whether a certain CM2 is pre-
supposed by ST2 without questioning the truth of that CM2 in cases where
ST2 itself is not considered true. But when ST2 is considered true, we
may question whether it presupposes CMZ; but we cannot question the
truth ofCM2 without questioning whether CM2 is presupposed by 5T2.

To make this position quite clear, let us consider a particular object-
'level discourse which contains a statement 3T2. Operationally we might
enlist a stenographer to write down all the statements made during a
particular discourse in a certain room between time t1 and time t2.
Looking at a stop watch, we tell her when to start writing and when to
stop. No other statements are written down but those that are stated
between t1 and t2. Our example statement ST2 is a statement that is
made between t1 and t2. Now as the stop watch stops at t2, we take the
pad from our stenographer and examine the statements written upon it.
After perusing the list, we may decide to order the statements logically

where any entail one another or other unmentioned claims; and we would

 

 

j. “HI :' IIII

272

write down these relationships on a second sheet of paper. When these
determinations are complete on our second sheet of paper, we find
mention of not only 5T2, but also of claim CM2 which ST2 presupposes.
If we look from our second sheet back to the first sheet, we find no
statement CM2 written on it. If we were questioned as to why CM2 does
not appear on sheet 1 but does appear on sheet 2, we would answer that
sheet 2 contains assertions about the statements written on sheet 1.
Thus on sheet 2, we find a statement MSTI1 "ST2 presupposes CMZ"; and
MSTi is a metalinguistic statement about STZ. We might equally well
write the metalinguistic statement "If ST2 is true then CM2 must be
true." When Collingwood says that a relative presupposition is one
that is subject to verification or is a statement which can be ques-
tioned, he is making a second-level meta claim MMSTZ, "'While 'ST2
presupposes CMZ' is true,’ it may be that ST2 itself is false; and
thus CM2 may be false," or, on another interpretation more in keeping
with Collingwood's confused identification of sufficient but non-
necessary hidden assumptions and presuppositions, the statement MMSTZ,
"'While 'ST2 seems to presuppose CM2,' is true,‘ ST2 also seems to
presuppose CM3, which is contrary to CMZ; and we must verify which CMx
is true since ST2 is true and can therefore only entail a presupposition
that is true." We will ignore this second second-level meta-statement
which arises from a basic error in Collingwood's discussion. He accepts
sense (1) of presupposition which violates his own definition of "pre-

supposition" which agrees with sense (2). Sense (1) captures the

 

vvrf

1Notation for first-level meta-statements.

273

meaning of "supposition"; and as the addition of a prefix suggests,
"presupposition" has a different meaning than "supposition." The
second second-level meta-statement is in fact the one that Collingwood
is discussing; but for interest's and completeness' sake, let us con-
sider the first second-level meta-statement MMSTI, "'While 'ST2 pre-
supposes CMZ' is true,‘ it may be that ST2 itself is false; and thus
CM2 may be false." This second-level meta-statement may be made about
any presupposition, i.e., any unstated claim CMx which is a necessary
condition for the truth of STX, when the truth of STx is not verified.
However, there are cases where certain statements that we make are
undeniably true, e.g., any true particular affirmative statement.1
Now if these true statements STx presuppose some CMX, then CMx must be
true. We could state this fact by making the first-level meta-statement
MSTZ, "Since ST2 is true, and ST2 implies CM2; CM2 must be true, i.e.,
is true necessarily," When a statement such as MST is true, we say
that CM2 is a presupposition which is necessary, i.e., a necessary
presupposition.

Mere presuppositions, which are commonly regarded as relative,
can be viewed as relative in the sense that they can be either true or
false because STZ can be either true or false; and we do not know which

it is, true or false. On the other hand, a necessary presupposition CM2

 

1For elaboration see N. R. Hanson, "A Note on Statements of
Fact," Anal sis, XIII (1952), 72. Factual statements are corrigible
in the light of further evidence but statements of fact cannot be
false.

274

can be characterized in terms of relative necessity since it is true
necessarily only in relation to the known truth of ST2 which presupposes
it. This is consistent with our view expressed earlier that all aSSEr-
tions of necessity are assertions of relative necessity. In view of
these confusing and competing senses of "relative," we shall not call
any presuppositions relative. We shall, however, remember that some
presuppositions may be true necessarily and that this necessary truth

is relative.

A necessary presupposition, i.e., a presupposition which is
true necessarily relative to some true statement ST2 which presupposes
it, may be what Collingwood calls an absolute presupposition. He
defines an absolute presupposition by saying that it is "one which
stands relatively to all questions to which it is related as a pre-
SUpposition, never as an answer."1

By this, Collingwood means basically that an absolute presup-
position is one which is "not verifiable."2 He explains:

This does not mean that we should like to verify them
but are not able to; it means that the idea of verifi-
cation is an idea which does not apply to them, because,
as I have already said, to speak of verifying a presup-
position involves supposing that it is a relative
presupposition.’

In the above passage, it seems that Collingwood's view coin-

cides with our own previously expressed one that questions of verifi-

ability are not pertinent to presuppositions. But where he sees that

 

1Collingwood, op. cit., p. 31.
21bid., p. 32. 3 '

3Ibid.

275

this is the case with respect to absolute presuppositions, we claim
that this is the case with respect to what we have described as
necessary presuppositions. In view of the fact (stated in the above
passage) that a presupposition (absolute in Collingwood's view) is not
subject to verification, Collingwood goes on to state that these pre-
suppositions are not propositions (statements) since the distinction
between truth and falsehood does not apply to them. He says, afterall,
that questions such as "Is it true?" "What evidence is there for it?"
"How can it be demonstrated?" are not relevant to absolute presupposi-
tions. There are many issues involved in the above Collingwood claim.
There are situations when the question "Is it true?“ or "What
evidence is there for it?“ is not or should not be asked because a
statement is obviously true. As Austin would say, if the pig is in the
room, we do not ask for evidence that it is true that the pig is in the
room. Likewise, when a proposition is presupposed by a statement ST2
which is known to be true, it is not in need of verification simply
because its truth is automatically assured by its being presupposed.
No proposition (statement) that is false or supposed false can be a
presupposition of other statements which are true. However, the nec-
essary truth of presupposed claims does not make them non-statements
even when statements are defined (as Collingwood defines them) as that
which may be true or false. That which is entailed by another true
statement is true and necessarily so. The necessary presupposition CMZ,
therefore, satisfies one of the disjuncts of being either true or false,

viz., being true and necessarily so.

276

We see then that presupposition statements are necessary1
because they are entailed by other true statements. But this necessity
is a logical matter, a function of the role of the statement relative
to other true statements which presuppose it; and this necessity does
not preclude the possibility of empirical origination (discovery) or
justification of this presupposition. A presupposition is not in need
of verification or justification when it is presupposed by a true
statement; fbr its truth is presupposed, but this is not to say that
presuppositions are immune to verification or falsification should
empirical findings (which, by the way, may not be sought after for
purposes of falsifying or verifying the presupposition) indicate that
what has been presupposed is in fact true or false. We do not question
our presuppositions or seek to prove them true, but if evidence comes
our way then a presupposition CM2 which it negates might well become
falsified.2 If this occurs, we would likely have to regard ST2 which
presupposes CM2 as false, i.e., we could no longer regard ST2 as true.
A falsified presupposition requires us to give up 5T2. It will be
recalled here that a falsified assumption or supposition does not

require that we abandon 5T2, for assumptions and suppositions in our

 

1In our non-modal sense, we mean.

2We may perceive some object emerge right before our eyes. We
have in no way moved our head and the perception of this object seems
to have just popped into existence. It is an anomaly which we cannot
integrate into any conceptual pattern fbrmed by our present causal laws.
If this kind of perception randomly reoccurred we would have observa-
tional evidence against the uniformity principle of causality and some
evidence to at least suspect the universality of causality.

277

reconstructed theory of presuppositions are not necessary for ST2 or
entailed by 5T2. Rather assumptions and suppositions imply ST2 while
ST2 implies not assumptions and suppositions but presuppositions.

Thus it is the case that a presupposition is not immune to
falsification. It could itself be falsified by evidence in which case
it could no longer be logically prior to a tppg_statement ST1 which
presupposes it. A presupposition CM2 might also be falsified or at
least suspected of being false if ST2 which presupposes it is somehow
falsified. In addition, to maintain that one statement is logically
prior to another is not to make any assertion as to the origin (empir-
ical or otherwise) of either statement. A presuppositional statement
nay originate (be suggested) via empirical research. It may even be
justified as true by empirical research, but when it is presupposed by
another statement its empirical suggestion or justification becomes
secondary to the role that it now plays as a presupposition in relation
to other statements within a deductive system which presuppOse it.

Let us summarize what can now be said concerning presuppositions.
A presupposition is:

l. logically prior to the statement(s) 5T2 that presupposes it.
2. not necessarily or even usually epistemologically (in either

a discovery or justificatory sense of the word) prior to ST2

the statement that presupposes it.

3. a usually unstated claim CM2 which does not occur within the
context of the discourse in which ST2 the statement which pre-

supposes it occurs.

4. in one sense, dictionary (1), much like an implicit or hidden
assumption which we take for granted as true.

10.

11.

12.

13.

14.

15.

16.

17.

18.

278

in another sense, (2), not only a hidden claim that is made
but one that must be made since its assumption (adoption as
true) is a necessary condition for the truth of ST2 the
statement which presupposes it.

perhaps best understood only in sense (2) adopted by Collingwood
since (1) is very general and can serve as the meaning of many
related terms such as supposition or assumption.

Suppositions are assumptions, and assumptions are suppositions.
Assumptions, may be explicit or implicit (hidden).

When assumptions are implicit, we are unaware that they have
alternatives or that there are alternatives we have not taken
into account.

A presupposition in sense (1) is the same as an implicit
assumption or a hidden assumption, viz., an unstated assumption.

All assumptions have alternatives which we can adopt in their
place either in sense (I) or (II); thus if a "hidden assumption"
can be established as lacking any alternative in sense (I), it
is a necessary precondition and therefore not an assumption at
all but a presupposition in sense (2).

Presuppositions can be discovered through experience though it
is not the case that they must be discovered through or arise
from experience.

Presuppositions can be falsified directly by empirical evidence
though perhaps not immediately, for their logical role may
incline us not to accept falsifying evidence.

Presuppositions can become suspect as untrue indirectly when
the statements STn which presuppose them are falsified.

Falsification of a presupposition demands that ST, which
presupposes it be abandoned as untrue.

Falsification of a supposition does not demand that ST2 which
it implies be given up or abandoned as untrue.

Questioning the truth of CM2 is no longer to presuppose it if
ST2 which entails CM2 is true.

If ST which presupposes CMz is regarded as false, any question-
ing oi CM2 is compatible with saying that ST2 presupposes it.

279

19. Statements that STx presupposes, CMX, are all metalinguistic
statements.

20. Virtually all presuppositions are considered necessary pre-
suppositions, for we rarely entertain or state false statements
knowing that they are false. Except for very speculative
disciplines, STx which presupposes CMx is usually considered or
thought or known to be true; however, it may not be.

With this much determined about presuppositions, it is well now

to return to our discussion of Kant.

Kant'spproof of an a priori principle

Kant claimed that the general law of causality (GLC), "Every
event has a cause," is a necessary presupposition of experience,1 if
the distinction between subjective perceptions and objective perception
successions (events), is to be made. To demonstrate that the general
law of causality is a necessary presupposition of experience, Kant
showed that GLC was entailed by various premise statements which are
true.

It is likely that by "necessary presupposition" Kant meant
merely “necessary condition." On our theory, of course, "necessary
presupposition" would constitute a redundancy; for "presupposition"
already means a necessary precondition. However, we have reserved
a non-redundant interpretation for "necessary necessary precondition,"
i.e., necessary presupposition, but is not only that the truth of a

claim CM2 is a necessary precondition of a statement ST2 but also that

 

1It is important to note that GLC is a necessary presupposition
of experience only if experience is equivalent to the discernment of
objective perceptions and subjective successions of perception. This
need not always be the case. Thus GLC is only a presupposition of this
or that kind of experience or part of experience.

280

CM,, as a necessary precondition, is also true necessarily since it is
entailed by ST,, a true statement. Whether Kant had this view in mind
when he used the phrase "necessary presupposition“ is not important,

for his argument in the "Second Analogy" is a classic proof of a CM2
which is a necessary presupposition in our sense of the phrase "neces-
sary presupposition." As we have already seen in the "Second Analogy,“ »
Kant provides a deductive argument as proof that "Every event has a
cause." This claim, which we have referred to as the general law of
causality, is the conclusion of an argument which Kant deduces from

five premises. The premises are several statements ST3 ,..., ST, which
jointly entail GLC. As such these statements ST3 ,..., ST, could be
interpreted as assumptions or suppositions from which GLC may be deduc-
ible. From previous close examination of Kant's argument in Chapter II,
we have concluded that it is the case not only that STa ,..., ST, jointly
entail GLC but also that ST3 ,..., ST, are true. Thus it can be said
that GLC must be true. We see up to this point that in our terminology
GLC is a necessary precondition of ST3 ,..., ST,; however, there is no
apparent reason yet for terming GLC a presupposition rather than a
theorem1 within some system of ordered discourse in which all the
statements of Kant's argument are asserted. The crucial aspect of
Kant's discussion becomes obvious when one interprets Kant's argument

as actually a metalinguistic discourse in which ST3 ,..., ST7 are

 

* 1To say that ST; ,..., ST, jointly entail GLC is not to say
that GLC is a presupposition of ST, ,..., ST, yet, for ST, ,..., ST7
could well be stated assumptions of an axiomatic system from which
GLC might be deduced as a theorem.

281

metaphysical or quasi—linguistic, meta-statements about the experiential
or empirical justification for the truth of any object-level statement,
st1 "This is an event A," or st2 "This is a happening," or st8 "I per-
ceive an objective successive sequence of percepts."1 It is Kant's
point that all these object-level statements that all of us make at
sometime or other are assertable only in virtue of the nature of our

ST , ST , ST . The

perceptions as described in the premises ST,, ST“. 5 5 7

conclusion of Kant's argument, GLC, is not a statement that is asserted
within the discourse of empirical or perhaps, as Kant might say, experi-
ential discourse. Thus on the object-level we have statements st1 ,...,
stn about this or that event. GLC is never an stx statement. No one
can say on purely observational or experiential grounds that they see
or perceive that every event has a cause. Thus GLC is not a statement
made within the object-level system of discourse. But the five premises
of Kant's quasi-metalinguistic argument which describe the empirical
given [also described by the object-level statements st1 ,........ stn]
imply the unstated claim GLC. It is in this way that GLC is a presuppo-
sition, an unstated claim whose truth is a necessary condition for the

truth of the conjunction ST,, ST STS, 5T6, and ST7 and thereby every

“9

 

lIf 5T3, ST“, ST, ST , and ST ,--the premises of Kant' s argument
were straightforward metalinguistic statements which mentioned object-
level statements we would have written them as MST3, MST,, MSTs, MSTs,
and MST7 as has been our method of notation so far. However ST, ST,
ST,, ST,, and ST, have an object- level appearance while serving what I
call a metaphysical metalinguistic function of describing those charac-
teristics of perceptions which determine the assertion of object-level
statements about reality. These objectelevel claims I choose to refer
to as st , st2 ..., stn to preserve in this particular case the differ-
ence in evels of discourse which I believe Kant' s argument exhibits.

282

st1 , ..... , stn about this or that event. In assessment of Kant's
argument, we might make the quasi—second-level metalinguistic claim

that "ST,, ST ST ST,, and ST, presuppose GLC."1 In sum, Kant's

“9 5’

move is to show that st1 ,..., stn each presupposes ST,, ST“, STS, STS,
and ST, and therefbre st ,..., stn each presupposes GLC. But the
force of Kant's argument is Still not fully apparent. If we examine
it again, we note also that, in as much as each of his premises are
true, GLC must be true, i.e., a necessary precondition. GLC is not
merely implied by five premises which are false or perhaps true or false
in which case GLC would be a precondition stillborn with logical but no
epistemological efficacy or role. Rather, GLC is a precondition of
premises which seem acceptable to all as true. Kant's premises describe
the empirically given at the low level of percepts; and he seems to have
given, as his five premises commonly acceptable truths (with the excep-
tion of premise I, which was easily modified without affecting the force
of Kant's argument). Thus it would seem on the analysis herein that
Kant has demonstrated that GLC is a necessary2 presupposition of object-
level statements about events. However, his argument has even greater
strength.

We have already claimed that non-necessary presuppositions, even
necessary ones are falsifiable. Falsification is always possible, for a

presupposition can stand or fall with the statements such as ST,, ST“,

ST,, ST,, and ST7 which jointly presuppose it. But as mentioned with

 

1Again this is in appearance a metalinguistic statement.

2This is not a modal sense of “necessary."

283

respect to Kant's argument, the statements ST3 ,..., ST, which jointly
presuppose GLC are quite certainly true.

Even more spectacular, however, is the fortification given GLC
by the fact that 5T3.,..., ST, are general statements which describe
the perceptual grounds for asserting an infinite number of stx state-
ments, "This is an event," "This is another event," etc. We have seen
from Kant's argument that not only his five premises entail GLC but also
all the st statements st1 ,..., stn which generally presuppose ST3 ,...,
ST, in non-philosophical discourse. Thus it is not only the case that
Kant's argument is convincing and strong but extraordinary in logical
force. Kant's argument exhibits the logical role of GLC which, serves
to make it almost impregnable to falsification. As a presupposition
even a necessary one, however, GLC is always subject to falsification.
There may beiaday when ST, or ST“ or STs or ST6 or ST7 no longer appear
true or when they are viewed as entailing a competing CMnl (one that
contradicts GLC) which displaces GLC as the presupposition of these ST
statements.

Let us return to our discussion of a priori judgments as
described by Kant. It will be recalled that Kant regarded a priori
principles as those which were necessary presuppositions of experience.
We have treated the argument which Kant gave in illustration of this
characteristic in the case of one of his prime examples of an a priori

statement the general law of causality. During the course of the

 

1The meaning of words in ST3 ,..., ST, may change.

284

treatment of Kant's argument, we have seen that presuppositions can

be characterized as necessary in a two-fold sense. First, the truth

of a presupposition is a necessary condition for the truth of the
statements which presuppose it; and second, when these statements

which presuppose some claim are true, the claim, a presupposition,

must be true. Both aspects of necessity are relative just as we have
claimed all concepts of necessity are. In view of what has been
previously maintained about statements of the form "All S are P,"

we shall omit from our view of a priori statements the claim that

they are universal since attributing this characteristic to a priori
statements is only a corollary of the assertion that statements of this
form are necessary in an absolute sense. On the contrary, we have tried
to show that some statements such as 'All S are P' are better considered
intensional contingent statements which establish necessary connections.
Thus we claim for a priori statements all the characteristics of neces-
sary presuppositions; and we identify a priori statements and presup-

* positions and, of course, necessary presuppositions as one and the same.

Can a priori statements be gynthetic?

Given the earlier reconstruction herein of analytic statements,
it should be clear that necessary presuppositions can very well be a
priori and synthetic. Again in Kant's paradigm example, we find several
premises ST3 ,..., ST, which jointly entail GLC. However, none of the
statements ST3 ,..., ST, are definitions (as we have explained defini-
tion); therefore, while GLC is presupposed by ST3 ,..., ST7 jointly,

GLC does not follow from ST3 ,..., ST, as an analytic statement which

285

would be entailed by definitional statements of its terms. Interestingly
enough, any claim that an analytic statement is a priori would be a
complicated claim indeed in terms of our theory of a priori since pre-
suppositions, unlike analytic statements, are not even stated within the
level of discourse where they are said to be presupposed. This is not
true, however, of analytic statements. Just the reverse is true; ana-
lytic statements are stated within a lower-level system of discourse
than the IC definition statements from which they follow. It might be
suggested perhaps that analytic statements presuppose various defini-
tions. However, a statement STX may be analytic via any number of
competing definitional statements in which case a given definitional IC
statement is not necessary to but only a sufficient condition for the
truth of STx, the analytic statement. Thus ICx entails STx but STx
would not imply ICX. For a definition statement ICx to be a presup-
position, STx would have to imply ICx. It might be claimed that ana-
lytic statements and presupposition statements are both deductively
entailed by other statements, either ICx or ST, ,..., ST,, and are
therefore of the same logical status, both necessary in the same way.
But we have tried to show that there is a difference between the nec-
essary and jointly sufficient presupposing premises of Kant's argument
and the non-necessary but sufficient explicit assumptions which are
analogous to the definitional IC statements from which analytic state-
ments are deduced. In the former (in terms of Kant's very strong

argument), we would say ST, ,..., ST, jointly presuppose GLC; and in

4—

 

. 286

the latter, we would write ICn entails STAn.1 However, we know from
lengthy discussion that in the former we have a metalinguistic state-
ment about a set of written or uttered object-level statements and an
unstated claim which they entail such that we can write the second-
level meta-claim "'ST, ,..., ST, presupposes GLC' is true"; but in the
latter, we have a second-level meta-statement to the effect that a
stated first-level metalinguistic definitional IC statement entails

a stated object-level analytic statement. The two-presuppositions and
analytic statements are the same only on a very superficial level of
investigation. In addition, it will be recalled that IC statements
which entail analytic statements are very different in kind from the
premise statements of Kant's argument.

Several writers have realized that there are possible recon-
structions of Kant's two distinctions the A-S and P-P such that a
priori statements need not be merely defined as analytic or vice versa.
N. R. Hanson, in his article "The Very Idea of a Synthetic A Priori,"
is one who stresses this point. He describes the A-S distinction as a
bisection of judgments which is exclusive and exhaustive in the follow-
ing way:

P is analytic if and only if its negation is of the form
(or leads to something of the form) Q --Q. . . . However,
a synthetic proposition is such that its negation, -P is

not of the form (nor does it entail anything of the form)
Q‘--Q.’ The knife called "a priori-a posteriori" however

 

1Notation for analytic statements.

2N. R. Hanson, "The Very Idea of a Synthetic A Priori," Mind,
LXXI (1962), 522.

287

cuts the class of pr0positions through quite a different
stratum.' To characterize aproposition as "a priori" is
to say nothing whatever about its formal structure or the
structure of its negation or consequences derivable there-
from. It is, rather, to remark the mode whereby the

truth of the proposition is discovered. A proposition

is a priori if its truth is established without recourse
to any possible experience (past, present, or future).1

In view of his beliefs expressed above, Hanson claims that
there is no reason for identifying 'analytic' with 'a priori' such
that these two terms would have exactly the same extension. Ultimately
Hanson adopts a non-falsifiability criterion for a priori propositions
but he does not adopt this characteristic (as we have not) as a mark of
analytic propositions:

To say that a pr0position is analytic is not to say just
that it is forever and always true. That P is forever
true follows from.(but is not identical with) the fact
that P's negation is self-contradictory.

Confusions result from failing to make this distinc-
tion. Thus when analytic propositions are said to be
those which are forever true and couldn't be false, the
ideazof analyticity collapses into that of invulnerabil-
ity.

In our present theory, we have tried to avoid such a collapse
not only by characterizing analytic and a priori statements in very
different ways but also by avoiding the adoption of the ambiguous
non-falsifiability (empirical independence) criterion as a mark of
either analytic or a priori statements. Unlike even Hanson who holds
our own position that it is possible for a statement to be a priori and
non-analytic, we have identified the a priori as a necessary presupposi-

tion, not necessarily of experience (as Kant would claim) but of other

1Ibid., p. 522.

 

21bid., p. 523.

 

288

empirical statements. The reason for our departure from Kant will be
evident in Chapter V where we will discuss a priori claims in science.
Were we to think only of the entire system of empirical discourse as
did Kant, then surely our presuppositions as unstated claims would have
to be non-empirical in at least the discovery and justification senses,
if not the content sense of "empirical." But we are interested in
discussing the empirical presuppositions of various subset paradigms

of empirical discourse that one finds in science. We have therefore
defined “a priori" in terms of its functional role as a necessary
presupposition and not in terms of its empirical status.

There is of course a second reason for avoiding the non-
falsifiability criterion as a mark of either analytic or a priori
statements which is simply that its formulation as a criterion by Kant
was inconsistent, confused, andconfusing. As such, it has led thinkers
(even those sympathetic to the very idea of a synthetic a priori) to
conclude that it is impossible or logically inconsistent that there be
any statements both a priori and synthetic. Hanson seems to be one of
the few who avoids this conclusion while adopting the non-falsifiability
criterion as a mark of a priori propositions (statements). However, in
as much as he does use this criterion, the extension of "a priori" state-
ments is limited so as to include only Kantian examples such as GLC. We

wish of course a broader application of the term "a priori."

289

Views on the synthetic-a4priori
The a prioriéa posteriori distinction as drawn by Kant is

typically interpreted by critics as a distinction between statements
that cannot be falsified by experience and those that can be so
falsified. This characterization of the difference between a priori
and a posteriori statements is basically the third criterion for the
distinction which was not formally proposed by Kant but nevertheless
seemed to be suggested by his elaborative remakrs as noted earlier
(page 226). Had Kant himself not provided some basis for choosing
empirical independence as the definitive criterion of a priori state-
ments, it could have been inferred from the two criteria which Kant
specifically and formally did give, universality and necessity, in the
following way, as pointed out by Jonathan Bennett:

Necessity and strict universality '. . . are . . . sure

criteria of' a priori knowledge! The context clearly

implies that necessity and universality are entailed

by a priority as well as entailing it. Thus, if the

judgment that all F's are G is a priori then experience

cannot render it false by yielding even a single F which

is not G. If it iS‘a posteriori, then it could be

falsified by experience.1

Jonathan Bennett capsulizes the problem of the above quotation

as one of deciding a sense for the term 'cannot' which is stronger than
(a) 'flatly against what we have so far discovered about how the world
inorks' but at the same time a sense other than that of (b) 'ruled out

by the meanings of the words involved.‘2 The suitable sense of 'cannot'

 

'Bennett, op. cit., p. 9.
21bido 9 pp. 9‘10.

290

must be stronger than that of 'flatly against what we have so far
discovered about how the world works,’ claims Bennett, since Kant
counts several judgments which have never had a counterinstance as
a posteriori judgments, e.g., every human body is larger than any ant.
At first glance andperhaps even at last glance, the problem
of finding a third sense of 'cannot' other than (a) and (b) above seems
insoluble. It is perhaps for this reason that many philosophers have
denied that there is any third sense of 'cannot' and that, since (a)
cannot be the correct sense of 'cannot,' (b) must be the only correct
sense of 'cannot' with regard to the impossibility of falsifying a
priori judgments. But (b) is the sense of 'cannot' in which analytic
statements are usually said not to be falsifiable, i.e., ruled out by
the meanings of the words involved." Therefore, the conclusion that
contrary to Kant's Critigue the only sense in which a statement is a
priori is the sense in which that statement is analytic has become
widespread. This conclusion rules out the possibility of any a priori
synthetic judgment which Kant sought to establish in the Critigue.
Prominent among writers who have considered a priori statements
to be empirically non-falsifiable is A. J. Ayer. In Lahgpagg_Truth and
£9312, Ayer refers to Kant's two explicit criteria for a priori truths,
universality and necessity, but quickly translates them both into the
third troublesome criterion of empirical independence much in the manner

of Bennett. By "independent of experience," Ayer means that the truth

 

1Ibid., p. 10.

rr'a-‘- ! .

 

291

of a priori propositions is not determined by empirical verification.

As he says with regard to the truths of mathematics and logic:
We must come to discover them through an inductive
process; but once we have apprehended them we see that
they are necessarily true, that they hold good for every
conceivable instance. And this distinguishes them from
empirical generalizations. For we know that a proposi-
tion whose validity depends upon experience cannot be
seen to be necessarily and universally true. . . . The
principles of logic and mathematics are true universally
simply because we never allow them to be anything else

. we cannot abandon them without contradicting our—

selves, without sinning against the rules which govern
the use of language and so making our utterances self-
stultifying. The truths of logic and mathematics are
analytic propositions on tautologies.

Within one page of text, Ayer moves smoothly and effortlessly
from the universal and necessary character of a priori propositions
whose truth does not depend on experience to the linguistic rule
governing necessity of analytic propositions whose validity does not
depend upon empirical verification. There seems to be no glimmer of
suspicion in Ayer that possibly there is more than one sense of "nec-
essary" at play in Kant's distinctions of the analytic and a priori.
How could there be any difference in the necessity which characterizes
both a priori and analytic propositions when neither are subject to
falsification by experience? Not surprisingly, Ayer explicitly con-
cludes what he implicitly assumed at the outset, i.e., that all a
priori propositions are analytic. Ayer proposes that analytic pr0po-
sitions be considered propositions whose validity depends solely on the

definitions of the symbols they contain and synthetic propositions are

 

1A. J. Ayer, Language Truth and Logic, p. 26.

292

those whose validity is determined by the facts of experience. Since
the validity of a priori propositions cannot depend on the facts of
experience, a priori pr0positions can never be synthetic. In short,
there are no synthetic a priori propositions.

Ayer's discussion is a classical case of the adoption of the
"empirical independence" criterion of a priori propositions which leads
inevitably, it seems, to their identification with analytic propositions
and the resulting denial that there are any synthetic a priori proposi-
tions. As has already been indicated by this author, the adoption of
this criterion is pppplly a fatal move because it is not only a confused
and confusing criterion but also because it blurs totally the central
insight for which the Critique of Pure Reason was written, though
unwittingly perhaps, that not all necessary propositions are necessary
in the same way.

C. D. Broad, in a symposium article "Are There Synthetic A
Priori Truths?" seeks to defend the position that there may be synthetic
a priori statements by reducing Mr. Ayer's discussion of the analytic a
priori to the position that all analytic propositions are a subspecies
of synthetic judgments. Thus if there are any a priori pr0positions at
all and they are all analytic, they will also all be synthetic and will
constitute examples of the sought for synthetic a priori judgments.
Broad suggests that one can say that a person "knows a certain prop-
osition if either it is self-evident to him or he actually sees it

to be logically entailed by premises all of which are self—evident

293

to him."1 It follows, he claims, that unless there are intuitably a
priori truths, there can be no demonstrably a priori truths. Broad next
proceeds to give his definitions of "analytic" and "synthetic" but first
remarks that ff"analytic" is to be applied only to the common examples
found in logic textbooks, e.g., "all equilateral triangles are equi-
lateral," then there are no analytic propositions (in the sense of a
proposition being that which is either true or false); and therefore,
if there are any a priori truths, they must one and all be synthetic.
Broad considers a wider view of analytic propositions which
he believes to have been expressed by A. J. Ayer, viz., that analytic
propositions are those whose validity depends solely on the definitions
of the symbols which they contain and that they call attention to
linguistic usages.2 In contrast, Broad notes in Ayer an independent,
unrelated definition.of "synthetic" that a "proposition is . . .
synthetic when its validity is determined by the facts of experience."
Broad claims that Ayer's definition of "synthetic" is the usual defini-
tion of the word "empirical" as applied to propositions and that Ayer
has really defined synthetic in such a way that all synthetic proposi-
tions will be empirical. By doing so, Broad concludes that Ayer has
predefined the case against the possibility of there being any synthetic
a priori since "empirical" and "a priori" are always understood to be
opposed to each other.3 Broad also concludes that there are no a priori

propositions on Ayer's view saying:

 

1C. D. Broad, "Are There Synthetic A Priori Truths?" Aristotelian
Society_Proceedings, suppl. vols. 14-16 (1935), 107.

2Ibid.
3Ibid., p. 106.

 

294

Now I think it is evident that no proposition which was

analytic in the second of Mr.-Ayer's senses could be

a priori in the sense just defined.‘ No proposition

about the linguistic usages of'a certain language could

possibly be seen to be true by mere inspection of its

terms and reflexion on them and their mode of combina-

tion. And no such proposition could be entailed by

premises all of which were of this nature.1

On a more charitable reading of Ayer, Broad surmises that
perhaps Ayer maintains that there are ostensibly a priori propositions,
i.e., propositions which seem prima facie to answer to our definition,
and that all ostensibly a priori propositions are really synthetic
empirical propositions of a certain kind, viz., announcements by the
speaker of his present intention to use certain words and phrases in
certain ways or statements about the current usage of certain words and
phrases in a certain language";2 and these "synthetic empirical proposi-
tions of this special kind are defined as "analytic propositions."3
It is doubtful that Ayer would accept Broad's analysis of his

view since it is obvious that Ayer does not think that analytic state-
ments which reflect linguistic usages as expressed by definitions are
empirical. Rather the definitions themselves are normative and arbi-
trary without descriptive or factual content, and analytical statements
determined by the definitions of the terms in them are also without
empirical content. Thus analytic statements could not be considered

a special sub-class of synthetic statements in Ayer's view if Broad is

 

1Ibi

O.

., p. 109.

M
0-4
0'
.1.
D.

c»
H
U'
_a
a.

295

correct in regarding Ayer to hold that only synthetic statements are
subject to empirical investigation in the determination of their truth
value. But even if one would grant Broad that definitions are not
merely normative but also and predominately descriptive and, therefore,
that analytic statements can contain factual content though they need
not be validated by investigation of the empirical world (the view
presented in this paper); there is no need to conclude that analytic
propositions in their reflection of linguistic usages are synthetic;
for there need be no unique identification made between the synthetic
and the empirical as has also been previously expressed in this work.
If certain statements are analytic because their validity can be deter-
mined by analysis of the meanings of their constituent terms and other
statements are synthetic because their validity cannot be determined in
this way, then no analytic statement can be synthetic no matter what
empirical information, old or new, it may express.

Though perhaps sympathetic to the possibility of synthetic
a priori statements, Broad, via his adoption of an empirical dependence
criterion for synthetic statements and a tacit adoption of an empirical
independence criterion which he believes is implicit in the self-
evidence conditions he sets for a priori statements, demonstrates that
if there are any a priori propositions at all they are only ostensible
a priori propositions and not real ones. This is hardly a defense of
the possibility of synthetic a priori statements, and it fails precisely
because one sympathetic to their cause nevertheless adopts the confused
falsifiability criterion not only for a priori statements but for

synthetic ones as well.

296

A more promising case, at least initially, was made by A. J. D.
Porteous who began his discussion in agreement with the present writer
on two major points:

1. He holds a relativist view of necessary truth for he says:

For no proposition can be seen to be selfevidently
true strictly in and by itself. A full understanding
of it requires consideration of the propositions which
define the meaning of its terms and relations, and it
is in the light of these defining pr0positions that
it is evident.

2. That the primary criterion for "a priori“ is the necessity of
a statement for he says:

. it was because of an alteration made by Kant

in the meaning of the term "a priori" that our ques-

tion was first introduced into the arena of philosoph-

ical discussion, and indeed appeared to Kant to sum up

the main problem of the Critical Philosophy; . .

since he offered universality and necessity as the

distinctive marks of them (the second bein more

important, since the first follows from it? he was,

in effect, divorcing the a priori from its connexion

with the self-evident.2

But while Porteous says in elaboration of 2 that he thinks it
important to follow Kant in regarding necessity rather than self-
evidence as the distinctive mark of the a priori, he adopts what he
calls a self-evidence criterion “in an epistemological sense," not a
psychological one, for analytic statements. In terms of it, an ana-
lytic statement is self-evident since it receives its meaning and
certainty within a system of logic or mathematics from a selected set

of independent postulates which serve jointly to limit the sc0pe of

 

#—

‘ 1A. J. D. Porteous, "Are There Synthetic A Priori Truths?"
Aristotelian SocietyProceedings, supp. vols. 14-16 (1935), 120.

2Ibid., p. 101.

297

possible interpretations of the variables (symbolizing elements or
relations) employed in its statement. Relative to the initial postu-
lates and definitions of the system, the analytic statement is self-
evidently true.1 He considers synthetic statements not to be self-
evident in this sense. Having decided to follow Kant's lead in
regarding the mark of a priori truths to be their necessary character,
Porteous seems not really to want to omit the old self-evidence crite-
rion which had often been assigned to a priori propositions; and so he
pins it on analytic statements, a totally unenlightening move. In
departure from his initial statements about the a priori, he begins
his discussion of the possibility of synthetic a priori statements not
really satisfied with Kant's necessity criterion and drags in once again
the non-falsifiability criterion when he says:

The term “a priori" I shall use in the extended Kantian

sense of “necessary propositions whose truth cannot be

established inductively or on empirical grounds."2

Though Porteous does not explain what he means by "established"

in the above criterion, we would have to conclude that on the basis of
these extended criteria that Porteous' answer to the question of whether
there are synthetic a priori statements such as the general law of
causality would be different from that of this paper. For, as we have
maintained, the general law of causality can be established on empirical

grounds as Kant did via various empirical premises in his argument in

 

1Ibid., p. 125.
2Ibid., p. 127.

298

the "Second Analogy" and yet be logically a priori instead of
epistemologically a priori within an organized system of knowledge
such as a paradigm of science.* It would seem that Porteous could not
hold such a principle to be a priori since it is established on empir-
ical grounds.1 But oddly enough at the end of his discussion, Porteous
is in agreement that the general law of causality is a candidate for
being a synthetic a priori statement; but one finds, once again just
before this conclusion, mention of his original narrowed criteria at
least in part when he says:
Must we assume that the class (of synthetic a priori
propositions) is empty--that in my sense of the words,
there are no propositions which state necessary truths
but are yet not analytic (evident)? I do not think so.
. . An obvious candidate for this class is the Prin-
ciple of Causality, formulated in some such form as
“Every occurrence must have a cause," or "All events
are causally determined."2
Interestingly, contrary to his statement of the extended version
of his original a priori criterion, he also says early in his discussion
that he wishes to "vindicate a place fbr synthetic a priori propositions
applying to nature." This is nothing different than the sense of being
an empirical proposition stated herein, i.e., that a priori truths may
be empirical statements which are subject at any given moment to empir—
ical falsification. Thus if by "established" in his extended version

of the a priori criterion Porteous means verified or falsified, then

there is no disagreement between his view and ours. The only objection

 

1The contradiction in Kant concerning pure a priori and non-
pure a priori judgments (p. 228) should be recalled)

2Porteous, op. cit., p. 137.

299

to his discussion would be that his introduction of self-evidence as

a characteristic of analytic statements is unnecessary to his real
contention that such statements can be validated by recourse to the
meanings of their constituent terms. In addition, its introduction is
misleading because of its previous association as a mark of a priori
statements. Secondly, if Porteous is not going to explain his use of
"applying to nature" as opposed to "established on empirical grounds,"
it would be best not to introduce any semblance of the falsifiability
criterion.

In an interesting article "On the Meta-Semantics of the Problem
of the Synthetic A Priori," Lewis White Beck warns that the problem of
whether there are synthetic a priori judgments should not be solved
by deduction from the definitions of the four kinds of judgments but
rather in light of the recognition that there are two principles of
division. As he says:

No real issue is solved by defining "a priori" in such
a way that all a priori judgments must be analytic.
Whether there are such judgments is a question of fact.

. Discussion of these facts and not tampering with
the definitions has been the fruitful way in which many
of Kant' 5 specific decisions concerning the status of
single judgments have been settled. But when the ques-
tion is discussed in greater generality, I think I detect
a tendency to modify Kant's definitions in such a way that
the general issue is decided against the possibility of

a priori synthetic judgments. It is this kind of argu-
ment which seems to me to be perverse.1

 

' 1L. W. Beck, "On the Meta-Semantics of the Problem of the
Synthetic A Priori," Studies in the Philosophy of Kant, p. 94.

300

Beck regards the distinction between a priori and a posteriori
as drawn by Kant as an epistemological distinction and the necessity of
a priori judgments as different from that of analytical judgments which
Beck regards as a logical necessity. If new definitions of "analytic"
and "a priori" are formulated, he believes that they should take account
of the historico-linguistic fact that "analytic" is a logical or a
linguistic concept and that "a priori" is an epistemological concept.1

On the contrary, however, Beck notes that most competent
original thinkers, particularly in English speaking countries writing
- during the period 1920-1950, concluded that there were no a priori
synthetic propositions so that by 1946 it was considered a dead issue.
But he claims further:

With the triumph of the point of view that there are no
synthetic a priori truths, there occurred, almost inevi-
tably, a subtle meta-semantic change. Analytic a priori
was seen as the only species of the a priori, and there-
fore “analytic a priori" seemed to be a redundancy . . .
the decisive word was "analytic" and "analytic" came to
function as a synonym for "necessary" which had previously
been regarded as synonymous with "a priori." At least
they were ascribed the same denotation, and since the
criterion of necessity came to be regarded as exlusively
logical or linguistic, nothing essential was omitted in
saying simply "analytic" instead of "analytic a priori."
And few people in the-main current of recent philosophy
talked about "a priori" at all.2

The foregoing situation led to what Beck calls a collapse of the

two distinctions into one which made any further discussion of whether

 

301

there were or are synthetic a priori statements very difficult. Any
proposed candidate for a synthetic a priori statement would not be
admitted as a necessary statement unless it was logically or linquis-
tically necessary; and once this was determined to be the case, the
statement was admitted at once to be not only a priori but analytic
as well.1
There is yet, however, another outcome of the collapse of
Kant's distinction, which Beck calls a meta-semantic shift, found in
recent discussions of "analytic." There is a modern theory of the
analytic which "holds that a statement is analytic to the extent or
degree to which it will be held impregnable against revision by
experience."2 In elaboration he states: '
Then there are two ways in which a proposition may be
found to be analytic . . . (l) by inspection of the
sentence itself, if it is logically or linguistically
true; and (2) by investigation of its role in an orga-
nized body of experience we call knowledge.’
The first sense of analytic determined by test (1), which Beck

regards as-a microscopic sense, he calls "analyticl. In contrast,
"analytica"is the name he assigns to the second sense of "analytic"
which he says is determined "simply by reference to the statement's
necessity (in an epistemological or methodological, and no doubt watered
down, sense which would not perhaps in some cases have been recognized

by Kant as constituting necessity at all). Then "analyticz' is formally

 

n-o
H
U'
do
0.

2Ibid., p. 96.

 

m
H
U"
.a
D.

302

equivalent to the old "a priori."1 Thus, as he observes, "analytic"
in the modern, broadened, two-fold sense of the term comes to mean not
only what it meant classically, "analytic,"; but it also comes to mean
"analyticz," the same as "a priori," with all the vagueness and impre-
cision of the latter. In conclusion, he makes the following interesting
point:

The former is the logical or linguistic analytic (analytic)

and the latter means any one or more of the following;

analytic truths which are factual or 'real,' the 'func-

tionally a priori,' the 'methodological a priori,' the

'hypothetically necessary,‘ the 'material a priori,' the

'regulative a priori'--and the 'synthetic a priori.'2

It is then for the very same reason today that the general law
of causality might be considered an analytic type statement in a par-
ticular language or body of science that Kant said it was synthetic a
priori, viz., that it is "not known to be true either by reference to
empirical facts or by inspection of its logical form but by reference
to its contribution to the 'possibility of experience.”3
In this chapter, we have sought to avoid any predefinition of

"a priori" which would automatically make it coextensive with "analytic."
We have tried herein and in Chapter III as well to follow the lines of
Kant's original A-S.and P-P distinctions to determine if they could

indeed be characterized independently of one another while at the same

time avoiding the many objections which his thought has provoked. We

 

1Ibid.. p. 97.
2Ibid.. PP. 97-98.

3Ibid., p. 98.

303

found that analytic statements and a priori statements have two points
in common. They are both necessary in a relative sense as we have
explained relative necessity, and they are or can be both empirical

in several senses of "empirical" which we have discussed.

Analytic statements cannot be false relative to the definitions
which entail them. We have chosen the phrase "true necessarily" as a
two place metalinguistic predicate which obtains between the conclusion
and true premises of a deductive argument. In as much as analytic
statements are true necessarily relative to true premises which entail
them, analytic statements can become false if the premises are falsified
and we have maintained that definitions, especially lexical ones, can be
either true or false. Thus a true definition which entails an analytic
statement can refer to the empirical world, and analytic statements may
be about the empirical world.

Contrary to Kant's contention that a priori truths are inde-
pendent of experience, we have found through close examination of his
demonstration of the truth of GLC that GLC (the general law of causal-
ity), which is "true necessarily” relative to the empirical premises
which presuppose it, is falsifiable, applies to the empirical world
and is established via empirical premises.

We come then to the conclusion that the declaration of empir-
ical independence for either analytic or a priori truths is false which
was our initial suspicion in Chapter I concerning Hume's a priori

argument.

304

With this much unraveled, we are now ready to present our
necessity theory of the law statements of science. In terms of our
above conclusions and our IC theory of metalinguistic statements as
well as our non-modal theory of necessity, we shall explain how it is
that law statements are contingent while involving claims of necessity,

and are sometimes a priori while expressing empirical fact.

CHAPTER V

A NEW NECESSITY THEORY OF LAW STATEMENTS

Preface to our theory

We began our discussion in Chapter I by mentioning two
controversial questions debated by modern philosophers of science.
They were:

1. What is the complete logical form of laws of science?
2. What are the differences between them and other kinds

of statements?

Empiricists answered these questions by formulating the regularity
theory of law statements. According to this view, laws describe
regular sequences of events. It can be concluded from their expanded
discussions of this empiricist tenet that regularity theorists regard
law statements as neither factual assertions which go beyOnd empirical
data nor as necessary assertions. As such, law statements are charac-
terized logically as material conditional statements in a purely
extensional language.

In continuing discussion we reviewed serious objections to
the regularity theory of laws which have made it untenable even to
most empiricists. The chief weakness of the regularity theory has
been its inability to provide for a distinction between true universal

statements which are laws and true universal statements which are

305

306

accidental generalizations, since both these kinds of statements are
formulated logically by regularity theorists as extensional material
conditionals. In literature as recent as the 1960's, Hempel and
Scheffler1 have concluded that law statements can only be distinguished
from accidental generalizations, or non-laws, by inspecting, as a whole,
the body of ordered scientific statements in which the statement occurs.
The role played by a given statement within such a system indicates
whether it is a law or not. However, the distinctive role law state-
ments play within a system of organized statements has still not been
clearly explained.

One supposed characteristic of the role of law statements within
'a system of scientific statements is that such statements are used for
prediction while their non-law counterparts are not. However, this
suggested characteristic is not acceptable since there are practitioners
of science, especially in the social sciences, who are willing to pre-
dict from statements not usually regarded as laws, e.g., regression
equations. Even if we ignore this consideration and content ourselves
that in the natural sciences only law statements are used for prediction
purposes, we might still ask why it is that natural scientists are only
willing to predict from some statements and not from others. Is the
statement "All pears in this basket are ripe," an accidental generaliza-

tion simply because no one uses it for prediction to future pears in

 

lCarl Hempel, As ects, p. 364; and Israel Scheffler, "Prospects
of a Modest Empiricism, e Review of Metaphysics, X (1957).

307

this basket or is the statement "All the pears in this basket are ripe,"
an accidental generalization for the same reason that we would not
predict from it? The latter would seem to be the case. If it would

be possible, therefore, to show that law statements have a logical

form which allows for predictions while non-laws do not, or cannot,

have a logical form which allows for prediction; then there might be
some deeper unpacking of this recourse to the role a statement plays
within an organized system of statements.

In the past, as mentioned in Chapter I, writers who pursued the
possibility of a differing logical form for law statements and non-law
statements introduced notions such as necessity into their theories.
The employment of this intensional term made their theories unamenable
to the expression of law statements in a first-order extensional lan-
guage--a major goal of contemporary empiricists. This has been consid-

ered a serious objection to necessity type theories.

Another_problem for necessity views

There is a further objection to typical necessity views more
formidable than the one just mentioned. Early writers such as Ewing
and Blanchard had only made vague suggestions toward a necessity view
of law statements, i.e., that law statements are somehow necessary or
have nomic necessity while non-lawlike statements are not nomically
necessary. Current writers who have tried to explain this necessity

have referred to laws in some cases as analytically true1 and hence

 

1Maxwell, op. cit., pp. l-25.

308

necessary or necessary as a priori‘truths.1 Unfortunately the serious
criticism of such views is that analytic statements as well as a priori
statements (we have seen in Chapter IV that these two kinds of state-
ments have been carelessly identified as one kind of statement by most
modern writers) are commonly regarded by philosophers as non-empirical.
Hence, the major drawback of current necessity views is that they render
the laws of science unempirical. Thus philosophic literature dealing
with law statements up to the present has offered only two unpleasant
alternatives with respect to the status of law statements. Either law
statements are compendious assertions of past observed facts framed as
material conditionals in a first-order extensional language which gives
rise to the vacuousness, hypothetical cases, mere coincidence, acciden-
tal generalization and subjunctive conditional objections as discussed
in Chapter I; or law statements are necessary, as all these objections
seem to suggest, but not expressable as first-order extensional state-
ments nor empirical. Faced with this dilemma, we have chosen to pursue
the direction indicated by the many serious objections of the regularity
theory toward a necessity view. But we have also tried to frame a
necessity theory which would not involve an intensional apparatus on
the object-level of discourse for expressing law statements or require
law statements to have a non-empirical status. Thus we undertook in
Chapters I, II, III, and IV to investigate the unclear but often used

notions of "analyticity," "a priority," and, indeed, "necessity."

 

lArthur Pap, The A Priori in Physical Theory.

309

We examined these notions to see if the major objection to a necessity
type view, i.e., law statements are noneempirical, could be avoided,
that is, to see if laws could be necessary or involve necessity somehow
and still be empirical.
In Chapter I, we analyzed this second critical objection to any

necessity type theory into the following three claims:

1. We cannot give a priori proofs of causal laws.

2. Causal laws are only probable, not certain.

3. No logical connection between cause and effect can be proved.

Each of these claims can be found in the writing of David Hume,
the father of modern empiricism. However, for each of Hume's arguments
in support of the above criticisms we found numerous rejoinders. Our
discussion of these in Chapter I ultimately indicated that empirical
statements might be known a priori and might even in some sense be
certain. With these historical pillars of modern empiricism undermined,
we proceeded to explore further the possibility that a statement would
be certain or necessarily true but nonetheless empirical. Thus in
Chapter II we explored Kant's "Second Analogy" where it has been claimed
Kant gives a proof that the uniformity principle is a necessary presup-
position of experience. If indeed Kant proved that some statement such
as "Nature is uniform" is presupposed but nevertheless about the empir-
ical world, it might well be that some statements are a priori (neces-
sary) and also empirical.

In Chapter II we adopted a minimal interpretation of Kant's

"Second Analogy" argument which was that Kant only proved the

310

universality of cause (GLC) to be presupposed by statements about
empirical events. However, this weaker principle, the universality
of cause, serves equally as well as an example of a presupposed truth
about the empirical world as would the stronger uniformity claim.

Kant showed that GLC is apresupposition of all those who
distinguish subjective perception successions and objective perception
successions commonly called events. In Chapter II we showed in detail
that GLC, deductively entailed jointly by five Kantian premises, conveys
empirical information, and is contingent (could be false in an absolute
sense) but certainly true relative to the premises from which it is
deducible and therefore, in a relative sense, necessary. Kant had
claimed that such a principle was synthetic and a priori. This agrees
with our assessment. But due to the long standing rejection of Kant's
synthetic a priori which historically reflects Hume's claim that no
factual statements are a priori, we sought to work from our factual a
priori statement (GLC) to theories which would underpin the possibility
of our having such a factual a priori statement. Thus we set ourselves
in Chapters III and IV to see if (1) analytic statements and a priori
statements are really only one distinct kind of statement, (2) to see
if a priori statements are necessary or if on the other hand they are
in some sense empirical, (3) to see if analytic statements are neces-
sarily true, yet if they might be considered in some respect factual
and (4) to see if necessary law statements need be either analytic or

a priori in order to be necessary.

311

In Chapter IV we examined the notion of "necessity." We have
already said that GLC is relatively necessary since it is deducible from
five true premises. However it might be rejoined that relative neces-
sity is nothing but the expression of a logical implication and not a
bonafide application of the term "necessary." For the term "necessary"
is commonly meant by philosophers to serve as the label of self-evident
truths which cannot be false in an absolute sense, not merely relative
to other statements from which they are derived. Such necessary truths,
it might be argued, are necessary in and of themselves.

As we discussed in Chapter IV, the distinction between absolute
and relative necessity was maintained by Aristotle. However, in our
examination of all fbur uses of the term "necessary" in current philo-
sophical literature, none were found to be cases of absolute necessity.
Each type of statement commonly called necessary was necessarily true
only in relation to some other statement(s) from which they were deduced.
We failed to uncover any examples of a statement which is self-evidently
necessary in and of itself, not even logical truths referred to by
Quine.1 It should also be noted that there are philosophers who use
the term "necessary" very specifically in a relative way. Carl Hempel
emphasizes that logical inference or implication is expressable as B is
necessarily true relative to A and that ”necessary" should not be con-

sidered a modal operator (as it would be if used in an absolute sense)

 

1See pp. 247-248.

312

but rather as-a'2'place'relational‘predicate.“'Wesley'Salmon says
much the same thing in his introductory logic book.2

In view of these considerations and our investigation of the
uses of "necessary" actually employed by philosophers, we concluded in
Chapter IV that all claims of necessity are relative, i.e., all neces-
sary truths are necessary due to their deducibility from other true
statements. Not surprisingly we found in Chapters III and IV that
analytic statements and a priori statements were in varying ways both
relatively necessary. We first examined analytic statements. We
claimed that a statement is analytic iff it is true or false by ref-
erence to the meanings of the terms within it. Given our metalinguistic
theory of meaning in which definitions are viewed as intensional, con-
tingent, quasi-legislative reports; we concluded that statements are
analytic when deducible from the definition statements of their terms.
It is in this way that analytic statements are true necessarily. They
are not statements absolutely necessary such that we would never give
them up or consider them false. Rather they are necessary relative to
the definitions which entail them and these definitions which have a
factual aspect can be falsified or changed.

Our investigation of the term "a priori" in Chapter IV revealed
that while a prior statements are also necessary, as Kant had claimed,

they are only necessary relative to statements which entail them.

 

1Carl Hempel, "Inductive Inconsistencies," Aspects of Scientific
gyplanation, p. 59.

2Wesley Salmon, Logic, p. 9.

313

However, we did not find that a priori statements are analytic
statements. The former statements are presuppositions. As such they
are unstated but entailed by nonédefinitional statements, often about
the empirical world. The truth of‘a presupposition, e.g., the truth
of GLC, is a necessary condition for the truth of the statements which
presuppose it. A priori statements can become false when the state-
ments which presuppose them are falsified. In contrast, analytic
statements are not presupposed by others. They are not unstated.
However, when uttered, or written they are deducible from definition
statements which are usually unstated. The truth of these unstated
definitions is not necessary for the truth of analytic statements but
only sufficient for it.

Ultimately we concluded from our research that there are two,
viable distinctions applicable to statements, i.e., the A-S and the P-P
distinctions. Both a priori and analytic statements are relatively
necessary in as much as both are entailed in circumstantially different
ways by other supposedly true statements. But when we return to our
original question about the logical form of law statements, we can now
see that while law statements are empirical and possibly necessary in
some way they are not necessary qua analytic or a priori. Law state-
ments qua laws are not entailed by a finite conjunction of other
statements of fact.1 Despite Hume and some regularity theorists it
seems clear that laws are not merely abbreviations for long strings

of factual assertions about past conjunctions of observed events.

 

1We mean non-vacuously deducible.

314

Law statements as empirical and theoretical generalizations make
assertions which go beyond scientific data. Indeed this characteristic
has been cited as a problem of induction. This open-ended aspect of
laws such that they are considered as applying to both actual and
possible cases is due to what we shall call the scientific act of
imputation.

Now we will explain imputation and how law statements, while
not analytic or a priori or relatively necessary qua law statements,
nevertheless involve an element of necessity and make assertions about
the empirical world. We shall also see how laws can sometimes function
as analytic or as a priori statements within a system but nevertheless
remain empirical as provided for in our previous discussions of

analyticity and a priority.

Law statements

 

In his interesting book Causality--The Place of the Causal

Principle in Modern Science, Mario Bunge lists a minimum of eight

 

different kinds of scientific law or determination. They are:

quantitative self-determination
causal law

interactive or reciprocal causation
mechanical determination
statistical determination
structural determination
teleological determination
dialectical determination.

CO'xiOiU'Ihwmc-I
o e o o u c c, a

An example of the first and lowest level type of law, quantitative self
determination, is one in which a functional relationship is expressed

between the successive positions of a freely moving macroscopic body and

315

its position and velocity at any prescribed instant of time.1 Another
example given by Bunge is the spontaneous transfbrmation of an isolated
thermodynamic system which leads to states of increasing entropy.2 As
Bunge says of type (1) law statements:

Quantitative self-determination is the category of

determination prevailing in the continuous unfolding

of states that differ from one another in quantitative

respects only; . . .3

The second type of scientific statement listed by Bunge, causal
laws, is familiar. The usual example of such a law, according to Bunge
is one in which change is produced by some external factor(s). He thus
considers cause in terms of efficient causation. A similar kind of law,
type (3) reciprocal causation, Bunge refers to as an interaction law.
For example, he notes that the orbits of the components of a double star
are determined by their gravitational interaction. As an example from
biological science, he gives the case in which every gland in the human
body depends on that of the remaining glands.“‘
The fourth type of law, which is partly an interactive and

partly a causal kind of law, is the kind Bunge calls mechanical deter-
mination. Such laws pertain to situations in which a force might modify

the state of motion of bodies. In further elaboration Bunge notes the

fact that "the streamlines in a fluid are determined by the latter's

 

1Mario Bunge, Causality--The Place of the Causal Principle in
Modern Science, pp. l7-18.

 

2Ibid., p. 18.
3Ibid.
“Ibid.

 

316

previous state, by the external forces acting upon it, by internal
friction (viscosity), and by internal pressure differences."1
The fifth type of law statement described by Bunge, called

statistical determination, is one in which a change results from the
joint action of independent or quasi-independent entities. Examples
of this sort of determination are familiar to all of us. We know that
the probability of throwing a six with a fair die is one sixth. As
noted with respect to the previous categories of scientific law, Bunge
maintains that this fifth type may also be in part other kinds of
determination:

As in the case of other categories of determination,

statistical determinacy may emerge from processes on

deeper levels, in which still other categories of

determination are involved.2

Structural determination, the sixth type of law cited by Bunge,

is illustrated by a molecule in a fluid where the behavior of an indi-
vidual molecule is determined by the "over—all structure of the collec-
tion to which it belongs."3 The seventh type of law, teleological law,
has been common in some disciplines such as biology. In such a disci-
pline, Bunge maintains, one might find talk of birds building their
nests in order to safeguard their young. In this connection, Bunge is

careful to state that goal-directed structures, functions, and behaviors

need not be purposefully planned by anybody.

 

H
H
U
do
D.

 

N
H
U-
do
0..

 

m
H
U"
_a
O.

., p. 19.

317

The last category of determination listed by Bunge, dialectical
determination, is exactly what its label suggests.' These laws obtain
in cases where there is a "qualitative self-determination of the whole
process by the inner 'strife' and eventual subsequent synthesis of its
essential, opposite components."1 Changes of state in matter, in bulk,
are produced by the interplay and final predominance of one of the two
opposite trendszthermal agitation and molecular attraction.2

Bunge delineates the above kinds of laws which express various
categories of determination in order to emphasize the point that causal
laws are but one kind of scientific law, i.e., expressive of only one
category of determination. However, he does admit that these various
types of determination are genetically connected with one another, the
higher types being dependent on the lower without being entirely reduc-
ible to them. He stresses:

. . no type of determination is found to operate in
all purity, to the exclusion of all others save in ideal
cases. To illustrate my first contention, take mechanical
determination, which is a peculiar combination of purely
quantitative self determination (in this case inertial
motion) and reciprocal action, which can often be polar-
ized into cause and effect. Or take statistical deter-
mination which emerges, with characteristics of its own,
as a result of the interplay of a large number of ele-
ments that are individually determined in accordance with
other types of determination (mechanical or teleological).3

In addition, all of the kinds of law statements mentioned above

have the common bond of expressing a determination of some sort; and

 

H
0—1
0"
-l
O.

l.-

 

3I id., p. 20.

318

fittingly, Bunge does explain what he means by the term "determination"
itself. He claims that there are two meanings of "determination":
(1) property and (2) constant connection. In sense (1) Bunge claims
that "determination" is synonymous with 'characteristic,' either qual-
itative or quantitative. It is this sense of "determination" which
comes from the post-Roman Latin, "determinatio," and this first sense
is the primary sense of "determination“ in various European languages.1
Bunge elaborates upon its meaning by saying:

In this sense, that is determinate which has definite

characteristics and can consequently be characterized

unambiguously; when applied to descriptions and defi-

nitions, 'determinate' is used as an equivalent of

precise or definite in contradistinction to vague.2

Sense (1) of 'determination,‘ which might be the general char-

acterization of all eight types of laws listed by Bunge which express
common types of determination (ways in which some thing, property or
event is rendered determinate), seems also to characterize a specific

type of scientific law not mentioned by Bunge. This ninth type of law

which is a phima facie determination in sense (1) is a classificatory

 

law. Campbell refers to such laws as the "laws of the properties of a
substance, or a kind of system, the law, namely, which asserts that
there is such-and-such a substance of such-and-such a kind of system

steel or magnets."3 In addition Campbell states:

 

1Ibid., p. 7.
2Ibid.

3Norman Campbell, What Is Science, p. 56.

319

These laws, in an elementary and imperfect ferm, are
the earliest laws of science and they retain their
peculiar significance through much of its consequent
development.’

We have just referred to this ninth kind of law as a prim;
jppig_determination in sense (1); for it is our contention in this paper,
as just suggested, that in a significant sense, every kind of scientific
law is indirectly, if not primarily, classificatory, i.e., a determina-
tion partial or complete in Bunge's sense (1) of "determination." We
add the qualification that determinations can be either partial or
complete because the "unambiguous characterization of things, properties
or events" is a phrase which can be understood in either a strong or
weak sense. If a thing, property or event is characterized unambig-
uously in the strong sense then the characterization will serve as an
identity mark of that thing, property or event. A set of characteris—
tics which taken together are sufficient for the unambiguous identifi-
cation of a thing, property, or event as a certain kind of thing,
property or event would not be just a set of characteristics but would
rather be taken aS'a mark of the thing, property or event. In such a
case a statement of the set of characteristics would be a complete
determination. On the other hand a law may present us with only one
of the characteristics of a thing, property or event. In so doing the
law characterizes the thing, pr0perty or event unambiguously in a weaker

sense. We are not given a mark of the thing, property or event but only

a characteristic which by itself may not be sufficient for identifying a

 

111:1 ., p. 56.

320

thing, property or event as exactly a certain kind of thing. There

may be in fact two or three kinds of things, properties or events

with this characteristic. Thus reference to the characteristic will
identify a thing, property or event as any of the three kinds. A law
which characterizes the thing, property or event unambiguously in this
weaker sense is only a partial determination. The complete determina-
tion of the thing, pr0perty or event will be given by a set of partial
determining laws which when taken together identify the thing, property
or event so unambiguously that it will be known as of one kind of thing
and no other.

Thus we claim that scientific law statements or sets of law
statements qua complete determinations unambiguously characterize things,
properties, or events and, in so doing, provide the identifying marks of
those things. We might clarify this claim by saying that things, prop-
erties, or events are defined via law statements; but we prefer to say
that; whereas words are defined (as described in Chapter III) when
characterized unambiguously; things, properties, and events are not
defined but given an identity when characterized unambiguously. All
eight types of law listed by Bunge express relationships among proper-
ties, things, or events; and these relationships which are constant
and unique1 connections provide or serve as a whole themselves at least

as identity characteristics for entities under scientific investigation.

 

1Here we mean connections hold in only specific ways for only
specific relata and no others and we mean specific in kind or quality
not number.

321

Indeed constant and unique connection itself is the second sense which
Bunge gives of the term "determination."' As he says:

But in science the most frequent use of the word

'determination' that is relevant to our concern

seems to be that of constant and unique connection

among things or events, or among states or qualities

of things, as well as among ideal objects.1

A constant and unique connection of this sort is expressed by
the following law for the thermal expansion of a metal rod: L(t) =L(o)
(l-+a t) where L(o) stands for the length of the rod at the temperature
t==o, a being the coefficient of thermal expansion.2 This statement of
law may appear much like one which merely expresses a functional rela-
tionship quantitatively, but as Bunge notes, it also expresses a con-
stant and unique relation “among the qualities length and temperature
and the disposition expansibility of metal rods."3 As such it serves
at least partially to characterize metal rods unambiguously. Let us
consider the case of a particular kind of metal rod.

Suppose we are presented with a rod made of a pure unidentified
metal. In order to at least partially identify the metal, we can sub-
ject the rod to varying temperatures and record its changes of length.
Given this information and the law for thermal expansion, we can obtain
some functional relationship which may happen to be a particular func-

tional relationship confirmed andaccepted as law for rods made out of

copper. The law statement which expresses the functional relationship

 

1Bunge, op. cit., p. 7.
2This example is from Bunge.

3Ibid., p. 9.

322

between the length of copper rods and their temperature gives us an
identifying characteristicof‘copper'rods.1 Heated rods which expand
according to this particular functional relationship and satisfy other
laws regarding copper are copper while those which expand according to
a different functional relationship are not. Therefore, by rendered
at least partially determinate, we mean in sense (I), being definitely
and precisely characterized. In this first sense, a property or a
functional relationship in toto expressed by a law statement can be
said to render some things, e.g., copper metal rods, at least partially
determinate. We see then that once constant and unique connections are
expressed by scientific laws, even functional ones, things or properties,
can also become partially or fully determinate; and the identity (whole
or partial) of a property, thing, or event is established. Indeed as
Richard Schlegel writes:

In Western science, this abstraction of certain determi-

nables seems particularly to be favored by the use of

mathematics in science. ‘The relations of a few selected

and defined aspects of nature--for example, electric

charge, electric current and magnetic field--can be

explored and understood with immense scientific fruit-

fulness, once an association is set up that permits all

the deductive power of mathematical reasoning to be
directed to the relationship.2

 

1There may be other laws about copper or some other metal may
have the same coefficient of expansion.

2Richard Schlegel, "Do We Know The World Through Science?"
Mind, Science and HistQEX, II, 176.

 

323

Properties which can serve to identity
In an attempt to provide an adequate explanation of how law

statements establish the identity of things, properties, or events,
we shall borrow some apparatus from Daniel Bennett's article entitled
"Essential Properties."1 We shall only employ one of his considerations
in a very modified way since it is not our present concern to discuss
essential properties but rather determinators (properties or functional
relationships which partially or completely identify things, properties,
or events) which are not in any historical philosophic sense essential
or natural. Determinators are properties or relationships referred to
in law statements which omnitemporally classify a thing, property, or
event as distinct from at least some other individual things, properties
or events but not always from all other things numerically distinct from
this thing, property or event if there are any. In explanation we pre-
sent the following claims concerning sortal properties:

We shall say that a property P completely sorts aniindi-

vidual (a) at a time t, iff it is true of (a) at t but of

nothingdistinct from (a) at t. The property of being

identical with Socrates sorted Socrates from all other

individual things and therefore individuated him through-

out his life. . . . In contrast a property P partially

sqrts an individual (a) at t iff it is true of (a) at t

and some things distinct from (a) but not of all things

distinct from (a).

We shall say that a property P classifies an individual
(a) at time t iff it partly sorts (a).

In view of the above, we now see that being a man and being a

philosopher classified Socrates at certain times in his life. However,

 

1Daniel Bennett, "Essential Properties," Journal of Philosophy,
LXVI (1969), 495.

324

being identical with Socrates sorted him so completely that it did

not classify him but individuated him from all other things including
all other men and all other philosophers. We have chosen Bennett's
treatment of "sortal properties" as suggestion for an appropriate
partial treatment of what we shall call determinator properties men-
tioned within various kinds of scientific laws precisely because some
sortal properties partially sort but do not individuate things,
properties or events. Scientists are usually not interested in
investigating samples of gold or copper for their individual differences
qua samples; but rather they are interested in establishing the common
properties of all sample ores which are to be considered gold or copper.
Scientists therefore are interested in discovering classifying sortals
but not individuating sortals.

Though we have found Bennett's account of sortal properties
helpful in describing the difference between classifying sortals and
individuating sortal properties, it is obvious that being a classifying
sortal property is not all that is required for a property to become a
determinator property within some law statement of science. Science is
not concerned with temporary full or partial determinations which might
be made by classifying sortal properties that are, as Bennett describes
them, phase properties which sort. Such a classifying sortal phase
property might be the pr0perty of being post-pubescent which classified
Socrates not in infancy or adolescence but during most of his life.
Classifying sortal phase properties which might serve to classify some
thing are not thought to be omnitemporal properties and are not ordi-

narily chosen as determinator properties in the framing of scientific

325

laws. Indeed if we recall Bunge's second definition of fdeterminate"
as constant and unique connection, it is apparent that sortals, which
classify individuals must also classify omnitemporally in order to be
adopted as determinator characteristics.

We now give our criterion for potential determinator
properties:

A property P or relationship R will be considered an
omnitemporal classifying property P or relationship
R iff:

i. anything that is P or characterized in terms of
R, for as long as it is P or characterized in
terms of R, is partially sorted by P or R.

ii. anything that is ever P or characterized in terms

of'R is, for as long as it exists, always P or

characterized in terms of R.

If a given property P or a relationship R satisfies the above
conditions, it will be an omnitemporal classifying sortal. It is a
necessary condition that a property be an omnitemporal classifying
sortal if it is to be a determinator property. But being an omnitem-
poral classifying sortal is not a sufficient condition for a property
to be a determinator property. The necessary and sufficient condition
for a property to be a determinator property is that it be an omnitem-
poral classifying sortal which is incorporated within a law statement
of science. There may be non-law statements which contain only omni—
temporal classifying sortals, but these sortals are not determinator
properties nor is the statement the expression of a determinator

relationship since such statements have not been accepted by the

scientific community as law. If non-law statements contain only

326

omnitemporal classifying sortals and are confirmed (they have never
been violated), they are lawlike statements which may some day for
various pragmatic reasons be accepted as law statements. But until
they are so accepted, the omnitemporal classifying sortals within

them are not determinator properties.

The form of law statements

The connection between the law statements of science and what
we have characterized as determinator properties or relationships
should now be clear. When a true universal statement is also part
of a statement of scientific law1 it is a statement which can be
mentioned in a metalinguistic statement such as "The true universal
'(x) (ex 3 Vx)‘ expresses a relationship which is tp_pg_determinate,"
(a relationship between a determinable thing, pr0perty or event and at
least one determinator). We might thus write Newton's Law of Inertia
as "'(x)(Bx 3 Ix)‘2 is to be a partial determination." It is not only
the case that all past examined bodies have had a tendency to remain at
rest unless something moved them and to continue moving in a straight
line unless something stopped them or caused them to change their
direction but all bodies are determined by this tendency. This meta-
statement claims that the true, universal statement "All bodies have
a tendency to remain at rest etc.," expresses a relationship between

bodies and their behavior such that the latter should be considered as

 

1The full expression of a law of science is a metalinguistic
statement saying of an object-level generalization that the relationship
it expresses is to be determinate.

2x is a body and x has an inertial tendency.

327

at least a partial identity mark of the former. The property of
inertia was found to be an omnitemporal classifying sortal, and it
became related to bodies as one of their identifying characteristics
by the law statement '”All bodies have a tendency . . ." is to be a
partial determination.‘

‘ This law statement in turn is a sufficient condition for the
framing of another metaéstatement, viz., "The term 'body' means in part
'having the tendency.. . .'" The first meta-statement, the law state-
ment, relates a body and its behavior. That behavior then is taken to
be a partial identity mark of a body; and once this relationship is
established between a thing and one of its properties, another meta-
statement can be asserted which establishes a meaning relation between
the words which refer to this thing and this property. This meta-
statement is a meaning statement of the kind we have described in
Chapter III, and it is a sufficient condition for asserting yet another
meta-statement "'(x)(Bx a Ix)"i ,tp_pg necessary." The IC intensional
contingent (the name we have given to our metalinguistic meaning state-
ments in Chapter III and which we now give to other quasi-reportive
legislative statements) law statement is a sufficient but non-necessary
condition for the framing of the IC definition statement because IC
definition statements can be framed for words that do not refer to any
things, properties, or events whether characterized unambiguously or
not. Thus the metalinguistic statement about the true universal gen-
eralization "All bodies have an inertial tendency," viz., "'All bodies have

an inertial tendency' is at least a partial determination," entails another

328

quasi-reportive legislative second IC metalinguistic meaning statement
which corresponds to the statement"'(x)(Bx a Ix)' is to be necessary."1
To reiterate, the full expression of a law of science is not
just an object-level true universal statement but rather a metalinguis-
tic statement about some true object-level generalization that the
relationship it expresses is to be a determinate one (involving a
determinator property or relationship). This metalinguistic statement
is both descriptive and legislative. It describes the fact that a
property P is an omnitemporal classifying sortal of some thing, property,
or event. But it also prescribes that this property P which is this
kind of sortal should be considered an identity characteristic of that
thing, property, or event. The metalinguistic law statements of science
qua IC statements are of course themselves contingently true, yet they
are sufficient for the positing of a metalinguistic definition IC
statement about the terms which refer to the determinable thing,
property, or event and the determinator property or relationship which
determines at least partially the determinable. These contingentz
definition statements in turn are the necessary and sufficient condition
for asserting a statement of the form "'(x)(Bx o Ix)' is necessary,"
which we have already maintained in Chapter III is a regular descriptive

entailment statement (if the antecedent is true the consequent must be

 

1This is sufficient for "is necessary," and necessary for
"is necessary."

2We use "contingent" here in an absolute sense. Some IC
statements when entailed by others are relatively necessary as we
have explained this notion.

329

true). Via a'long'chainof‘sufficient‘condition1 statements, a
non-necessary metalinguistic scientific statement of law ultimately
leads to an assertion of necessity, i.e., something x must have a
tendency . . . or "'x has a tendency . . .' is true necessarily," in
those situations in which x is said to be a body and the true universal
"All bodies have a tendency . . ." is said to be at least a partial
determination, the provision of a determinable thing, body with an
identity characteristic or mark.

Confusion concerning the kind of necessity involved in state-
ments of law arises because the long chain of sufficient conditions
between the contingent metalinguistic law statement and the particular
prediction statements of science which often employ the term 'must' is
elliptically omitted. The term 'necessary,‘ as we have maintained
earlier, characterizes a relationship between verbal expressions.
Application of the term "necessary“ to this relationship occurs first
in the metalinguistic statement "'(x)(Bx 3 Ix)' is to be necessary"
which is entailed by the quasi-reportive, legislative, metalinguistic
IC law statement "'All Bodies have a tendency . . .' is to be at least
a partial determination." 'The final particular prediction statement
(a lawful statement2 as Opposed to a statement of law), "The next body
we find must exhibit the inertial tendency," is true necessarily much

in the manner of any analytic statement of a natural language discourse.

 

1One statement is necessary as well as sufficient.

2A lawful statement is any statement non-vacuously deducible
from a law statement.

330

A partial meaning of the term 'body"i5'9having an inertial tendency."
One can always verify the prediction empirically, but checking the
prediction is perfunctory unless the well confirmed statement of law
which together with the statement "x iS'a body" which entails the
prediction statement is in some jeopardy being perhaps incompatible
with other more powerful laws or new unexplained phenomena. Once the
implicit dictionary of a particular science includes within it a defi-
nition, such as "'body' means in part 'having an inertial tendency,'"
it is not difficult to see why there is a necessary aspect of lawful
prediction statements which are uttered in particular contexts by
individual scientists on various occasions, e.g., "The next body must
have this tendency. . . ." This latter statement token which is true
necessarily is so because it is an analytic statement.

It is important to remember that these lawful statement tokens
made on particular occasions may be analytic and may serve either to
predict or to clarify the meanings of various scientific terms, as
Garver suggested;1 but the initial statements of scientific law which
are found in scientific texts are never analytic or necessarily true.
As the initial statement of‘a newly presented law, they are not deduced
from some finite class of empirical statements; and thus, law statements
in textbooks are neither analytic nor a priori. There are times, how-
ever, when statements expressed by sentences exactly like the sentences

used to express the initially stated non-analytic or non-a priori law

 

1See pp. 165-166.

331

statements, are analytic or a priori. If a scientist is using a law

statement token by writing it or uttering it after its initial pre-

sentation, it may well be analytically true in the same way as A,

stated at time t, is deducible from A1 initially presented at time t,.

In such a case At, is analytically true when At1 is true. At1 might

be the statement of Boyle's law in a physics text on page 8 when it

is being presented for the first time as a law. Later on page 14

Boyle's law may be rewritten exactly as before within the context1

of some discussion or deductive work. The law statement token on

page 14 is an example of At, an analytic statement deducible from At1

while At1 is not deducible from any other particular statement which

precedes its utterance or inscription on page 8.2 Sometimes of course

a low-level law is deducible from some other more general law. In this

case even an At1 law statement may be analytically true but not because

it is a law rather only because it is a certain kind of low-level law.
At times, some very general and powerful law statements may also

be a priori and therefore relatively necessary. A law statement which

at times may be presupposed in some context of minute or specialized

 

1We are not saying here that every time Boyle's law is uttered
or written after Boyle's initial propounding of it, it expresses
statements which are analytic. It will depend on the context in which
an assertion of Boyle's law is made whether the writer or speaker is
propounding it once again in the legislative way in which Boyle proposed
it or whether the writer or speaker is making a statement of Boyle's law
in a context where the hearer or reader is thought to have access to or
knowledge already of this partial convention--Boyle's law.

2Again we mean non-vacuously deducible.

332

scientific work will be a priori in that situation. The orderly
scientist who writes the laws he will be using at the top of his
calculations may not write down the law that he is presupposing.
Arthur Pap recognizes the possibility of such situations in explaining
how laws may be functionally a priori.1 Pap writes:

e.g., in the context of experimental inquiry in which

a certain measuring instrument is employed the law in

accordance with which the indicators of that measuring

instrument are interpreted is no doubt a priori in the

sense of being irrefutable by the results of the exper-

imental investigations. But no physicist would regard

such a law as an arbitrary "rule of procedure" and deny

to it its empirical foundation and contingency. The only

way to solve this apparent paradox is to recognize the

development of what are results of experience at one

stage into "constitutive conditions of experience” in

Kant's terms at another. . . . A proposition which is

a priori in one context of inquiry may be a posteriori

in another context.2

Pap's description of the possibility we have raised concerning

the occasional a priori status of scientific laws succinctly represents
our own position herein. However, unlike Pap we claim that the law
mentioned is a priori because it is irrefutable by the results of
experimental investigation. In terms of our theory of presuppositions
in Chapter IV, we claim that this presupposed law statement is true
necessarily (in a relative sense) but it can be falsified whenever the
statements which entail it are falsified by experience. We also claim

that lawful statements which are analytic as just described can also be

 

1We have given a functional explanation of a priori in Chapter
IV and therefore do not employ the term 'functional as an adjective as
does Pap.

2Arthur Pap, The A Priori in Physical Theory, pp. Vii-viii.

333

considered empirical in that their terms can refer to objects in the
empirical world and that they can be falsified whenever the IC law
statements entailing them are falsified by experience.1 Thus we see
that even when law statements are occasionally analytic or a priori
they are necessarily true and they also are empirical. This was our
initial suspicion in Chapter I, which is supported by our research in

Chapters II, III and IV.

Law statements as imputations
Our theory of scientific law is in complete agreement with the

remarks of Nicholas Rescher in his book Scientific Explanation. Surely

 

it is widely recognized by now that, as Rescher claims, "Lawfulness is
a matter of imputation."2 As he so clearly writes:

When an empirical generalization is designated as a law,
this epistemological status is imputed to it. Lawfulness
is something which a generalization could not in principle
earn entirely on the basis of warrant by the empirical
fact. Men impute lawfulness to certain generalizations

by according to them a particular role in the epistemo-
logical scheme of things, being prepared to use them in
special ways in inferential contexts (particularly
hypothetical contexts) and the like.3

In terms of our own theory, the imputation of lawfulness is
expressed by our formulation of laws as intensional, contingent, meta-
linguistic statements, e.g., "'All bodies have a tendency . . .' is at

least to be a partial determination." This metalinguistic statement

 

1See Chapter III, p. 166.

2Rescher, op. cit., p. 107.
3Ibid.

334

imputes to the true, universal statement "All bodies have an inertial
tendency," that it is at least a partial determination, i.e., that the
thing body is a determinable determined by the determinator inertial
tendency which omnitemporally classifies a body. The intensional

aspect of such statements is their imputation, viz., that they establish
complete or partial identity criteria until further notice (as might be
given by future counter evidence or changing pragmatic considerations).
Just as Rescher notes, it is certainly the case that if we look at "the
explicit formulation of the overt content of a law, all we find is a
certain generalization."1 But when one characterizes laws as metalin-
guistic IC statements, as we have done, both the descriptive and legis-
lative aspects of these statements of law is explicitly apparent. As
Rescher claims, laws are not discovered, they are made. The form we
have chosen to express these law statements reflects the act of their
making. When any generalization is classed as a law, we always find

a man-made element, an imputation of determination;2 for the "evidential
basis must always be grossly insufficient to the claim actually made"
since this evidence relates to the past and present while a scientific
law relates to the future and the hypothetical.3 However, the imputa-
tions that men make are not made at a formal meeting or council of

scientists but rather at that informal moment or interval of time

 

1Omnitemporal classifying sortal properties are proclaimed
determinators.

2Rescher talks about imputation of nomic necessity. We have
tried to unpack what could be meant by nomic necessity.

3Rescher, op. cit., p. 107.

335

when scientists decide that a given confirmed statement is acceptable
as a law. The when of imputation is asvague and informal as the when
of becoming a law. However, hypotheses are accepted as laws by scien-
tists somehow at some point; and the uniformity of physics texts
suggests that agreement is fairly widespread among scientists though
there are not formal meetings convened in order to make imputations.

The time of imputation is vague; but once it occurs, it is a
yes-no process. Either a statement is a law or it is not. There are
not laws of varying degree. While it may be possible to talk about
degrees of confirmation of hypotheses, it is doubtful that we would
want to say that any partially confirmed hypothesis is a law. For
example, if we decide that some hypothesis will be confirmed after 100
violation free test trials, we would say that this hypothesis is con-
firmed to some degree after 10 successful trials and to a greater degree
after 30 successful trials. Only after the goal of 100 successful
trials is reached would we be willing to consider the imputation of
our test hypothesis and perhaps not even then if our hypothesis lacks
explanatory power, simplicity or some other pragmatic qualities. Never-
theless in any event, we would not impute the hypothesis a law until our
testing program was carried out completely. Lawfulness is imputed at
the last stage of scientific work; and when imputation takes place, the
status of a hypothesis is entirely changed to that of a law. At the
point of possible imputation, a hypothesis either remains a highly con-
firmed non-law or, via the imputation, it becomes a full-fledged law

statement. No imputed statement is 40 percent a law or 80 percent a law.

336

All laws are 100 percent law even though some laws may be more important
than others. More general laws can sometimes play more expansive or
pervasive roles within a total body of laws. We will be less willing
to give these more general laws up since we have greater need of them
for our future theorizing and predictions. But nevertheless, all law
statements with the same logical form, that of a metalinguistic IC
statement, are precisely that--the same.

The view we have just presented is a necessary view of law
statements which claims that they are contingent, intentional, meta-
linguistic statements. Laws are not necessarily true but they inform
us of what predicates are to be considered as necessarily related.
While legislating a necessary relationship, laws can nevertheless be
empirical in several ways. Though law statements are meta-statements
‘they are not ultimately only about object-level expressions. Meta-
linguistic statements can indirectly be about things of the empirical
world to which the object-level terms within them refer.1 Law state-
ments also find their source in experience, and they can always be
falsified by experience. It is in this way that laws are still empir-
ical and non-necessary though they involve nomic necessity. The major
objection to necessity views, i.e., that laws are not empirical, is
avoided. Nor have we resorted to an object-level intensional language
in which to formulate law statements. Our theory also explicates the

dual aspects of laws: (1) that they are based on a finite number of

 

1See page 200.

337

factual statements (constant conjunctions as Hume might say) but yet

(2) they assert more than some claim about past observed cases. In

our theory laws make claims about future and hypothetical cases as well.
Thus our theory not only avoids the major objections to typical neces-
sity views; but our metalinguistic necessity view also avoids the major
objections to the regularity theory as well.

Our theory circumvents objections to

the regularity view

It is clear that the imputation which occurs during the final
stage of scientific work is only applied to constantly conjoined events
within the justifying era of a law's biography. Hence our theory avoids
the mere coincidence objection to the regularity theory. It is not any
old regular invariable sequence which looks like a law. Only sequences
that develop in experimentation or scientific observation will ever be
imputed as metalinguistic IC law statements.

Our theory also enables us to distinguish accidental generaliza-
tions from laws. Mere accidental generalizations can still be written
in a frist-order extensional language as material conditionals. But
law statements as metalinguistic IC statements impute that some hereto—
fore true non-strict universal should now be considered a strict uni-
versal. Mere accidental generalizations expressed by object-level
statements such as "All pears in this box are ripe," may not be used
for prediction or in the deduction of counterfactual statements simply
because we have not imputed to them that they are at least partial

determinations as we do when stating the metalinguistic, intensional,

338

contingentstatements which express the law statements of science.
These latter mention a true universal statement and legislate that

it is to be at least a partialdeterminatiun. Since the object-level
statement “All pears in this box are ripe," merely purports to report
that certain pears are ripe without legislating also that it is at
least a partial determination that the property ripeness shall be
considered a property which plppyp classifies pears in this box, it
cannot be said that any pear which is not now in this box would, if

it were in this box, also be ripe since ripeness has not been imputed
to be an identity characteristic of pears in this box.1 If ripeness
had been imputed orlegislated as an omnitemporal identity character-
istic of pears in this box, as it would be were "All the pears in this
box are ripe" a part of-a metalinguistic law statement, then of course
anything not now a pear in this box, if it were a pear in this box,
would have to be ripe also; for if it were not ripe, we would be unable
to identify it aS'a pear in this box since it would lack one of the
identity characteristics of being a pear in this box. Thus the impu-
tation of certain properties as determinators is the difference between
true universal statements that are laws and those that are not. The
difference between accidental universals and strict law universals in
our theory is clear. Mindful of Hempel's observations about the re-

strictedness (construed as finite extension) of some law statements

 

1N.B. ripeness is not an omnitemporal classifying property
thus it is not a potential determinator property and will be found
neither in law statements nor lawlike statements. This will be
explained more fully in succeeding pages.

339

and the essentially generalized form of some accidental generalizations,
we, as Popper, construe restrictedness to mean closed extension rather
than finite extension. 'Law statements which express the imputation of
some generalization express that the extension of the generalization
shall be considered open such that contrary-to-fact conditionals and
subjective claims can be deduced from them. These contrary-to-fact

and subjunctive conditionals would not only be those which claim that

if some nonépear in this box were numerically the same as one of the ob-
jects which is a pear in this box it would be ripe but also that if some
non-"pearsin-thisabox" were added to the extension of ”pears in this
box" as a new object not before mentioned by the statement "All pears

in this box are ripe," it would be a ripe pear. The openness or
closedness of the extension of terms in laws and accidental general-
izations is indicated in our theory by the words "is a determination"

in the metalinguistic law statement. Since law statements are inten-
sional metalinguistic statements, the extensions of the terms in a true
universal object-level statement can vary without fear of truth value
change while the extensions of terms in accidental generalizations
expressed as first—order extensional object-level statements cannot.

Our theory therefore provides at once for a way to distinguish
accidental generalizations from law statements and for an answer to the
subjunctive case, vacuosness, and hypothetical cases objections. The
characterization of laws as metalinguistic intensional statements allows
for the introduction of the intensional notions of necessity and possi-

bility. If'a quality or event is imputed as the mark or identity

340

characteristic of another then this necessary relationship of course
will apply to all possible cases. As mentioned in Chapter I,1 if we
can say that laws apply not only to actual cases but possible ones as
well, then the vacuosness objection is easily avoided. We merely say
of laws, such as Newton's forceafree-body law, that it is a statement
of possibility such that if there were any bodies on which no forces
were acting then they would behave according to Newton's law. While
the regularity view maintains that law statements cover only what is
actual as Ayer concludes,2 our theory states that law statements cover
what is possible as well thus avoiding the vacuosness objection.

Our theory also providesfor hypothetical cases such as that
cited by Ayer. A functional law such as the one which legislates and
reports the relationship between a volume of gas at a given temperature
under a given pressure written as x = Fy is one in which the range of
the variable y covers all possible values of the quantity in question.
In terms of our theory laws express a relationship which holds for all
cases actual and possible since the posit of even a partial determina-
tion is the decision to characterize every x actual and possible as an
Fy until further notice.

All objections to the extensional regularity view are dissipated
by our intensional view. Yet we have not resorted to a first-order
intensional language on the object-level of discourse. In addition our

theory talks about laws legislating necessity without being necessary.

 

1See page 5.
2See page 6.

341

Therefore, our laws are not selfeevidently true. They are based on
experience. They can be falsified while nevertheless providing for
the relative necessity of particular prediction statements which often
contain the term "must." It is in this way that we can say that laws
have nomic necessity (establish certain connections as necessary) and
yet are empirical while avoiding the typical objections to both the
necessity and the regularity views of law statements.

It might still be objected that in our theory laws as man-made
imputations are more conventional than empirical. We claim however
that the idea of imputation is perfectly reflective of scientific
practice. In our necessity view, statements of scientific law are not
arbitrary or strictly conventional. Rather they are, as Rescher states,
well-founded imputations based both on observational evidence and upon
the "systematic-coherence of fitting the generalization into a fabric
of others that in the aggregate constitute a rational structure, an
integrated body of knowledge that constitutes a "branch of science."1
In view of these conclusions it might well be wondered what exactly is

to be understood by "science"?

A view of science

The views of reduction, realism, and instrumentalism are well-
known, much-debated doctrines about the nature of scientific theories.
The limitations of each view are also well known. However, in pre-

senting the view of science which will yield a possibly adequate

 

1Rescher, op. cit., p. 107.

342

account of it, we wish to consider instrumentalism as expressed in
Duhem's The Aim andStructure of Physical Theory_since it is the most
suggestive starting place for our own view. Duhem writes:

A physical theory is not an explanation. It is a system
of mathematical propositions, deduced from a small number
of principles, which aim to represent as simply, as
completely, and as exactly as possible a set of
experimental laws.1

In further elaboration he says:

1. Among the physical properties which we set ourselves
to represent we select those we regard as simple proper-
ties, so that the others will supposedly be groupings or
combinations of them. We make them correspond to a cer-
tain group of mathematical symbols, numbers and magnitudes,
through appropriate methods of measurement. These mathe-
matical symbols have no connection of an intrinsic nature
with the properties they represent. They bear to the
latter only the relation of sign to thing signified.
Through methods of measurement we can make each state of
a physical property correspond to a value of the repre-
sentative symbol, and vice versa.

2. We connect the different sorts of magnitudes thus
introduced by means of a small number of propositions
which will serve as principles in our deductions. . . .
These hypotheses may then be formulated in an arbitrary
way. ‘The only absolutely impassable barrier which limits
this arbitrariness is logical contradiction either among
the terms of the same hypothesis or among the various
hypotheses of the same theory.

3. The diverse principles or hypotheses of a theory are
combined together according to the rules of mathematical
analysis. . . .

4. The various consequences thus drawn from the
hypotheses may be translated into as many judgments
bearing on the physical properties of the bodies. The
methods appropriate for defining and measuring these
physical properties are like the vocabulary and key
permitting one to make this translation. These judgments
are compared with the experimental laws which the theory
is intended to represent.2

_—

1B. Brody and N. Capaldi, eds., Science: Men,_Methods and
Goals, p. 81.

2Ibid.. PP. 81-82.

 

343

Theories in the above sense are useful, as Duhem points out,
for classification of experimental laws and their ordering into a
smaller number of principles. However, it is also part of such an
instrumentalist view that scientific theories do not explain or describe
reality and even in the view of some writers that scientific theories
are not true or false but only useful or unuseful. Osiander, in his
preface to Copernicus' On the Revolutions of the Celestial Spheres,
claims that “it is not necessary that these hypotheses should be true
or even probable; but it is enough if they provide a calculus which
fits the observations."1 Brody and Capaldi make the same claim when
they say, according to instrumentalism,"a theory is neither true or
false."2 However, we find no reason to draw such a conclusion from
the instrumentalist position. A scientific theory may be an economical
way of classifyingexperimental knowledge, and it may also guide further
research in the discovery of novel experimental laws. But this is not
to say that such theories do not explain observed phenomena or that
they are useful but not true. Obviously disagreement in this regard
turns on the interpretation that is given to the words "true" and
"explain."

We accept the interpretation of "true" which Duhen (an instru-
mentalist as we have already seen) must have had in mind when he spoke

about theories being true and false. He says:

 

1Ibid., p. 74.
2Ibid., p. 11.

 

344

Thus a true theory is not a theory which gives an

explanation ofphysical appearances in conformity

with reality;it is a theory which represents in a

satisfactory manner a group of experimental laws.

A falsetheory is not an attempt at an explanation

based on assumptions contrary to reality; it is a

group of propositions which do not agree with the

experimental laws. Agreement with experiment is

the sole criterion of truth for a physical theory.1

More explicitly Duhem writes:

The words "truth“ and "certainty“ have only one

signification with respect to such a theory; they

express concordance between the conclusions of the

theory and the rules established by the observers.2

It would seem that even in terms of a correspondence theory of

truth Duhem's account is a perfectly acceptable theory of truth. It
is true that I am writing with a penbecause I observe the pen in my
hand as I write. ‘My statement corresponds with what I observe. So
too, a theory can be considered true as long as it corresponds with
the experimental data and it is the theory actually adopted.3 Several
theories may be compatible with the experimental data, and one of them
may be adopted because it classifies or catalogs the experimental laws
more economically or leads to more research or has a wider predictive
scope. Any or all of these considerations may be the rules established
by observers to which Duhem refers. When a theory is compatible with
observations and satisfies these other rules of economy, etc., it is

the one theory accepted as true at least until the rules are changed

 

1Ibid., p. 82.

 

2Ibid., p. 172.

3There may be other competing theories which are confirmed but
not accepted or adopted. InasmUch as they are not adopted, we would
not say that they are true.

345

or contrary observations are made. Since other considerations besides
experimental evidence enter into the decision to adopt a particular
theory and theories extend well beyond the realm of observation, it

is readily understandable why it is fairly useless to speak of a
particular theory being the only possibly true description of the real
world. 'Theories are corrigible, varying descriptions of the world;
and each in its turn, as adopted, constitutes our present view of what
is or is not the real world. C. D. Broad expresses the situation well
when he says:

Presumably there is a system of ojbective fact, with a
structure of its own which is quite independent of us
and our thoughts andour languages. Its structure may
be compared with the intrinsic nature of a developable
surface of finite uniform curvature or of a surface of
finite variable curvature. In our cognitive activities
we are trying to think about this system of objective
fact and to know as much as we can of it. But in order
to do this in any detail we have to use language and
other systems of symbolism. . . . An immense number of
alternative languages and other systems of symbolism
are no doubt available. . . . But presumably the objec-
tive structure of the system of fact imposes some
limitation on the alternative systems of language or
symbolism which are capable of representing it. . .
Thus if we could disentangle and formulate any features
common to all systems of symbolism and languages we
would have a clue to the structure of the objective
system of fact. And it is doubtful whether we could
discover anything about the latter except in this
rather roundabout way. Moreover to be intelligible

the formulation would have to be made in some partic-
ular pre- existing language or system of symbolism.

As we also have noted, the existence of competing theories does

not mean that the one chosen according to certain rules as acceptable

 

1C. D. Broad, “Are There Synthetic A Priori Truths?" p. 112.

346

and true does not describe what in some way1 really exists. In.
stressing this point, Arthur Pap claims that the question of whether

a theory is a true description of reality or merely an instrument of
prediction is a pseydo-question that "owes its longevity to pictorial
thinking."2 Pap maintains that the statements that "'There is a table
in the room' describes reality," and "'There is a table in the room' is
a hypothesis confirmable by the predictions it yields," are compatible.
He claims indeed that the latter explicates the former. As he states:

If one feels that the former means more than the latter,
the feeling has the following psychological origin; when
nobody is there to perceive it, you yourself have in
your mind an image of the table in the room, and this
image is distinct from the purely hypothetical state of
affairs that tablelike sense-data would occur if certain
conditions were fulfilled. But any conceivable evidence
for the existential proposition is evidence for the
hypothetical proposition and conversely since the prop-
ositions are logically equivalent, have the same factual
content?3

Pap raises this argument to combat the unsophisticated belief
that only what is pictorial can explain phenomena or describe what is
real.

. . . Physical theories are intuitively satisfactory only
if they gain pictorial content through models. Where
such models are lacking, as in the relativisitic theory
of geodesics in curved space and the quantum theory of
probability waves, the feeling may arise that useful
conceptual mathematical constructions have replaced
descriptions of physical reality.“

 

1We apply the phrase "really exists" to different entities as
our theories change.

2Arthur Pap, An Introduction to the Philosophy of Science, p. 350.

3Ibid., pp. 355-356.

1'Ibid., p. 355.

 

 

347

Thus, Pap is concerned to show that physical theories can be
viewed both as descriptive of reality and as instrumental.
More extremely than Broad's pronouncement, we might agree with
Thomas Kuhn that the theories, which are adopted by scientists as most
instrumental, actually present us with the real. As theories or
paradigms change so does reality. It is not merely our view of some
absolute hidden reality behind our data which changes. As Thomas Kuhn
writes:
One perceptive historian, viewing a classic case of a
science's reorientation by paradigm change, recently
described it as "picking up the other end of the stick,"
a process that involves "handling the same bundle of
data as before, but placing them in a new system of
relations with one another by giving them a different
framework. Others who have noted this aspect of sci-
entific advance have emphasized its similarity to a
change in visual gestalt: the marks on paper that
were first seen as a bird are now seen as an antelope,
or vice versa. That parallel can be misleading. Sci—
entists do not see something as something else; instead,
they simply see it.1
In terms of a metaphor of a boat we claim likewise that from
the shore a philosopher of science might say that the ship of science
follows different courses at different times on the ocean of reality
but, as Kuhn would say, the scientist on the ship just sees on any
given course the ocean--the real as it is at a given time.
Kuhn's description of science not only agrees with our
instrumental view of science but also more particularly with our
metalinguistic theory of law statements. Not surprisingly his words

are much like our own when he says:

 

1Thomas Kuhn, The Structure of Scientific Revolution, p. 78.

 

348

. we shall see that the chemical law of fixed
proportion, which before Dalton was an occasional
experimental finding of very dubious generality,
became after Dalton's work aningredient of a defi—
nition of chemical compound that no experimental work
could by itself have upset.1

The above quotation reflects the act of imputation whereby
an object-level true universal generalization is declared to be a
determination by a metalinguistic intensional contingent statement
such that the terms which previously referred to accidentally corre-
lated events or properties become definitionally joined by a second
metalinguistic IC definition statement. A property or set of prop-
erties serves as a definitional mark of some investigated phenomenon
until further notice--until there is an anomaly which "evokes a crisis."2

When a crisis arises definitions and the laws which provide us
with them are changed. But as the word crisis suggests this is not a
smoothe transitional change. Definitions do not blend into one another
or build on one another so much as they replace one another.

In further discussion about the change of laws and their
corresponding definition statements as when whole paradigms change
Kuhn writes:

Each of them (changes of theories) necessitated the
community's rejection of one time-honored scientific
theory in favor of another incompatible with it. Each
produced a consequent shift in the problems available
for scientific scrutiny and in the standards by which

the profession determined what should count as an
admissable problem or as a legitimate problem-solution.3

 

 

 

349

The replacement of laws and theories by competing laws and
theories serves to indicate the on-off process of imputation as we
have described it. However, as we have also maintained the process
whereby a law is imputed or refuted is not a clear-cut empirical
activity. One well confirmed hypothesis is accepted or imputed while
its competitors are not because such a hypothesis is more simple than
they, more fruitful or more explanatory.

As Kuhn says again of the acceptance or rejection of sets of
laws:

The decision to reject one paradigm is always simulta-
neously the decision to accept another and the judgment
leading to that decision involves the comparison of both
paradigms with nature and with each other.1

The above passage points once again to an instrumental con-
ventionalist view of science such as the one we adopt herein and this
instrumental view permits us to talk of definitions and identifying

marks without resorting to some brand of essentialism so stultifying

to scientific progress.

Essentialism

 

We have, in the present account of statements of scientific
law and of science itself, sought to avoid all suggestions that science
either searches for the essences of things or deals with essential
properties. Mr. Bennett in his discussion of sortal properties took
the subject of his discussion to be essential properties, and he

regarded sortal properties as necessary or natural properties. We

 

1Ibi ., p. 77.

350

have avoided adopting his characterization of non-phasal, sortal
properties as necessary pr0perties; and we have also avoided all
mention of the term "essence" and its derivatives. The reason for

our eschewment of the term "essence" should by now be clear. Ours

is a relativistic, instrumentalistic, and conventionalistic (in the
best sense of these terms) theory of science and its laws. Our theory
is relativistic in that scientists choose to establish varying identity
relationships depending on their varying interests. Man is identified
by the biologist as a featherless biped, by the sociologist as a social
agent, by the psychologist perhaps as a rational animal. The identifi-
cation of man in the various branches of science is as varied as the
interests of their practitioners. Our theory is instrumental in that
the relativity of identity relationships reflects the truth that the
possible true courses of a scientific ship upon the ocean of what is
are many while the observational lighthouse guides are few. Many
alternative theories employing diverse languages and symbolism may
yield explanations of phenomena without being violated by any obser-
vational data. The theory (the ship's course) which is the pragmat-
ically most workable is usually the one chosen as the true theory (best
course of the ship). The pragmatic and epistemic considerations of 1
systematic force, explanatory force, economy, and fruitfulness for
further research, which are the basis of imputation to certain general-
izations of law status, illustrate well the conventional though non-

arbitrary character of our theory. This much stressed, it should be

 

1Karl R. Popper, The Conjectures and Refutations, p. 104.

351

obvious that we argue neither for nominal essences or natural kinds
when maintaining that identity relationships are conventionally
established by scientists relative to their needs. IC law statements,
which are instrumental in permitting scientists to make predictions
about the phenomena they have chosen to study, are based partly on
experience just as any tenable empiricism would require. Popper
describes the theory of essentialism so contrary to ours by saying:

. . essentialism maintains that the best truly scientific

theories describe the 'essences' or the essential natures

of things--the realities which lie behind the appearances.

These theories are neither in need nor susceptible of

further explanations and to find them is the ultimate

aim of the scientist.1

Newton, who adapted from Descartes the view that the essence

of a thing must be a true or absolute property of the thing (a property
which does not depend on the existence of other things) such as exten-
sion, did not regard gravity as an essential property of matter.2 But
as an essentialist, Newton did try to find an acceptable ultimate
explanation of gravity by trying to deduce the square law from the
assumption of a mechanical push. Such a solution, had he found it,
would have been regarded by him as the ultimate explanation of gravity.

But as Popper notes, the question of "Why can bodies push one another"

can still be asked today.

 

1Karl R. Popper, The Conjectures and Refutations, p. 104.
2Ibid., p. 107.

352

We now believe that they push one another because Of

certain repulsive electric forces. 'But Cartesian and

Newtonian essentialism might have prevented this

question from being raised. '5 ‘belief in essences

creates obstacles to thought.1

Popper believes that essentialism must be discarded since the

world Of each of our theories may be explained by further theories Of
a higher level Of abstraction or universality and Of testability.2
However, he also says:

I do not think that a language without universals

could ever work; and the use Of universals commits

us to asserting and thus at least to conjecturing

the reality of dispositions though not of ultimate

and explicable ones--essences.3

All that has been said in this paper concerning the statements

Of scientific laws has been said in agreement with the above. It can
even be said that our theory is just the Opposite Of an essentialist
doctrine. We do not suppose that in some arm chair manner a rational-
ist scientist can sit down, divide reality into parts and wholes, and
then by reflection decide which properties are determinators, i.e.,
which will sort various properties, things, or events from other things,
properties or events. Rather we mean to maintain quite oppositely that
scientists Observe, hypothesize, test experimentally, and confirm
various hypotheses. Then for various reasons Of conceptual consistency,
economy, etc., scientists impute to various confirmed generalizations

the status of;1aw. The properties mentioned in these laws and not the

 

1Ibid., p. 115.

2

 

H

bi

Q

 

3Ib

a—Io

d., p. 119.

353

properties mentioned in the non-imputed (but perhaps confirmed lawlike
generalizations) become the determinator properties of various things,
events, or other properties. It may even be the case that determinators
found in different law statements are clustered to fOrm the identifying
locus of some thing which was not considered a thing at all before the
clustering was undertaken.1 In this way the omnitemporal classifying
properties that are chosen as determinators may actually affect the
claims that scientists make about what is or is not a thing, property
or event. Thus there is more to the real than what meets the eye. Our
theory is not, therefore, purely empirical as is the regularity view
nor purely conventional nor rationalist as previous necessity views.
Our metalinguistic necessity view of law statements is, if anything,

rationally empirical.

 

1Horned angles example, p. 235.

WORKS CITED

__ I- _. ———.._ ..._.—_

WORKS CITED

Ayer, A. J. The Concgpt of a Person. New York: St. Martins Press,
1963.

. Foundations of Empirical Knowledgg, London: Macmillan &
Company, 1964.

Language Truth and Logic. New York: Dover, 1946.

Alston, W. P. Philosophg of Language.- Englewood Cliffs, N.J.:
Prentice a , 4.

"The Quest for Meanings." Mind. LXXII (1953). 79-87.

Aristotle. The Basic Works Of Aristotle. Edited by Richard McKeon.
NewaOrk: Random HOuse, 1

Beck, Lewis White. "On the Meta-Semantics Of the Problem of the
Synthetic A Priori." Studies in the Philoso h Of Kant.
Edited by Beck. New YorE: 80555 MErrill, 1365.

Bennett, Daniel. "Essential Properties." Journal Of Philosophy, LXVI
(1969), 487-499.

Bennett, Jonathan. Kant's Analytic. Cambridge: University Press,
1966.

Bird, Graham. Kant's Theoryyof Knowledge. London: Routledge and
Kegan Paul, 1962.

Black, Max. "Making Something Happen." Determinism and Freedom.
Edited by Sydney Hook. New York: New York University
Press, 1958.

Models and Metaphors. Ithaca, N.Y.: Cornell University
ress, 1962.

. "Necessary Statements and Rules." PhilOSOphical Review,
LXVII (1958), 313-341. '

Bohnert, Herbert. "Carnap on Definitions and Analyticity." The

Philoso h of Rudolf Carna . Edited by Schilpp. LanTTe,
Ill.: gpen Court, 1963.

354

355

Broad, C. 0. "Are There Synthetic A Priori Truths?" Aristotelian
Society Proceedings Supplement. XIV-XVI (19351, 102-177.

. The Mind and Its Place in Naturg, New Jersey:
Littlefield Adams Company,

Brody, B., and N. Capaldi, ed. Science: Men,_MethOds and Goal.
New York: W. A. Benjamin, Inc., 1968.‘ 7’

Buchdahl, Gerd. "The Kantian Dynamic of Reason with Special Reference
to the Place Of Causality in Kant's System." Kant Studies
Today. Illinois: Open Court, 1969.

Bunge, Mario. Causality. Cleveland: The World Publishing Company,
1963.

 

Campbell, Norman. What Is Science? New York: Dover, 1921.

Cavendish, A. P. David Hume. New York: Dover Publications, Inc.,
1958.

 

Church, Ralph. hpme's Theory Of the Understandihg, U.S.A.: Archon
Books, 1935.

Collingwood, R. G. An Essay on Metaphysics. Oxford: Clarendon Press,
1940.

Copi, Irving. Introduction to Logic. 2nd ed. New York: Macmillan
Company, 1961.

Dewey, John, and A. F. Bentley. "Definition." Journal of Philosophy,
XLIV (1947), 281-306.

Donagan, Alan. The Later Philosophy_of R. G. Collingwood. Oxford:
Clarendon Press,_T962.

Ducasse, Curt John. Causation and the Types Of Necessity, New York:
Dover, 1969.

Ewing, A. C. Fundamental Questions Of Philosophy. New York:
Macmillan, 1952.

. Idealism: A Critical Suryey, London: Methuen and
Company Ltd., 1934.

Kant's Treatment Of Causality, U.S.A.: Anchor Books, 1969.

Fischer, Kuno. A Commentar on Kant's Critique Of the Pure Reason.
London: Langmans, reen and COmpany, 1866.

356

Flew, Anthony. Hume's Philosophy Of Belief. London: Routledge
and Paul, I961.

Frege, G. The Foundations Of Arithmetic. Illinois: Northwestern
University Press, 1968.

Garver, Newton. "Analyticity and Grammar." Kant Studies Today.
Edited by Louis Beck White. LaSalle, 111.: ’Open Court,
1969.

Gewirth, A. "The Distinction Between Anal tic and Synthetic Truths."
The Journal of Philosophy, L (1953 , 397-425.

Goodman, Nelson. Fact, Fiction and Forecast. 2nd ed. U.S.A.:
Bobbs-Merrill Company,l965.

Hall, Roland. "Presuming." The Philosophical Quarterly, XI (1961),
10-21.

Hamlyn, D. W. "Analytic and Synthetic Statements." Encyclopedia
of Philosophy, lst ed., Vol. I.

. "Contingent and Necessary Statements." Encyclopedia of
- Philosophy. lst ed., Vol. II.

 

Hancock, Roger. "Presupposition." Philosophical Quarterly, X (1968),
73—78.

Hanson, N. R. "A Note on Statements Of Fact." Analysis, XIII (1952),
24.

. Patterns Of Discovery. Cambridge: Cambridge University
Press, 1958.

. "The Very Idea Of a Synthetic A Priori." Mind, LXXI
(1962), 521-524.

Hempel, Carl. As ects of Scientific Explanation. U.S.A.: The Free
Press, 96 . 7’7

Hospers, John. An Introduction to Philoso hical Analysis. Englewood
Cliffs, N.J.: Prentice Hall, 1953.

Hume, David. A Treatise of Human Nature. Edited by Selby Bigge.
Oxford: Clarendon Press, 1888.

Essential Works Of David Hume. Edited by Ralph Cohen.
New Yor : antam 00 s,

357

Kant, Immanuel. Critique of Pure Reason. Edited by Norman Kemp Smith.
London: Macmillan X Cdmpany Ltd., 1964. ‘

. Prolegomena to Ahy Future Meta h sics. Edited by L. W. Beck.
New York: The Liberal Arts Press, l950.

Kneale, William. "Universality and Necessity." British Journal for
Philosophy_of Science, XII (1961), 89-102.

. "Natural Laws and Contrary-tO-Fact Conditionals." Analysis,
X (1950), 121-125.

Probability and Induction. Oxford: Clarendon Press, 1949.

. "Truths of Logic." Aristotelian Society Proceedings, CXLVI
* (1945-1946), 207-235. ,

Kuhn, Thomas. The Structure of Scientific Revolutions. Chicago:
University OfChicagO"Press,’1964.

Laird, John. Hume's Philosophy of Human Nature. London: London
Methuen & Company, Ltd., 1932.

Lovejoy, Arthur 0. "On Kant's Reply to Hume." Kant; Disputed
guestions. Edited by Moltke S. Gram. Chicago: Quadrangle
OO 5, 67.

Malcolm, Norman. "Are Necessary Propositions Really Verbal?" Mind,
XLIX (1940), 189-203.

Martin, R. M. "On Analytic." Philosophical Studies, LXI (1952), 41-47.

Massey, Gerald. Understanding Symbolic Logic. New York: Harper Row,
1971.

Mates, Benson. "Synonymy." Semantics and the Philosophy of Lan ua e.
Edited by Leonard Linsky. urbana, I11.: University of
Illinois Press, 1952.

Maxwell, Grover. "The Necessary and the Contin ent." Mhhnesota Studies
in the Philosophy Of Science, III (19621, 398-404.

Maxwell, Nicholas. "Can There Be Necessary Connections Between
Successive Events?" British Journal for Philosophy Of Science,
XIX (1967), 1-25. 1

Milmed, Bella. Kant and Current PhilQSOphical Issues. New York:
New York UniversityPress, 1961.

358

Pap, Arthur. An Introduction to the Philosophy Of Science. New York:
Free Press, 1962.

. "Indubitable Existential Statements." Mind, LV (1946),
234-246.

. "Regularity Theory of Causation." Journal Of Philosophy,
XLIX (1952), 657-666.

The A Priori in Physical Theory, New York: Kings Crown
Press, 1946.

 

Passmore, J. A. The Philosophy of David Hume. Cambridge: Cambridge
University ress,

Pepper, Stephen. "The Descriptive Definition.“ Journal Of Philosophy,
XLIII (1946), 29-36.

Popper, Karl. The Con'ectures and Refutations. 2nd ed. New York:
Basic Books, ICES.

. The Logic Of Scientific Discovery, London: Hutchinson,
1959.

. "A Note on Natural Laws and SO-Called Contrary-to-Fact
Conditionals.“ Mind, LVIII (1949), 63-66.

Prichard, H. A. Kant's Theory Of Knowledgg, Oxford: Clarendon Press,
1909.

Porteous, A. J. 0. "Are There Synthetic A Priori Truths?" Aristotelian
Societyh§upplement, XIV-XVI (1935), 118-140.

Quine, W. V. "Two Dogma's of Empiricism." From a Logical Point of View.
Cambridge: Harvard, 1953.

Reichenbach, Hans. The Rise of Scientific Philosophy, Berkeley:
University O?_Calif6rniaPress,1963.

Rescher, Nicholas. Scientific Explanation. New York: The Free Press,
1970.

Robinson, Richard. Definition. Oxford: Clarendon Press, 1954.

. "Necessary Propositions." Mind, CLXVII (July, 1958)
289-304.

Russell, Bertrand. Analysis Of Mind. London: George Allen and
Unwin Ltd., 192T.

359

. Mysticism and Logic. London: George Allen and Unwin
Ltd., 9 7.

Rynin, David. "Donegan on Collingwood: Absolute Presuppositions,
Truth and Metaphysics." Review Of Metaphysics, XVIII
(1964-1965), 301-333.

Salmon, Wesley. Introduction to Logic. Englewood Cliffs, N.J.:
Prentice-Hall, I963

Scheffler, Israel. "Prospects for a Modest Empiricism." Review
of Metaphysics, X (1957), 383-400.

Schlegel, Richard. "00 We Know The World Through Science?" Mind.
Science and Histogy. Edited by H. E. Kiefer_and M. K. Hunitz.
Albany::State Ufiiversity Of New York Press, 1970.

Sellars,]Wilfred. "Presupposing." Philosophical Review, LXIII (1954),
97-215.

Strawson, P. F. Introduction to Logical Theory. London: Methuen
and Company [td., 1952.

. "Necessary Propositions and Entailment Statements." Mind,
LVII (1948), 184-200.

Stroll, Avrum. "Presupposing." Encyclopedia Of PhilOSOphy. lst ed.
Vol. VI.

Suchting, W. A. "Kant's Second Analogy Of Experience." Kant Studies
Today. Edited by Lewis White Beck. LaSalle, I11.: The Open
Court Publishing Company, 1969.

Van Fraassen, Bas C. "Presuppositions, Implication, and Self-Reference.
Journal of Philosophy, LXV (1968), 136-151.

Walters, R. S. "Laws Of Science and Lawlike Statements.“ Encyclopedia
Of Philosophy, lst ed., Vol. IV.

Webster's New Collegiate Dictionary. 12th ed., 1961.
Whitehead, Alfred North. Symbolism. New York: Macmillan Company, 1927.

Wilbur, James, and Harold Allen. The Worlds Of Hume and Kant. New York:
American Book Company, 1967. T

Zabeeh, Farhang. Hume Precursor of Modern Empiricism. The Hague:
Martinus NyhO ,

175 013

3 1293 03

l

 

llllllll