PARADOX, PREDICTION AND SELF-VETiATiNG
HYPOTHESES

Thesis For tbs Beam of Ph. D.
MiCHiGA‘N RATE UNIVERSITY

Aifred Iackson Simmer
1963

This is to certify that the

thesis entitled
PARADOX, PREDICTION AND

SELF-VITIATING HYPOTHES ES

presented by

Alfred Jackson Stenner

has been accepted towards fulfillment
of the requirements for

Pth: degree in PhilOSOEhy

4w“; 0. //
MajoGrofessor U

Date May 15, 1.263

0—169

LIBRARY

Michigan State
University

 

ABSTRACT

RARADGX, PREDICTION AND SELF-VITIATING HYPOTHESES

by Alfred Jackson Stenner

Predictions in the social sciences often have a
puzzling effect: the utterance of the explanandum of the
prediction brings about a state of affairs which is
described by the contradictory or a contrary of the
explanandum. These phenomena have been called by a number
of names: e.g. “self-stultifying belief“, "suicidal
belief", "self-destroying hypothesis“, “Oedipus effect",
etc. The term employed in this essay is “self-vitiating
hypothesis".

The occurrence of self-vitiating hypotheses has
led some writers to claim that important methodological
differences obtain between the social and non-social
sciences. It has also been urged that self-vitiating
hypotheses make an exact science of history logically
impossible. This essay examines several of these related
claims and attempts to show that they are not cogent.

In order to deal with the various claims concerning

self-vitiating hypotheses, a formal explication of the

term is undertaken. This is followed by an analysis of the

related predicates "self-vitiatable hypothesis" and "self-

fulfilling hypothesis". While the occurrence of self-

Alfred Jackson Stenner

vitiating hypotheses is, at present, predominantly a
social science phenomenon, such hypotheses may also be
problematicfbr the non-social sciences.

The solution of the problems encountered in dealing
with self-vitiating hypotheses requires a theory which
will enable us to predict which of our utterances are, and
which are not self-vitiating. At present no such theory
is available and the difficulties which will attend the
development of such a theory are formidable. But the
difficulties are empirical and not logical.

If and when a theory of self-vitiation is developed,
other puzzles may arise. The theory of self-vitiation
will not necessarily be immune to the kind of difficulties
which prevent the social scientist from accurately
predicting the future. Predictions about certain utter-
ances being self-vitiating may themselves turn out to
contain self-vitiating explananda. These difficulties
are shown to be empirical, and they may be resolved by the
development of a more comprehensive theory of self-
vitiation.

Predictions of onefls own behavior may also contain
self-vitiating explananda. While the problems encountered
'in these instances differ in some respects from problems

considered previously, they are not beyond the possibility

of resolution.

The essay concludes with a brief examination of

some of the relations which may obtain between a theory

Alfred Jackson Stenner

of values and a theory of knowledge due to the phenomena

of self-vitiating hypotheses.

PARADOX, PREDICTION AND SELF-VITIATING HYPOTHESES

by

Alfred Jackson Stenner

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Philosophy

1963

;_believe in_order that ;_may know. -- St. Anselm

;§_you don't make any assumptions, you can't draw
any conclusions. -- Henry Leonard

;§.w§_are ready Eg_tolerate everything §§_understood,
there i§_nothing left tg_explain; while i£_wg_sourly
refuse tg_take anything, even tentatively, §§_clear,
ng_§xplanation can b§.given. -- Nelson Goodman

PREFACE

The following essay concerns the phenomenon of
self-vitiation, a subject of considerable importance for
the philosophy of science. Yet it is a kind of phenomenon
which has received but scanty attention in the literature.
The fact that self-vitiation is often overlooked in
discussion of science may be due in part to the fact that
we have so little understanding of the phenomenon in
question. It is hoped that this essay will contribute to
an increased awareness of the problems which are posed by
self—vitiating utterances, and also to an increase in time
and other resources which will be required in attempting to
resolve some of the problems in this area of inquiry.

I am grateful to the members of my guidance committee:
William J. Callaghan, Ph.D., Paul M. Hurrell, Ph.D., Richard
S. Rudner, Ph.D., and John F. A. Taylor, Ph.D. for their
helpful comments and encouragement.

Those who are familiar with the recent philosophical
literature will recognize my indebtedness to Professors
Nelson Goodman, Carl Hempel, Henry Leonard, and Paul

Oppenheim.

Above all, I wish to thank Professor Richard S.
Rudner who has guided this work from its inception to its
completion. My conversations with him about the material
in this essay have all proved singularly rewarding. His
comments and suggestions have brought a degree of organ—
ization and clarity to the following material which would
not have been otherwise achieved.

Finally, I wish to express my gratitude to my
family, my wife, Jean, and our three sons, Jack, Jim and
Rob, without whose encouragement, sympathy and tolerance
through a trying period, this manuscript would never have

been completed.

ii

TABLE OF CONTENTS

CHAPTER ' PAGE
I. A PRESYSTEMATIC DISCUSSION OF SELF-VITIATION..1
The Nature of the Problem...................l
Adoption of a uniform Terminology...........3

Some Illustrations of Self-Vitiating
Hypotheses..................................lS

The Relevance of the Issue..................24

II. AN EXAMINATION OF SOME OF THE ARGUMENTS IN
TIE LITERATIJREOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO28

IntrOdUCtionOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO28
The Arguments of John Venn..................28

Perturbance Effects in the Physical and
SOCj-al SCierlceSOOOOOOOOOOOOOOOOOOOOOOOOOOOOO34

"Predictions of Future Knowledge and

Knowledge of the Future" -- The Arguments

Of PopperOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO38
A Critique of Popper's Views................42

The Arguments of Feuer and Gewirth..........57

III. A FORMAL ANALYSIS OF THE PREDICATE 'SELF-
VITIATING'OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO71

IntrOdUCtionO O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O 7]-
FirSt StepSO O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O.72
smarYOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO93

IV. SELF-VITIATING AND SELF-FULFILLING HYPOTHESES
COMPAREDOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO95

IntrOductiOnOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO95

A Definition of 'Self—fulfilling'...........95

iii

iv

'Self—vitiating' and 'Self—fulfilling' as
Applicable to Generalizations...............lOl

“Boundary Conditions“ and Self-vitiating
HypOtheseSOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOlls

Summary.....................................126
V. TOWARD A THEORY OF SELF-VITIATION.............127
Introduction................................127
A New Predicate, 'Self—Vitiatable'..........128
The Problem of Disposition Predicates.......132

Pervasiveness of Self-vitiation and Self-
Vitiatability.OOOOOOOOOOOOOOOOOOOOOOOOOOOOOO145

Goal-directed Phenomena and Self-vitiation..l49
Sign-behavior and Self-vitiation............155
Summary.....................................l67
VI. RECAPITULATION................................l68
Introduction................................l68

Does Self-vitiation Make Knowledge of the
Future Impossible?..........................168

A Re-examination of the Arguments of Feuer
and GewirthOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO178

An Empirical Requirement for Successful
PrediCtionOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO0188

Self-vitiating Instances of a Theory of
Self-VitiationOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO197

Predicting One's Own Behavior...............201
Summary.....................................205
VII. SUMMARY AND IMPLICATIONS......................207
Summary of the Analysis.....................207
Some Implications of the Preceding Analysis.212

BIBLIOGMPHYOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO221

CHAPTER I
,A PRESYSTEMATIC DISCUSSION OF SELF-VITIATION

Iptroduction

 

One of the most important function, if not thg_most
important function of scientific knowledge is that it
enables us to predict, with a high degree of success, what
the morrow will bring. Sometimes, however, the very
utterance of the prediction seems to bring about an event
which is described by the contradictory of the utterance.
In this chapter we shall adopt a terminology which will
enable us to describe, with modest precision, the phenomena
in question, and then give some examples of the phenomena
which will give some indication of the range of application
of the predicate 'self—vitiating'.

The Nature of the Problem

Throughout the recent philosophic literature, a
particular argument -- or what could be considered as a
set of related arguments -- has received considerable
attention. The claim has been made that methodology and
content of the social sciences differs in certain specifiable
ways from the methodology and content of the physical
sciences. And it is further claimed that these differences
are of such a nature that for purposes of methodological

analysis it makes little or no sense to speak, in Quine's

la/

phrase, of ”The whole of science". Among the concerns with
which the present essay will be concerned is whether the
proponents of this view have made good their claim.

Not all of the arguments which have been advanced
in favor of this thesis will, of course, be dealt with in
what follows. We will not, for example, be specifically
concerned with the arguments which have frequently been
made from verstehen; nor will we be concerned with arguments
concerning the uniqueness of, or peculiarly teleological
character of, social science phenomena. These arguments
have been effectively answered elsewhere1 and need not
detain us, except to note that it is this author's conviction
that each of these arguments suffers from a fatal flaw.

The argument to which we shall direct our attention,
has appeared less frequently in the literature. It con-
cerns a class of phenomena which has been called by a number

of names to characterize a peculiar feature shared by all

 

For a criticism of verstehen as a methodological
principle, see Theodore Abel, “The Operation Called Verstehen”,
American Journal of Sociology, 54:211-218, 1948-1949, re-
printed in Edward W. Madden, Th§_Structure 9f;§cientific
Thought (Boston: Houghton Mifflin Co., c.d. 1960) pp. 158-166.
For a discussion of the place of law-like statements in
historical explanation, see Ernest Nagel, Thg_Structure 9§_
Science, (New york: Harcourt, Brace, and‘World, Inc., c.d.
1961) pp. 547-606. An extensive bibliography for treatments
of these issues is found in Madden, 22, gi£,, pp. 372-376.

 

2
members of this class. These have been called "self-
destroying prophecies",2 "suicidal predictions",3 "self-
destroying beliefs",4 while Richfield and Copi refer to
them as "predictions...which are vitiated".5 We shall
employ the predicate 'self-vitiating hypothesis' to
characterize the phenomena in question. One of the
important facets of the explicandum we are here con-
sidering is its suprising recalcitrance to very clear
explication. Accordingly, a major effort in the
discussion to follow will be given over to an attempt to
furnish a fairly rigorous and precise explicatum for
this term. For the present, however, we shall begin
with an informal characterization of the phenomena to
which we intend to apply the term and then give a

number of illustrations which indicate the scope of the

 

2Alan Gewirth, "Can Men Change The Laws Of Social
Science", Philosophy p£_Science, 21:229-241, July, 1954.

3Ernest Nagel, pp, cit., pp. 468-473.

4R. K. Merton, Social Theory and Social Structure,
(Glencoe, Ill.:The Free Press, 1957, Revized and enlarged
edition.) pp. 121-130.

SJerome Richfield and Irving M. Copi, "Deciding
and Predicting", Philosgphy 9; Science, 28:47-52,
January, 1961.

term's application.
Adoption of a Uniform Terminology

In order to keep our presystematic discussion of
self-vitiation as free of ambiguity and vagueness as
possible we shall find it convenient to adopt a uniform
terminology.

By hypothesis we shall mean any statement or
utterance of a statement.

We shall use the term 'statement' neutrally to
avoid any unnecessary commitment to the existence of
propositions, meanings and like entities.6 Thus construed,
the term 'statement' may be understood as "a declarative
sentence—type insofar as it is used to indicate a certain
proposition"7 or as an individual composed of a number
of sentences "catalogued under a single label",8 i.e.,
an individual composed of all the replicas of a particular
sentence or the sentential part of an individual sign—

event.

 

6 . . . .
The reasons for this dec181on are given below.

7Henry Leonard, Principles p§_Right Reason (New
York: Henry Holt and Co., c.d. 1957), p. 615.

8Nelson Goodman, The Structure p£_éppearance
(Cambridge, Mass., Harvard University Press, 1951), p. 290.

 

4

The following analysis shall be of such a nature
as to be serviceable in a system which interprets
'statement' in any of the ways described above. At least,
such is our hope. It should be clear, however, that such
a hope will be fulfilled only if the cogency of our
construal of the term 'statement' is not dependent on (but
is compatible with) the "existence” of propositions or
meanings. Hence, although we claim to use the term
'statement' neutrally, this claim should not be taken to
mean that we are using the term ambiguously.

In particular, we shall use the term 'statement'
so that it may be interpreted in the nominalistic sense
indicated by Nelson Goodman,9 i.e., any number of sentences
which may be construed as catalogued under a single label
will be called replicas of one another; and all the
replicas so included in a catalogue under a single label
may be construed as a spatially and temporally discrete
individual which we call 'a statement'. 'We are contending
that such an explication of the term'statement' will also
be adequate to a so-called platonistic system which

recognizes propositions and meanings as values of its

 

9loc. cit.

5

variables. It is therefore only in the sense just
described that our usage of 'statement' is to be under-
stood as neutral.

It will be necessary to recognize two different
usages of the term"hypothesis'. .According to one usage
we shall understand simply an individual utterance or a
number of utterances of a particular statement. When
we wish to symbolize a particular hypothesis in this
sense we shall use the lower-case letter 'h}. When we

wish to speak of an hypothesis in the sense of a statement

 

(i.e. as an individual composed of all_the replicas of a
given sentence) we shall use the upper-case letter {Hf .
When the distinction is not required for the purposes of
the analysis, or when the context makes it clear which of
the two usages is being employed we shall speak simply of
an hypothesis. When, in the discursive portions of the
text, the distinction is required, we shall speak, on the
one hand, of utterances of an hypothesis or utterances of
a statement, and, on the other hand, of statements of

an hypothesis.

We shall use the term 'utterance' or 'statement-

 

utterance' to refer to particular sign-events in which a
replica of a given sentence is produced. We shall use

the term 'sentence—token' or 'tokens of a_particular

6

statement' or 'inscription' to refer to the physical
concrete objects so produced.

The term 'prediction' calls for extended comment.

 

We shall use the term to refer to those linguistic entities
which conform to the requirements for constituting a
prediction laid down by Carl G. Hempel and Paul Oppenheim.10
In particular, only those entities will be considered as
predictions which fulfill the following requirements:

1. The explanandum must be a logical consequence
of the explanans.

2. The explanans must contain general laws, and
these must actually be required for the
derivation of the explanandum.

3. The explanans must have empirical content.

4. The sentences constituting the explanans must
be true.

The above requirements do not avoid all the

difficulties involved in determining what does and does
not constitute a scientific explanation or prediction.
But the Hempel-Oppenheim paradigm is, at present, the most

definitive statement available. Consequently, we shall

perhaps be justified in adhering to these requirements

 

10Carl G. Hempel and Paul Oppenheim, "Studies in
the Logic of Scientific Explanation", Philosophy p£_Science,
15:135-175, April, 1948.

7
until a more definitive statement is forthcoming.ll What
we are explicitly affirming is the logical symmetry of
explanation and prediction. The characteristic which
distinguishes prediction from explanation is a temporal
one. qugediction contains an explanandum which is
actually deduced and uttered prior to the event described
by the explanandum. An Explanation contains an explanandum
which is deduced and uttered after the event described by
the explanandum.

.A large class of utterances which shall concern us
do not conform to the requirements of the Hempel-Oppenheim
thesis and yet are of such a nature as to be of concern
to the social scientist. These utterances may be shrewd
guesses, educated guesses, prophecies, “predictions" (in
the vulgar sense) and like utterances. These utterances
are of concern to the social scientist since they too may
have a self—vitiating effect, presystematically understood.
These utterances we shall refer to as 'prophetic-hypotheses'
or as '"predictions"'. The use of double quotation marks
serves to indicate that the entities in question do not

conform to requirements 1-4. Those statements which can

 

11For a recent defense of the Hempel-Oppenheim
thesis, see Adolf Grunbaum, "Temporally-Asymmetric
Principles, Parity between Explanation and Prediction,

Mechanism Versus Teleology”, philosophy p£_Science 29:
146-174, April, 1962.

8
serve as explananda of scientific predictions we shall
call 'fore-dictions'.
Another class of utterances needs to be considered.
This kind of utterance, at first glance, appears to be one
which should be subsumed under those labeled 'fore-dictions'
but which on further analysis turn out pp£_to be deducible
from an explanans for one of the following (or other)
reasons: they contain mispelled words, or are grammatically
incorrect, or are amphibolous or contain simple ambiguities
or vague expressions. Hence we shall treat these malformed
hypotheses as.a sub-class of prophetic-hypotheses. In
Chapter III we shall consider ways of dealing with prophetic-
hypotheses which we would presystematically understand to
be self-vitiating. But the interesting cases which shall
be our especial concern are fore-dictions. We shall there-
fore restrict the predicate 'self—vitiating' to an hypothesis
which has the following characteristics:
a) Predictions of which i5_is an essential
component (i.e. the explanandum) are not borne
out when the hypothesis is uttered.
b) The hypothesis and other circumstances are
such that we are warranted in asserting the
counterfactual: "If the hypothesis had not

been uttered, then the prediction would have
been fulfilled".

9

The requirements 1-4 listed above together with
the informally indicated defining characteristics a) and
b) serve at the outset to mark off a reasonably precise
area for investigation. It is therefore hoped that the
ensuing discussion will be marked by a rigor and precision
which has been lacking in some of the discussions in the
literature.12 we may note how some of the vaguenesses
and ambiguities which have characterized such terms as
'suicidal belief' and 'self—destroying prophecy' are
removed by these requirements. Characteristics a) and b)
serve to discriminate between those hypotheses which we
want specifically to consider and those which are simply
false. While we may intuitively grasp the distinction
between the two types of hypotheses we shall have occasion
to note, as was remarked above, that the formal characteri-
zation of the term 'self—vitiating hypothesis' poses some
difficult problems.

Requirements 1-4 enable us to distinguish between
"shrewd guessing“ or "prophesying” on the one hand, and
predicting, on the other. It may be the case that guessing

and prophesying also may be self-vitiating in some

 

12See Chapter II for a criticism of certain attempts
to deal with the problem of self-vitiation.

10
presystematic sense of the term. If so, then the features
which characterize self-vitiating fore-dictions will also
be features of certain guesses and prophesies. But while
guessing and prophesying about the future may often turn
out to be either felicitous or to share other character-
istics of fore-dictions, they have no scientific status.
The peculiar puzzles created by certain prophetic-hypotheses
shall be examined and dealt with in Chapter III. To repeat,
the predicate 'self—vitiating' shall be reserved for fore-
dictions.

Merton's use of the term 'self—destroying belief'
is particularly ambiguous due to the difficulty of
determining the precise character of the phenomena in
question. One of his examples illustrates this particular
difficulty:

In the dark days of 1862, when McClellan was
stalemated and the armies in the west immobilized,
Lincoln did not issue a public call for the
desparately needed thousands of new troops, ex-
plaining, "I would publicly appeal to the country
for this new force were it not that I fear a
general panic and stampede would follow, so hard

is it to have a thing understood as it really
is".13

 

l3Merton, pp, cit., p. 128.

11

This particular passage appears in the section, of
Merton's book, dealing with self-destroying beliefs. What
is unclear from the illustration and Merton's treatment
is exactly what is being destroyed or would be in danger
of being destroyed had Lincoln issued his call for troops.
Would it have been the union? The morale of the northern
troops? The morale of the citizens in the northern states?
Or would it have been Lincoln? His prestige? This partic-
ular kind of difficulty is Obviated if we make clear from
the outset the kind of entities that shall be our concern.
The kind of entities which shall concern us are linguistic;
and they have the peculiar property or characteristic of
being causally related to the event which is described by
their negation.

We need to discriminate even further and distinguish
statements from utterances of statements or statement-
occurences. Likewise we need to distinguish those utter-
ances of statements which enter into a causal relationship
with the event described by the negation of that statement
and those utterances of the same statement which do not
enter the causal relationship. An example may show the
need for these distinctions. Suppose that a particular
statement §.describes §_and that as a result of certain

utterances of g, ~gcomes about. Now as we have defined

12

'statement' we may note that it is not gll_the utterances
of §_which are causally efficacious in bringing about "p_
but only some of them. If, for instance, some of the
utterances of §_occur after ~gwe shall probably not wish
to say that these later utterances entered the causal
sequence which brought about "§,l4 .A more detailed
elaboration of these points will be undertaken in Chapter
III. AAt this point we need only remark that the predicate
'self—vitiating' will be applied to utterances of state-
ments. Utterances, it should be remembered, are construed
as sign-events. And hence utterances may function as values
of variables in a causal explanation.15.

One further distinction needs to be noted. we
distinguish an utterance of g, here construed as a psycho-
social event, from the sentential components or parts of
the utterances of S, The latter are not events but
physical objects which are parts or constituents of events.

Follow1ng Nelson Goodman, we may draw the distinction by

 

14The only cases in which this particular assertion
will not hold will be those in which a11_the utterances of
§_predede ~gand each and every utterance of § enters the
causal relationship.

15Our construal of the term 'causal explanation'
is delineated in Chapter III.

13

noting that utterances contain times whereas their sentential
parts do not.16 This is not to suggest that sentence-tokens
are not fully concrete entitites. It is to suggest that for
the purposes of analysis we attend only to those qualia of
tokens that are other than time-qualia. We could have
treated sentence-tokens as containing time-qualia and hence
as events. And presumably an analysis operating on this
assumption would also prove adequate. It may therefore
seem that our method of distinguishing between utterances
and tokens is somewhat arbitrary. It is arbitrary only
in the sense that we need to make some decision concerning
what may serve as values of the variables in any adequate
theory of self-vitiation. And it is entirely possible that
an analysis which allowed tokens (here construed as events)
to serve as the value of the variables in a theory of self-
vitiation would also be adequate for explaining the phenomena
in question.

‘We shall attempt to avoid making essential appli-
cation of the predicate 'self—vitiating' to such entities
as meanings, propositions, states of affairs, and statements

(when these are construed as abstract entities), although

 

1692. cit., pp. 283-287.

14
it appears likely that so nominalistic an orientation may
become exceedingly difficult to maintain in attempting to
resolve some of the puzzles posed by the phenomena we are
examining. To some, our reasons for imposing such rigid
restriction on our analysis may be a source of wonder.
Why, it may be asked, when meanings play so obvious a role
in the phenomena under consideration would one be constrained
to avoid reference to meanings when such reference could
presumably aid in the analysis? In reply we may say that
abnegations in philosophy, no less than in religion and
morals, have their own peculiar rewards. The basic notion
here is that of simplicity. We wish to keep our universe
as uncluttered as possible. But at the same time we wish
to be able to say all that is worth saying concerning the
phenomena in question. What we shall therefore attempt
is to show how we can deal with the designative role of
utterances in relationship to the events they purport to
describe without introducing 'meaning' as a primitive or
undefined term. We do not ignore the question of meaning,
but for the sake of simplicity we shall attempt to avoid
giving ”meanings" -- and other similar entities --
ontological status. In other words, meanings will not serve

as values of the variables of lowest type in any system

15

based on the following analysis. If we can carry this
plan through to completion, our analysis will be simpler
than one which allows the introduction of a greater number
of entities as values of its primitives.
Some Illustrations of Self-Vitiating Hypotheses

The phenomena in question are fairly common occur-
rences in daily life. And the following illustrations are
not meant to exhaust the various types of hypotheses in
question but only to suggest the kind of entities with
which we are concerned in this essay.

A. An economist predicts, on the basis of reliable
information and highly confirmed economic laws, that, due
to a shortage of a particular commodity 9, companies g,

X; and §_which are presently engaged in producing g_will
realize a profit g_at time 3,17 Subsequent events, however,
show that the explanandum of the prediction turned out to
be false. Those companies which had expected, on the basis
of the prediction, to make a profit, instead sustain a loss.
What has happened is the following: when the fore-diction

was itself made a piece of public information, other

companies, hoping to profit by the production of §_began

 

17Throughout the remainder of this discussion, the
requirements that what are acceptable as predictions are
those which conform to points 1-4 on page 7 of this essay
will be tacitly understood and will not be reiterated. .An
important qualification is made in Chapter V, p. 154.

l6
flooding the market with this item. The result is that
§11_the companies engaged in producing g_are faced with
large inventories and loss of profits. Wherein lies the
difficulty with the fore-diction? If we are warranted in
asserting the counterfactual, “If the hypothesis (i.e. the
explanandum) had not been uttered, then companies g, X,
and g_would have made a profit at time 3," we may predicate
'self—vitiating' of this particular hypothesis.

B. The National Safety Council publishes a statement
to the effect that the Council predicts at least four
hundred Americans will die as the result of automObile
accidents during a particular holiday, H91, This statement
is published in newspapers, and carried on spot television
and radio announcements. The explanandum of the prediction
is not borne out by a subsequent examination of the facts.
We may “explain" this sequence of events by noting that
the prediction may have had the observed effect through
its own allusion to the dangers involved in travelling
during £91,; i.e., many Americans, learning of the fore-
diction, had remained at home, and others had exercised
more caution than usual during their travels. From the
evidence it seems plausible to assume that predictions by
the National Safety Council often do have just such an

effect. That their fore—dictions do not have this

l7
self-vitiating character more often may be due to the
fact that those who make the predictions hope that their
campaign, coupled with the utterances of their hypotheses,
will actually have such a self-vitiating effect. It may
also be the case that those responsible for making the
predictions actually do take account of the possible
self-vitiating character of such fore-dictions and revise
their predictions accordingly.

C. At times this self-vitiating effect may be
used with deliberate intention to deceive. A commander
of a regiment in time of war predicts, on the basis of
information gathered from prisoners, espionage, etc. that
the enemy is about to make a move M, Discovering that he
is ill-prepared to defend against M, he allows his fore-
diction to ”leak" back to the enemy camp in the hope that
his opponents will abandon MLas a plan of operation.

D. An economist predicts that a certain event §_
will take place relative to conditions of the stock market.
He publishes his prediction in one of the leading economic
journals; and the statements describing the future occurence

of §_are thence published in the Wall Street Journal where

 

they are subsequently read by most of the major investors
in the stock market. As a result of the prediction, the

activity of the investors is altered. §_does not occur.

18

Common sense would again lead us to assert the counter-
factual, “If the fore—diction had not become known to the
buying public, §_would have taken place."18

E. In 1948, most of the major public opinion
polls predicted that Thomas E. Dewey would become the next
President of the united States. That this fore-diction
was not borne out is a fact of history. Some of those
who had supported Mr. Dewey blamed the pollsters for1
engaging in "pseudo-science“ and basing their predictions
on insufficient evidence and poorly confirmed laws. A
few voices, however, suggested that one possible cause of
the outcome of the election had been the effect of the
utterances of the fore-diction itself on Republican voters
who simply stayed away from the polls, convinced as they
were of a Dewey victory. The minority voices were saying
in effect that the fore-diction was self-vitiating; i.e.
they were asserting the counterfactual, "If the fore-
diction had not become public information, Mr. Dewey would
have won the election.“ If we raise the question as to
why the predictions of presidential elections in 1952,

1956, and 1960 did notshare this self-vitiating characteristic,

 

18This example is a paraphrase of one suggested

by Popper in Poverty p£_Historicism.

19
common sense has a ready answer: "The general public has
become aware of the self-vitiating character of such fore-
dictionsand has acted accordingly."19 We shall later have
occasion to note that the foregoing answer of common-

. 20
sense leads to some puzzling consequences.

 

19An interesting feature of the 1960 presidential

election, however, concerned just this self-vitiating feature.
N.B.C. and C.B.S., which televized the election returns,
made use of gigantic computers which determined the odds of
a Kennedy or a Nixon victory. A few minutes after the first
returns began to be fed to the computers, the odds in favor
of a Kennedy landslide were seven hundred to one. These
odds remained for almost an hour after the polls had closed
in the East. In the West, however, the polls remained open
for three hours after they had closed in the eastern quarter
of the nation. When the polls had finally closed in the
West and the votes had been tabulated, Kennedy's odds had
been so reduced that the difference in popular votes cast
for him and Mr. Nixon was but a few thousand. That the
computer's "predictions" had such a self-vitiating effect
seems plausible, even in the absence of empirical evidence.

For one thing it tends to vitiate the self-
vitiating character of certain fore-dictions. For another,
it seems to be based on some general law of human behaviour
which may be roughly phrased: "Whenever the general public
becomes aware of the self-vitiating character of a pre-
diction, it will act in such a way as to vitiate the self-
vitiating character of the prediction; or it will act in
such a way as to bear out the initial prediction." Such
general law-statements may raise puzzles in their own
right. On the basis of such a law, we can make certain
predictions. But then we will wish to know whether
utterances of this law and hypotheses deducible from it are
self-vitiating. See Chapter VI for an extended discussion
of this problem.

20

F. Karl R. Popper has argued that it is precisely
this self-vitiating character of our predictions about
the future course of human history which makes such
predictions “logically impossible”. At this point, we
may simply take note of his argument which in substance
is as follows: Knowledge is for doing. And insofar as
what we do is dependent upon our knowledge about the world,

just so far is it necessary for us to predict what knowledge

 

g§_shall possess ip_the future if we are to successfully
predict our future behavior. But it is not logically
possible to predict our future knowledge; since to predict
that at some future time, 2, we would have a specific
piece of knowledge 5, would be to vitiate the prediction.
This would be due to the fact that we would need to possess
K prior to E_in order to make the prediction. Hence all
predictions about action which is dependent upon knowledge
of what our future knowledge will be is in effect self—
vitiating.

While Popper's arguments have an initial plausibility,
we hope to show that they will not bear examination; and
that much of the plausibility is due to an ambiguity in

the term 'knowledge'.

 

21Karl R. Popper, Logic p§_Scientific Discovery,
(London:Hutchinson, 1959).

21

G. Another, and perhaps related example is
offered by Richfield and Copi.22 "Suppose I agree to make
a decision between alternative courses of action éLand §_
next Wednesday, and that I have promised to delay my
action until that day." Suppose further that on the basis
of knowledge about myself, my previous decisions, certain
tendencies to act in a certain way under certain speciable
circumstances, I predict that I will decide in favor of
A_on next Wednesday. One might argue that to make such a
prediction is in effect to make the decision now and hence
to vitiate the fore-diction.

Professor Henle, for example, has argued24 that
“any prediction I make about myself is vitiated by the fact
that I make it...While I may predict the weather or coming
political events or your conduct, I do not predict at what
time I shall have lunch or go to bed. These are questions

which I simply decide.P Here the notion of predicting

 

22Jerome Richfield and Irving M. Copi, "Deciding and
Predicting" Philosophy p£_Science, 27:47-51, January, 1960.

23Although the example is due to Richfield and Copi,
they are concerned to deny that the prediction is self-
vitiating.
24 .
Paul Henle, "Do We Discover Our Use of Words?“
Journal p£_§hilosophy, 54:750, November 7, 1957.

22

events in daily life requires considerable sharpening
before we can examine the logical issues involved.25

H. Another example:26 Nation M_is at war with
nation M, M_has all of the military advantages at its
disposal; it possesses money, manpower, and arms together
with the means for producing more of everything than does
M, M_is lacking in financial resources; its manpower is
depleted; it lacks food and arms in sufficient quantities
to wage a successful war against M, The odds for a
victory by M_are so overwhelming that social scientists
begin making predictions to the effect that M_will win
the war. The causal consequences of the utterances of
the social scientists are the ensuing lethargy and lack
of interest on the part of the populace of M_in regard to
the war effort and renewed determination on the part of
the populace of M to win the war. The final result: M

wins the war. In such circumstances we believe ourselves

to be warranted in asserting the counterfactual: "If the

 

25Thus the notion of prediction here seems not to
conform to the requirements indicated above although it
may do so in specific instances.

26This example is a paraphrase of one given by
Alan Gewirth, pp, cit.

23
fore-diction, 'M_will win the war.‘ had not been uttered,
M_would have won the war,"
John Venn was one who held that this class of
phenomena enables us to distinguish between the social

and the physical sciences. He writes:

Any person can see that to draw inferences
about a thing, and then to introduce a disturbance
which was not contemplated when the inference
was drawn is to invalidate the conclusion we have
obtained. But when the inference is about the
conduct of human beings it is often forgotten that
in the inference itself, if published, we may
have produced an unsuspected source of disturbance.
In other words, if the results of our investi-
gations be given in the form of statements as to
what people are doing and what they will do, the
moment these statements come before their notice
the agents will be subject to a new motive, which
will produce a disturbance in the conduct which
had been inferred...

The publication of the Nautical Almanac is
not supposed to have the slightest effect upon
the path of the planets, whereas the publication
of any prediction about the conduct of human
beings (unless it were kept out of their sight, or
expressed in unintelligible language) almost
certainly would have some effect. The existence
of thnsdistinction renders all such physical
illustrations entirely inapplicable when we thus
attempt to explahithe way in which it is supposed
that human conduct can be studied and foretold.27

 

27John Venn, Principles pf Empirical p£_Inductive

 

Logic (London, New York: MacMillan and Company., 1889),
pp. 575-576.

24

Venn's general position seems to be one which is
popularly subscribed to, although it is one which, as we
hope to show, will not prove to be completely satis-
factory; nor are all of his arguments rendered plausible
when subjected to careful analysis.28
The Relevance of theygpgue

The issues raised by the foregoing discussion may
be discussed under two broad headings which, loosely
speaking, may be labeled "practical" and ”theoretical".
We shall simply take note of the former and throughout
the course of our discussion it will be tacitly assumed
that the latter is the point of our inquiry. This is not
to suggest that the practical aspect of the issue is of
less importance in the total scheme of human affairs
than the theoretical. For it is entirely possible that
a self-vitiating hypothesis could have disastrous historical
consequences. The point is rather that we would prefer
to leave empirical questions concerning any particular
self-vitiating hypothesis or set of hypotheses as issues
‘with which the sociologist, psychologist and historian

are better equipped to deal. We may note that the kind

 

28In Chapter II we shall hve occasion to examine

Venn's position in greater detail.

25

of examples offered by Merton29 and Gewirth3O are sufficient
to dispel any belief that self-vitiating hypotheses are
of little moment in regard to the concerns of everyday
life. The practical politician and statesman appear to
be well aware that predictions or fore-dictions on which
public domestic policy and international diplomacy are
based would have disastrous effects if they became known
prior to the time of the occurence of the events predicted.
And much of the secrecy surrounding the activities of the
State Department in Washington and its various counter-
parts throughout the world can be understood or "explained"
not simply as attempts to guard military secrets and
scientific information; but also as attempts to prevent
other nations from discovering what we "predict” they will
do in the future. Having noted the practical relevance
of the phenomena of self-vitiation, we may leave the task
of explaining instances of this type of phenomena to the
social scientist.

Our principal concern will be with the relevance
of this class of phenomena for epistemological or "theo-
retical" considerations. We need first of all to note that

29 See p. 11 of this essay.

30See pp. 22—23 of this essay.

26
a part of the program of empiricism from John Stuart Mill,
through the Vienna Circle, and up to the present time has
been the attempt to show that our knowledge about the
world is not simply comprised of isolated facts and theories,
but that there is, in Quine's phrase, such a thing as "the
whole of science". We need not engage in the debate as to
whether this scientific knowledge constitutes the "whole
of knowledge" but rather whether this scientific knowledge
consists (even ideally) of a unified and integrated whole.
If it should turn out that self-vitiating hypotheses
constitute a class of phenomena with which an empiricist
understanding of science is incapable of dealing, and
moreover that such hypotheses are indigenous to the social
sciences, then those who hold that there are methodological
differences between the social sciences and the physical
sciences will have additional ammunition for their guns.
If, however, we can show, as we hope to, that self-
vitiating hypotheses offer no insuperable obstacle to an
empiricist program, then we shall have made some small
contribution to the task of showing that the claim for the
unity of knowledge about the world is not discredited on
these grounds. Specifically we hope to show that the claim
that self-vitiating hypotheses are social phenomena which

enable us to mark a distinction between the social and the

27
physical sciences is untenable. In a more positive vein
we shall attempt to clarify some of the issues which will

need to be considered in the development of a theory of

self-vitiation.

CHAPTER II
AN EXAMINATION OF SOME OF THE ARGUMENTS IN THE

LITERATURE

introduction

Although the phenomena of self-vitiation have
seldom received detailed analysis from an epistemological
or logical point of view, the argument that this kind of
phenomena is indigenous to the social sciences has not
been without its supporters. In this chapter we shall
examine some of the arguments which have been advanced
in favor of this thesis and attempt to evaluate the
cogency of these arguments.
The Argpments of John Venn

The issue as to whether or not self-vitiating
hypotheses give us adequate grounds for marking a dis-
tinction between the social and the physical sciences is
at least as old as the late nineteenth century. John
Stuart Mill believed and argued that, while he had little
hope for the scientist's ability to predict with any
degree of accuracy individual behavior, he did believe
that with the advancement of the social sciences it would
be possible to predict human behavior on a large scale

or in the mass, i.e., general trends in social movements.

28

29
Mill had no doubts or reservations regarding the possi-
bility of making sociology as exact a science as astronomy.l
JOhn Venn took issue with Mill precisely on the

point of self-vitiating hypotheses:2

If the results of our investigations be
given in the form of statements as to what people
are doing and what they will do, the moment these
statements come before their notice the agents
will be subject to a new motive, which will
produce a disturbance in the conduct which had
been inferred. we may make what statements and
criticism we please about the past conduct of
men, but directly we commit ourselves to any
statements about the future, or, in other words,
begin to make predictions, we lay ourselves open
towthatdifficulty just mentioned. The predictions
can be made seems to be held by many of those
who have adopted the application.of logic now
under consideration. They do not, of course,
claim to be able to foretell the particular
actions of individuals, but they assert that it
is quite possible that we may some day be able
to foretell general tendencies, and the results
of the conduct of large masses of men.

Between Mill and Venn we see the battle line
clearly drawn between those who claim that prediction of

human behavior is often self-vitiating and therefore

impossible and those who claim that prediction is

 

1JOhn Stuart Mill, A,System p§_Logic (New York:
Harper and Bros., Publishers, 1874), pp. 550 ff.

2John Venn, pp, cit., pp. 575-576.

3O
always possible in spite of ostensibly self-vitiating
characteristics. Lest the discussion be obscured by
unnecessary ommissions we hasten to add that although
Mill and Venn do not explicitly say so they are both

discussing successful prediction. ~Neither of them would

 

wish to deny that from a particular set of law-like
statements and well evidenced statements describing
antecedent conditions it would be possible to deduce
certain conclusions according to valid rules of infer-
ence. The point at issue is whether utterances of such

deductions are of such a nature that gp_p;inciple such

 

predictions cannot be successful.

In regard to Venn's arguments several points need
to be noted. First, it is not clear in what sense an
inference's conclusion may be invalidated by subsequent
events. The purport of his objection seems clear enough.
But the ambiguity in the term 'inference' makes suspect
his precise point. Suppose we have a set of law-like
statements p_and a set of statements g_describing certain
antecedent conditions and that the conjunction of p_and
g logically imply M, ‘We might say then that from p_and
§_taken in conjunction we can infer p, In what sense

can this inference be invalidated? Surely not in the

logical sense of the term 'to infer'. For no matter

31

what we add to the premises of the argument, §_will
still be deducible from the conjunction of p_and g, It
will be recalled that Venn said "Any person can see that
to draw inferences about a thing, and then to introduce
a disturbance which was not contemplated when the infer-
ence was drawn is to invalidate the conclusion we have
obtained.“3 In what sense is Venn holding that the
conclusion is invalidated? Surely it is not gpp_deductive
inference. Venn as an able logician would hardly have
been guilty of such an elementary blunder.

Another initially plausible construal of his
claim might be that it is directed to inductive inference
in which, by adding an additional premise to the set, we
invalidate the conclusion. Thus, for example, a set of
premises 'Fa', 'Fb', ... 'Fn' is confirmatory for the
hypothesis '(x)Fx' while one statement such as '“Fy'

will tend to disconfirm the hypothesis, ceteris pgribus.

 

But consider the following inductive arguments:

 

3Loc, cit.

32
A 2

Fa . Fb ... Fn Fa . Fb ... Fn . ~Fy
(x) Fx (x) Fx

 

Now if we are interpreting Venn correctly he seems
to be suggesting that argument MLoccurs at a time p_and
that an event fp occurs at a time later than 3, a time

~

which we may call 30. We assume that the event p_is

|~

described by the statement Fy'. .Argument §_presumab1y
occurs at a time later than 30 say 31' Argument §_we
assume to be invalid by any standards of inductive
validity. But in what sense does this render the con-
clusion of argument M;invalid? Yet if we understand
Venn's argument he seems to be suggesting that since the
premise of g, i.e. '"Fy', is presumably true (since it

~

describes p), therefore the conclusion of argument A;
which occured prior to ~pis invalidated. But this
certainly appears to strain the sense of 'validity' to a
point where the term will fail to apply to anything.
Perhaps what Venn had in mind when using the term
'invalidate' was simply that the conclusion of argument A
is false. In this case his argument loses much of its
force. For the scientific status of the conclusion to

argument Miwill depend on a number of factors; e.g.

whether or not we are willing to accept as true the

33

statement '"Fy' and whether or not we are willing to
give up the generalization '(x)Fx' after considering the
cost involved if we were to abandon it.

A second difficulty arises in regard to Venn's
use of the term 'statement'. He says, "The moment these

statements (i.e. self-vitiating hypotheses) come before
their notice, the agents will be subject to a new notive.“

The question as to what constitutes a statement and
whether or not it is the statement or the occurrence of
a statement which is involved is a matter of some impor-

tance. A statement-occurrence can be considered as a
psycho-social event having certain causal consequences.

If, however, what is intended is the assertion that

"meanings“ are causally efficacious, then certain ontological

questions would need to be raised and answered before the

problem could even be clearly articulated. What we shall

suggest is the the important issues in regard to the

phenomena in question can be answered without raising
ontological questions of this kind. For the present we
may simply observe that Venn is not clear, in his account
of the matter, what he intends us to understand by the

term 'statement'. And until this ambiguity is resolved

it is difficult to assess the weight of his argument.

34
gerturbance Effects in the Physical and Social Sciences

Up to this point we have been overlooking the
major point which Venn wished to make: i.e. that whereas
such publications as the Nautical Almanac are not supposed
to affect the orbits of the planets, publications about
the behavior of human beings certainly do have some effect.
Phrased somewhat differently what is being claimed is
that fore—dictions in the social sciences have self-
vitiating instances whereas fore-dictions in the physical
sciences do not. we must now attempt to determine whether
or not this claim can be cogently upheld.

Using Venn's example we may ask whether the pub-
lication of the Nautical Almanac could have self-vitiating
instances. More specifically we may inquire whether any
predictions or fore-dictions contained within the Almanac
can have replicas such that they are constituents of a
sign-event which is causally efficacious in bringing
about an event whose occurence disconfirms the prediction
or fore-diction.

Prior to the space age we more than likely would
have been content to take Venn's arguments at face value.
Today, however, with the unleashing of the power of the
atom and the possibility of space flight, there does not

seem to be evidence to support the claim that the orbit

35
of Mars could not be altered. Consider, for example, a
particular statement §_which describes the elliptical
path of Mars around the Sun. Suppose that replicas of
§ are produced by exploding cobalt bombs propelled by
rockets into the vicintiy of the planet Mars. These
bombs explode in such a fashion as to spell out replicas
of §_and have the causal consequence of changing the
orbit of Mars. It follows that the fore-diction concerning
the orbit of Mars is self-vitiating. Now a number of
objections may occur to the reader regarding the uncon-
ventional tokens employed and the failure to take
cognizance of the difference between the use of language
for communication and its use for moving the orbits of
planets. Many of these objections we hope to consider
in later chapters. At the present time we may simply
note that the example conforms well with our presystematic
discussion of what shall be called a self-vitiating
hypothesis. That is, §_is deducible from an explanans
which conforms to requirements 1-4 listed on page 7.
Also, predictions of which it is an essential component,
i.e. the explanandum are not borne out when uttered
(although, to be sure, this is a somewhat unusual mode
of utterance). And finally the hypothesis and other

circumstances are such that we are warranted in asserting

36
the counterfactual: "If §_had not been thus uttered
then the prediction would have been fulfilled." Hence
§_fu1fills all the requirements for the label 'self-
vitiating'.

Now the point of our example needs to be kept
clearly in mind. We are not claiming that there is no
distinction between certain kinds of self-vitiating
hypotheses. At this stage we are only concerned to show
that Venn's example, and the claim based upon it, lack
the cogency usually attributed to it.

Ernest Nagel gives an example of self-vitiatinn
which is perhaps intuitively closer to the kind of
phenomena to which we would ordinarily be willing to
apply the predicate. His example also, however, is taken
from the physical sciences:

The pointing and firing of an anti-aircraft
gun can be effected by means of a purely physical
mechanism. Such a mechanism includes, we may
assume, a radar for locating the target, an
automatic computer for calculating the direction
in which the gun should be pointed to hit the
target as reported by the radar, an adjusting
device for pointing and firing the gun, and
some system for transmitting the calculations of
the computer as a series of signals to the
adjusting apparatus. Let us now suppose that,
were the gun fired in accordance with the cal-
culations of the computer on a given occasion,
the target would be hit: but let us also suppose

that the signals transmitting these calculations
have disturbing effects (whether on the adjusting

37

apparatus or on the target) for which the computer
has made no allowance. Accordingly, although the
gun is set and fired in accordance with calcu-
lations that were correct at the time they were
made, it nevertheless fails to hit the target
because of the changes introduced by the process
of transmitting those calculations.

This example also conforms to our presystematic
requirements concerning what shall be included in the
extension of the term 'self-vitiating'. In this example
it is the utterance of the calculations which are causally
efficacious in bringing about the result. And if there
are "essential" differences between this example (and
others of its kind) and those indicated in Chapter I,
these differences have not yet been clearly delineated.

Returning to the arguments of John Venn we see
that either his arguments are too weak or too strong.
They are too weak if he assumes that self-vitiating is
a phenomenon which concerns the social scientist only.
But if he is claiming that the phenomenon of self-
vitiation makes successful prediction of the future
impossible, then his arguments are too strong. For, as
we have seen, examples of self-vitiation may occur in a

branch of science as sacrosanct as that of celestial

mechanics.

 

4pp, cit., p. 469

38

"Predictions of Future Knowledge and Knowledge of the
Future" - The Arguments of Popper

The concept of scientific methodology which is
adopted for the purposes of this essay is essentially
that which is advocated by Karl Raymond P0pper. The
requirements for what constitutes a scientific explanation
or prediction are in most respects due to the early work
of this man. What we have called "the Hempel-Oppenheim
thesis" is sometimes referred to as“the Popper—Hempelian
thesis“.5 It therefore is somewhat anomolous to have to
deal with his arguments for the thesis that "no society
can predict, scientifically, its own future states of
knowledge".6 This claim seems to conflict with his
frequent assertions to the effect that the methodology
in the social sciences is identical to that of the physical

sciences:

 

5S. A. Grunbaum, pp, cit.

6Popper's arguments are scattered throughout
three major publications: The Poverty p§_Historicism,
(Boston: The Beacon Press, c.d., 1957); The Open
Society and its Enemies (London: Routledge &.Kegan Paul,
Ltd., 1945); The Logic pf_Scientific Discovery (London:
.Hutchinson, 1959). The above quotation is from gpp'
Poverty p§_Historicism, p. xi.

 

39

From the point of view of scientific method,
a social hypothesis suggested by self-intuition is
no different position from a physical hypothesis
about atoms. The latter may also be suggested
to the physicist by a kind of intuition about
what atoms are like. And in both cases, this
intuition is a private affair of the man who
proposes the hypothesis. What is 'public', and
important for science, is merely the question
whether the hypotheses could be tested by ex-
perience, and whether they stood up to tests.

From this point of view, social theories are no
more 'subjective' than physical ones.7

Popper does not believe, however, that this point
of View is cogent for p;£_social science. That is to
say, history, and predictions about the future course
of human history are excluded. Since Popper gives his
own summary of his argument, we may perhaps be pardoned

for allowing him to speak for himself:

;_have shown that, for strictly logical

reasons, ip_i§ impossible for p§_pp_predict the
future course pf_history...

I propose to give here, in a few words, an
outline of the refutation of historicism. The
argument may be summed up in five statements, as
follows:

(1) The course of human history is strongly
influenced by the growth of human knowledge.
(The truth of this premise must be admitted
even by those who see in our ideas, including
our scientific ideas, merely the by-products
of material development of some kind or other.)

7The Open Society, p. 308, n 14.

4O

(2) we cannot predict, by rational or scientific
methods, the future growth of our scientific
knowledge. (This assertion can be logically
proved, by considerations which are sketched
below.)

(3) This means that we must reject the possibility
of a theoretical history; that is to say, of
a historical social science that would cor-
respond to theoretical physics. There can
be no scientific theory of historical develop-
ment serving as a basis for historical pre-
diction.

(4) The fundamental aim of historicist methods...
is therefore misconceived; and historicism
collapses.

The argument does not, of course, refute the
possibility of every kind of social prediction; on
the contrary, it is perfectly compatible with the
possibility of testing social theories - for example,
economic theories — by way of predicting that
certain developments will take place under certain
conditions...

The decisive step in this argument is statement
(2). I think that it is convincing in itself: if_
there i§_such p_thing p§_gpowing human knowledge,
then yp_cannot anticipate to-day what we shall know
only tomorrow.8
Before examining this argument in detail we may

show that it is directly related to the problem we are

considering in this essay. One of the crucial claims

has to do with the assertion “We cannot anticipate today

8
The Poverpy f Historicism, pp. ix - x.

41

what we shall know only tomorrow." What Popper
apparently has in mind is something like the following:
suppose that we predict that in the future we shall know
a general law Q, But if we can formulate p, this means
that our knowledge of p_is not future buty present.
Therefore we cannot predict that our knowledge of p_will
be limited to the future. In our terms such a fore-
diction would be self-vitiating. Such fore-dictions may
be treated as a small sub-class of self-vitiating hypotheses.

Popper, however, also deals with the more general
phenomena of self-vitiation in his examination of what
he calls the “Oedipus effect":

The idea that a prediction may have influence
upon the predicted event is a very old one.
Oedipus, in the legend, killed his father whom
he had never seen before; and this was the direct
result of the prophecy which had caused his
father to abandon him. This is why I suggest
the name 'Oedipus effect' for the influence of
the prediction upon the predicted event...whether

this influence tends to bring about the predicted
event, or whether it tends to prevent it.

 

Historicists have recently pointed out that
this kind of influence may be relevanttn the
social sciences; that it may increase the diff-
iculty of making exact predictions and endanger
their objectivity. They say that absurd conse-
quences would follow from the assumption that
the social sciences could ever be so far developed
as to permit precise scientific forecasts of
every kind of social fact and event, and that
this assumption can therefore be refuted on
purely logical gnbunds. For, if such a novel

42
kind of scientific social calendar were con-
structed and became known...it would certainly
cause actions which would upset its predictions.
Just how much of the preceding paragraph Popper
himself will subscribe to is by no means clear. Later
statements suggest that he subscribes to the thesis that

"an exact scientific calendar of social events, compar-

able to, say, the Nautical Almanac, has been shown...

 

to be logically impossible".10

A Critique of Popper's Views

There are a number of difficulties involved in
Popper's arguments which need to be unravelled prior to
evaluating the cogency of these arguments.

First of all, the indicator term 'we', which
appears in his statement (2) above, is not particularly
relevant to the argument.11 The point at issue is
whether or not a rational beipg could predict the future
course of human knowledge and hence the future course

of human history dependent upon that knowledge. When

 

91bid. p. 13
lOIbid. p. 37
11

We are not criticizing Popper's attack on
historicism with which we are in full agreement; but
only his claim that it is logically impossible to
predict the future course of human history.

43
the question is thus rephrased much of the puzzle which
Popper sees in the issue simply disappears. In fact it
seems that his argument is in some respects based on a
tautology: "Future knowledge is not present knowledge,
and conversely.“ In this age of space exploration, our
imagination concerning “possibility" (however construed)

need not be limited to human knowledge pf human knowledge.

 

There is surely nothing logically impossible in the
assumption that somewhere in the vast reaches of the
Milky Way there exists some rational being or race of
rational beings who, with a knowledge of a thousand
civilizations on as many planets, could formulate a
theory which, given the antecedent conditions of human
civilization up to the present, could with some high
degree of accuracy predict the future course of human
knowledge. Popper's logical impossibility thus turns
out to be a phantom. In fact all his argument seems to
say which i_s tenable is "For any human being A, if A
does not know p, then A does not know Q," And this as
we noted above is simply a tautology. Hence part of the
puzzle disappears if we remove the indicator term 'we'
from Popper's arguments.

Then the question as to whether or not we know

now what we do not know now is seen to demand a negative

44
reply. But this reformulation, rather than enhancing
his argument also simply renders it innocuous. And we
cannot infer from this argument that therefore the future
course of human knowledge is unpredictable due to a logical
difficulty.

Third, whether or not any event can be success-
fully predicted depends in large measure on the body of
theory from which the statements in the explanans are
derived. The same argument holds in regard to predictions
of the future course of human knowledge. Whether or not
human knowledge is predictable will depend in large
measure on whether or not there are sufficient law-
statements which are highly confirmed and which, taken
together with certain statements describing relevant ante-
cedent conditions will logically entail that in the
future a particular piece of knowledge will be had by
human beings. On the basis of our present understanding
of human and natural history it appears plausible to
assert that the first atomic pile was produced in the
twentieth century. Now according to Popper's thesis it
would have been impossible for any scientist to predict,
say, prior to 1850, that by mid-twentieth century we
would a) know certain properties of the uranium atom, or

b) know how to produce atomic fission, or c) that President

45

Truman would decide to use this power to destroy two
Japanese cities. Thus not only is prediction of human
knowledge impossible, according to Popper; the same will
hold for future history. The question, however, is, given
evidence of a number of other civilizations of rational
beings plus a creative imagination, would it be logically
possible for a rational being to construct a theory which
could predict the future course of human history? There
seems to be no legitimate reason for a negative reply to
this question. If the results of the prediction were
made public would the hypotheses on which it was based
prove to be self-vitiating? The answer to this question
will in large part depend on the kind and specificity
of the event being predicted.12

Suppose that a rational being with a body of
sociological and psychological theory and evidence con-
cerning human civilization were to predict that human
civilization would possess a body of laws El ... L not

—n

prior to the year 2000 A.D. Suppose further that this

 

12We shall have occasion to note that Popper's
thesis fails to make clear the highly ambiguous character
of the term 'knowledge'. And until this ambiguity is
removed it is difficult to assess clearly his claim re-
garding the inability of any society to predict,
scientifically its own future states of knowledge.

46
fore-diction becomes known to the scientific community in

1963. Now we may ask, "Under these circumstances, would

the fore4diction be self-vitiating?" The answer to this
question will depend on the kind of designators that are

used in the prediction to refer to El ... Ln. Do '21. ...‘pn'
in the prediction simply name the laws (e.g. "Boyle's law“,
"Newton's laws of motion")? Or do they characterize the

laws (e.g. "laws regarding the rate of disintegration of the

uranium atom", “laws of the refraction of light waves")?

Or do they explicitly state the laws (e.g. "E = MCZ",

"Every body at rest or moving in uniform rectilinear motion
continues at rest or in uniform rectilinear motion until
acted upon by some external force.")? If the prediction
simply names the laws or characterizes them,the self-

vitiating effect probably would not appear. Even if the

laws are explicitly stated it is not clear that even in
this case the fore-diction would be self-vitiating.

Whether or not it would be self-vitiating would depend on

the kind and degree of understanding which existed on the
part of the scientific community in regard to what the laws
indicated.

This latter point may be made clear by a somewhat

simple-minded example. we may predict in regard to some

Child A, that on Afs twenty—first birthday, and not before,

47
he will know the law “Ontogeny recapitulates philogeny".

If the child become acquainted with tokens of the fore-
diction we may ask whether the fore-diction was in his
case self-vitiating. The answer will, it is clear,

depend on what is meant by the term 'know'. At this point
we can distinguish several ways in which the term 'know'
can be understood in this context.

1. The term means merely that the individual is
acquainted with (i.e. has had a sensory experience of) a
certain set of tokens.13 If this is the meaning of the
term in the illustration, then clearly the fore—diction
is self-vitiating as soon as the child hears the tokens
produced by the predictor.

2. The term means that upon hearing or seeing
the tokens, one will be able to reproduce replicas of the
tokens or reasonable facsimiles thereof. If this is the
sense of the term being employed, then whether the fore—
diction is self—vitiating will depend on whether the child
is able, because of having heard the fore-diction, to

repeat the words ”Ontogeny recapitualtes philogeny' prior

 

13Later we shall need to distinguish between tokens
and the names of tokens. This distinction shall prove
important for an understanding of some of the issues
involved in the problem before us. At this stage of the
analysis we will assume that the individual is acquainted
with the tokens rather than the names of the tokens.

48
to his twenty-first birthday, and in such a way that the

tokens he produces are recognizable replicas of the
original. If, after hearing the fore—diction, 5.15 able
to reproduce the tokens in such a manner we may say that
the fore-diction was self-vitiating.

3. The individual, on or after his twenty-first
birthday, is acquainted with the semantical rules of the
language in which the fore-diction is made; and consequently
he would be able to use the individual terms of the fore-
diction in other statements according to the semantical
rules of the language. In our example the fore-diction
would not be self-vitiating unless M_were extremely
precocious.

4. The individual, on or after his twenty-first
birthday, is provided with a body of evidence sufficient
to confirm the law statement in question. If this is what
the fore-diction intends by the term 'know' then most
likely MLdoes not Mppy_the law ”Ontogeny recapitualtes
philogeny" when he first hears the fore-diction. Hence
the hypothesis is not self—vitiating.

5. The individual, on or after his twenty-first
birthday, is acquainted with the scientific procedures and
the kinds of evidence required to confirm or disconfirm

the law statement in question. Again, the fore-diction

49
would not be self—vitiating if the term 'know' is

construed in this way.

In regard to 4. it may be remarked that we have
various ways of confirming particular beliefs. Or put in
a slightly different way, there are a number of grounds
on which we warrant our acceptance of certain statements.
In common parlance we are often said to kngw_certain
statements on the basis of some authority. Most of our
knowledge about the world is of this character. I may be
said to know that Castro is the present leader of the
island of Cuba, that the Chinese mainland has a communist
government, that John Glenn was the first American to
circle the globe in a space capsule. In this sense of
'know' I am not directly acquainted with (i.e. have not
had any sensory experience of) any body of evidence which
would directly confirm or disconfirm the statements in
question. Nor am I clearly conscious of the procedures
which would be required to confirm or disconfirm the
statements. I know in the sense that an authority has
informed me. This is not to suggest that the authority
is beyond challenge. He may be mistaken, deceitful, deluded.
And hence his statements may be false. Yet the point is
that in this sense of 'know' the authority remains unchallenged;
and I may or may not be acquainted with evidence or procedures

Which would tend to confirm or disconfirm his assertions.

50

The interesting case occurs when the predictor
may be such an authority. Whether or not the fore-

diction in such a case would be self-vitiating (in the
sense of 4.) will be determined by the attitudes (i.e.
the rationality) of A_and also his conviction of the
reliability, in the sense just indicated, of the person
making the prediction or uttering the fore-diction. If
the present attitude is one of rational trust in an
authority, then the fore-diction in this sense will be
self-vitiating.

These several sensesof the term 'to know' make
Popper's claim quite uncompelling in the absence of any
precise specification by him of a sense in which pre-
~dictions about the future course of human knowledge are
impossible. As we have seen, whether or not such a
prediction contains a self-vitiating explanandum will
depend on which sense of 'know' is intended. And it
has been shown that there are a number of senseSin which
the term may be employed in predicting future knowledge
which do not have the self-vitiating effect.

These five meanings of the term, however, are
by no means exhaustive of the possible ways in which it

is used. The following schema is meant to be suggestive

of some of the other senses in which the term 'know‘ can be

understood. This schema is based on the distinction

51
between knowing that and knowing how pp,l4 This schema
will also bring to light the extremely ambiguous character

of Popper's claim that we cannot predict future knowledge.

6. M predicts that at time p, M_will know the
statement _$_.

7. §_predicts that at time p, M will know that
the statement §_is true.

8. M_predicts that at time 3, M will know how
to do g,

A number of other examples can be treated as
special cases of 6 - 8. A special case of 6. is:

9. M_predicts that at time p, M_will know the
statement "—1 is true“.

Another special case of 6. which shares certain

features of 8. is:

10. M_predicts that at time p, M will know the

statement "At time El: leill know how to
do Z".

A special case of 7. which shares features of 8.

is:

11. X predicts that at time t, M_will know that
the statement “At time t1, XLW will know how
to do Z" is true.

 

14
The distinction is made by Gilbert Ryle, The

Concept p£_Mind, (New York: Barnes and Noble, Inc., c.d.,
1949) pp. 25-61, we hold no brief for the distinction
which may prove to be untenable. Our concern is merely to
point out some of the ambiguity in the expression 'to know'
and to point out also how this factor is of considerable
importance for the claim that knowledge of our future
knowledge is impossible.

52

The preceding schema enables us to produce examples
of ever-increasing complexity. By re-writing variables
and combining elements of 6. and 11. we can get:

12. g_predicts that at time p, M;will know the
statements "M_predicts that at time El: M
will know that the statement 'At time 52,
M_will know how to do g} is true“.

In this example, the verb 'to know' is used in a
variety of ways. The first occurrence is subject to the
demand for clarification in terms of l. - 5. above. That
is to say, it may mean merely that ELWill be acquainted
with a set of tokens which are replicas of those which
follow the term 'statement' (first occurrence) in 12. Or
it may mean that M;is acquainted with a body of evidence
which will tend to confirm or disconfirm the statement
(i.e. the expression beginning with '"M,predicts...“')
embedded in the illustration. Or it may be taken in one
of the other ways in which the senses of the term 'to know'
is delineated in 1. - 5.

The second occurrence of the term 'know' in the
above example seems to accord more closely with the sense
characterized by 4. and 5. The third occurrence of the
term is used in the sense of 8. But even this occurrence
is not without ambiguity. Consider, for example, the
locution, “M_knows how to play the piano." In the most
common sense of this term it means that M_has the ability
to play the piano. But a number of qualifications are

needed to make clear what is intended. M_may have broken

53
some of his fingers or one of his arms. Or his hands may
be crippled with arthritis. What the locution then intends
is the counterfactual, "If M_were not crippled by arthritis
(or suffering from broken fingers or arms) then he would
be able to play the piano".

The need for other qualifications is equally apparent.
A composer, for example, himself a violinist, may know how
to play the piano in the sense that he is acquainted with
the rules of musical composition and can discriminate be-
tween playing a piece of piano music properly and improper-
ly according to the rules of accepted piano technique.

He may also have acquaintance with the various names of the
notes of the piano-forte and be acquainted with the rules
which prescribe which fingers to use in striking certain
combinations of keys, yet he himself be totally unable to
play the piano,

On the basis of these considerations it is readily
seeithat 12. is filled with ambiguity. And whether any
such fore-diction is or is not self-vitiating will depend
first of all on specifying the exact sense in which the
hypothesis was intended. until this is done, the question
of whether or not such an hypothesis is self-vitiating
cannot be answered.

Similar arguments apply to the other forms 6 - 11.

A teacher, on the bais of past experience and certain
general laws of learning, may successfully predict when a

given student will have mastered a particular piece of music.

54
Likewise, on the basis of past experience, a teacher may
be able to predict that a given student will know at time
p_that the statement §_is true, whereas formerly the student
was acquainted with tokens of §_ but was not aware that g
is true and may even have believed that §_is false.

What is evident from the fore-going analysis of
Popper's claim is that certain fore-dictions of human
knowledge but not every such fore-diction, will be self-
vitiating. But whether any particular fore—diction is in
fact self-vitiating will depend on a number of factors
including the question of what was intended by the fore-
diction.

Even more important than the above considerations
is the question whether there are logicallyéunavoidable
cases of self-vitiation. It can be shown that the kinds
of cases we have been considering offer a number of possi—
bilities for avoiding the self-vitiating effect. Venn
suggests a variety of ways when he says, "The publication
of any prediction about the conduct of human beings (unless
it were kept out of their sight, or expressed in unintell-
igible language) almost certainly would have some effect".15
A fore-diction §_could be uttered in a foreign language
which was not understood by the person(s) about whom the

prediction was being made. This move, in effect, prevents

 

1
5%. cit., p. 576

55
such a person or persons from knowing the prediction except
perhaps in the sense of l. and possibly in the sense of
2. In the sense of 3 - 5 however, since the prediction
is made in an unfamiliar language and presumably the persons
about whom the prediction is made are unfamiliar with the
semantical rules of the language, the self-vitiating
feature is effectively removed. But it is possible, surely,
that if the fore-diction were uttered in English and con-
cerned the linguistic behaviour of a person who spoke and
understood only the French language, then the fore—diction
would not be self-vitiating even in the sense of l. and 2.
This claim can be made clear by way of an example. llmay
predict that a French-speaking child §_will be acquainted
for the first time with the French translation of the law—
statment "Ontogeny recapitualtes philogeny" on his twenty-
first birthday. Even though §_is presumably acquainted
with the tokens contained in my fore-diction the hypothesis
is not vitiated. For what the hypothesis refers to are
certain tokens in the French language rather than the
English.

If 'know' is taken in the sense of l. or 2. there
are other similar ways in which we can predict events in
the manner of 6. and at the same time avoid the self-
vitiating character of the fore-diction. We may, for
example, instead of predicting that at time p, M will

know the statement "Ontogeny recapitualtes philogeny",

56
use the elliptical form:

At time p, M_will know that the statement in-
dicated by three words: the first of which is
composed of the fifteenth, fourteenth, twentieth,
fifteenth, seventh, fifth, fourteenth and twenty-
fifth letters of the English alphabet; the
second of which is composed of the eighteenth,
fifth, third, first, sixteenth, ninth, twentieth,
twentyfirst, twelfth, first, twentieth, fifth,
and nineteenth letters of the English alphabet;
and the third of which is composed of the six-
teenth, eighth, ninth, twelfth, fifteenth,
seventh, fifth, fourteenth, and twenty-fifth
letters of the English alphabet.

Here again the particular difficulty of self-
vitiation is obviateiin one sense; although in another
sense it may not be. Whether or not the individual, M,
will be motivated to look up the words in question in a
standard dictionary will be a matter of some concern pre—
sumably to the psychologist. we have shown, however, that
no inherent logical difficulties remain in this formulation
of the fore-diction. As long as the prediction concerns
Mfs acquaintance with tokens, we can always find suitable
means for avoiding the use of those tokens; i.e., we can
name the constituents of the tokens or use definite

descriptions or other descriptive or naming devices to

designate the tokens referred to in the fore-diction.l6

 

l6Whether such devices are acceptable options for

preventing the self-vitiating effects in a society which
values the free interchange and exchange of information is
a highly debatable question. And in our opinion it is

not debated with sufficient frequency. Merton writes, "To
be sure, in misanthropic mood, or in excessive devotion

to the values of social science above all other human
values, or in the self-defined role of a scientific samurai,

57
These remarks are not meant to indicate that certain
fore-dictions may not have causal consequences such that
the events they predict are not borne out. Indeed many
(if not all) self-vitiating hypotheses turn out to have
precisely this characteristic. All we have been concerned
to show up to this point is that Popper's logical thesis

is unsubstantiated.

The Arguments of Feuer and Gewirth

Others have made similar claims and have gone on to
argue for the difficulties or impossibilities of making
predictions about the future in the social sciences. Lewis
S. Feuer has argued for this position in the field of
economics.

What the directions of economic development
are, whether toward stagnation, collapse, or

prosperity, depends primarily upon the emergence
of new industries. After the laws of economics

 

the social scientist might write out, seal and safely
deposit his prediction of impending unemployment or war
or internecine conflict, bringing it to light only after
the predicted events had come to pass. But this would be
almost as reckless of the body politic as of his own
corporeal self. When one considers the profound objection
of many individuals to being regarded as psychological
guinea pigs, one can roughly imagine the aggregated fury
of an entire population upon discovering itself trans-
‘mogrified into one immense sociological guinea pig." Qp_
g;£,, p. 130. The conditions under which an entire
population would react in the manner described by Merton
are not altogether obvious. But one thing appears quite
clear: the kinds of alternative solutions to the problem
of self-vitiation we have listed raise questions of value
of considerable importance to a "free" society.

58

have been formulated, laws concerning prices,
wages, the rate of interest, it is still not
possible to foretell what the human intelligence
will discover during this stage of economic
development. There was no forecast from any
economic school that the fission of the uranium
atom would be discovered, and that a new mode

of industrial energy would become available.17

While Feuer's comments are meant to apply principally
to capitalism as an economic system they are relevant for
problems of wider scope. They seem to represent a wide-
spread conviction that human creativity is unpredictable

in principle.

Creative response, as a social phenomenon then,
introduces variables which do not fall within a
determinate set of economic equations. A surd
factor enters into the course of technical evo-
lution. The directions of economic development
therefore cannot be predicted from the principles
of economic theory.18

He concludes with the following paragraph:

Most important, there is no way of fore-
seeing whether a given era will be marked by
the dearth or the emergence of new great in-
dustries. It remains indeterminate therefore
whether investment waves will be forthcoming
which would enable the capitalist system at
any given period to surmount the pressures
of underconsumptionist disequilibrium.19

 

17Lewis S. Feuer, "Indeterminacy and Economic
Development", ghilosgphy p; Science, 15: 234, July, 1948.

18Ibid., p. 235.

19Ibid., p. 241.

59

These arguments somewhat parallel Popper's claim
concerning the logical impossibility of predicting future
knowledge. Feuer, however, simply claims that we cannot
predict what creative responses will be forthcoming in the
future. But this claim seems to be based on a non sequitur.
The fact that no economist did in fact predict the fission
of the uranium atom is hardly an argument that such pre-

dictions are impossible ip_principle. More generally, the

 

fact that we, at present, have no body of theory which
would enable us to predict so-called creative responses is
hardly a sufficient argument to substantiate the claim

20 One of the

that no such body of theory is possible.
crucial questions which any such claim must answer is "What
is meant by the term 'possible'?". For if it is logical
possibility which is intended then, as we have seen, the
claim is untenable. If it is empirical possibility (or

impossibility) which is intended, then the claim that a

particular activity or piece of knowledge is beyond the

 

Often in the literature the position here
described offers us more rhetoric than argument. Charles
Hartshorne, for example, writes,

"The idea that to know is to be able to pre-
dict, so that any inability to predict must
mean partial ignorance, is, I hold, untenable.
The ultimate function of knowledge is not to
foresee, but to create. The two are by no means
coincident. We predict most perfectly astron-
omical phenomena where we have no creative con-
trol. We control most perfectly our bodily be-
havior, but how vague and uncertain are our pre-
dictions of this behavior!...

60
"possibility" of science or human reason is perhaps only
to baptize our ignorance, or to give evidence of an infertile
imagination, or to exhort us to stop trying to reach "beyond
our grasp". Baptism of ignorance is hardly a noble philo-
sophical pursuit; and claims that our reason is limited in
such and such a way are usually short-lived or quickly dis-
proven by new discoveries. We may readily admit that the
human mind is so constituted that it cannot know everything.
But it does not appear to be particularly warranted to
claim that one has discovered what the precise limitations

of the human mind happen in fact to be.21

 

"Is the goal of psychology to be able to write
the poet's poem beforehand by predicting it, to
compose the composer's music by telling him what
it is going to be? Should the psychologist be
able to foresee, and thus make the creative decisions,
and their expression in speeches and laws, by which
statesmen, or an entire people resolve some political
tangle? How extravagant can the theory of science
become? I suggest that, since creative activity is
that which leaps unpredictably from its causal base,
the proper function of psychology in regard to
creation is to increase our freedom from the com—
pulsions and fixations which often make behavior
only too largely predictable.”

"Freedom Requires Indeterminism and universal
Causality." Journal of Philosophy, 55:800-801, September 11,
1958. -_-

 

21
The kind of arguments we have been considering

seem to be motivated by an inordinate fear of scientists'
being able to predict the creative response of human
individuals. It is as though if we were able to predict
such responses we would automatically dry up the well-
springs of creative endeavor. If we were to respond to
rhetoric with rhetoric we might reply in dis way:

"How many Beethovens and Mozarts have gone to
their graves because no one taught them to sing

61
In a later article, Feuer makes it clear that the
kind of possibility he has in mind is not that of logical

possibility:

A directional law of evolution is not logically
impossible. The law of entropy in physical science
asserts that for closed systems there is a definite
direction that energy transformations will take.

A directional law of social change cannot there-
fore be excluded on purely logical grounds. The
law of entropy, however, holds only for closed
systems; if there were an influx of energy from
other parts of an unlimited universe, a given
region might not actually evolve toward thermo-
dynamical equilibrium. And laws of history are
beyond our grasp precisely because human initiative,
human intervention, upsets any approximation toward
a closed social universe. The factor of human
creativity corresponds to the influx of energy
into the physical system. It makes moreover for

the unpredictable aspect of historical causation.22

unless we are misconstruing Feuer's comments it
appears that the phrase ”closed social system" is function-
ing as a metaphor in the argument. The same appears to be

true in regard to the use of the phrase "law of entropy"

 

the melodies ringing in their souls? And how

many Newtons and Einsteins have perished without
giving to the world the answers to some of the
burning problems of mankind? And how many Children
have died of cancer because a Harvey or a Lister

or a Pasteur died without performing the experiment
that was in him?

"And how many of these perished, these unknown
untutored geniuses of art and science, because no
one could look into their souls and predict that
here is a genius who will contribute immeasurably
to the welfare and well-being of mankind in the
following specific ways.“

2Lewis S. Feuer, "Causality in the Social Sciences",
Journal pf_Philosophy, 51: 694, November 11, 1954.

62
when it is applied to a closed smial system. If we are
not misconstruing his arguments and these terms indeed
function metaphorically then I must admit that the force
of his objections escapes me. If, however, Feuer means us
to take these comments as empirical statements about matters
of fact, then as far as I am aware, his arguments find
little if any support from the physical p£_the social
sciences. In fact, prima facie it appears that his claim
that "human intervention upsets an approximation toward a
closed social universe" may indeed be false.23 At any
rate, Feuer has, it seems, failed to justify his thesis
that human creativity makes ”for the unpredictable aspect
of historical causation."

Alan Gewirth also has argued that certain fore-
dictions of social events can be vitiated by the creative
response of the individuals about whom the prediction is
made. He treats self-vitiating hypotheses as a sub-class
of phenomena which he calls the “reflexive action of men
on social laws". He writes:

Now the general position I wish to present is
the following. In dealing with social phenomena,
social science deals largely if not entirely with
things which impinge directly on men's values -
wealth, power, various kinds of interpersonal

relations, and so on. The aim of social science
may be said to be te- attain knowledge of the laws

 

For an interesting and illuminating discussion
of the concept of entropy in the physical sciences,see
A. Grunbaum, pp, cit.

63

of these matters - that is, of their cause-effect
relations. Since, however, man as conscious
voluntary agent is in large part both the knower
and the subject—matter of these laws, his knowledge
of their impact on his values may lead him to

react on the laws reflexively in order to change
them. Consequently, the laws of the social sciences
cannot have the same fixity or permanence as the
laws of the natural sciences.

This reflexive reaction of men on social laws
has interesting logical as well as social conse—
quences. One of its familiar aspects is found in
such reflexive situations as the "self-destroying
prophecy."24

In a footnote he attempts to give two formal
characterizations of this type of prediction:
1. If there is a prediction §_that a certain kind
of action ALwill result in a certain consequence
p, then this prediction has the result that the
contrary...of action A_occurs and that it
results in the occurrence of the contrary of p,
2. If there is a prediction §_that a certain kind
of action A .will result in a certain consequence
p, then this prediction has the result that
action A; and also other action A2 occur, (sic)
and that these actions have the result that the
contrary of consequence g_occurs.
In ChapterVI we shall have occasion to examine his
arguement in more detail. At this point we can remark two
difficulties which appear, at first glance,to render his

formalization inadequate. First it is by no means clear

what is meant by the phrase 'the contrary of an action'.

 

24Alan Gewirth, "Can Men Change Laws of Social

Science?", Philosoppypp£_Science, 21: 230, July, 1954.

25Ibid.

64
Perhaps this is simply an elliptical way of saying "the
contrary of the statement describing action pf. If our
rephrasing meets Gewirth's intent, then another puzzle
appears: what can be meant by the term 'ppp_contrary' in
this case? For example, what would constitute the contrary
of a sentential function 'Fx'? Presumably p_contrary
would be one which is a universal generalization of this
function. In other words the function universally general-
ized would be '(x)Fx'. And a contrary of this statement
would be '(x)"Fx'. But Gewirth's own symbolization prevents
this interpretation, since he uses an existential quantifier
in such a way as to preclude it.26 The opacity of his
symbolism, which accords with no logistic system with which
this author is acquainted, makes his intent impossible to
fathom.

Second, he in no way specifies what are to be the
values of the variablejAj. Unless the range of values of
this variable is clearly specified we are in no position
to discriminate between those predictions whose explanans
contains faulty laws or unreliable evidence and whose

explanandum, although in no way causally related to the

 

2
6Gewirth's symbolism of l. and 2. above is as
follows:

1. (ax) (ma—ac] )——9{(ay) (ZAy)-> (32) (2a)}
2. (EXHPX EAec] )—>{_C(ay)(Aly) . (az)(Azz):J-7(m<zcw)}

The double tilde is meant to represent, not the negation,
but the contrary of the expression(?) which follows.

65

subsequent turn of events, turns out to be false; and those
self-vitiating predictions whose explanans contains hypotheses
which we would accept as true, at least prior to the demon-
stration of the self-vitiating character of the explanandum.27

The body of Gewirth's article, however, attempts to
show that self-vitiating fore-dictions illustrate man's
ability to change social laws; and if Gewirth's arguments
are persuasive, we shall have to abondon the empiricist
claim that no demonstrable differences have been shown to
obtain between the physical and the social sciences. This
possibility is hardly at issue, however, inasmuch as he

seems merely to be proposing a new usage of the term 'social

law'.

Let us consider three different relations
which man may bear to scientific laws: namely,
that he can exchange the laws, apply the laws
and create the laws. Man exchanges laws when
he replaces the conditions under which one
set of laws - i.e., one correlation of vari-
ables - Obtains, for another set of conditions
under which a different correlation obtains.
Man applies laws when he uses a correlation

 

27Somewhat later we shall find it necessary to
modify the fourth requirement of the Hempel-Oppenheim
thesis for our purposes. Their requirement is that the
statements in the explanans be true. Now this require-
ment may indeed be retained as a criterion for determin-
ing what constitutes a scientific explanation. For our
purposes, we need to weaken the criterion to admit of
those explanantia which contain statements which we would
accept as true prior to the determination of their self-
vitiating character. It should be noted that we are not
advocating a weakening of the criterion for a scientific
explanation, but only of the criterion for what shall
constitute a self-vitiating fore—diction. The reason
for weakening this requirement is simply this: if we re-
tained the stronger criterion, the definition which we
shall propose would turn out to have a null extension.

66

which exists independently of his decisions

to produce an effect which would not have occurred

if he had not made use of this correlation.

Man creates laws, finally, when by means of

his free decisions and consequent action, he

causes a correlation to exist which did not

exist before.28

Gewirth's error arises out of his failure to realize
the tenselessness and non—existential import of a lawelike
"correlation“. It is therefore likely that the distinction
which he wishes to draw cannot be maintained according to
his criteria. Two examples will illustrate this contention:

Assuming that the first self-propelled heavier-than-

air craft was flown in the first decade of the twentieth

century we may claim that in this instance man exchanged

 

laws. In this instance man replaced the conditions under
which one set of laws obtained, i.e.,the laws of gravity,
for another set of conditions under which, in addition, a
different correlation obtains, namely, the conditions
described by the laws of aerodynamics. But this example
also conforms to Gewirth's description of what it means
to apply a law; i.e., in this instance men used a corre-
lation which exists independently of any decision on their
part to fly an airplane. What Gewirth wishes to maintain
is that in social science phenomena what often happens

is that man "causes a correlation to exist which did not
exist before" whereas in physical phenomena (such as that

in the example of the airplane) the correlation of

 

289p, cit. p. 234.

67

variables is not created but existed prior to its discovery.
Causing an instance of such a correlation is what Gewirth
means by creating laws. Yet we may inquire as to whether
or not the flight of the first airplane cannot also be
subsumed under this classification. But perhaps the matter
can be put more illuminatingly.

On the basis of our assumption, no heavier-than-
air craft had flown prior to the twentieth century. The
problem then may be stated in the form of a counterfactual:
"If someone had built a craft of the type known as a
Piper Cub, prior to the twentieth century, and had been
able to supply it with fuel, it would have flown”. Now
apparently Gewirth wishes to claim that we are warranted
in making some such counterfactual claim in regard to
natural laws but that we are not warranted in making them
in regard to social laws. But if this is the case, then
what we need are other criteria for distinguishing between
natural and social laws. We cannot make the distinction
on the basis that in a social law man creates a correlation
which did not previously exist unless we have independent
evidence for determining whether the correlation was
confirmed -- e.g., evidence constituted of something other
than the described human intervention. Otherwise we are
in peril of doing one or the other of two things: we
determine that the correlation did not exist previously
because the law in question is a social law, and hence

involve ourselves in a circular argument; or we treat the

68

assertion "If p_is a social law, then the correlation of
variables which obtains in p_was not instantiated prior

to human intervention.“ as analytic. In the latter case
we have done little more than formulate a rule governing
our use of terms: but we have hardly shown how to dis-
criminate between social laws and natural laws or between
correlations which are confirmable prior to specified kinds
of human intervention and those which do not. If Gewirth's
contention is to be maintained, what is required is a

means for determining when a correlation of variables
"exists prior to human intervention" and when such a
correlation is created gp_novo. It is by no means clear
how such a question could be answered and Gewirth's
analysis offers us no guide. And inasmuch as the question
appears unanswerable, Gewirth cannot be said to have made
good his claim that in self-vitiating hypotheses we have

a class of phenomena which enable us to distinguish
between social and natural laws.

Gewirth makes a related claim concerning the logical
relationships which obtain between the terms of social and
natural laws:

It is recognized that the relations of im-
plication which hold between antecedent and con-
sequent in natural laws are different from those
which hold in the case of purely logical relations
like those of mathematics. My point then is that
the relations of implication in social laws
are different from both of these. For both
within the antecedent and between the antecedent
and the consequent in social laws, human decisions

do or may enter. Therefore, even the kind or
degree of eternal truth found in natural laws

69

which may be said to hold only for our preignt

cosmic epoch, is not found in soc1al laws.

Whether or not we accept his enthymatic inference
as valid willlargely depend upon our willingness to accept
his suppressed premise, yig. “If human decisions do or
may enter within the antecedent or between antecedent and
consequent of any law 9, then the kind or degree of
eternal truth found in natural laws will not be found in
pf. But this premise itself is plagued by the same
obscurity which confounds other parts of his argument.
What kind of entities are human decisions that they can or

may enter the antecedent p£_between the antecedent and

 

consequent pf ipyp? Gewirth may have some kind of

spatial or temporal entity in mind. For example, if some—
one propels a billiard ball toward another, human decision
may prevent the first from striking the second by causing
the placing of a hand or other object between them
immediately after the initial impact. In such a case what
we encounter is an example of a whole set of problems
raised by our failure to make explicit the auxiliary
assumptions regarding boundary conditions which are
tacitly assumed in making a specific prediction. In the
second half of Chapter IV we shall have occasion to examine
this problem in greater detail. At this point it may
suffice to note that our example of human intervention
involved physical or natural laws rather than social laws.

But if we were to accept Gewirth's classification, such

 

291bid., p. 237.

70
phenomena as that of moving billiard balls would need to
be considered as social rather than as physical phenomena
inasmuch as human decisions seem to be involved in up-
setting predictions of the balls' behavior.,

In this chapter we have examined what appear to be
some of the representative arguments in the literature
concerning the phenomena of self-vitiation. We have
attempted to show that these arguments fail for a number
of reasons not the least of which is the ambiguity and
obscurity of the terminology employed in the arguments.

we shall now turn our attention to a more rigorous
examination of the problem and attempt to explicate the
notion of self-vitiation in a formal and more precise

manner 0

CHAPTER III

A FORMAL ANALYSIS OF THE PREDICATE 'SELF-VITIATING'

Introduction

The problem of self-vitiation involves two rather
distinct features of hypotheses. It involves first of
all a feature of those hypotheses which, due to certain
characteristics, have already been demonstrated to be
false. In this instance what we require is a manifest
predicate which can be applied to the hypotheses in
question.1 On the other hand, there are those hypotheses
which, on the grounds of their similarity to other

manifestly self-vitiating hypotheses, we have reason to

 

believe would be self-vitiating if they were uttered under
certain conditions. What we require in this instance is

a disposition predicate, i.e., a term which refers to a
latent property or characteristic of hypotheses. The

term which we shall employ in referring to the manifest
characteristic of hypotheses is 'self—vitiating'. The
term we shall employ in referring to the latent character—

istic of hypotheses is 'self-vitiatable'. A formal defin-

 

ition of the manifest predicate 'self—vitiating' presents
no particularly difficult obstacles. Defining its
associated disposition predicate 'self—vitiatable', how-
ever, presents problems which are not easily resolvable.

In addition to the general problems of defining

 

1we use the terms 'manifest' and 'disposition'
according to the explication of Nelson Goodman. See
Epct, Fiction and Forecast.

71

72
disposition terms, we face the difficulty that 'self-
vitiatable' unlike such terms as 'magnetic' or 'soluble'
is not found in any well-confirmed body of theory. Thus
in attempting a formal definition of the dispositional
predicate 'self—vitiatable' we face a double task: we must
show not only how the term may be clearly defined in a way
which will specify the range of application of the term;
but also how the term may be incorporated into sociological,
psychological and physical theory. The problems of de-
fining 'self—vitiatable' will be treated in Chapter V. In
the present chapter we shall confine ourselves to defining
the manifest predicate 'self—vitiating' and problems related

to this task.

First Steps

In this chapter we wish to mark off for investi-
gation one particular class of phenomena which, until now,
has been considered together with another important class.
The distinguishing characteristic of this class of phenomena
is that of a causal relationship which obtains between an
utterance and the event described by the utterance. This
formulation of the distinguishing characteristic is itself
faulty as will become clear in'a moment. What we wish to
note at this point is that we are not considering that
particular class of hypotheses which are better considered
as a sub-class of the traditional paradoxes (syntactical
and semantical). It would appear that the failure to

mark this distinction between these two broad classes of

73

hypotheses (i.e. between those which are vitiated by
entering into a causal relationship with the event described
by the hypothesis, and those which are vitiated due to
their self-referential or to their syntactical character)
in the past has been a factor in preventing any clear and
concise notion of what could be meant by 'self-vitiating',
'self-destroying', etc. By making the distinction as
clear as we can, we may hope to mark off those areas of
investigation which belong primarily to the social and
physical scientists, on the one hand, and those areas
which belong primarily to the logician, on the other.

The following symbols are introduced as formal
aids in the analysis of the predicate 'self—vitiating'.

Let

'h' stand for a particular utterance
or an individual composed
of a number of utterances
containing 2 members
(where 'p} is any positive
integer) each of which
contains as a sentential
part a) a replica of the
sentential parts of each
of the others, or b) that
which is logically equiv-
alent to the sentential
parts of each of the others,
or c) that which may be
construed as an adequate
translation of the senten-
tial parts of each of the
others. And further that
each such utterance con-
stitutes a fore-diction
that p,

'H' stand for the class of all the
utterances which conform

74

to the conditions
specified under the des-
cription of p,2

'1' stand for an utterance which is the
explanans of an explanation
from which p_is deducible.
The requirements for p_
apply mutatis mutandis to
the components of l,

'L' stand for the class of all utterances
which conform to the re—
quirements for membership

in l,

'xl'...'xn' stand for particular events in a
causal sequence.

'Cxixj' stand for 'xi causes xj'.

'"pf stand for the event such that the
substituend of '“pf is
a replica of the negation
of the sentential part(s)
of the value of p,

'IYh' stand for 'y logically implies pf.

'Fh' stand for 'M_is false'.

'Vh' stand for 'p_is self-vitiating'.

'Self—vitiating' then may be contextually defined
as indicated by the following schema:
Vh = df. (3x1) (3x2) . . .(sxn) (an) (31.)(heH . (y) (yeL 3 th) .
Cth o CXlXZ e o o o e CXn e e Fh)
Rendered in English this may be read: "To call a

‘particular utterance or set of utterances p self-vitiating

is to say that there exists a set of utterances M such that

 

In a sense p_and M are pseudo-classes.

75
every member of §_1ogically implies p; that p_is part of
a causal sequence which eventuates in “p; and that M_is
false.“

A number of terms and conventions are adopted here
which require some comment. The values of M_are utterances
which have sentential parts that are tokens. Utterances
are, of course, events, where 'event' is here construed in
its broadest sense; i.e., an event may be composed of a
number of other events. Thus, for example, several members
of M may be understood to compose a single spatially or
temporally extended event. The single event so construed
is termed an 'individual' in the sense of the calculus of
individuals, i.e. as a sum of individuals.3

The term 'replica' is understood here as follows:
"§_is a replica of y_if and only if x_and y_share certain
specifiable physical characteristics“. For example, 'DOG'
is a replica of 'dog' as well as of 'D O G' insofar as
these word-tokens are each composed of letter-tokens of the

fourth, fifteenth, and seventh letters of the English

 

3Nelson Goodman, The Structure p§_Appearance, and
Henry Leonard, Singular Terms, typescript, Widener Library,
1930. A later version by these two authors is found in
"The Calculus of Individuals and its Uses", Journal pg
Symbolic Logic, 5:45-55, June, 1940.

76

alphabet in that order.4 We have expressly avoided the
introduction of the notion of “sign-types“ in order to
simplify the problem before us.5

We have added the phrase "or c) that which may be
construed as an adequate translation of the sentential
parts of each of the othersi? in order to make clear what
may not be clear by dealing solely with the notion of

logical equivalence.6

It is by no means always clear that
an expression §l is logically equivalent to §2 if the latter
is an adequate translation of the former. Of course
'adequate translation' may be so defined that p2 is an
adequate translation of §l only if §2 is logically equiv-
alent to pi.
would be far too restrictive for our purposes. Suppose

But, it would appear that this definition

that one or more individuals make up p, each of which
contains tokens we would say are misspelled, syntactically
incorrect according to the rules of sentence formation, or
otherwise defective as an English sentence. The question
as to whether this sentence (i.e. a sentential part of p)

was logically equivalent to some other English sentence

 

4We can extend the notion of "replicahood" to other

senses by introducing the notion of an ideal observer: a set
of phonemes P is a replica of an inscription W if an ideal
observer 0 were to utter P upon being asked to read W aloud.
Similar remarks obtain for braille inscriptions and sign-
language.

5See Chapter II for the reasons for this decision.

6We shall later find it convenient to drop clause c).

Reasons for this decision will be given presently.

77
which lacked the grammatical errors of the first, is
beside the point. The question at issue would be whether
or not the poorly constructed sentence entered a causal
relationship with certain events in such a manner that the
claim made by the utterance of which this set of tokens
is the sentential part turns out to be false. Or put in
aeiightly different way, the question would be whether the
receiver (whether the intended second party or not)7 of
these tokens could construe them in such a manner that he
would understand them in a "similar enough" manner to the
correctly formulated English sentence intended by the pro-
ducer. At issue is whether or not such a poorly constructed

sentence would contribute pp_the self—vitiatipg effect.

 

If such an utterance would, under suitable circumstances
produce the effect of self-vitiation, then it would appear
that we need to include it in p_even though the task of
construing it as logically equivalent to the other
individuals in p_seems to strain the notion of "logical
equivalence". But more of this point below.

'M} is to stand for the class of utterances which
conform to the conditions specified under the character-
ization of p, we have used the class calculus purely for

reasons of convenience. And we have assumed that it will

 

See Henry Leonard, Principles p§_Right Reason;
(Henry Holt and Co., New York), 1957, pp. 89 ff. for a
discussion of sign-events. We shall use Professor Leonard's
terminology except in those cases where we have reason to
believe that a more nominalistic terminology will be help-
ful in our analysis. These latter circumstances Will be
indicated.

 

 

78
be possible to translate our definition (and any proofs
dependent upon it) into the calculus of individuals. It
may also be remarked that the way in which classes are
introduced into the definition, no assumptions are made
regarding what causal effects classes can have upon events.
In other words, the ontology is left "open“ in such a way
that the definition would be serviceable in a system which
admitted classes as values of its variables as well as in
a system which admits only individuals as values of its
variables.

We shall often wish to speak of the sentential

 

parts of cerain utterances and of individuals composed of
several utterances. It will be therefore at times conven-
ient to speak of the sentential part(s) of p_to write
'p.(sp)'. This will make it clear that we are speaking of
the tokens as physical concrete objects and not of the

total sign-event (itself a complex bit of socio-psychological
behavior) of which they are but parts.

The remarks concerning p_and M apply equally to.L
and Q, But in this case certain difficulties regarding
logical implication become evident. Consider a case in
which an individual M.on the basis of a certain if "deduces"
M? which constitutes a fore-diction of p, We assume that,
lf(sp) and pf(sp) have, however, one or more defects:
either 1f contains a syntactically incorrect formulation
of an explanans which is to constitute part of an expla-

nation; or pf(sp) is not logically deducible from ;f(sp);

79

or Mf(sp) is not syntactically correct. But the subse-
quent course of events warrants the counterfactual: 'If
pf had not been uttered, p_rather than ~_e__would have
occurred.‘ Now we are faced with a choice. Either we
can drop this kind of linguistic event from the extension
of the term 'self—vitiating' and attend solely to those
linguistic events which do not share these difficulties.
Or we can lift some of the formal restrictions on the
values of the variables. If we choose the first alter-
native, we are faced with the problem that we have so
restricted the application of the term that its service-
ability to the social scientist may be likewise restricted.
If we accept the second alternative we seem to allow the
notion of logical implication to be used in violation of
certain logical laws, or in a sense which is at considerable
variance from any well-established logical theory. If
pf(sp) and ;f(sp) are not well—formed expressions, then
according to logical theory it is false to say that 1?
implies 11* . 8

These two alternatives are not exhaustive, however.
The problem is to determine what may be construed as

proper parts of p.and ;_and whether or not mal-formed

 

8We shall use certain logical predicates in such

a way as to allow them to apply indifferently either to
utterances or to the sentential parts of those utterances.
Hence we may write either 'M_is false' or 'pjsp) is false'.
In like manner we can write 'l_implies M} or 'ifisp) im-
plies pfisp)‘. We may not, however, use this procedure
with the predicate 'causes'. The reasons will presently
be made clear.

80
expressions may be so construed. Another alternative
would be to attempt to determine what an utterer, p,
intended when he produced pf. we assume then that if the
utterer p_intended p_when he said M} theipf is to be in-
cluded in M, Likewise with 1?. We could then re-write
our definition in such a way as to preserve a rigorous
sense to the term 'logical implication'. Logical
implication would hold only between ;_and p_and not be-
tween ;f and pf. The question as to what an individual
intends by his utterances is one which we assume can be
explicated by psychological theory rather than by logical
analysis of the tokens.

Then, under this interpretation, for any two
sentences g, and §_, §2 is to be considered an adeguate
translation of §l if an only if a person p,who utters _S__l
intends to signify that which is signified by p2,?) If,
that it is theoretically possible to discover what;p
intends by an utterance §l’ then if §1 is syntactically
incorrect or otherwise defective, it will be possible to
faithfully translate §. by §_ such that §_

1 2 2
and signifies that which p intended by pl.

is well-formed

The kind of utterance we are here considering, how-
ever, may still prove troublesome even if psychological

theory is capable of explicating the sense of the expression

 

9We use the term 'signify' here in a sense similar
to that of Professor Leonard. A statement §_expresses
the concern of the utterer p_and indicates his topic of
concern.

81
what p_intends by §1'. It may be the case that p_is
not consciously aware of what he intends by §1, and that
psychological theory is capable of demonstrating that
fact. Then, if an utterance of _S__l should have some causal
consequence such that §l turns out to be false, we would
seem to be committed to including §1 among the members of
M, Presystematically, at least, _S__l seems to conform to
the requirements of what would constitute a self-
vitiating hypotheses.

None of the three alternatives proposed leaves us
in a particularly felicitous situation. Later we shall
propose a more adequate means for dealing with recalci-
trant utterances such as M}. Before proceeding, however,
we need to consider another type of diffiCulty which is
closely related. This has to do with ambiguous express—
ions. In such a case we are uncertain as to whether a
particular utterance ought to be included in M, In the
case of an ambiguous expreSsions §3, according to one
translation p3 will turn out to be self-vitiating,While
according to another translation it will ppp_be self-
vitiating. An interesting example of this kind of situ-
ation is afforded by Shakespeare's account of the witches
prophecy in 'Macbeth'. The "prediction" that Macbeth

would never be slain by one of woman born, if interpreted

as MacBeth interpreted it, could be construed as

82

self-vitiating.10 Madbeth believed that the witches
intended by this expression to predict that he would
never be slain by a mortal. The belief in this "pre-
diction" led Macbeth to act with an utter lack of caution
for the safety of his own person; thus construed the
"prediction" seems, in some presystematic sense, to turn
out to be self-vitiating. If, on the other hand, the
"prediction“ is understood to mean that, as subsequent
events make clear, Macbeth will die at the hand of one
who "from his mother's womb was untimely ript“, then the
"prediction" on this interpretation (but, note, not as
thus construed by Macbeth) is not self—vitiating.

Our initial problem which brought up all the above-
mentioned difficulties involved specifying the criteria
whereby we could determine whether an expression pf ought
or ought not to be included in the extension of M, ‘We
were led to suggest that if pf(sp) could be adequately
translated as pjsp) (i.e. what the utterer p_intended by
M? was M) then M? is to be included in the extension of
M, It should be noted that, as we have been using 'ad—
equate translation', the term 'logical implication' (i.e.
the 'I' in our definition) may hold only between p_and L
and not between if and M? even though the former are
adequate translations of the latter. This particular

outcome is not a happy one. For if the same criteria

 

10I am indebted to Mr. Peter Meloney of Michigan
State UniverSity for this example.

83
are to be applied to the members of p, then the definition
we have proposed for 'self—vitiating' will fail. It will
fail due to the fact that there will be at least one member
of p, namely, if regarding which the relation of logical
implication between explanans and explanandum will not
hold: i.e. it is false that (x)(xeL':)Ixh).

The solution we propose to all of these diffi-
culties is the following: we drop clause c) in its entirety
and construe any utterance of pf or if which are causally
efficacious in effecting the event "p as a constitutent
of the value of some 'xi' rather than as members oflg
and p_respectively. In regard to this proposal we can
retain our rigorous formal requirements concerning member-
ship in M_and p_and at the same time avoid ignoring non-
well-formed expressions which may be constituents of the

causal sequence resulting in "p,11

Any 'M,' can take utterances or non-linguistic

i
events as values. The expression 'xi causes xj' symbol-
ized 'Cxixj' is to be construed as meaning that xj is

described by a statement constituting an explanandum in

a scientific explanation and x;

l is referred to by an

 

ll . . . . .
This criterion Simply restricts the values of

p to well-formed contingent hypotheses. we hold that
any contingent hypothesis can serve as the explanandum
of an explanation or prediction. That is, say, no event
is inexplicable. By this we mean that at least the
following statement is not self-contradictory: 'p_is the
explanandum of some explanation.‘

84

essential constituent in an explanans from which the
statement describing xj is logically deducible.12

We are now able to deal with the recalcitrant
utterances M} and if. we deal with them on the basis of
an assumption: namely, that M} and if are causally de-
pendent upon some other pair of utterances p and A, An
elaboration of a previous illustration may help to make
this point clear. Suppose the National Safety Council
predicts that on a given holiday Mpi, four hundred
Americans will die in automobile accidents. Let us call
this fore-diction 'pl'. Subsequently a number of replicas
of p1(sp) are published in newspapers and magazines;
other replicas are contained in spot radio and television
announcements. Let us call these utteranCes containing
replicas of pl(sp) 'MQ' ... 'pn'. As a result of hearing
or reading these replicas and believing them to be true
fore-dictions of the future, a number of individuals A,
p, ... M_are led to make remarks which are replicas of
one or more of the following:

1. There's gonna be an awful lot of people killed
this comin' weekend.

 

2 . . . . .
Soientific prediction (or explanation) may be
composed of various combinations of utterances:

a) All the utterances which are replicas of a
particular utterance or set of utterances con-
forming to the H.O. requirements.

b) An individual composed of some of the members
described in a).

c) An .individual composed of some members des-
cribed in a) plus some members of M.

d) any combination of a) - c).

 

85

2. I see by the papers that four hundred people
are going to get it this weekend.

3. John said that he heard on T.V. that four
hundred people are going to be killed over
the holidays.
4. They're predicting another slaughter this
weekend.
And so forth. Let us call replicas of these sentences
'hfl' ... 'an'. Although the various Mfi's are in
some sense "translations" of the Mi's, the adequacy of
our definition is in no way dependent on such a construal.
The term 'self-vitiating' can be applied to the members
of M and withheld from the various Mfi's. The Mfi's are
understood to contribute to the self-vitiating effect
of M; but are not construed as members of M, In regard
to our definition they are construed as being values of
some 'Mi', i.e. they are essential constitutents of the
causal sequence which begins with the utterance of p_and
eventuates in the event "p, The question then as to
whether or not the various Mfi's are adequate translations
of p_is of no importance for our analysis. What is
important is whether or not these M? utterances are
causally related to p_and "p, And the task of answering
this question we suppose can be left to psychological
investigation and some body of psychological theory.
While this proposal does solve the particular
problems concerning the various pfi's, it does not appear

to solve the problem of the logical relationship which

holds between a particular utterance p_and translations

86

of Mfisp) into certain foreign languages. Presumably the
predictions of the National Safety Council are of suf-
ficient importance to be included, in translation, in
the various foreign-language publications in the Uhited
States. And presumably these translations also enter
into the causal relationship which obtains between the
fore-diction p_and the outcome ”p, Ought we to attempt
to show that an English language statement Me logically
implies a translation pf into the French language? An
affirmative answer to this question would lead us into
difficulties with which we need not be concerned. One
solution which is immediately apparent is that of treat-

ing all such translations of h as members of the class

—e
of Mfi's. Such an alternative will not be open to us

only in those situations in which predictions in English
and predictions in some other language, say French, are
deduced independently of each other and subsequently
published in their respective languages. We may perhaps
be justified in assuming that this kind of situation occurs
only infrequently. And even when it does occur, it should
pose no particular problem. In this case, if both the
utterance in English, Me, and the utterance in French,

pf, are required to bring about the self—vitiating effect,
we can remove any ambiguity from the application of the

predicate by saying that the individual composed of the

utterances Me and pi is self-vitiating. Whether or not

87

either Me or pf taken by itself would be sufficient to
bring about fp is a question which needs to be investi-
gated by the social scientist.13

Perhaps enough has been said at this point to
warrant dropping c) as a sufficient condition for member-
ship in M_or as sufficient condition for inclusion in
M, All utterances such as the Mfi's which contain
neither replicas of nor expressions logically equivalent
to the initial utterance p_are to be considered as values

of the variables 'Ml' ... 'xn', and not as parts of M_

nor as members of M, Therefore, let

'M} stand for a particular utterance
or an individual composed
of a number of utter-
ances Containing 2
members (where 'p} is
any integer) each of
which contains as a
sentential part a) a
replica of the sentential
parts of each of the
others, or b) that
which is logically equiv-
alent to the sentential
parts of each of the
others. And further,
that each such utter-
ance constitutes a fore—
diction that E,

The translation criterion is dropped, and trans-

lations of individuals composing p_are construed as values

of one or another '§i'°

 

13Some of the problems confronting the social

scientist in attempting to answer such questions will be
considered in Chapter V.

88

One further word needs to be said concerning the
values of an 'Efi'- We have remarked that any translation
of p_which is not logically equivalent to p_is to be
treated as a value of some 'fii' according to the require-
ments of a scientific explanation outlined in Chapter I.
Thus a value of an 'Mi' mpy_be, but it pppg pp; pp_a
linguistic event. It may, for example, be a certain
other psychological event which is causally connected
with particular utterances. These psychological events
may, in turn, be "the cause" of some other utterances.
What is to be observed is that the values of 'M}, 'fp'
and 'Mi' are spatio—temporal events. The substituends
of these variables will be the designators of these
events.

A rather peculiar relationship exists between M_
and ~p_which demands some clarification. '"pf takes as
substituend a replica of the negation of the sentential
part of the value of 'Mf. There may appear to be a hint
of a violation of type theory here. Events and names of
events both appear on the same type level. That this
difficulty is only apparent may be made clear by way of
an example.

Suppose an economist 5 predicts at time t that

—o

a particular stock §_will advance one point between time

89

14 And suppose that as a causal consequence of

grand En'
this prediction, the stock fails to advance.15 The sub-

stituend of 'Mf will be the expression:

The utterance by M, at 30, '§ will advance one

point between time p_and pn'.
The substituend of '"pf will be the expression:
§_will not advance one point between time p_and
En°
The fact that there are certain physical similarities
between the substituends of 'Mf and '"pf does not create
any type difficulties. It is true that the quoted ex-
pression in p_appears in the meta-language and the ex—
pression which described ~p_appears in the related object
language. But both p_and ~p_appear on the same linguistic
level. The quoted expression embedded in p_can actually

be eliminated. 'M} could take as substituend the express-

ion:

The utterance of M_at go.

This effectively denotes the utterance although it does
not show what it is. The quoted expression which appears
in 'M} serves to characterize the relationship which the

utterance has to the event "g, It is inserted in M_

 

l4
tenseless.

15

The verb 'will advance' is here understood as

The example, but not the analysis, is due to
Popper.

90
simply for convenience; i.e., to make clear that one of
the conditions for M being self-vitiating is fulfilled.
But even this point can be made clear by other circum—

locutions. 'M} could take as substituend:

The utterance of M.at which predicts p,

120
Or 'M} could take as substituend:

The utterance of A at t which predicts that S

will advance one point—getween p_and En; '—

The convention of indirect discourse here effect-
ively removes any hint of a violation of language levels.
Thus the kinds of contradictions which grow out of such
linguistic entities as the Grelling paradox, do not arise
in our definition of 'self—vitiating'. What has been
construed as a logical difficulty in regard to the phenomena
of self-vitiating hypotheses does not, at least, depend
upon this particular feature of the phenomena in question.

The convention adopted for the predicate 'logically
implies' is for the purposes of keeping the various lang-
uage levels clear. Instead of the symbolism 'th' we
could have used the more familiar 'y.fi>h'. Our convention,
however, makes obvious the relationship between the two-
place predicate 'M} and the one-place predicate 'Ef
(which stands for 'is false'), and the predicate '2}. It

is hoped that our convention makes this fact clear.16

 

6
We write 'Fh' for the expression 'h is false'.
We could not write 'TM' for this would presumably have

91

Some interesting and important consequences of
our analysis follow. First of all, it is only specific
utterances which are self-vitiating. For any M only
particular members of M can truly have the predicate
applied to them. (In the literature, this point has not,
to my knowledge, been clearly made.) 'Self—vitiating',
as so far construed, is a predicate Which applies to objects
or events which also happen to be linguistic entities
or events. We note here an important difference between

'self—vitiating' and such predicates as 'logically implies'

 

meant something entirely different from what we intend.
It will be remembered that 'p} takes as substituends the
names of particular expressions. Thus '"Mf would either
be not well-formed or else would refer to the logical
complement of M neither of which we intend. Following
Tarski, we may write .

'Snow is black' is false if and only if snow is
not black. '

NOW'Mjsp) will be expressions similar to those of "'Snow
is black'" rather than 'Snow is black'. Hence if we wrote
'""Snow is black"' this expression would either refer to
the logical complement of that which is named by the

name '"Snow is black"' or else it would not be a well-
formed expression. It is for this reason that we have
chosen the convention of applying the predicate 'is false'
to an utterance and writing 'Fh' for the statement 'M_is
false'.

When we wish to contrast a pair of utterances con-
taining sentential parts which are mutually contradictory
we shall write 'Mf and LE}. Here what we intend is that
the sentential part of M.is the negation of the sentential
part of M, The reasons for this convention are similar
to those just stated.

On the other hand, when we write 'éf, this express-
ion is to mean the same as '"pf and is used only for con-
venience where 'pf is to appear as a constituent of a
more complex expression.

 

92
or such predicates as 'is true' or 'is false'. Given any
pair of tokens Mjsp) and ijsp), if Mfsp) logically im-
plies Mjsp) then for any other pair of tokens Mf(sp) and
;f(sp) if Mf(sp) is a replica of Mfisp) and Mf(sp) is a
replica of ijsp) then ;f(sp) logically implies Mf(sp).
But if a particular p_is self-vitiating it by no means
follows that every member of Mgis self-vitiating. We can,
of course, use some locution as 'Every member of M_is

derivative y self-vitiating provided that some member of

 

' M_is self-vitiating.' This would make clear that we were
not applying the predicate 'self—vitiating' to utterances
other than those which begin the causal sequence in
question. But, I believe, there are good reasons for
avoiding this locution. What we wish to make clear is the
distinction between those hypotheses which are simply
false, and those which are false due to the fact that they
enter into a particular causal relationship which brings
about a state of affairs which exhibits their falsity.
This distinction, we maintain, can be upheld if we dis-
tinguish between those members of MQWhich are part of the
causal sequence and those which are not (e.g., members of
M uttered pgpp£_"p).

Another consequence of our analysis shows that we
can assert unconditionally:

(h) (Vh D (X) (er D Fx))

That is to say, if any member of M_is self-vitiating,

then every member of MLis false.

93

Another point has to dO‘With "replicahood" of
utterances as a condition for membership in a particular
ML Consider a pair of utterances pi and MJ each of which
contains a sentential part which is a replica of the
sentential part of the other. This is not a sufficient
condition for their joint membership in MR. One of the
requirements for such joint membership would be that both
hi and Mj contain as sentential parts expressions whose
negations describe "p, The following example may elucidate
this point:

John Doe: 'I will eat the last ice-cream bar

in the freezer.‘

Mary Doe, sister of Jehn Doe, overhears her

brother, runs to the freezer, saying, 'I will eat

the last ice—cream bar in the freezer.‘ And she
promptly eats the ice-cream.

Although the two fpredictions" are replicas of each
other they are not members of the same M, The indicator
term 'I' in the first sentence refers to Jehn, and the
second occurrence of the indicator term refers to Mary.
Here we may note that, in some presystematic sense, the
first utterance is self-vitiating while the second is not.
Similar remarks could be made concerning elliptical ex-

pressions using tensed verbs where the time or the utter-

ance and the time of the event is not clearly specified.

Summary

In this chapter we have attempted to show how the

term 'self—vitiating' may be clearly explicated. 'We have

94
shown how certain recalcitrant utterances (such as pf
and if) may be handled without doing violence to our
usage of terms such as 'logical implication', 'scientific
prediction' and the like. Before proceeding to an analysis
of the associated disposition predicate 'self—vitiatable'
it will prove helpful to consider several related problems,
among which will be the analysis of the notion of a self-
fulfilling hypothesis, an examination of certain problems
connected with self-fulfilling and self-vitiating general-
izations, and an attempt to show how the phenomena of
self-vitiation and self-fulfillment may be subsumed under
a class of phenomena which have to do with unexpressed or

tacitly assumed auxiliary assumptions.

 

CHAPTER IV

SELF-VITIATING AND SELF-FULFILLING HYPOTHESES COMPARED

lntroduction

We need now to examine a phenomenon which is
closely related to that of self-vitiating utterances and
which presumably would need to be accounted for in an
adequate empirical theory of self-vitiation. We shall
attempt to determine whether the mode of analysis used in
explicating the term 'self—vitiating' will also be adequate
for explicating the term which will characterize the
phenomenon of self-fulfilling utterances. The logical
relationships between the two predicates will be examined.
And finally we shall attempt to show that the phenomena
of self-vitiation may be treated as subsumed under a more
general class of fore-dictions whose failure is due to
the lack of certain statements in the explanans or to the
lack of certain qualifications of statements in the ex-

planans.

W-fuléflling'

Another kind of hypothesis which is closely related
to those of the self-vitiating type is variously called
'self-justifying belief', 'self—justifying prediction',
'self—fulfilling prediction', etc. We propose the term
'self—fulfilling hypothesis' to distinguish this kind of
entity from those which we have labeled 'self—vitiating'.
A few examples of this kind of hypothesis, which are in

some respects analogous to those of the self-vitiating

type already considered, are as follows:
95

96

A. An economist predicts that a particular stock
§_will increase p_points or more between time p_and Eo-
Publication of the fore-diction causes increased demand
for the stock and subsequently the stock increases p_
points. Circumstances are such that we may feel warranted
in asserting the counterfactual: 'If the fore-diction had
not been published, the stock.would not have increased p

points'.

B. A student M_is discouraged due to his previous
failure to master material M, A teacher M_"predicts" in
the presence of M_that M_will master M_by time 3, Having
his self-confidence restored, M then proceeds to master M
by time 5, Again we may feel warranted in asserting a
counterfactual in regard to this utterance: "If M_had
not 'predicted' Mfs mastery of M, M_would not have mastered

M_by time pf.

C. A sociologist predicts that a small group of
persons p_will exhibit behavior M_under circumstances 9_
at time p, The individual members of the group being in
a cooperative mood proceed, as a group, to exhibit be-
havior M_when circumstances 9 obtain at time p, This type
of phenomena is often referred to in popular parlance as
"the power of suggestion".

D. A number of interesting examples related to p,

are to be found in the practice of hypnosis. A hypnotist,

M, “predicts" of a particular subject, M, a heavy smoker,

97

that §_will not smoke for a period of time p, The "fore-
diction“ is made in the form of what is called 'a post-
hypnotic suggestion' and is made while §_is in a hypnotic
sleep. On being awakened §_resists all attempts to get
him to smoke during 3,1 In this example we may have
ample warrant for claiming that if the "fore-diction" had
not been uttered by M, M_would have smoked during p,

Several important differences appear between self-
vitiating and self-fulfilling hypotheses. But the method
for analysing the manifest predicate 'self—vitiating' turns
out to be adequate for the analysis of the manifest pred-
icate 'self—fulfilling'. Let the symbols employed for
the definition of 'self—vitiating' be used in the definition

of 'self-fulfilling'; and in addition let:

'SFh' stand for 'M_is self-fulfilling'.
'Th' stand for 'M_is true'.
'e' stand for The event such that the

substituend of 'pf is a
replica of Misp).
Then the predicate 'self—fulfilling' may be defined

contextually as indicated by the following schema:

SFh =df (axl)(sx2) ... (3xn)(sH)(sL)(heH . Chxl .
Cxlx2 . ... . ane . Th . ((y)(yeL :)th)

v (y)(ye L :)Iyfi)))

 

1we shall not attempt to enter any of the intricate

problems involved in distinguishing conscious from uncon-
scious motivation. But we may suggest that a more adequate

98
In English this may be read: "To call a particular
utterance M_self—fulfilling is to say that p_is part of a
causal sequence which eventuates in p; that p.15 true;

~

and that either M_or an utterance p_is deducible from a

set of utterances L,2
The formal character of this definition differs from

that of the definition of 'self—vitiating' in regard to

the disjunctive expression in the definiens. This express-

ion asserts that either M.or M_is deducible from a body

of scientific laws.3 This element of the definiens enables

us to speak of an hypothesis as self—fulfilling when

eitherafi it is deducible from a set of law-statements and

antecedent-condition-statements and enters causally into

a sequence of events which culminates in the event described

by the hypothesis; or b) its contradictory is deducible

from a set of law-statements and antecedent-condition-

statements and ;p_itself enters into a causal sequence

which culminates in the event described by the hypothesis.
It may appear, at first glance, that this definition

includes within it an assertion which is logically equiv-

alent to the definiens of the definition of 'self—vitiating'.

But this seeming equivalence is deceptive. The difference

 

understanding of the character of hypnotic suggestion
would be of considerable help in the development of a
theory of self-vitiation.

2For the usage of the negation sign above the 'Mf
see foot-note 16, p. 90.

That this expression is not a tautology will be
made clear presently.

99

lies in the non-logical rather than in the logical
characteristics of a particular utterance. Suppose, for
example, that according to our definition of 'self-
vitiating' we have grounds for applying this predicate to
a particular utterance M, Are we then able to assert that,
since 11 is self-vitiating, fl is self—fulfilling? We are
not. And this is the case because even though M may have
entered a causal sequence in such a manner as to bring
about the event ”p, this is not sufficient grounds for
the assertion that M_has entered any such causal sequence.
If M has proved to be self-fulfilling, this fact will be
described by an empirical assertion which is neither
logically equivalent to nor deducible from the quite dif-
ferent assertion that MDis self-vitiating.

A second point needs to be noted. The disjunct
in the definiens is not a tautology. It does not, as does
a tautology, exhaust all the cases. It does not, for
example, take into consideration those cases such as the
Mfi's dealt with in the preceding chapter. What the dis-
junct does is to limit the class of utterances to which
the term 'self—fulfilling' is applicable to those contain—
ing sentential parts that are well-formed. That the
disjunctive expression is not required in the_definition
of'Self-vitiating' is seen by the following considerations:
consider an utterance M such that M_is deducible from a
set of law-statements and antecedent-condition-statements;

yet M_is a member of a causal sequence eventuating in “p,

100
In some presystematic sense we may wish to say that M.is
self-vitiating. Yet an important difference Obtains be-
tween this M_and those which conform to the requirements
of the definiens of the definition of 'self—vitiating'.
In regard to this M_we are prepared prior to its entering
the causal sequence to brand it as a false utterance. It
is an uninteresting case inasmuch as it in no way adversely
affects any body of theory. We shall be prepared, pre-
sumably, to hold almost precisely the same beliefs after
M_has entered the causal sequence as before.

But these arguments do not seem cogent in regard
to self-fulfilling hypotheses. The first member of the
disjunct poses problems of some importance for a theory
of confirmation. When an utterance tends to bring about
an event which can be predicted from a particular body of
theory, to what extent can this event be counted as con-
firmation of the theory in question? But we need not
argue this point further. If eventually it should prove
feasible to include Mi's such as those characterized above
within the extension of the term 'self-vitiating' we now
have a means for accomplishing this. Instead of the ex-
pression '(y)(yeL :)th)' we include in the definiens
of the definition of 'self-vitiating' the expression
'(y)(yeL :)th) v (y)(yeL :)th)'. Having taken note of
this problem, we may proceed to considerations more germane

to our inquiry.

101
i§elf—vitiating' and 'Self-fulfilling' as Applicable to
Generalizations
Up to this point we have tacitly assumed that self-
vitiating and self-fulfilling hypotheses were singular
statements. Further analysis, however, will show how
utterances of generalizations may also be self-vitiating,
and with an important qualification, self-fulfilling.
This assumption will in no way affect the way in which
these two terms have been defined. What is required is
that we characterize the 'Mf in our definition in such a
manner as to clearly exhibit the relationships which obtain
between an utterance of a universal generalization M_and an
event fp. In the case where M_is a universal generalization
it does not appear to be an appropriate way of speaking
to say that M.describes a specific event, p, Instantiations
of M may, however, describe p, when appropriate antecedent
conditions are specified. p_may, for example, be described
by the statement 'Fa.Ga'. Thus if M_is an utterance of
the generalization '(x)(Fx :)Gx)' and we are prepared to
assert that Fa, then we can deduce 'Fa.Ga' from the con—
junction of '(x)(Fx :)Gx)' and 'Fa'.
With this example in mind we may now characterize
M_in such a way as to include within its scope universal
generalizations. we now let:
O'Mf stand for a particular utterance
or an individual com-
posed of a number of
uttErances containing
M members (where 'Mf is

any positive integer)
each of which contains

102

as a sentential part a)

a replica of the sentential
parts of each of the others,
or b) that which is log-
ically equivalent to the
sentential parts of each
of the others. And fur-
ther, that each such
utterance constitutes a
fore-diction that p; or M_
is a universal generali-
zation such that there
exists an M1 and h , such
that the conjunction of M_
and M1 implies M2, and
either M_ or M_ . h con-
stitute a fore-diction
that p, it is not the
case that h? is deducible
from Ml alone, and
finally, M1, is true.

In a number of instances M_l will simply be an in-
stantiation of the antecedent of the universal generalization.
This need not be the case, however. And the only logical
requirement is that the utterance M2 (or the conjunction
M1 and M2) which describes p_must be logically deducible
from the generalization M when M.is taken in conjunction
with M_l.4

We may now examine some of the characteristic
features of the two predicates 'self—vitiating' and 'self-
fulfilling'. Suppose that Miis a universal generalization,
one of whose utterances M conforms to the definition of
'self-fulfilling hypothesis'. That is, there is some M

2
describing p, which is logically deducibe from the

 

4For the requirements concerning instantiation
which shall be tacitly assumed throughout this essay, see,
Nelson Goodman, Fact,_§iction andpgprecast, p. 68ff.

103

conjunction of M_and Ml. By definition, to say that M.is
self—fulfilling is simply to say that there is some
utterance of M which causally culminates in a confirming
instance of M, It is not to say that the utterance oij
causes M_to become true. Now there may be some cases in
which the locution 'the utterance of M causes MDto become
true' would be used without difficulty. These cases,
however, would involve non-law-like rather than law-like
generalizations. For example, I could spray the rug in
my study with gasoline in such a manner that when I touched
a match to the rug the flames would spell out a replica
of the sentence 'All the fibres in the rug in A. J. Stenner's
study will burn up at time M]. It might be held that this
is a generalization, although a non-lawelike one, which
on its flaming occurrence causes itself to become true in
some sense of "cause". But there are sufficient reasons
for avoiding such locutions. And it can be shown that
other ways of speaking enable us to say what we wish to
say without encountering the difficulties which arise
when such locutions are employed.

Now the interesting cases we wish to consider are
those law-like generalizations which are not restricted
in such a way as are so-called non-law-like generalizations.
And granted that establishing of a criterion for distin-
guishing between law-like and non-law-like generalizations
is difficult, the difficulty, again, is not one which is

peculiar to our subject matter. We may therefore simply

104
note the difficulty; for it does not seem to be one
whose solution would affect our analysis one way or the
other. On the contrary, it would appear that any criteria
which are finally adopted for distinguishing between law-
like and non-law-like generalizations would need to take
cognizance of some of the points made in our analysis of

self-fulfilling and self-vitiating hypotheses.5

 

5It may turn out that the kind or the amount of
evidence needed to determine whether or not a given
utterance of a generalization is self-vitiating or self—
fulfilling will also enable us to determine whether the
generalization in question is law-like or non—law-like.
What is involved is the following: in the case of some
generalizations the expression 'Th' in the definiens is
redundant; whereas in other generalizations it is not
redundant. (The proof of this assertion will be indicated
presently.) That is to say, if from the other components
of the definiens we are able to determine the truth of
the assertion 'Th', then 'Th' is redundant and the gen-
eralization in question is one we would consider as non-
law-like. If, however, 'Th' is not an expression whose
truth can be determined by reference to the truth of the
other components alone, then 'Th' is not redundant and
the generalization is a law-like one.

Before this problem can be adequately dealt with, what is
required is a definition of 'self—fulfillable' and 'self-
vitiatable' both of which are disposition predicates.

The reasons for our inability to define these terms at
present will be given in Chapter V. But in the event
that such definitions are forthcoming we may now indicate
how they may help to solve the problem of determining the
law-likeness of certain generalizations. Suppose that an
utterance of a generalization M_ is one such that under
certain circumstances p_we would apply the predicate
'self-fulfilling' to M_. This is to say that h .is self-
fulfillable. Now the uestion as to whether or not h is
law-like or non-law-like could be determined by decidgng
whether or not the expression 'Thg' would or would not be
redundant under circumstances 9,

The question of the redundancy of 'Th' will be in part
dependent on the kind and specificity of the value of 'p},
and the logical relationships which obtain between 'pf
and M. Consider the assertion: 'Thsp'. In regard to

105

In regard to law-like generalizations thepredicates
'self-fulfilling' and 'self-vitiating' are logically in-
compatible. It might seem presystematically that we
could have two utterances of a single M;one of which would
be self-vitiating and the other self-fulfilling. It will
indeed be the case that two utterances containing sen-
tential parts which are replicas of each other will, or
at least probably will, have different causal consequences.
But we can show that this creates no particular difficulty
for our analysis which is not also encountered in other
theories.

Logically considered, no amount of evidence can
ever completely verify a contingent universal generali-
zation. But one disconfirming instance, if it is con-
firmed to a high degree, may lead us to reject the general-
ization, provided that "other things are equal". This is
not to say that we do not have means for preserving those
generalizations which have proved exceedingly useful and
which we are loathe to give up as false. It is to make

a simple logical point: namely, that the assertion

 

a generalization h if we are prepared to assert that

h is true if and nly if p_then we will also be prepared
to assert that h is non-law-like. For presumably law-
like generalizations do not "describe" p_in the same
manner that non-law-like generalizations do. The problem
then boils down to two questions: 1) What can serve as
the value of 'p}? and 2) What constitutes confirmation of
p? We can answer the first question; only actual events
may serve as values of 'p}. The answer to the second
question waits upon an adequate theory of confirmation.

106

'Fa . Ga' is logically incompatible with the assertion
'(x)(Fx :)"Gx)'. And if we wish to hold that the former
is true we must then hold that the latter is false.6
Whether we give up the generalization or find some means
of modifying it, or whether we discount the evidence for
the singular statement will depend on considerations which
are other than logical ones. The point we wish to make
is logical. According to the way in which 'self—viti-
ating' and 'self—fulfilling'have been defined, a partic-
ular hypothesis M cannot have both predicates truly
applied to it. And it follows that there cannot be two
members of a particular M such that one is self-vitiating
and the other self—fulfilling. If an hypothesis is self-
vitiating, then by definition it is false; and we noted
that if one member of MRis false, then every member is
false. On the other hand, if a particular hypothesis
is self-fulfilling, then by definition it is true. It
follows, therefore, that there can be no two members of
an hypothesis one of which is self-fulfilling and the
other self-vitiating.

From this analysis an important difference between
the two predicates emerges. The grounds on which a
particular utterance of a law-like generalization can

have the predicate 'self—vitiating' applied to it are

 

6Unless, of course, we are so bold as to be willing
to give up the law of contradiction.

107
clear, from a logical standpoint: one disconfirming in-
stance of the generalization which conforms to the re-
quirements of the definiens of the definition will be
sufficient to apply the predicate 'self—vitiating' to
the hypothesis. But one confirming instance of a law-
like generalization will Mpp_be sufficient evidence for
applying the predicate 'self—fulfilling' to a law—like
generalization.7 We need to know on other grounds that
the hypothesis is true; for just as one swallow does not
make a summer, neither does one confirming instance of a
universal generalization render that generalization true.
One of the consequences of this fact is that to say that
a particular utterance of a generalization is self-
fulfilling, is to say that it never has~had, does not now
have, and never will have either a disconfirming instance
or, more narrowly considered, a self-vitiating instance.

But to call an hypothesis 'self-vitiating' is not to

 

7We are, of course, overlooking some notorious
difficulties in confirmation theory. Our willingness to
accept an assertion such as 'Vh' will depend on a great
many factors among which will be the strength of the
evidence for each of the assertions in the definiens
and the cost of abandoning M_as a useful generalization.
These difficulties do not appear to be ones which will
adversely affect our analysis. We are concerned specifi-
cally with the problem of what follows from the truth of
assertions such as 'Vh' or 'SFh'; and not the problem of
what would constitute confirming evidence of these as-
sertions. The latter problem is one which would require
the cooperative endeavors of a number of the sciences
and a rather highly developed theory of self-vitiation
as well as an adequate theory of confirmation. We may
perhaps then be pardoned for apparently ignoring these
issues.

108

claim that the hypothesis has not had any confirming
instances. On the contrary; presumably the overwhelming
majority of generalizations to which we would be willing
to apply the predicate 'self-vitiating' will be general-
izations which have been confirmed to some degree and

in many cases have been confirmed to a high degree.

To see what is involved we may consider an example

frequently used in the literature:
M, All ravens are black.

We may inquire as to some of the conditions under
which this generalization may be self-vitiating or self-
fulfilling. There are several ways, in fact a large
number of ways, in which the generalization could be self-
vitiating. An utterance of M, could be inscribed on the
body of a female raven M, The instruments used for the
inscribing could be tatooing needles containing some
radioactive substance M, M_we assume is of such a nature
that its presence in the body of ravens brings about
certain genetic changes in the offspring of M, i.e, the
offspring of M_are not black. Hence the utterance of M,
would be self-vitiating.

Another way in which M, could be self-vitiating
would be in the event that tokens of M, were produced by
spraying a chemical M} into the air in the vicinity of
Mfis nesting place. The chemical M} would be one which

changed the nature of the pigmentation in the feathers

109

of M and the causal consequence of this spraying would
be that M,was no longer black. A more fertile imagination
than our own could undoubtedly conceive of many other
possibilities.8

we may also inquire concerning the conditions under
which utterances of M, may be self-fulfilling. (We need
to keep in mind that one of the requirements of a self-
fulfilling hypothesis is that it be true.) An interesting
example, and one which also shows some of the relation-
ships which obtain between the predicate 'self-fulfilling'
and the predicate 'self-vitiating' would be the following:
rave» M, previously innoculated with chemical, M, is
innoculated with a new substance, Mf', which counteracts
thefipossible" effecuaof M_prior to the actual birth of
any offspring of M, The innoculation, as in the previous
instance, is done in such a way as to spell out tokens

of M, The causal consequences of these innoculations of

M with Mf' is that all of the offspring of M_are born

 

8If these examples seem remarkably far from what we

would usually construe as self-vitiating hypotheses we
may indicate some our reasons for choosing these examples.
First, they are simple enough to avoid the kinds of com-
plications which arise in regard to purposive behavior.
Second, they serve to show that the use of the phenomena
of self-vitiation as a means for distinguishing the
social from the physical sciences is without forundation,
inasmuch as our "far-fetched" examples conform to both
our pre—systematic discussion of self-vitiating and our
systematic requirements. Third, the discussion based
upon these examples serves as an introduction to the
problems discussed later in the chapter under the heading
of 'Boundary conditions'.

110
black and remain black. In this latter instance the
second utterance of M, interrupts the causal sequence
which was begun by the first utterance of M, Hence,

provided that M_is true, the predicate 'self—fulfilling'

 

may be applied to this second utterance of M, In such an
event the predicate 'self—vitiating' would not be truly
applicable to the first utterance of M,

The underlined expression in the above paragraph
describes a condition which is of crucial importance in
distinguishing the predicates 'self—vitiating' and 'self-
fulfilling' as applied to generalizations. It may be
shown that the expression 'Fh' in the definiens of our
definition of 'self-vitiating' is redundant. That is,
if we are able to assert that the other expressions in
the definiens are true, then we can show that 'Fh'
logically follows from these other assertions and that

hence we may apply the predicate 'self—vitiating' to M,9

 

9The proof is as follows:
Let 'Chg' stand for '(x)(Gx :)Kx) . (x)(Kx :)"e) . Gh'
l. '"e' is true E ~e By semantic definition of

'true' and 'false'.

2. h2 is false 2 '"e' is By semantic definition of
true 'true' and 'false' and the
relation of M_to TM,
3. Chg Premise
4. (x)(Gx Kx) . By replacement for 3.

(x) (Kx "e) . Gh

5. Gh.:)Kh 4, simp., spec.

111

The reason for keeping 'Fh' in the definiens
despite its redundancy is that of preserving a notational
symmetry between the predicate 'self-fulfilling' and the
predicate 'self-vitiating'. But the expression 'Th' in
the definiens of the definition of the predicate 'self—
fulfilling' is by no means always redundant. That this
is the case is readily seen by observing that being able
to assert 'Fa.Ga' is not sufficient grounds for asserting
'(X)(FXDGX)'.

The distinction we are here considering between the
two predicates 'self-vitiating' and 'self—fulfilling' may
be summed up in the following way:

1) ‘If the assertion '(Exl)...(EXn)(EM)(3H)(heH .
(Y)(Y€L :)IYh) . Chxl . ... . ane)' is true,
then 'Fh' is true. And hence 'Vh' is true.

2) If the assertion'(3x1) . ... (EX2)(3L)(EH)(heH .

Ch~x1... . ane . ((y) (yeL D M) v (y) (yeL D
th)))' is true, it may or may not be the case

 

6. Kb :)"e 4, simp., spec.

7. Gh :)”e 5, 6, syl.

8. Gh 4, simp.

9. ~e 7, 8 m.p.

10. '”e' is true 1, 9, subst.

ll. h2 is false 2, 10, subst.

12. (h .h1)~?h2 Premise

13. (h . hl) is false 11, 12, by definition of tfi>'
l4. hl is true Premise

15. h is false 13, 14, by definition of '.'

112
that 'Th' is true and hence it may or may
not be the case the 'SFh' is true.

It can be seen that different grounds are therefore
necessary for determining whether the two predicates
'self-fulfilling' and 'self—vitiating' will apply in any
given case of a generalization.

Up to this point we have been considering 'SF' as
applied to generalizations. When we come to the consid-
eration of singular statements, however, we discover
that the difference between the two predicates just
mentioned does not obtain. For in the case of an utterance
of a singular statement we discover that the expression
'Th' in the definiens of the definition of 'self—fulfilling'
is also redundant. The proof for this is similar to the
one given for 'self—vitiating'. We need simply note that
if the event predicted by an utterance of M comes about,
then by the semantic definition of truth we are able to
assert that M_is true.

We may now inquire as to whether or not there are
any self-fulfilling generalizations such that the expres-
sion 'Th' in the definiens is redundant. The answer of
course will be dependent on a vast array of empirical
evidence, evidence which presumably we do not have at our
disposal at present. But we can easily see hOW"Th' may
be redundant in a given instance by considering an example
previously dealt with in regard to the burning of all the

fibres in a particular rug. In this example the value of

113

the sentential part of M_is a replica of the substituend
of 'pf. When this particular relationship obtains between
M_and p, in this instance we are able, by the semantic
definition of truth, to deduce 'Th' from the other com-
ponents in the definiens.10

This example differs in an important respect from
another which shares certain features with it. Consider
a case in which raven M.is innoculated with a substance
M}. The result of this innoculation is that a mutated
species is produced which we shall call 'croven'. The

innoculation is again performed with tatooing needles

spelling out an inscription which is a replica of:
M, All crovens are blue.

We assume further that there are good empirical
grounds for believing that all crovens are in fact blue. Now
suppose that after three generations of crovens all the orc-
vens thus born of M_are destroyed. Are we in a position to
claim that the assertion "'All crovens are blue' is true"
is redundant in the assertion that M_is self-fulfilling?

The answer to this question appears to be in the negative.11

 

10See footnote 5, pp.104-105 of this essay for the

relevance of this distinction to the problem of distinguish-
ing law-like from non-law-like generalizations.

11Unless, of course, we wish to define 'croven' as
'decendent of M3 rather than specifying other characteris-
tics as the defining characteristics of croven. But such a

move does not appear to be one which has proved the most
fruitful in the sciences.

114

One reason for this fact would be that the example does
not rule out the possibility that other crovens may be
born under similar circumstances or the offspring of an-
other raven Mf may mutate by ”natural" means. In the case
of the rug fibre example then we are prepared to claim
that the flTh' in the definiens is redundant, whereas in
the croven example we are not prepared to make this
claim. Wherein lies the important difference?

As we have previously noted, in the rug fibre ex-
ample the value of Mfisp) is a replica of the substituend
of 'p}. But in the croven example this relationship does
not obtain. Speaking loosely, we suggest that in the rug
fibre example the generalization describes only actual
fibres whereasin the croven example the generalization
speaks “about" possible as well as actual crovens.

We are not, of course, specifically concerned to
solve the problems encountered in a characterization of
law-like generalizations. We have merely been attempting
to note that this problem is not unrelated to the problems
encountered in an analysis of self-vitiation. And pre-
sumably the solution of the one will contribute to the
solution of the other. In the next section we shall
attempt to show hOW the problems of self-vitiation are
directly related to another, broader class of problems

in the sciences.

115

:Moundary Conditions" and Self-vitiating Hypotheses

In this section we wish to show that the class of
self-vitiating and self-fulfilling hypotheses can be
construed as a sub-class of a certain type of phenomena.
The characteristic feature of this class of phenomena
concerns the nature of the so-called boundary conditions
which must be fulfilled if the event is to be successfully
predicted from a body of law. The hypothesis to be ad-
vanced in the course of our argument is that the same
characteristic features are present in those utterances
which we have called 'self-vitiating' and 'self-fulfilling'
as are present in other predictions whose success or
failure is dependent on making the boundary conditions
explicit.

In our example of the disruption of the orbit of

Mars12

‘we noted how the production of tokens of a certain
type could bring about an event such that the fore-diction
exhibited by the tokens turns out to be false. Now it

can be easily seen that in the physical sciences, at least
(and we hope to show that the same will hold true for the
social sciences) this class of phenomena is easily subsumed
under a more general class of phenomena. What occurs in
these cases is that the so-called "boundary conditions"

are not clearly specified in the theory in question but

are tacitly assumed in any scientific explanation of

 

116

events or in the deduction of certain general laws from
more general laws. A few examples may serve to show how,
in the physical sciences, instances of self-vitiating
hypotheses may be subsumed under this more general class.

Consider the following deductive explanation for
the fact that a particular iceberg, M, floats:

.Q. All ice will float in water.

All icebergs are ice.

M_is an iceberg.
M_is in water.

 

.'. M floats

Now suppose that there is a particular object 9,
which we would designate as a glacier, which has, in
moving down a mountain side in Alaska, picked up a large
amount of glacial debris which is embedded in its interior.
p_slides into the Pacific Ocean, and as it moves into the
Japanese currents begins to melt. As it melts it sinks
to the bottom of the ocean due to the large quantities of
glacial debris embedded in its interior and the consequent
increase in the mean density of p_as it melts.

From the set of premises in M,, together with the
premise 'p is an iceberg' we can deduce that p floats when
placed in water. Now this conclusion contradicts what by
assumption are the facts. Hence we must modify or reject
at least one of the premises. There are a number of ways
in which the required modifications can save the laws
stated in the premises. One such way would be to define

'ice' so that an object such as 9 would fail to qualify

117

as ice. For example, we could specify that an object
would be considered to be ice only if its mean density was
less than that of water. p_would then not be treated as
an instance of ice. It also could not be considered as
an instance of an iceberg. This might place seemingly
artificial restrictions on popular ways of speaking, but
such speech, as is indicated by the problems arising from
example §,, often need clarification if we are to avoid
the kinds of puzzles which thwart our attempts to under-
stand and explain phenomena.

Another method of saving the appearances would be
to deduce the law 'For all M, if M_is ice, theaM has a
mean density less than that of water' from physical theory.
9 may then be excluded as being an instance of ice on the
grounds of its failure to conform to the law in questbn.
That is, by modus tollens we conclude that p_is not ice.
But this method possibly can lead to contradictory results
when taken with other law-like statements of physics and
chemistry. Consider
J, All water freezes (i.e. becomes ice) at 329M

—' or below.

p_is composed of water and its temperature is
29°F.
290; is below 320;.

 

.’. Q is ice.

Since the conclusion to example M, contradicts our
former conclusion that p_is not ice, other modifications

need to be made in the theory. we can, for example, define

118

'water' in such a way that a particular sample of liquid
will be called 'water' only if the sample in question con-
tains no more than a specified proportion of a foreign
substance, i.e. a substance other than one whose chemical
properties are characterized by the formula 'HZO'. An-
other way of proceeding, one probably more in accord with
actual chemical practice, is to define water simply as
'HZO'. The fact that there are few if any samples of
this pure mixture in nature will demand that some other
set of auxiliary assumptions be included in any predictions
about the behavior of water under actual empirical con-
ditions.

In many cases these auxiliary assumptions are not
required for the successful predictions of the behavior
of water under certain conditions. Auxiliary assumptions
are usually required in cases similar to those of our
example of the iceberg. The notorious ceteris paribus
clause can always be invoked when recalcitrant phenomena
such as the occurrence of p.becomes problematic. But to
invoke such a clause is seemingly only to make clear that
the necessary auxiliary assumptions or definitions have
not been included in the premises from which the explanan-
dum was deduced. Once these auxiliary assumptions have
been introduced, the falsity (or perhaps the ambiguity)
of one of the premises is determined. And from the revised
set of premises the false explanandum is no longer deducible.

With these points in mind we may now consider an-

other example. .Instead of an iceberg, let 'pf' stand

119
for an object which has been formed in the following
manner: a number of pieces of lead buckshot are placed
on the bottom of a refrigerator tray, arranged in a

manner to spell out tokens which are replicas of:
M. 9_' will float in water.

The refrigerator tray is then carefully filled with
water and placed in the freezing compartment of a refrig-
erator. When the liquid has frozen solid, it is removed
together with the embedded buckshot and placed in a tub
of water. As the mean density of the composite object Q}
is increased due to the melting of the ice, pf subsequently
sinks to the bottom.

It can be readily seen that the generalizations
functioning as auxiliary assumptions required to explain
the fact that 9 did not float are precisely the generali-
zations required to explain the fact that pf did not float.
One difference between 9 and Q} concerns the arrangement
of the non-aqueous material contained within the two
objects. The glacial debris in p_presumably is not arranged
in a manner such that the debris constitutes a prediction
of the behavior of 9; while the lead shot in Q} is arranged
to spell out a prediction of the behavior of Q}. But even
this difference need not obtain. Statistical theory would
lead us to assert that given an indefinitely long period
of time and the production of a vast number of icebergs,
there is an overwhelming probability that there would

appear an iceberg which exhibited the characteristics of

120

p_and in addition had its debris arranged in such a manner
as to spell out a set of tokens which were replicas of
the sentence 'pf' will float in water.’ Let us call
such an object 'p}". The explanans which is required to
explain why such objects sink when immersed in water
contain the same general laws in the cases of p, pf and
Q}'. The fact that the differences between these three
objects may be important for the explanation of other
characteristics of these objects should not obscure the
fact that in the particular explanantia we are here con-
sidering these differences are irrelevant.

Perhaps it will be objected that in the social
sciences the shape of the tokens involved constitutes
one of the relevant factors in any explanation of self-
vitiating hypotheses, whereas, in the examples just
given, the shape of the tokens obviously is unimportant
for predicting the behavior of the objects in question.
For example, if the Uhited States' State Department

predicts:

M, Russia will attack West Berlin at time p,

this utterance and others containing as sentential parts
replicas of it will have,presumably, vastly different

causal consequences than a set of utterances of the form:

L}. West Berlin will attack Russia at time p,

121
Here it is the difference in the order (or what we might
call 'the shape') of the tokens which will determine the
actual outcome of events. No theory of self-vitiation
which failed to make this point clear could possibly
have predictive value. M} since it contained all and
nothing but replicas of the word-tokens in M would be
treated as having the same causal consequences as,M
which we know presystematically is not likely to be the
case. Now whereas in the cases of p and Q} the arrange-
Mppp_of the debris is an irrelevant factor in explaining
the causal consequences of the ice's containing foreign
objects, in the case of the utterances of M_and M} the
arrangement of the tokens is of crucial importance.

In replying to these objections we may agree that
any theory of self-vitiation will need to make clear
when the shape of the tokens is relevant and when ir-
relevant. But it is by no means clear presystematically
what is to constitute the relevant shape characteristics
of the objects in question. Suppose, for example, that
the iceberg pf' contains all of the debris embedded in
the outer layer rather than within the central portion.
Then when the iceberg, pf', reaches the warm waters of
the Japanese currents, the glacial debris will begin
to drop off so that the mean density of the iceberg re-
mains less than that of water and pf' continues to float
rather than sink to the bottom. Here we can see that
the shape of the debris is of considerable importance

to the outcome of the predicted event.

122

It may be objected that what is involved here is
not so much that of shape but that of the position of
the debris, whereas in the case of predictions M_and M}
what is involved is the shape of the tokens, and that
hence we need to distinguish between those predictions
such as would be falsified by an appearance of an object
such as pf', in which the important characteristic is
the position of the debris and not the shape. Ignoring
the Obscurity of the objection for a moment, we may note
that the relative position of utterances M and M} would
also be an important consideration in any attempt to
predict what the causal consequences of M_or M} might be.
If the only extant replica of M_occurred on the head of
a pin or was inscribed on the underside of President
Kennedy's desk, the causal consequences of such an
utterance of M_would be presumably much different than
if the replicas of M appeared in prominent positions on
the front pages of the leading newspapers of the world.

Our example of disrupting the orbit of Mars
points up also how important the shape of the tokens may
be in determining whether or not a prediction of a
physical event is self-vitiating. Suppose that the co—
balt bombs are built to explode in such a fashion as to
spell out a sentence token which is a replica of:

M, Mars' orbit will not be altered by a factor
a at time M.

123

Suppose further that there is one bomb for each of the
word-tokens and one for each of the substituends of the
simple literals 'x} and 't}, and that these bombs must
explode in a precise order such that if they explode in
any other order the alteration of Mars' orbit will not
take place. Now whether or not the fore—diction in
question will or will not be self-vitiating will de-
pend principally on the shape of the tokens produced by
the exploding bombs.

It seems therefore clear that position, shape,
and order of tokens are or may be equally important in
a self-vitiating hypothesis in physics as in a self-
vitiating hypothesis from sociology or psychology.
These characteristic features of the sentential parts
of self-vitiating utterances, therefore, are not those
which enable us to make any significant discrimination
between the social and the physical sciences.

A further objection might take the following
form: it is granted shape and position of the tokens
may be factors in predictions of events whether these
have to do with the behavior of icebergs and planets
or the behavior of human beings. But there is one
important difference between social and physical phe-
nomena which has been overlooked. In the case of the
cobalt bombs the same effect could be achieved by means
other than that of having the bombs spell out the par-
ticular tokens in question. Whereas in the case of‘;

the causal consequences are dependent on certain meanings

124
attached to the tokens, in the case of disrupting the
orbit of Mars meanings are irrelevant. The bombs could
explode in the shape of an American Flag, or a picture
of Santa Claus, and provided that the bombs exploded
according to certain laws of mechanics the result would
be the same whether or not they exploded as tokens of
a fore-diction or whether they exploded according to
some other pattern. But in regard to the prediction
concerning Russia's behavior in west Berlin the tokens

must be replicas of L_or similar to L, otherwise the

 

causal consequences will not be the same as they would
be if‘utterances of g, were alone in question.

NOW'We need not, I believe, resort to "meanings“
in order to deal with the problem in question. The
force of the objection is lost due to the obscurity of
the term 'similar'. By introducing the notion of tokens

which are similar to L. one only begs to question as to.

 

how similar the tokens will need to be in order for them
to have the game_causal consequences as L, we need also
in the case of Mars' orbit to consider how similar the
proposed different explosions need to be to the ones
initially proposed in order for the same causal con-
sequences to result.

If one does not resort to the notion of 'similar
to' and instead attempts to draw the distinction on the
basis of the "meanings" and the causal consequences of

these "meanings" it is by no means clear that what can

125

be explained by an appeal to the meanings of the tokens
cannot also be explained without resorting to such
entities. If what one wishes to do is to distinguish
between the consequences of linguistic events involving
purposive behavior and those involving icebergs, ice-
cubes and exploding bombs, there may be certain distinc-
tions which can be made. We shall consider some of these
in the next chapter. But it is not prima facie obvious
that an appeal to meanings helps clear up any obscurity
in the distinction. As we noted before, the paucity of
our imagination may be the only thing which is lacking
from attributing meaning to any object or event whatso—
ever.13 That is to say, any event or object may be
interpreted as a sign which indicates some other event.
In some construal of 'meaning' which is yet semantical
(rather than valuational in the sense of 'importance')
any event whatsoever can have a meaning.

The point of these remarks is n9t_that the des—
ignative function of utterances has no bearing on the
outcome of events generally nor on the phenomena of self-

vitiating sign-events in particular. All we are

 

13"Whatever the author intended the token to be
a clew to, that is what the token means." Henry Leonard,
Principles 9£_Right Reason, p. 138 Cf. Also, Lewis Carroll,
Through the Lookinnglass, (Random House, New York, 1946,
special edition) "'When I use a word,‘ Humpty-Dumpty said,
'it means just what I choose it to mean - neither more nor
less.'" Analogously we may extend the notion of 'token'
so as to include objects or events of any size, shape or
description whatsoever.

126
suggesting at this point is that the introduction of
"meanings" into our universe of discourse does not help
but may actually hinder our understanding of the problem
of self-vitiation. This problem we take to be that of
the development of a theory which will enable us to ex-
plain (in the sense explicated in Chapter I) occurrences

of self-vitiating hypotheses.

Summary

In this chapter we have attempted to explicate
the notion of 'self—fulfillment' in a manner directly
analogous to that of the notion of 'self—vitiation'. We
have found that the method of analysis used for expli-
cating the latter is adequate to the analysis of the
former. we went on to show some of the logical relation-
ships Obtaining between the two terms. And finally, we
tried to show how the phenomena of self-vitiation may
be understood as subsumed under a more general class of
phenomena in which prediction of the future is unsuccess—
ful due to failure of the prediction to make clear
certain boundary conditions. We attempted, at the same
time, to answer possible objections to our analysis.

Throughout this chapter we have been concerned
primarily to explicate the manifest predicate 'self-
vitiating'. Next we must consider what would be involved
in a definition of the associated disposition predicate
'self—vitiatable' and related problems. It is to this

task that we now turn our attention.

CHAPTER V

TOWARD A THEORY OF SELF-VITIATION

Introduction

Up to this point in the discussion we have attempt-
ed to characterize, more precisely than seems hitherto
to have been achieved in the literature, the phenomena
of self-vitiation. We have attempted to define a predi-
cate which could be applied to these phenomena with a
minimum risk of ambiguity and vagueness. Now we must
examine a much more difficult problem, the problem of
determining the conditions under which we shall apply
the disposition predicate associated with the manifest
predicate 'self-vitiating'. We hold that the task of
adequately defining this particular disposition predicate
will depend crucially on empirical considerations. But
there are nevertheless certain formal problems which must
be considered before it is possible to construct what
would be acceptable as an adequate theory of self-vitiation.
Among such considerations is the classification of
various types of self-vitiation. We shall claim that for
these and other purposes the distinctions which prove the
most useful are not those which traditionally are held
to obtain between the social and the non—social (or
physical) sciences. ‘We are primarily concerned with those
types of self-vitiation in which important factors are
those of ggal-directed-behavior or purposive-behavior or

sign-behavior.

127

128
A New Predicate, 'Self-vitiatable'

So far we have seen how the predicate 'self-
vitiating' can be applied to particular utterances. This
predicate is, however, applicable only in those cases in
which the characteristic feature of self-vitiation is
actual. That is to say, the predicate {self-vitiating'
is correctly applicable to just those utterancesin which
it is the case that the utterance in question is causally
efficacious in bringing about "2, We now wish to be able
to describe those cases in which an actual utterance has
not been self-vitiating but which nevertheless is such
that we believe it would be self-vitiating if it were to
be uttered under certain particular conditions. we are,
in short, raising the question of how to explicate
'potentially self-vitiating' as applied to hypotheses.

In raising this question we confront the problem of suit-
ably defining an appropriate disposition predicate. The
specific "potentiality" with which we shall be concerned
is that particular feature of those utterances which,
under certain circumstances, would enter a causal sequence
such that an event described by their negations would
result. The predicate which we shall use to characterize
this feature of utterances is the predicate 'self-
vitiatable'. In other words we wish to direct our atten—
tion to the assertion, “If there is a member h.of a par-

ticular statement g such that if p_were uttered under

suitable circumstances, 9, h_would be self-vitiating, then

129
h_is self-vitiatable." Several points need to be noted
concerning this characterization of the predicate 'self—
vitiatable'. We shall state each point briefly and then
discuss each one in greater detail.

First, the characterization is not a formal defin-
ition. It gives one of the sufficient butrnne of the
necessary conditions for the application of the term
'self—vitiatable'.

Second, it does not specify what are the "suitable
circumstances“ under which hywould be self-vitiating.
Presumably the specification of these circumstances would
depend upon results of empirical investigation. It would,
for example, be required, prior to a formal definition
of the general predicate 'self—vitiatable', to examine
a large number of actual utterances which were self-
vitiating, to determine specific features shared in common
by all of these examineiinstances, and then to determine
which of these common features are relevant for character-
izing the phenomena of self—vitiatability.

Third, the characterization, given above, is ex-
pressed in the form of a counterfactual. This character-

ization, of course, is not the only type available to us.1

We could translate this counterfactual characterization

 

1Much of the material in this section is dependent

on the analysis of dispositions given by Nelson Goodman in
Fact, Fiction, and‘Forecast, (Cambridge, Massachusetts:
Harvard University Press, 1955).

130

in such a way that the antecedent and consequent of the
statement appear in the indicative mood. Translated in
this manner, our characterization would read, "Since
the statement, 'There is a member h_of a particular state-
ment HLWhich is uttered under suitable circumstances 9]
is true, the statement 'h_is self-vitiating' is true."
This translation does not replace a counterfactual by a
material conditional. The translation also has the
counterfactual form for, by hypothesis, its "since-clause"
is contrary to fact. But it deals with certain semantic
relations which are claimed to hold between the two
indicative components. "The disposition term ... is
eliminated without the introduction of any such trouble-
some word as 'possible'; only non-dispositional predicates
appear to remain, even if they are slightly jaundiced
with a modal inflection."2 This translation is one which
may make us "feel that we have exchanged an ontological
problem for a linguistic one“.3

Fourth, as in the case of its associated manifest
predicate, the term, 'self—vitiatable' is applicable to

utterances of statements rather than to statements.4

 

 

292, cit., p. 40.

 

31bid.

4See footnote 15, p. 12 of this essay for a
qualification of this assertion.

131

The disposition term is a predicate only of actual
utterances (i.e. past, present, or future ones) which are
members of H and not of possible members of E. This is
a matter of some importance and it will be dealt with in
more detail in a moment. We shall not ignore "the pos-
sible but non-actual" members of H;but shall treat these
as being in the extensions of predicates that apply to
persons; i.e. as dispositions of persons. Thus, a
possible, non-actual, self-vitiatable h.is an actual
disposition to utter self-vitiating hypotheses.

Sixth, this treatment of the predicate 'self-
vitiatable' avoids any crucial ambiguity in the antecedent
of the conditional characterization. Note that the
counterfactual is phrased "If there i§_an h_which is a
member of HT and not "If there 3232 an h_..."

Seventh, it follows from the fifth point that the
extension of 'Hf is not null. That is to say, if there
is some actual utterance of h_then there is at least one
member of §_and hence 'gf is not an empty description.

We are thus able to avoid having to deal with problems
created by allowing as substituends of the variables names

of non-existent entities.5

 

5An example given by Henry Leonard is the following:

§anta Claus lives at the North Pole
a. Someone lives at the North Pole

Whether one accepts this inference as valid will depend
on a number of issues with which we need not be concerned.
The example is from Philosophical Studies, H. S. Leonard,
"The Logic of Existence", 7:52, June, 1956.

132

Eighth,we have been fortunate in finding a def-
inition of"self-vitiating' which is serviceable in the
characterization of the predicate 'self—vitiatable'.

This move is not always open to us in attempting to show
the relationship which obtains between a specific manifest
predicate and its associate disposition predicate. When
we are fortunate enough to discover this relationship,

we are then able to see more clearly the direction we
must take in defining the disposition predicate.

Ninth, by making sign-events (rather than tokens
or statements) values of our variables we are better able
to deal with those situations in which there exists a
relevant "time lag" between the production of tokens and

the receipt of the tokens.

The Problem of Disposition Predicates

We shall assume that it is not required that we
recount the history of attempts to deal with the problem
of defining disposition predicates. we need only note
that the material conditional and bilateral reduction
sentence are as inadequate to our particular task as
they are to the problem of stating the criteria“ for
defining disposition terms generally. Since no general
solution to the problem of defining disposition terms has
yet been found, we have been forced to make use of the
counterfactual form of statement in order to characterize

the predicate 'self—vitiatable'.

133

But the counterfactual form itself poses some
difficult problems. Several of these problems as they
are related to the counterfactual we are here considering
may be indicated.6

Part of our particular problem lies in the fact
that the antecedent and consequent are both false in the
counterfactual characterization of 'self—vitiatable'.

But so are the antecedent and consequent of the following
counterfactual:

If h_were made of blue cheese, then h_would be

self-vitiating.

Now in the event that we wish to accept our char—
acterization of 'self—vitiatability' we shall have to do
so on grounds other than those of the truth values of the
components since the latter counterfactual also has both
a false antecedent and a false consequent: and presumably
we shall wish to reject the latter counterfactual as
false while accepting the former as true.7 To simplify
our problem, we shall assume the existence of p_and hence,

the specific counterfactual we shall consider (now no

 

6Throughout this portion of the discussion we
shall assume that we understand the nature of the "suit-
able circumstances, Q? indicated in the characterization.
Needless to say we do not know what these circumstances
are. And one of the tasks for an empirical theory of
self-vitiation would be to discover the nature of these
circumstances. The point is that the statement, describing
whatever conditions constitute suitable circumstances,
need not contain dispositionals. More of this below.

7We are here assuming, of course, that our theory
of self-vitiation will show being made of blue cheese to
be unconnected with self-vitiation.

134
longer an “indicative“ counterfactual,) is the following:

If h_were uttered under suitable circumstances,

Q; then p_would be self-vitiating.

When we assert of any utterance p_that it is self—
vitiatable what we intend by this assertion is that if
h_were uttered under suitable circumstances, 9, then h_
would be self-vitiating.8

This characterization does not enable us in general
to distinguish between what we would take to be a true
counterfactual in constrast to a false one such as the one
containing 'h_is made of blue cheese' as antecedent.9

Now the problem of determining the criteria for
discrimination between true counterfactuals and false
ones is not “a fussy little grammatical exercise."lo
expecially in the case we are considering. A major
difficulty concerning the predicate 'self—vitiatable'
lies in the fact that it appears at present in no
recognized and well-confirmed body of theory. The sugges-

tion, that a true counterfactual is one which rests upon

 

8It should be remarked that although our pre-
systematic characterization of the phenomena which is
described as self-vitiating involved the use of a counter-
factual, our definition of 'self—vitiating' used only
indicative assertions.

 

9
As Goodman has shown. See Fact, Fiction and
Forecast, Chapter 1.

0
Goodman, pp, cit., p. 13.

135
a well-confirmed law-like connection between its ante-
cedent and consequent, will here not help us in the least.
For the fact is that there is available no set of law
statements at present from which, in regard to any partic-
ular h, we can deduce the conclusion '§_is self-vitiating'.

These assertions, about the state of our current

empirical knowledge, are made by way of showing why it

is not possible, at present at least, to clearly specify
the criteria for application of the term 'self—vitiatable'.
Rather than defining 'self—vitiatable' we must content
ourselves with showing what steps must be taken in order

to arrive at the kind of definition we are seeking.
Fortunately, we do not need to wait, as Goodman has shown,11
for a resolution of the general problems of disposition
terms before undertaking the task of defining specific
disposition predicates. The task before us is illuminated
by Goodman's example of the disposition term 'flexible'.
He writes:

In dealing with a particular disposition, say
flexibility, we may start with such predicates as
'bends' and '(is) under suitable pressure'. If
both apply at one time, then the predicate 'bends
under suitable pressure' applies; while if 'under
suitable pressure' applies when 'bends' does not,
then the predicate 'fails to bend under suitable
pressure' applies.... We may also hereafter ab—
breviate 'bends under suitable pressure' as 'flexes'
and 'fails to bend under suitable pressure' as

'fails to flex'.

Now 'flexes' and 'fails to flex' are mutually
exclusive, and together they exhaust the realm of

 

11
92, cit., pp. 44ff.

136

things that are under suitable pressure; but
neither applies to anything outside that realm.
Thus from the fact that 'flexes' does not apply
to a thing, we cannot in general infer that 'fails
to flex' does apply. Within the realm of things
under suitable pressure, however, the two predicates
not only effect a dichotomy but coincide exactly
with 'flexible' and 'inflexible'. What the dis-
positional predicates do is, so to speak, to
project this dichotomy to a wider or even to the
universal class of things; and a predicate like
'flexible' may thus be regarded as an expansion
or projection of a predicate like 'flexes'. ...

we can define 'flexible' if we find an auxiliary
manifest predicate that is suitably related to
'flexes' through "causal" principles or laws. The
problem of dispositions is to define the nature of
the connection involved here: the problem of
characterizing a relation such that if the initial
manifest predicate 'Qf stands in this relation
to another manifest predicate 'Afl then '5} may be
equated with the dispositional counterpart --
'Qrable' or 'Qd' -- of the predicate 'Q}.12
Goodman's treatment of 'flexes' offers us a clear
method of proceeding. The two predicates with which we
are concerned are 'uttered under suitable circumstances'
and '(is) self-vitiating'. If both of these apply to
a given utterance, then the predicate 'is self—vitiating
when uttered under suitable circumstances' also applies.
And in a case in which 'uttered under suitable circum-
stances' applies and 'is self-vitiating' does not apply,

then likewise 'is not self-vitiating under suitable cir-

cumstances' applies.

The problem of defining 'self—vitiatable' is the

problem of finding an auxiliary manifest predicate that

 

121bid., pp. 48—49.

137
is suitably related to 'is self-vitiating when uttered
under suitable circumstances' through causal laws. One

such law might take the following form:

(mum . Bh.Dh:D. Cthh)

where 'A}, 'g}, and '2] stand for certain specific
characteristics of p_and 'g} stands for the predicate

'is uttered under suitable circumstances'. Provided
this statement is unviolated, supported, and uneXhausted,
we have the means for defining the disposition predicate

'self-vitiatable'.13 The definition would read as follows:

V*h =df. Ah . Bh . Dh

where [y*' stands for 'is self-vitiatable'. Now provided
that 'Af, 'g}, and '2] are themselves manifest predicates
(or can be defined in terms of manifest predicates
according to the method suggested by Goodman) we are then
able to define 'self-vitiatable' without employing a
dispositional and without reference to any such trouble-

some words as 'possible'or 'potentially'.

 

13The confirmation of such a statement raises
other problems in confirmation theory; e.g. what con-
stitutes violation and what constitutes support of a
particular generalization? Although these problems
would need to be considered in the development of any
theory of self-vitiation they are not problems peculiar
to our subject matter. we discussed certain other problems
in confirmation theory in Chapters III and IV; (see also
Goodman, Ibid., Passim.).

138

We have also eliminated the need for a counter-
factual characterization of the phenomena of self—
vitiatability. And the two predicates 'self-vitiating'
and 'self—vitiatable' enable us to say what is cognitively
significant about the phenomena in question; that is to
say, we can find confirming evidence for the assertion
'h_is self-vitiatable' by determining whether or not the
assertion 'Ah . Bh . Dh' is true.

we have remarked that the predicates apply only
to actual utterances of H and not to possible ones. But
there are surely occasions on which we would wish to
speak of ppssible members of H;having the characteristic
of self-vitiatability. Suppose, for example, that H1
has only one member, hi, but this member does not have
characteristics A, g, and 2, Now we may feel that we
are justified in an assertion such as the following:

If there were a member of H h , such that h

had characteristics A, g, Ei’and Q, then pa ’2

would be self-vitiating.

The justification for such an assertion would be
its being supported by the law: (h)(Ah . Bh . Dh : :).
Ch :)Vh). But it may be noticed, the above counter-
factual speaks of only possible members of H1 and not
actual members 0f.§l- And hence our definitions of
'self—vitiating' and 'self-vitiatable' would appear to
be inadequate; for they do not allow us to speak of

possible members of HLbut only actual members.

139
While the actual solution to this particular

problem would undoubtedly require vast expenditures of
time and energy in empirical research, the outline of
the solution is again relatively clear from the fore-
going analysis. The difference lies in the nature of
the disposition in question. This can be made clear by
rephrasing the counterfactual:

There is a person p_such that p_has the dis-
position to utter self-vitiating replicas of

hi (Sp) .

Here the disposition term in question is not a
predicate of possible, non-actual utterances but rather
a predicate of actual persons. What we are suggesting
is that the same method of analysis which has been used
for the disposition terms 'flexible' and 'self—vitiatable'
would be adequate for an analysis of the term 'has the
disposition to utter self-vitiating replicas of 21' and
also the more general term 'has the disposition to utter
self-vitiating hypotheses'. What would be required is
the specification 0f certain characteristics of persons
(or perhaps machines, computers, etc.) which, when
present under suitable circumstances, lead by law to
their uttering in those circumstances, self-vitiating
hypotheses. And this empirical task would presumably be
one for the psychologist or sociologist.

An extension of this method could take care of

cases such as that in which we wish to speak of possible

140
persons uttering self-vitiating hypotheses. This case
could be dealt with by specifying the conditions under
which actual persons would have procreated in such a
fashion as to produce such a possible person. Needless
to say such an extension of method would be considerably
difficult and complicated. But our point is simply that
the method offers no insuperable theoretical obstacles
to saying all that we wish to say using only manifest
predicates which have no hypostatized entities in their
extensions.

Up to this point we have been dealing with two
classes of utterances: those which actually are self-
vitiating and those which although not self-vitiating
are self-vitiatable. Now there are a number of utterances
which seem to "hover" between being self-vitiating and
being self-vitiatable though not self-vitiating. An
example may make this clear. Suppose that a scientist
makes a fore-diction during 1962 concerning an event, g,
which is supposed to occur during the year 2000 A.D. The
fore-diction is not published at the time that it is
made but is discovered in the effects of the scientist
when he dies in the year 1990. The fore-diction is then
published in 1990. Another scientist observes that the
publication of these replicas of the sentential parts of
the original fore—diction has the characteristics A, A,
and Q, The question then is: "Shall the initial utterance

of the fore-diction be called 'self—vitiating' or

141
'self—vitiatable'?" The answer to this question will
depend on whether or not the predicate '9] is applicable
to the utterance. But here we note that a certain am-
biguity occurs in our usage of the term 'utterance'. We
have not clearly specified whether the sign-event which
we have called 'an utterance' is to be construed as

simply the ppoduction of tokens or whether it is to be

 

construed as the production and reception of tokens.

Let us, in the former case, speak of the utterance
as being incomplete and, in the latter case, of a complete
utterance. In an incomplete utterance the tokens have
not been, so to speak, received. If the tokens have been
received (either read, heard, felt, etc.) then the utterance
will be called complete.14 We suggest that the predicate
'self—vitiating' be applied only to complete utterances.
We are assuming that prior to 1990 the fore-diction
mentioned above was part of an incomplete utterance.

But the question of its eventual completeness is what is

 

14'Leonard draws the distinction between undirected
and directed sign-events on the basis of there being an
intended second party or being no such intended second
party. The distinction is inadequate for our purposes.
Some sign-events, although having an intended second party
will be incomplete in our sense (e.g. letters addressed
incorrectly and eventually winding up in the dead letter
office of the U.S. Post Office.). The question of eventual
completeness is what is important for our analysis and
not the quite different question as to whether the pro-
ductive sign-event has an intended second party. Similar
remarks obtain in regard to receptive sign—events. By a
complete utterance we mean one in which some second party
actually receives the tokens whether this second party is
the producer or not. The only events then which are to
be classed as incomplete are those in which the tokens

are not received. See Leonard, Principles p£_Right Reason,
pp. 89-107.

 

142
at issue here. And this issue presumably can, theoretically
at least, be resolved by law. In fact the only time in
which incomplete utterances are of any interest to us
would be in those cases such as the fore-diction mentioned.
For once the tokens are read or heard or received in some
other manner, the utterance is, by definition, complete.
And once it has been shown to be complete, whether or not
it is self-vitiating will depend upon the causal conse-
quences of its having been completely uttered, that is,
produced Apg.received.

Thus to call such an utterance 'self—vitiating'
is to say that the utterance either did have, does now
have, or will have a self-vitiating consequence. In
other words, 'self—vitiating' is a predicate which is as
tenseless in its application as 'is true'.

Interestingly enough, the predicates 'self-
vitiating' and 'self—vitiatable' differ in some important
respects from predicates such as 'is magnetic'. For ex-
ample, of a particular iron bar we may say that it is
magnetic or that it attracts iron filings at time p, we
may also say that while it was magnetic it is magnetic

no longer.15 But it will not make sense to say that

 

15One of the principle reasons for the differ-

ence between 'is magnetic' and 'is self-vitiatable' is
that in regard to the former we have no special term which
is its manifest correlate or counterpart. So we use the
same term for both the manifest property and the dis-
position. Our point, however, has to do only with the
dispositional use of the term 'is magnetic'.

143
while p_was self-vitiatable, it is no longer self-
vitiatable, and especially so when p_is a generalization.
One of the characteristics of the sentential parts of an
utterance of a generalization is that replicas of the
generalization may be freely reproduced. Hence the
initial utterance will enter into a causal sequence such
that many ensuing utterances may become instantiations
of some 'xi' in our definition. And each replica, pro-
vided it has the characteristics A, A, and 2, will also
be construed as self-vitiatable.

In the case of singular statements certain of the
replicas will presumably not be construed as self-
vitiatable; for example, those utterances produced after
~ghas occurred. But law-like generalizations will not
be so restricted in regards to their replicas.l6

In regard to self-vitiating hypotheses we may
cogently apply the predicate to those which have proved
to be self-vitiating or to those which according to the
laws exemplified in the expressions 'Cxixj', we may
predict will cause "g, This is not to suggest that our

predictions may not sometimes turn out unsuccessful.

 

Non-law-like generalizations may also be so
restricted both in regard to 'self—vitiatable' and in
regard to 'self-fulfillable'. Prima facie, the restric-
tion seems to be dependent upon the relationship of p_to
2, Where the occurrence of p_may be said to “completely"
confirm p_then if p_is a generalization it is non-law~
like. After §_has occurred, any future utterance ofgg
will presumably not be causally efficacious in bringing
about g, See Chapter IV.

144
Any initial theory of self-vitiation, we may expect,
will need to be continually revised prior to its being
able to give us a preponderance of successful predictions.
The point is that we will be justified in applying the
predicate 'self-vitiating' to utterances both which have
been "proven" to bring about ”p_in the sense explicated,
or which according to well-confirmed laws may be pre-

~

dicted to bring about g, The point here is one which
is directly analogous to the Hempel-Oppenheim paradigm
of scientific explanation. We do not distinguish logic-
ally between explanation and prediction. In like manner,
when we say 9A_is self-vitiating" we are not saying
whether the event 7§_is past, present, or future. The
sense of our claim that p.13 self-vitiating is that its
occurrence leads by law to "p,

we need also to make clear some of the relation-
ships which obtain between the predicates 'self-vitiating'
and 'self—vitiatable'. If p_is self-vitiating, then it
will also be self-vitiatable. And should a particular
p_prove to be self-vitiating without at the same time
being self-vitiatable, then presumably something is wrong
with the choice of the set of characteristics, A, A, and
2, However, what would presumably have happened in such
a case is that p_failed to exhibit one or more of these
characteristics although it actually is self-vitiating.
What may then be needed is a revision of the set of

characteristics Which have been used as the defining set

for the term 'self-vitiatable'.

145
It is also clear that not every utterance which
is self-vitiatable will be self-vitiating. The differ-
ance Lies in the fact that while a particular utterance
p_may have characteristics A, A, and p_it will fail to
eXhibit characteristic g_(i.e. being uttered under
suitable circumstances) and hence will not, according to

well-confirmed law-statements, be self—vitiating.

Pervasiveness of Self-vitiation and Self-vitiatability

The claim that the phenomena of self-vitiation
is one which enables us to distinguish between the social
sciences and the non-social or physical sciences is
apparently untenable. What we have tried to show is that
the phenomena in question are pervasive of all the sciences
and are not limited to any one set of sciences. There
are important distinctions yet to be made; but the dis-
tinction here in question is not one of them.

The phenomena of self-vitiation may be treated as
a sub-class of two other classes of phenomena. First,
it may be treated as a sub-class of what are sometimes
called "perturbance effects". Second, it may be treated
as a sub-class of phenomena such that predictions of
their occurrence sometimes fail due to the fact that
certain auxiliary assumptions regarding boundary con—
ditions have not been made explicit. Certain phenomena
in the physical sciences exhibit effects similar to those

of the phenomena of self-vitiation. It is a well-known

146
fact that the measurement of the position and velocity
of atomic particles creates a state of affairs such that
the simultaneous measurement of both position and velocity
does not seem to be possible according to present measuring
techniques and theoretical physics. The actual obser-
vations themselves cause the particles to react in ways
in which they would not react if they were not being
observed.

.A similar phenomenon occurs in social science. A
sociologist may, in observing the behavior of a small
group, 9b unwittingly introduce a disturbing factor into
the group, and the members of §_may then act in a manner
different from that in which they would act if they were
not being observed.

These two examples tend to ShOW'that the kinds
of disturbance one encounters in the physical sciences
and in the social sciences in measuring certain features
of the world do not eXhibit differences which prima
facie enable us to make relevant distinctions between the
two kinds of sciences. we are contending that self-
vitiation can be treated as a specific kind of problem
whose resolution may in some respects be hastened by the
development of a general theory (in the sense of a
systematic philosophical analysis) of perturbance.
Whether or not the contention is justified is a question
whose answer will need to wait upon the development of
such a general theory. But there do not appear to be

reasons why such would not be the case.

147

In regard to the second point, self-vitiation may
also be treated as-a sub-class of those phenomena such
that the predictions of their occurrence fail to be
successful due to a failure of the prediction to include
certain auxiliary assumptions regarding boundary con-
ditions. Self-vitiating hypotheses also exhibit this
particular characteristic. But whereas in the case of
physics we are frequently able to specify the unaccounted—
for boundary conditions after the failure of the pre-
diction, in the case of self-vitiation in the social
sciences we are by no means so fortunate. Again, what
is required is a scientific theory of self-vitiation
which will enable us in each case to specify the value
of each 'xi' in regard to a particular utterance. But
this distinction is irrelevant to our thesis. It amounts
simply to saying that in the one case (that of physics)
we have a more adequate theory than we do in the other
case (that of psychology and sociology). It is the same
as saying that the task of constructing a theory of self-
vitiation remains. And this is precisely what we have
contended throughout the course of this essay.

On the other hand, we need remark that the ex-
tension of the term 'self—vitiatable' is not so broad
as to be otiose. At least two major points may be noted.
First there are a large number of fore-dictions in
celestial mechanics which would not be construed as self-
vitiatable due to laws of relativity physics as presently

understood. A scientist §_may utter a fore-diction h_

148
at time p_to the effect that an event p_will occur in
the vicinity of the constellation Andromeda at time En-
If we assume that the time required for light to travel
from §_to Andromeda is greater than p, then A,will not,
according to any adequate theory of self-vitiation, be
construed as self-vitiatable. This is due to the fact
that presumably the speed of light is a limiting velocity
in the universe; and no matter how the tokens of p_were
transmitted they could not arrive in time to have a causal
effect on 2,

Second, it seems quite likely that an adequate
theory of self-vitiation will include as defining charac-
teristics of 'self—vitiatable utterance' a set of charac—
teristics such that they apply to only a small percentage
of actual utterances. We may note presystematically that,
in regard to our example of disrupting the orbit of Mars,
no actual utterances concerning Mars' behavior would be
construed as self-vitiatable. In the social sciences
presumably a much larger class of utterances would be
self-vitiatable. But even in the social sciences we may
expect the predicate to apply to only a small percentage
of actual utterances. The actual extension of the term,
however, will in part depend on the conventions employed
for defining the term 'self-vitiatable'. On page 137
we suggested that a law of self-vitiation might take the

form:

(h)(Ah .Bh.Dh::). ChDVh)

149
And from this analysis we suggested that 'self-
vitiatable' could be defined in terms of the predicates
'Af, 'Af and '2]. Now we may easily see that the above
statement is logically equivalent to

(h)(Ah.Bh:3 Dh.D.ChDVh)

And if the law were formulated in this latter form,
'self—vitiatable' could be defined in terms of the predicates
'Af and 'p}. And since the inverse relation between exten-
sion and intension would presumably hold in such a case,
the extension of the alternative definition of 'self—
vitiatable' would be greater than would the extension of
the former. ‘We are not claiming that such would necessarily
be the case. We only use this example to show that the
choice of predicates for the definiens of 'self—vitiatable'
could vary according to the purposes of the definer. And
as the purposes varied, so also could very the extension of
the term 'self—vitiatable'. At any rate it seems clear that
the extension of the term would not necessarily be so broad
as to apply to every utterance or even to a major proportion

of all utterances.

Goal_Qirected Phenomena and Self-vitiation

We have now reached the point in our analysis
where we can make distinctions which may prove useful in
the development of an adequate theory of self-vitiation.
The first of these distinctions has to do with goal-

directed phenomena.l7 Certain physical or biological

 

17The treatment of oal-dire ted henomenaijs
dependent upon Nagel's anal $13 in T e St ucture 0

Science, pp. 401-428.

150
systems "exhibit in varying degrees adaptive and reg-
ulative structures and activities" which enable them to
preserve themselves in some relatively steady state or
process of development in the face of environmental
changes or changes in the behavior of one or more of the
parts of the system. Thus, for example, the human body
tends to preserve an internal temperature which ranges I
between 970 F. and 990 F. We may call such states "goal-

states" or 'G-states' or 'property g}.

Let §_be some system, §_its external environ-
ment, and §_some state, property, or mode of be-
havior that §_possesses or is capable of possessing
under suitable conditions. Assume for the moment
(this assumption will eventually be relaxed) that
§_remains constant in all relevant respects, so
that its influence upon the occurrence of §_in §_
may be ignored. Suppose also that §_is analyzable
into a structure of parts or processes, such that
the activities of a certain number (possibly all)
of them are causally relevant for the occurrence
of g, For the sake of simplicity, assume that
there are just three parts, each capable of being
in one of several distinct conditions or states.
The state of each part at any given time will be
represented by the predicates 'Axf, '_yf, and
'95}. respectively, with numerical values of
the subscripts to indicate the different partic-
ular states of the corresponding parts. Accord-
ingly, 'Axf, 'Ayf, and '95] are state variables,
though they are not necessarily numerical vari-
ables since numerical measures may not be available
for representing the states of the parts; and the
state of S that is causally relevant to G at any
given time will thus be expressed by a speciali-
zation of the matrix (AxByCz) ...

One further important general assumption must
also be made explicit. Each of the state variables
can be assigned any particular "value" to charac-
terize a state, provided the value is compatible

151
with the known character of the part of §_whose
state the variable represents.18

We assume that §_is a deterministic system; and

also that the value of each state variable at any given
instant is independent of the value of each of the

other state variables §p_that same instant. Only certain
values of the state variables will presumably be causally
effective in bringing about a G-state in g, The parts

of §, however, are so related that a change in one of

the state variables causes a change in the other state
variables. Thus if a change in §_occurs such that the
value of Ax_deviates to the point that §_is no longer in
a G-state, §_is so organized that the valuesof Ay_and
g§_change to the point that §_returns at a later moment
to a G-state.

The prima facie distinctive character of so-
called "goal-directed" or teleological systems is
thus formulated by the stated conditions for a
directively organized system... What may be called
the "degreeof directive organization“ of a system,
or perhaps the "degree of persistence" of some
trait of a system, can also be made explicit in
terms of the above analysis.19

By relaxing the restrictions on §.and allowing it

to vary we simply add additional variables to the

 

18Nagel, 9p. Cit., pp. 411-412.

19Ibid., p. 417.

152
analysis and now consider a new system g} which is com-
posed of §_and g,

Our summary of Nagel's analysis of goal-directed
systems outlined above will enable us to mark a distinc-
tion between two classes of utterances: those involving
goal-directed receivers and those ppp_involving goal-
directed receivers. we believe that it is the former
class which is of particular interest to us rather than
the latter. The latter class includes such "far-fetched"
examples as those we recorded earlier regarding icebergs,
ice-cubes, and exploding cobalt-bombs. (Any uneasiness
we may have felt in regard to including these examples
in the extension of the term 'self—vitiating' may be
accounted for by noting that we had not yet distinguished
between those utterances involving goal-directed receivers
and those which do not.) Henceforth in our discussion
we shall limit ourselves to a consideration of only
those utterances involving goal-directed receivers.20

Using the symbols above we may now attempt to
analyze in some detail the process of self-vitiation.

Let p_be an utterer of an hypothesis A, p_be the receiver
of A, and §_be the system composed of p, ;, Ajsp) and
any other relevant factors in the environment of p, p,

and pjsp). We assume that p_is a goal-directed system

 

20Hence we do not feel obliged to coin a neo-
logism to distinguish the two classes here considered.

153
and that Ajsp) constitutes a fore-diction that p,
Further, we assume that §.constitutes a G-state of 5,
Now in the event that p_should turn out to be self-
vitiating, our problem is that of explaining this fact.
An explanation would consist in showing hOW’E_in uttering
Ajsp) causes certain changes in one or more of the state
variables of 5, The result of these changes in the values
of the state variables is that 72 occurs. In other words
p_is taken out of a G-state.

Now the analysis becomesmore complicated when we
note that systems such as p_(e.g. if p_were a human being)
may be considered as exhibiting "self—regulatory behavior
with respect to several §]s at the same time, alternative
(and even incompatible) gfs at different times, a set of
Qfs constituting a hierarchy on the basis of some postu-
lated scale of 'relative importance' or more generally
a set of gfs whose membership changes with time and
circumstance“.21 Now the difficult cases for a theory
of self-vitiation would be those in which two incompatible

gfs appear to be involved §p_the same time. Suppose that

 

in the above example Ajsp) constitutes a fore-diction
that p_will be in some G-state at time 3, Let us call
this G—state '§1'- Let us suppose also that fg_is a
necessary condition for the occurrence of another G-state

in p_at 3, Let us call this G-state 'QQ'. Now it is

 

21Nagel, pp, cit., p. 416.

154

obvious that it is not possible for p_to be in both g1
and g2 at 3, Hence reference to the original p_which
predicts that p_will be in g; must be inadequate as an
explanation or prediction of the phenomenon in question.

We must now take note of a fundamental difficulty
with our characterization of what shall constitute a
self-vitiating hypothesis. The difficulty involves
requiring that A.must be deducible according to the
requirements outlined in Chapter I, i.e. from a set of
p£p§_statements. What the foregoing shows is that this
requirement must be relaxed to read:

p_must be deducible from a set of statements

which we would be warranted in believing to

be true prior to the determination of Afs
being self-vitiating.

22 unless we make this

As we pointed out earlier,
modification the extension of the term 'self—vitiating
hypothesis' will be null.

We can now see that the theory on which the
deduction of the claim that p_will exhibit g1 at p_is
faulty and that it is the self-vitiating character of A_
which exhibits the faultiness of the claim. In Chapter VI

we shall examine some possible ways of deciding between

two such incompatible predictions (e.g. between p_and 'Vh').

 

22See footnote, page 64.

155
At this point we wish merely to clarify some of the
difficulty involved in the event that a system 3 may
exhibit incompatible G-states.

Nagel suggests a solution. We need to develop in
the case of p_and all like systems a list of "gfs con-
stituting a hierarchy on the basis of some postulated
scale of 'relative importance'." In the case of our
example the failure of p_may be traced to the fact that
some such scale of importance regarding the relation of
g; to g2 in p_was not considered by p_in the deduction
of A, It may be expected that the more complex the
system involved and the greater number of G-states to be
considered the greater would be the difficulty in
specifying the "relative importance" of the various gfs.
Again, we are not minimizing the magnitude of this
empirical task but simply attempting to show that there
are not theoretical obstacles in the way of developing

and adequate theory of self-vitiation.

Sign-behavior and Self-vitiation

The above analysis effectively discriminates
between the kinds of self-vitiation involved in the Mars'
example and those involving goal-directed behavior.
There is, however, another equally important distinction
to be drawn. This distinction involves those goal-
directed systems which exhibit what we may call "sign-
behavior" and those which do not. It is perfectly

possible that a self-vitiating utterance may involve a

156
receiver such that the receiver may be understood as a
goal-directed system and yet not be one which exhibits
sign-behavior.23 The interesting cases involve not only
utterance reception (as thus far construed) but gggpf
interpretation. It is this latter feature which creates
the most puzzling (from a presystematic standpoint) and
cognitively important instances of self-vitiation. In
this section we shall not undertake to give anything like
an adequate account of the phenomena of sign-interpre-
tation but shall limit our discussion to what we consider
to be the points most relevant to our problem.

To begin we may return to our previous character-
ization of what constitutes a replica. we introduced
the notion of 'replica' in Chapter III as follows: Wx_is
a replica of y_if and only if x_and y_share certain
specifiable physical characteristics." This character-
ization enabled us to explicate the concept of a self-
vitiating utterance without having to take account of
all the problems at once. Now we must note that this
particular characterization is inadequate to our task.
This inadequacy is due to the fact that two spoken ut-
terances may indeed contain sentential parts which we
would wish to include in a single p_or ALfrom a pre-

systematic standpoint. Yet the physical characteristics

 

23See our example on p. 108, Chapter IV.

157
of these sentential parts might, from the standpoint of
phonemic analysis, be quite unlike. For example, John
F. Kennedy will, in his ideolect, use a quite different
set of phonemes for the production of the word 'idea'
than will Dwight D. Eisenhower. And the set of phonemes
thus produced by Kennedy will probably share, in
Eisenhower's ideolect,24 more physical characteristics
with 'I, Dear' than with 'idea'. This example serves to
show that, while the notion of replicahood as so far con-
strued is adequate perhaps for the written language, it
is quite inadequate for the spoken language. And pre—
sumably we shall wish to include spoken as well as written
utterances in a specific A, Due to the fact that great
varieties of phonemic differences occur in different
dialects and different ideolects, some means must be
available for determining when a particular verbal utter-
ance shall be included in a particular A,

Two solutions to this problem are immediately
apparent for our purposes. The first is to specify all
the relevant semantical and syntactical rules of the
language. These would enable us to determine not only

when a pair of utterances Al and A2 are describing the

 

24The term 'ideolect' is taken from Paul Ziff,
Semantic Analysis, (Ithaca, New York: Cornell university
Press, c.d. 1960) who in turn has adopted the usage of
U.‘Weinreich, "Is a Structural Dialectology Possible?".
Ainguistics Today, p. 269. It refers to the character-
istic features of a language as that language is spoken
by one particular person.

158
same event Eb but also when they are or are not well-
formed expressions. The determination of which utter-
ances shall be members of a specific Aion the basis of
semantical and syntactical rules would undoubtedly prove
to be an exceedingly complicated and, in most cases,
difficult task. A second solution may prove to be more
easily manageable. This solution is the one we indicated
in Chapter III. It involves the introduction of the
notion of an ideal observer. An ideal observer would
presumably (for our purposes) have the following (among
other) characteristics: a) be able to read and write
the language A; b) have a "trained ear" so as to be able
to "understand“ the various ideolects and dialects of A;
and c) be absolutely fastidious in spelling and punctu-
ation. We can now deal with the problem indicated above
on the basis of our second solution: if Al(sp) is transcibed
by an ideal Observer p_as A3(sp) and A2(sp) is translated,
by the same (or possibly different) Observer, as A4(sp),
and.AB(sp) and A4(sp) are replicas of each other; then
Al(sp) is a replica of A2(sp) and conversely.

Our test does not eliminate the need for semantical
and syntactical rules in determining whether a pair of
utterances are members of some particular A, Our seman-
tical rules will still be required in determining whether
or not a specific A_describes 2} and syntactical rules
will be required in determining whether or not a specific

A_is well-formed. But our test does eliminate the need

159
for these rules in determining whether or not verbal
utterances are to be construed as replicas of one
another.
It may be Objected that our test is highly
inadequate inasmuch as what will usually be involved
is not a comparison of Al and A2 (when these are spoken

utterances) but a comparison of what is remembered as

 

Al and A2 or else a comparison of a new set of utter-
ances A5 and A6 which are presumed to be replicas of Al
and.A2 respectively. We seldom have ideal observers
recording all those utterances or even a small percentage
of those utterances which we would wish to consider as
instances of self-vitiation. In most instances we shall
be dependent upon someone's memory as to what was said
and what was meant in regard to a specific utterance.

We may readily admit that in regard to spoken
utterances this factor presents a formidable problem.
But it is one which would attend any investigation of
self-vitiation whether one used semantical and syntactical
rules for determining the question of replicahood in any
specific case or whether one used the test which we have
proposed. Our test does not circumvent this particular

problem. But it does provide an effective means for

160

circumventing other problems in linguistic analysis and
hence simplifies the task at hand.25

The preceding considerations concerning phonesis
point up a difficulty considered earlier inthis essay.
In regard to utterances which are vague, ambiguous, or
mal-formed we decided to treat these as values of some
'Ed' in our definition rather than as members of A, We
must now consider ways of dealing with those utterances
such that the receivers do not understand or else mis-
understand what is signified by the sentential parts of
the utterances. These utterances are complete, but due
to ideolectical difference between producer and receiver
(or other factors) the receiver does not understand or
misunderstands what is indicated by the utterance. A
number of kinds of circumstances could bring about such
a state of affairs. An utterer A_could stutter, lisp,
or perhaps have suffered a tracheotomy; or he could speak
the language in a dialect which differed so greatly from
that of the receiver p_that p_would be unable to determine
what is indicated by the set of phonemes employed by 2,

Yet an ideal Observer p_presumably would be able to

determine that which is indicated by pfs tokens. On the

 

25The effectiveness of the test of course will
depend on the linguistic ability of the observer. We
have not done away with the need for the rules in question;
we have simply allowed ourselves to trust the judgment
of a presumed authority.

161
other hand such utterances may turn out to be self-

vitiating, even though the receiver p§_the tokens did

 

not understand what was indicated Ay the tokens.

The way in which we may deal with these utterances
(which to distinguish them from A} utterances we shall
call 'Af* utterances') will depend on certain other cir-
cumstances. Three cases may be distinguished: l) Af*(sp)
is a replica of Ajsp) and A_is uttered prior to Af*.

~

Further, A_is a cause of A?* as well as A, Provided

that Af* enters the causal sequence eventuating in g3

it may be treated as a value of an 'Ai'. In this case
the predicate 'self—vitiating' would apply to A_but not
to A?*. In this case we assume that the important dif-
ference between A_and Af* is that the former is under-
stood by the receiver of A whereas the latter is not
understood by the receiver of A?*. 2) In this case the
occurrence of A?* precedes that of A, and is a cause of
A, Provided that A_is a cause of ”g, Af* may be ignored
and the predicate 'self-vitiating' applied to A, 3) In
this case there is no A which is a replica of A}* and
which leads by law to ”g, Yet Af* does lead by law to
"g, In this case the predicate 'self—vitiating' would
be applied to Af*.

Now the third case, even though it involves goal-
directed receivers will be treated as a somewhat uninter-

esting case of self-vitiation. Inasmuch as the tokens

of A?* are not understood by the receivers of these

162
utterances (i.e. the Af* type as in 3), this class may
be dealt with in a manner similar to that of the raven
and Mars examples.

The cases such as those in 3) do, however, point
up the characteristic feature of the kind of self-
vitiating hypothesis which is cognitively important in-
asmuch as this feature is lacking in utterances such as
those considered in 3). This feature may be described
as follows: the receiver £,of_tokens Ajsp) produced by
p_understands the semantic relations which obtain between
Ajsp) and the state of affairs which would be the case
if A_were true. The term 'understand' used in the above
description may prove to be subject to the kind of am-
biguity discussed in relation to the term 'know' discussed
in Chapter II. But this problem need not detain us. We
may use the term 'understand' in the following $3186:
"A_will be said to understand the semantic relations
which hold between A.and that which would be the case if
A were true if and only if p_believes that the statement
'Fh a "pf is true. And further, the assertion '"Fh a fig"
is true' is true. Less formally, it is not only the case
that £.believes that certain semantic relations hold be-
tween A_and that which would be the case if A_were true,

but that such semantic relations _d_g £1. fact hoid.26

 

26we have once again been forced to resort to the
counterfactual form in order to speak of the semantic re-
lations concerning A, We are assuming, however, that the
same mode of analysis used above would be adequate for

163

Now it may appear that our classification of self-
vitiating hypotheses will enable us to discriminate be-
tween the social and non-social (or physical) sciences
due to the fact that we have introduced the notion of
'belief' as a characteristic of receivers of the class
of utterances here considered. Believing we usually take
to be a characteristic of sentient and rational creatures.
But the term 'belief' need not be so narrowly construed.
we may (paraphrasing Leonard) define 'x_believes y} as
'§_has a disposition to act as though y;were true'.27
And Nagel's example quoted in Chapter II shows that
certain mechanical devices may have such dispositions.
Applying the term 'belief' to machines may seem to strain
customary speech. But we are not primarily interested
in defending "ordinary language" although such a defense
may at times be warranted. And we may always avoid any
seemingly unconventional "connotation" which the term
'belief' has in such contexts by employing the expression
in the definiens of the definition of the term 'belief',
namely, 'A_has a disposition to act as though y were true'.

This digression should not Obscure the fact that

the belief in question is not belief in the truth of§A

(although in some instances this fact also may be

 

disposing of the problems encountered in this regard.
This would Obviate any need for using an expression 'p}
when 'p} does not denote.

27See Principles p£_Right Reason, p. 46.

I. .11.... IE1. :

164
important) but in the truth of the assertion that certain
semantic relations hold between A_and What would be the
case if A_were true. In a number of cases the receivers
may actually attempt to act in such a way that A turns
out to be false.28
The simplest instances of the class of self-

vitiating hypotheses here being considered are those in
which what A_purports to describe is such that a receiver
A_would be taken out of a G-state if what A_purports to
describe were in fact the case. If p_is a goal-directed
receiver then, provided other things are equal, we may
presume that p_will act in such a way as to attempt to

~

bring about e. We shall, of course, need to stipulate that

pfs attempts will have some probability of success, even
though this may in some cases be small. But in this case

of self-vitiation, we are here considering that what

occurs is a success by p_in bringing about "A,

 

23A comprehensive theory of self-vitiation would

need to consider certain empirical questions such as the
manner in which human beings learn to speak a particular
language A, and how they learn to discriminate between
various homonyms. Linguistic theory may be considered
to have a relation to a theory of self-vitiation somewhat
similar to that which obtains between a theory of physics
and chemistry on the one hand and biology on the other.
Though the biologist is dependent upon the findings of
physics and chemistry, it is not required that he become
a physicist or chemist in the pursUit of his special field.
In like manner, one involved in the development of a
theory of self-vitiation would not be required to solve
all the problems in linguistic theory before undertaking
his task, even though a solution to some of the problems
in linguistic theory will aid in the solution to problems
in a theory of self-vitiation.

165
More complicated situations may be dealt with
taking into account certain considerations elaborated
above. When more than one G-state is involved we may
need a scale of I'relative importance" before we can
determine whether or not p_will attempt (or to what extent

~

p_will attempt) to bring about A, If, for example the
occurrence of that described by A would take p_out of
ga, but efforts to bring about 7§_would take p_out of g2,
and_G__2 (according to some scale of "relative importance",)
is more important than 91: then presumably p_will not
(provided p_understands certain of the causal consequences
of his action and provided he acts in a rational manner)
attempt to prevent p_from occurring.29

In sum, we are suggesting that the most important
class of self-vitiating hypotheses are those which in-
volve goal-directed receivers at least one of whose G-
states would be displaced by another state (not a G-state)
if A were true.

Where a given A_involves a number of receivers it
may be the case that a large number of G-states may be
also involved. In some instances these may even be mut-

ually incompatible inasmuch as it may be physically "im-

possible'for receiver E:

1 to remain in g1 while receiver

Ed remains in pa.

 

29What is needed at this point is an empirical
theory of value; i.e. a theory which would enable us to
predict in any given instance how men actually choose
when confronted with a "conflict of values". We have
suggested elsewhere in this essay how normative questions
of value, i.e. how men ought to choose, are also important

166

All of the foregoing remarks lead us to make two
observations. First there are no theoretical obstacles
to developing a theory of self-vitiation. On the other
hand, the actual construction of such a theory will
depend on the availability of a vast array of evidence
and on the confirmation of many laws which are not yet
available to us. There are no insuperable obstacles to
the development of such a theory; and yet the Obstacles
which do remain are patently formidable.30 This con-
tention is underlined by Popper's arguments against
historicism and historicists; even though we cannot agree
with his arguments that knowledge of the future is
logically impossible, we can certainly agree that at
present we have very little in the way of acceptable
theory which would justify any present claims that we
can Agy;predict, with any probability of success, how
our fellows will behave in certain situations. The import
of our remarks is to the effect that without AA_adeguate
empirical theopy g£_self-vitiation the social scientist
has little chance for anything but meager success in

predicting what the morrow will bring. That the physical

 

in regard to the phenomena of self-vitiation. Normative
questions, however, are not at issue here.

30We may secure some idea of the magnitude of the
task if we attempt to discover all of the relevant G-states
of human beings and then construct a scale of relative
importance by which we could determine in a given set of
circumstances how any individual receiver of tokens would
act. It is belaboring the obvious to remark that a great
deal of work yet needs to be done.

1'. 1.1.1 Illll

167

scientist is not plagued by this problem to as great an
extent seems due to the fact that in most cases he is
not required to predict the behavior of goal-directed
receivers who can understand the fore—dictions which he
utters. But as more and more use is made of such things
as computers and electronic brains, we may expect that
the physicist also will become increasingly concerned

with this problem.

Summary
In this chapter we have attempted to explicate

the sense of the term 'self—vitiatable' and then to show
why it is not possible at present to clearly define this
particular disposition predicate. We then went on to
show how the cases of self-vitiation which are cognitively
significant are those involving goal-directed receivers
and especially those receivers which have the disposition
to interpret signs. Neither of the two distinctions we
considered, however, enable us to discriminate methodo-
logically between the social and physical sciences. And
we hope that this particular argument may be safely laid
to rest.

In the next chapter we shall re-consider some of
the arguments considered earlier (in Chapter II); we
shall attempt to show in a more precise manner than was
formerly possible that these arguments lack the cogency

usually attributed to them.

CHAPTER VI

RECAPITULATION

Introduction

In this chapter we wish to re-examine, in the
light of our last results, some of the arguments in
previous chapters (especially Chapter II), review the
difficulties they involved and suggest some solutions to
these difficulties. we shall turn our attention primarily

to the arguments put forward by Gewirth and Popper.

Apes Self-vitiation Make Knowledge of the Futurepgmpossible?
We recall that Popper claims that successful pre-

diction of the future is impossible due to the fact that

our knowledge of the future is dependent on knowledge of

future knowledge and that this is logically impossible.

NOW‘We wish to make his point even more forcefully before

proceeding to show that there are ways of circumventing

the difficulty it apparently presents. Let:

"gf stand for Al ... L conjoined to 91'... 9n
(where any pi is a general law-
statement characterizing self-
vitiating hypotheses and any Si
a statement describing the rel-
evant antecedent conditions con-

cerning a particular hypothesis
L)

Then according to some theory of self—vitiation:

G(tt‘.h)“’>“’"n

168

169
where 'g} is a second degree predicate which holds be-
tween the conjunction of law-statements and antecedent-
condition statements on the one hand and utterances on
the other.

In other words, when the law-statements are taken
in conjunction with certain statements describing ante-
cedent conditions the conjunction of these statements
logically implies the statement that A_is self-vitiating.
The fore-diction in question then is the statement [yA'.

Now suppose that a social scientist A_publishes a
fore-diction regarding a particular utterance A1 to the
effect that A1 is self—vitiating. Let us call this
utterance of A_at time p_‘Ak'. This utterance Ak contains
as a sentential part a replica of the tokens 'A_i is
self-vitiating'.

Suppose further that A_publishes his fore-diction
AR. We can significantly inquire whether his utterance
'Ai is self-vitiating' is itself self-vitiating. We may
now note some of the advantages of taking utterances
rather than tokens, or propositions, etc., as values of
our variables. For the kind of paradoxes which are
generated in such instances as the Russellian or the
Grelling do not arise in regard to the predicate 'self-
vitiating'. The values of the variables are not tokens
but utterances which may be characterized by employing

replicas of the tokens in question but need not Ag, we

 

can always avoid any self-referential feature by giving

 

170
names to these utterances which in no way resemble AA§_
sentential parts 2E the utterances ;A_guestion. Hence
we need no recourse to a theory of types in order to deal
adequately with the phenomena in question.

The question then as to whether or not Ak is self-
vitiating is one which may be answered by an appeal to
empirical evidence and to the laws of self-vitiation.

If Ak is self-vitiating then Ak is false; but this then
means that at least one of the Ai's or one of the gi's
is false. This is due to the following considerations:
if EAk then it is false that yAi; and if 'yAi' is false,
then 'G(T,hi)' is false; but then at least one of the
statements composing 1 must be false.

In such a case we would say that our theory of
self-vitiation is inadequate. A more Serious difficulty
arises, however, in the following circumstances. Suppose
that scientist §.on the basis of the §AA§_theory of self-
vitiation used by scientist A_together with other ante-
cedent condition statements predicts that the utterance
of scientist p_is self-vitiating. That is, he is asserting
that EAR. Now the difficulty is as follows: insofar as
two fore-dictions are related as described above, it will
never be logically possible for both of these assertions
to be true. For if 'yAi is true then by definition of
'2] 'yAk' is false since the sentential part of Ak is a
replica of 'yAi'. And if 'yAi is true then it cannot be

self-vitiating. If, on the other hand, 'th' is true then

171
it is false that yAi. For if 'yhi' is self-vitiating,
then it is false.

An example may make these formal points easier to
keep in mind.

1. The National Safety Council publishes a fore-
diction to the effect that four hundred Americans
will die on Ag, Let us call the individual composed
of these utterances 'Ao'.

2. Scientist A_publishes a fore-diction on the
basis of a theory of self-vitiation T_that A0 is
self-vitiating. Let us call the individual composed
of certain of these utterances Am. (We may note that
the sentential parts of these utterances will be of
the form 'on'.)

3. Scientist A_publishes a fore-diction on the
basis of the same theory T_that Am is self-vitiating.
Let us call the individual composed of these utter-
ances 'An'. The sentential parts of these utterances
will be symbolized 'yAm'.

Now since all the above fore—dictions have been
deduced from well confirmed law—statements and highly
evidenced antecedent condition statements we may inquire
whether we have grounds for choice between them. We
cannot hold all three of them to be true. While 1 and 3
can both be true, 2 and 3 cannot both be true nor can 1
and 2 both be true. Fortunately, there A£g_grounds for

choosing between them. One such set of grounds has to

do with the nature of the auxiliary assumptions and the

172

boundary conditions discussed in earlier chapters. On
this basis we would choose 2 in preference to l inasmuch
as the explanans of 2 contains assertions about relevant
boundary conditions not contained in 1. Likewise (and
this is where the presumed difficulty lies) we would
choose 3 in preference to 2 since 3 makes certain other
boundary conditions explicit while 2 does not include
statements concerning these toundary conditions although
it does contain statements not included in 1.

It becomes immediately obvious that this "schema"
enables us to produce a rather lengthy sequence. For

example:
4. Scientist A_predicts that th.

In this case we would choose 4 over 3 since it
takes into account boundary conditions not described in
3, namely the results of the publications of tokens
which are replicas of 'yAm'. What this boils down to is
the question of which of the fore-dictions in question
(i.e. l, 2, 3, or 4) is better evidenced. And here
important value considerations intrude themselves. We
need to be able to specify when a particular fore—diction
is better evidenced than another. And we would need this
criterion before we could choose between 1, 2, 3, and 4.
Also it seems intuitively clear that simply the addition
of statements describing other boundary conditions is not

sufficient grounds for choosing a higher number in the

173
sequence unless we are prepared to assert that each of
the statements describing the antecedent conditions in
A_is as well-evidenced as each of the statements in 2:1:

It may appear that the "schema" described above
will always lead us into an infinite regress and hence
any prediction of the future will never have any degree
of certainty. Two points are at issue here. First it
is not the case that such a schema leads to an infinite
regress. Predictions and publications of fore—dictions
are the kinds of things that are carried out in time.
And prior to the occurrence of any predicted event p_(or
"§), there will only be a finite amount of time available
for the filling out of the schema. Hence the number of
fore-dictions about fore—dictions ... about fore-dictions
will always be finite.

We must not suppose, however, that the last fore-
diction in the sequence will be supported by better and
more conclusive evidence than the others. Being the
last member in the sequence then is not a sufficient
condition for our accepting it over any of the prior
ones in the sequence.

Also, there might well be certain psychological
limitations involved although their exact nature might
be a difficult problem for a present-day psychologist to

resolve. The limitation may be illustrated as follows:

174

Publication of 1 would presumably have some causal
effect. Publication of 2 could have a much smaller or
less important effect.1 Publication of 3, since it might
presumably be of interest only to a small coterie of
social scientists would have little observable effect on
the population as a whole in the relevant time span. If
some such psychological limitation is indeed present
then it would seem that the number of predictions needed
in any actual sequence for a successful prediction would
be small. A general theory of self-vitiation presumably
would take cognizance of this fact and would be substan—
tiated by an examination of instances of self-vitiation
on a number of "levels".

The second point at issue concerns our degree of
certainty or confidence that a theory of self-vitiation
will give us nothing but successful predictions. If what
is required of a theory before it is accepted is that it
come attached with some kind of built-in guarantee that
it will give us only successful predictions about the
future, then obviously we will accept no theory whatso-
ever. To expect certainty is patently to misconstrue
the nature of science. Moreover, no theory at present
can take into consideration all the possible boundary

conditions without considering, e.g., every possible

 

1In any of the pertinent sensesof 'importance'
explicated by Nagel, pp, cit., pp. 582 ff.

175
configuration of atomic particles in the universe. Most
of these configurations, however, are not relevant, or,
at least, unlikely to be relevant or even to occur. What
a theory, therefore, needs to account for are the relevant
and probable boundary conditions. And a theory of self-
vitiation will need to do only this and be well-confirmed
to be acceptable. If there is any noticeable difference
between, say, a theory of physics and a theory of
psychology it would be that there are fewer relevant
boundary conditions in physics (at least as presently
understood) than in psychology which need to be taken in-
to account in any actual prediction. But what constitutes
a relevant boundary condition is, itself, not wholly
clear. And until an adequate definition of 'relevance'
is forthcoming, we cannot use relevant boundary conditions
as a criterion for discriminating between the social and
physical sciences.

At any rate it is clear that no theory offers us
certainty. Whether or not we accept a theory of physics
or a theory of self-vitiation will depend in large measure
on how well—evidenced the theory is and not on whether
it comes to us packaged with a label guaranteeing only
successful predictions. In short, a theory of self-
vitiation offers us only contingent knowledge of the
world. To wish -for more or to demand more of any theory

seems to be a desire for that which no empirical theory

176

of self-vitiation can provide. It is something like
requiring a round square and is based equally on a con-
fusion about the object.

With the foregoing discussion as background we
may now deal more fully with Popper's claim that knowledge
of future knowledge is impossible. ‘We have already seen
that it is tautological to assert that future knowledge
is not present knowledge and that insofar as Popper's
thesis comes to this it is surely true -- but, as surely,
quite trivially unportentious. However, interesting cases
would be posed by thosesituations in which a fore-diction
about future knowledge is actually self-vitiating.

Suppose scientist E predicts that in 2000 A.D.
and not before the scientific community will possess a
law 5, Here 'Af, we may assume, characterizes the law
rather than merely names it or states it.2

Whether or not the fore-diction of scientist §_
would be self-vitiating could be determined by an adequate
theory of self-vitiation. What would be involved here

is the question as to whether the characterization of the

 

2These are the only interesting cases we need to
consider. If '53 names the law, then the only thing which
the fore-diction is claiming is that there will be some
law which will have that name. But it doesn't say what
this law will be; i.e., it does not give the import of
the law. If, on the other hand, the fore-diction states
the law, then chances are that this case will fall under
those of the tautological variety: i.e., present knowledge
is not future knowledge and conversely. These comments
are, of course, subject to the qualifications made in
Chapter II.

177
law generated sufficient interest among scientists to
motivate them to attempt to formulate and confirm the
law characterized by '5} in advance of 2000 A.D.

If this fore-diction should turn out to be self-
vitiating, then a scientist g;presumably could predict
that the utterances would have such a consequence; more—
over, by taking more of the relevant boundary conditionss
into account p_could more accurately predict when the
scientific community would possess §L(in some suitable
sense of 'possess', e.g. have confirmed ALto a high
degree. See Chapter II, pp. 45-56).

Here we see that the same kind of sequence can
obtain which we noted in regard to predictions of fatal-
ities on Api, The major difference between the two
examples involves only the difference in the substance
of what is predicted. In the latter case (i.e. the one
concerning fatalities) the time isspecified and then the
character of the event which is to take place at that
time is predicted, whereas in the former case, the char-
acter of the event is specified and the time at which it
will occur is predicted. In this instance, actually two
predictions are being made simultaneously: 1) that the
event will actually occur; and 2) that it will occur at
such and such a time. But these differences should not
obscure the fact that the kind of explanation or prediction
which should prove adequate in the two cases will not

significantly differ. In both cases what particular

178
prediction will constitute the final one in the sequence
will be determined in part by the amount of time elapsing
between the initial prediction and the event predicted.
And whether we accept the final member in the sequence
or some other as the most likely candidate for success
will depend on how well-evidenced are the statements in
the prediction and how many of the relevant boundary

conditions are included in the explanans.

A Re—examination of the Arguments of Feuer and Gewirth
we recall that Gewirth and Feuer have made claims
similar to those of P0pper. Feuer claims that human
creativity makes "for the unpredictable aspect of his-
torical causation," and Gewirth has claimed that human
decisions "may enter the antecedent or between the ante—
cedent and consequent of laws". we have had occasion to
point out the inadequacies of their analyses and the con-
clusions which they have drawn from them. But the kind
of phenomena with which they are concerned is important
for an adequate empirical theory of self-vitiation. We
shall in this section make a case for their position as
strong as possible, attempt to show precisely where the
difficulty lies and then suggest some considerations
which would be involved in a solution to these difficulties.
Suppose we now have a theory of self-vitiation.
It is presumably adequate in the following senses: all

of its statements are well supported, unexhausted (when

[_L

h

i

‘ 5 . PL.

179
these are generalizations), and unviolated. Further,
it has proved highly successful in predicting when any
given utterance is self-vitiating. we assume also that
the theory is able to account for the self-vitiating
character of utterances on any of the "levels" of pre-
diction discussed earlier.

Now let us suppose that armed with our theory we
encounter a particularly obstreperous individual whom we
shall name 'gf. g;knows something about the logic of
scientific explanation. On the other hand QLintensely dis-
likes having us able to predict his behavior successfully.
He feels that such prediction constitutes an unwarranted
intrusion on his freedom. He therefore candidly tells us
that if we predict that he will act in a manner described
by an utterance A, then he will act in a manner described
by an utterance A, And if we predict that he will act in
a manner described by A.then he will act in a manner de—
scribed by A,3

If our theory is adequate, it will have taken such
individuals into account. What we presumably could deduce

from such a theory is that under these circumstances

 

3Only certain values of A and A (here treated as

variables) need to be considered. Utterances describing
certain physical "impossibilities" are excluded from our
consideration as uninteresting cases. For example if A
contains as sentential part a sentence which is a replica
of 'J will jump over the moon at time t without the aid
of mechanical devices' we shall not consider the pair of
utterances A_and A in this analysis. Our reason is simply
that there does not appear to be any way in which such an
utterance could be self-vitiating. What we wish to deal

with are those cases where both alternatives lie within
the realm of "possibility" for the actor g,

180
that both .11 and A are self-vitiatable. Or we could deduce
that the predictor p_has a disposition to utter self-
vitiating replicas of Ajsp) or Ajsp).

It should be noted that, although the sentential
parts of A_and.A_are of the form '2} and '73} respec-
tively, the prediction in question is not a tautology
(i.e. of the form 'P v~p') nor of a contradiction (i.e.
of the form 'P . ~P'). This isso even though the
sentential parts of A_and A_are logical contradictories.

Our theory enables us to deduce:
V*h . V*A

And from this assertion it does not appear possible to
deduce a contradiction. we recall that the definition
of self-vitiatable will contain as definiens certain
predicates (non-logical) which apply to sign-events.
And there is nothing contradictory in asserting that
both A_and A_have the characteristics A, A, and A,

Our theory also, presumably, would enable us to

deduce under these circumstances:
(thvh') .~(Vh.vh')

That is to»say, although both A and A_contain a member
which is self-vitiatable, only one of the pair contains
a member which is self-vitiating. The first component

of this conjunction is not a tautology. We recall that

181

the definiens of the definition of 'Vh' contains a
number of empirical assertions. And it could well turn
out that neither A_nor A_is self-vitiating. Thus far our
theory presents no logical difficulties. But the task
of predicting gfs behavior remains. If we take our task
of predicting gfs behavior as that of determining whether
A_or A_is true, then what our theory has been able to
deduce up to this moment is inadequate for this task.
What we wish to know is whether gjwill act in a manner
described by A_or in a manner described by A,

To make the requirements as stringent as possible
we demand that whatever fore-diction is uttered by p_
must be understood (as explicated previously) by g, (It
will turn out that this requirement constitutes the
major difficulty in regard to successful predictions con-
cerning this kind of self-vitiating utterances.)

Given the conditions specified above we might
expect that our theory would enable us to deduce the

following:

6.1 If p_predicts then g;will act in a manner

h
described by A:-
6.2 If p_predicts A_then g;will act in a manner

described by A,

This is to say that if p_predicts A_then A_is self—

~

vitiating and if 2 predicts A then A is self-vitiating.

182

At least two points need to be considered. First,
we assume that p_would not predict both A_and A, What
we require for a prediction is that the explanans con-
tains statements which we are warranted in accepting as
true.4 And if B were able to deduce both Ajsp) and Ajsp)
from a body of theory we assume that he would not be
warranted in accepting all the statements in the explanans
as true; and hence either one or the other or both of the
utterances would not be considered a fore-diction.
Second, the locution "p_predicts A? is intended to mean
that A_is deducible according to the requirements for a
fore-diction, and that A_is a complete utterance in which
g.is the receiver. These requirements are already im-
plicit in our use of the literal 'Af; therefore we could
rephrase 6.1 and 6.2 as follows:

6.11 If u utters h then J will act in a manner

des‘c‘ribed by-A. _

6.21 If p_utters A_then p;will act in a manner
described by A,

~

Here we assume that while A_and A_may be actually utter-

 

ances their sentential parts will not both be deducible
from the same body of theory.

So far our theory has told us nothing about 'gjs

behavior which he himself has not claimed to be the case.

 

4See p. 154.

183
What ALhas claimed is precisely what 6.11 and 6.21 state
will follow as a consequence of certain utterances.

Suppose now that p_has a body of psychological
theory from which he can deduce Ajsp). Let us call this
theory '2}. But he is also armed with a theory of self-
vitiating g} from which he can deduce 6.11. He now knows
that if he utters a replica of Ajsp) (which he had deduced
from his body of psychological theory) g;will act in a
manner described by A, On the basis of if p_concludes
that E_is inadequate. It is inadequate since presumably
an utterance of a fore-diction deducible from E_will be
self-vitiating. Now p_has made some strides in under-
standing, but he cannot yet predict gfs behavior. All
he knows is that if he utters A then A_is true. The
same arguments apply mutatis mutandis to 6.21.

It may seem that we really do know how A will act
in such circumstances and that our problem is a bogus one.
After all, it may be argued, if I can deduce 6.11 from
T) and p utters A, I can almost certainly successfully
predict that g;will act in a manner described by A,

This fact should not be minimized. And while it does
suggest that such a theory of self-vitiation would be
extremely valuable, we have not yet come to grips with
the central problem. The difficulty lies in what one
takesto be the fore-diction which is used to predict gfs
behavior. In our example ALis taking the fore-diction

to consist of A_since this is the utterance of which he

184
is the receiver. Our objector takes the fore-diction to
consist of A_since this is deducible from 2] together
with the assertion that p_utters A, But we cannot allow
the solution to our problem to hinge upon a bald-faced
ambiguity. Nor can we allow the solution (at least at
this point) to depend upon deceit. If our fore-diction
actually consists of A which we deduce from 6.21 plus
the assertion 'p_utters A] then we have simply refused
to play according to the rules of the game which require
that g understand the fore-diction in question. We shall
later note that there is substance to the Objection.
But before examining this we need to take note of another
difficulty.

We remember that pfs problem is.to successfully
predict gfs behavior. 6.21 and 6.11 do not, by themselves,
enable p_to predict what g;will actually do. What is
required is that the antecedent circumstances also be
specified. But in these two cases what is required is
that p_actually utter p_statement which ;p_false, or
more specifically a self-vitiating hypothesis, for the
antecedent of 6.11 or 6.21 to be fulfilled. It might
therefore be suspected that statements from which other
statements such as 6.11 and 6.21 were deducible would be
unsuitable. we need only note, however, that any theory
of self-vitiation would be required to enable us to deduce
statements of the form 6.11 and 6.21 if the theory would
be adequate in the sense of enabling us to make a pre-

ponderance of successful predictions. What we shall

185
attempt to show is that the difficulty which we encounter

in such cases as that of QLis an empirical and not A_

logical one.5

The solution we propose depends on our lifting
the kinds of restrictions which we have purposely imposed
on the prediction of gfs behavior by p_in order to bring
out the nature of the problem involved. What our example
has shown is App_that gfs behavior is unpredictable or
that there is a logical problem which is unavoidable in
such cases. The difficulty lies in making the require-
ments for a successful prediction so rigid as to introduce
the very perturbance which creates the difficulty. In
this regard our Mars example may prove illuminating.
Suppose, for example, that we required that the astro-
physicist in making predictions of the behavior of Mars
were to utter his fore-dictions only in the form of
explosions of cobalt bombs as described in Chapter II.

This absurd requirement would mean that very few, if any,

 

5We may admit that at present this kind of
difficulty has little practical import. It may not be
one which will occur with any frequency. Yet it is
precisely such "theoretical" difficulties which when
resolved help to increase our understanding.

On the other hand, it is by no means certain that
a theory of self-vitiation, if used with any frequency,
would not tend to increase the occurrence of such cases
as that of g, It may even be suspected that such a
theory if used in certain areas of human concern would
tend to generate an attitude of defiance on the part
of those whose behaviOr was being predicted.

186
of the predictions concerning Mars' behavior would turn
out to be successful. Now the requirement that predictions
of the behavior of individuals like QLbe known and under—
stood by the individuals in question seems as absurd, from
a logical or epistemological standpoint as that of requir-
ing that predictions of behavior of Mars contain only
those fore-dictions which explode in the fashion described
earlier.

We are not claiming that the requirement is absurd
E§£_§p, The difference lies in the fact that human
beings, unlike planets, dislike being used as guinea pigs.
And it may well be an empirical fact that under certain
circumstances the goal of attaining a successful pre-
diction may need to give way to other goals with which
it is empirically incompatible. We do not wish to dispute
or to deny this possibility. As we indicated earlier,
we are not so much concerned with the practical as with
the theoretical aspects of the problem of self-vitiation.
Any attempt to do justice to the practical aspects would
perhaps require another treatise. Without wishing to
deny that the cost of successful prediction may be very
high in terms of human values depending on other circum-
stances,6 we are confining our attention to the question
as to whether there are any logical difficulties to pre-

dicting human behavior.

 

6See Merton's comments quoted on p. 56.

 

187

Suppose now that we have an unseen observer 9 who
watches the drama between p_and g, In the event that Q_
has evidence of the assertion that p_utters Afsp) we
note that he can predict (and presumably successfully
predict) gfs subsequent behavior. He does so on the
basis of 6.11 and the fact that p_has uttered Ajsp).
Similar remarks would hold if p_uttered Ajsp).

Involved here is the possibility of a sequence
which is very like that of the one elaborated concerning
fatalities on AAA, If p_makes his prediction known to
g, QLmay choose to vitiate pfs prediction; but in so doing
he will confirm pfs prediction insofar as the two fore-
dictions in question are contradictories. Likewise A,
may predict that pfs fore-diction is self-vitiating; and
if this latter fore—diction is not known to A, it may be
presumed to be more acceptable than that of the other
two inasmuch as it considers boundary conditions not con—
sidered by p_and 9, namely thefi fact of their utterances
including p_as a receiver.

Again we note that self-vitiating phenomena do not
give us grounds for asserting that any event is unpre-
dictable ;A_principle. Whether or not we can successfully
predict the behavior of any individual or group of individ—
uals on the basis of a theory T_will depend on whether or
not we are able to include the relevant boundary conditions

in the explanans of the prediction.

188
Whether we choose to accept pfs fore-diction or
pfs fore-diction will depend primarily on which of the
two fore-dictions we consider to be better evidenced.
Again what is required is a theory of confirmation which

will enable us to answer this question.

Anigmpirical Requirement for §uccessful_§pediction

The term 'successful prediction' is subject to all
of the difficulties encountered in those terms Which
share a process-product ambiguity. 'Prediction' may be
understood as referring to the process whereby a partic-
ular scientist or group of scientists deduce certain
conclusions from a set of law-statements and antecedent
condition statements. It may also refer to the product
which is the outcome of this process: i.e., the set of
tokens thus produced. The term 'successful prediction'
likewise demands clarification. A prediction may be
considered successful if the conclusion actually follows
from the premises and unsuccessful if it does not follow
from the premises or if the process is interrupted and
the conclusion never derived. It is not, of course,
this sense of the term with which we are primarily con-
cerned. The way in which we have been employing the
term is as follows: A prediction A_is successful only
if the fore-diction A_of §.is true.

The empirical requirement for a successful pre-
diction with which we are concerned may be stated as

follows:

189

6.5 A prediction A_is successful only if it does

not contain a self-vitiating fore-diction.

6.5 stipulates a necessary (but not a sufficient)
condition for a prediction being successful in the sense
explicated. If a prediction is successful we know that
it does not contain a self-vitiating fore-diction. But
the fact that a prediction does not contain a self-
vitiating fore-diction is not sufficient grounds for
believing it is successful. Predictions fail for reasons
other than that of containing self—vitiating fore-dictions.
(For example, they may also fail because they contain a
self—vitiating premise.)

By this stage in our analysis the requirement for
successful prediction may appear so obvious as to be
trivial. But should it appear trivial we suggest that
its apparent triviality is due in part to our having been
able to shed a little light on the problems of self-
vitiation.

To return to p_and his obstreperous receiver we
can see that the problem is precisely that of attempting
to make a successful prediction while ignoring requirement
6.5. What is being demanded of p_is that which is pre-
sumably not empirically (and in one sense, logically)
possible -- namely, that p_successfully predict gfs
behavior ppgpg_p_self-vitiating fore-diction. As we
pointed out earlier, such a demand is as unreasonable as

a demand that astrophysicists make all the utterances of

190

of their fore-dictions in a manner which would upset
the causal sequence anticipated by the fore-diction.

We have suggested that p_would not predict both
A_and A_on the basis of the same theory since this would
demonstrate that theory to be inadequate. But suppose
that p_unmindful of certain logical niceties decides that
the outcome of his prediction is not particularly impor-
tant in the "total scheme of things" and acts as follows:
he utters Ajsp) which he has deduced from a body of
psychological theory. But then on the basis of the theory
of self-vitiation g] and the fact that he has already
uttered Ajsp) (i.e. on the basis of 6.11) he utters Ajsp).
Thus far he has uttered a contradiction. But he has done

so on the basis of two different theories. He knows

 

therefore that there is something wrong with at least

one of the theories in question. Just which theory is

to be accepted will depend on which of the two theories

in question is better-evidenced. And this problem as we
noted previously is one which at present has no acceptable
solution. But now on the basis of 6.21 plus the fact

that he has uttered Ajsp) he can also deduce Al(sp) where
A1(sp) is a replica of Ajsp). But now he has deduced a
contradiction on the App;p_p£_just one and the same theory,
2]. Hence 6.11 and 6.21 are inadequate as they stand.

One possible solution to the problem is as follows:

6.13 l: 6.11 . 6.21:) if~and only if A does not
utter both A_and A,

191

That is to say, we can assert the conjunction of 6.11 and
6.21 if and only if p_utters only one of a pair of contra-
dictory sentences. This new formulation 6.13 does at
least two important things. First, it prevents the kind
of contradictions which we have seen could be deduced
from if if 6.11 and 6.21 were allowed to stand unqualified.
Second it preserves 6.11 and 6.21 for the theory. As we
have seen, any adequate theory of self-vitiation would
need to include 6.11 and 6.21.

But while 6.13 enables us to preserve the theory
it does not enable p_to predict the behavior of g, p_is
"right back where he started." He knows that either
utterance he makes will be self-vitiating. .Atthis point
he may give up in despair, strike QLin frustration, or
utter contradictions until he is tired of the game. 6.5,
however, elucidates his problem. One cannot utter a
fore-diction of a successful prediction when the utterance
of such a fore-diction is self-vitiating. This is not
to say that p_cannot successfully predict the behavior
of g, It is only to say that he can successfully predict
the behavior of ALWhen QLdoes not understand or mis-

understands or is not aware of what p_has predicted.7

 

7A number of interesting variations on this
theme are possible. Suppose that u utters A. But then
to himself and unbeknownst to J he _deduces A on the
basis of 6.11 and the fact that he has uttered A but
not A, Now suppose that J is aware of the laws _of
self-vitiation and is also aware of 6. 5 as a requirement.
He will now be puzzled as to how to act. If he acts
as A describes, he has failed to make good his claim.

But _if he acts as A describes, then he is prepared to

192
we have not discovered anything new in 6.5. Venn
was perfectly aware of this solution when he said:
The publication of any prediction about the
conduct of human beings (unless it were kept out

of their sight, or expressed in unintelligible

language) almost certainly would have some effect.8

Venn's comment is made parenthetically and it is
not clear as to whether he would be willing to allow
that certain scientific predictions ought to be expressed
in unintelligible language or kept out of the sight of
those about whom the prediction was made. The point is
similar to our own: Successful predictions do not con-
tain self-vitiating fore-dictions. We do not, however,
draw Venn's conclusion. He believed that this problem
made it difficult (impossible?) to predict the behavior

of human beings.9

He also believed that the phenomena
in question enabled us to distinguish the social from
the physical sciences. We can agree that self-vitiation

is a difficulty in the social sciences. But we believe

it is a difficulty to be overcome, not one to be tolerated.

 

have A_exhibit 6.11 and 6.5 as the basis for his "real"
prediction. The puzzler has now become the puzzled.

Analogous problems are dealt with in the theory
of games. .

892. cit., p. 576
gIbid.

193

Merton, while apparently more concerned with
value questions, has also pointed out the solution we
have indicated in 6.5.10 But Merton is perhaps so con-
cerned with the value questions that he has failed to
note some of the implications of his position. Some of
these implications may be suggested by way of an example.

Suppose that a social scientist A_after investi-
gating racial tensions in the united States up to 1962
concludes that all the major prOblems of race relations
will be resolved by 1970. As a scientist he wishes to
publish his findings. Scientist §_is more cautious. He
notes that the fore-diction of _I_<_ has characteristics A,
A_and A, i.e. it is a self-vitiatable utterance. And
should circumstances Q_Obtain, the utterance will be
self-vitiating. He urges gznot to publish his results
inasmuch as the mere publication will cause the racial
tensions to be prolonged. 5, however, as a scientist,
claims that he has an obligation to make known what he
believes to be the truth. 3, however, points out that
the fore-diction of 5:13 not the truth. What is the
truth is better expressed as, "Either racial tensions

will be minimized by 1970 or the utterance of Akin 1962

regarding the minimizing of racial tensions in 1970 is

self-vitiating if ngublishes replicas of the sentential

 

10See footnote, p. 56, this essay.

194
parts of his utterance." In other words, gfs utterance
is true only if certain antecedent conditions obtain.
And the most important, presumably, in this example, is
that replicas of the sentential parts of his utterance

not be published.11

 

11538 utterance contains as sentential part a
replica of:

6.7 All the major problems of race relations will
be resolved by 1970.

Suppose that E were then to publish:

6.8 fifs utterance is true only if no replicas
of it are published prior to 1970.

Whether or not 6.8 is self-vitiating would depend
on the laws of self-vitiation and other antecedent cir-
cumstances. The answer to this question would depend in
part on the form of gfs utterance. Does it characterize,
name or state what is contained in gfs utterance? Cer-
tain semantic problems may be involved also. Suppose E
publishes his assertion in the following form:

6.81 538 utterance that all the major problems
of race relations will be resolved by 1970
is true only if no replicas of it are
published prior to 1970.

Does 6.81 constitute publication of gfs utterance?
What would be needed in order to answer this question is
a clarification of what is included in the extension of
the term 'publication'. A.possible difficulty could be
that of the self—referential feature of 6.81. It speaks
of the publication of replicas of 6.7 while apparently
containing a replica of 6.7. This self-referential
feature could present interesting problems in itself.
But we have already shown how it is always possible to
avoid this kind of difficulty by using names of tokens
(or utterances) rather than replicas of them. (See our
discussion of this issue in Chapter II.) And presumably
a scientist more fastidious in his choice of tokens than
§_would seek to avoid the additional problems created by
making fore-dictions in the form of 6.81.

The question of the self—vitiating character of 6.8,
however, would be in the final analysis an empirical one.

195

For fikto publish his utterance would then be
irresponsible on two counts: he would be publishing that
which he had empirical reasons for believing was a false
utterance (i.e. after §_had confronted him with the theory
of self-vitiation and the facts) and he would be contrib-
uting to the increase rather than the decrease of racial
tensions. In neither case would the publication of the
results appear to be justified.

This analysis would suggest the rather far reaching
conclusion that the interests of science are not always
furthered by the publication of scientific results. Nor
are the human values we mean to preserve in our day to
day affairs always preserved by knowing what purports to
be the truth. The implications for value-theory cannot
be overestimated. In one sense the scientist appears
to be in the middle of a dilemma. He knows the truth.

But if he publishes what he knows as the truth, the ante-
cedent conditions regarding boundary conditions are
changed and it is no longer a true but a false sentence
which he has uttered.

Apparently, until an adequate theory of self-
vitiation is forthcoming, each scientist must make his
decisions on the basis of his own intuition and what
empirical evidence is available in regard to self-
vitiating utterances. The contemprary social scientist
is in somewhat of a predicament. He may suspect that some

of his utterances will have self-vitiating consequences

196
under conditions of publication. Yet the present set
of values by which the scientific community is apparently
guided would lead him to publish his results provided
that he had good empirical evidence for his conclusions.

But while the phenomena of self-vitiation puts
the scientist inia predicament, the phenomena of self-
fulfilling hypotheses may offer some modicum of reprieve
from the tensions engendered by the phenomena of self-
vitiation. Since we at present do not have a theory of
self-vitiation, we cannot do more than suspect or guess
or prophesy that a particular utterance will be self-
vitiating. It may turn out fortuitously, however, that
some of the utterances of the social scientist will tend
to be self-fulfilling and that unless A_can produce an
adequate theory for his assertions concerning Afs utter-
ance, 5 may, in the absence of empirical evidence,
publish his results with no more than a normal amount of
compunction attendant upon the publication of an empirical
hypothesis.

We are not suggesting that the problem is there-
fore unimportant. Our whole essay has been concerned to
point up the great importance of this phenomenon for
science and derivatively for humanity at large. Our
point is simply that there is as yet no empirical justi-
fication for keeping a large number of hypotheses secret.
We do not know;on the one hand, which utterances are

self-fulfilling and which self-vitiating and on the other

197

hand, which have relatively little causal significance
or import. And until an adequate theory is available,
we are perhaps best guided by those values which have
guided the scientific community in the past. In the
event that an individual scientist suspects that his
hypothesis would be self—vitiating if uttered, then he
shall have to weigh the possible costs involved in
publishing or in refusing to publish. Perhaps this is

one price of being human.

Self-vitiating Instances of a Theory of Self-vitiation

 

An interesting problem arises out of the preceding
discussion. We can see that a theory of self-vitiation
itself is not necessarily immune to the problems con-
sidered in this essay. we have had occasion to note
that predictions concerning the self-vitiating character
of specific utterances may themselves be self—vitiating.
And if the §§m§_theory of self-vitiation is used in pre-
dictions which occur on two successive levels, we recall
that one or the other of the two fore-dictions is false.
we have observed that in such an instance what will
usually be the case is that a "higher-level" prediction
will usually take account of certain relevant boundary
conditions not considered by the "lower-level" prediction.

But any time such a set of circumstances obtain the

198
theory in question will be considered inadequate since
it will be possible to deduce a contradiction from the

theory.12

Since this difficulty presumably would be one
which any theory of self-vitiation would have to deal
with we may inquire as to whether or not there are any
general principles which can be applied to a theory of
self-vitiation which will prevent the kind of difficulty
here encountered. An obvious solution is suggested by
6.5 which constitutes an empirical requirement for suc-

cessful fore-dictions. This requirement can be extended

to cover the generalizations of a theory as follows:
6.6 (x) (er 3 Tx) 3 -(Ex) (er . Vx)

Requirement 6.6 constitutes a necessary condition for
the truth of a generalization. It does not constitute
a sufficient condition for it does not take account of
§;;_the possible boundary conditions which could be con—
sidered as disconfirming evidence of the truth of A5
Requirement 6.6 is useful in any theory. We are
prepared to assert the truth of a generalization only if
it contains no member which is self-vitiating. But 6.6
is especially important to a theory of self-vitiation
inasmuch as we may expect that such a theory would tend

to generate predictions on several levels.

 

l2See pp. 169-171 for an outline of a proof.

199

One implication of this requirement is that for
the purposes of deduction no generalization in a theory
of self-vitiation would be asserted without the qualif-
ication of 6.6. This statement may appear to be trivial
inasmuch as an §;containing a self-vitiating member would
not be considered to be true.13 But there appears to be
a self-referential puzzle here involved which is by no
means trivial. Consider the following circumstances: a
scientist wishes to know the truth of a generalization
H in a theory of self-vitiation T]. One of the con-

—9’

ditions for the truth of Ag would be that it not have

any self-vitiating members. How shall he proceed to

determine whether or not A? has any self-vitiating members?

He shall have to use theory 2} in order to determine this.
But in making this determination he may have to rely on

the truth of Hg, due to the fact that is part of 2].

fig
He is thus involved in a circular quest after truth.

An immediately apparent solution to this problem
is one we have encountered earlier. What is now needed
is a theory T]' by means of which one can determine the
truth of Ag. Tag
it is part of T]. Hence both the self-referential features

we assume is not part of T}' although

of the problem are avoided as well as the circular nature

of the quest.

 

13These statements are of course subject to the
qualifications made earlier regarding the cost of aban-
doning a generalization when it has proved exceedingly
useful in the past.

200

And should it be necessary to determine the truth
of a specific law-statement of gf' we would then be
required to make this determination on the basis of a
theory Tf". Just how far we would be prepared to go in
developing theories which explain theories which ... ex-
plain theories would depend in large measure on how
important for our purposes we considered the enterprise
to be. And, as we indicated earlier in connection with
predictions on various levels, there may be certain
psychological limitations which would not make it required
that we go much beyond Tf" in determining the truth or
falsity of the generalizations whereby we make our pre-
dictions concerning self-vitiating phenomena. And we
may suspect, presystematically, that generalizations of
g}" would have few if any self-vitiating instances.l4

Again we note that in ouranalysis we have dis-
covered no theoretical obstacles to the development of

an adequate empirical theory of self-vitiation.

 

4
1 Another solution to this particular problem

would be to develop a general theory of the theories of
self-vitiation. But this general theory would be sub-
ject to the same kinds of questions raised above. Per—
haps, however, such a general theory would need to
proceed to fewer levels to achieve adequacy than does
the schema presented above. But this argument has, we
may fear, gone beyond practical significance. At present
we would be grateful for a theory such as 3]. Once this
has been highly confirmed we may then undertake the con-
struction of T}' and then perhaps a general theory which
explains certain features of both. Until we possess gr
much of what we are considering in this chapter is too
speculative to be of much value. We are only concerned
to show that these difficulties are not theoretically
insurmountable.

201

Predicting Our Own Behavior

One final problem needs to be examined; it involves
the possibility of predicting our own behavior. We have
ignored this general area until now because previously
our analysis was incapable of dealing with the problem.
The problem can be phrased in the form of a question:
"Can I, as a being who is capable of making certain
voluntary choices, predict how I will act under certain
conditions and circumstances?“ Henle claimed that we
cannot predict our own behavior because such a prediction
is tantamount to a decision.15 Richfield and Copi, in
replying to Henle, have argued that decisions and pre-
dictions of our own behavior represent two distinct
classes of phenomena. The distinction in question is
clarified by an example:

Suppose I decide to eat lunch early tomorrow, for
one reason or another. Then I remember that tomorrow
morning is scheduledbe very busy |:sic:] , and I pre-
dict that the press of work will be such as to make
me decide then to delay my lunch to a later time.
Still I hope that the work can be got through early,
and I renew my decision to eat lunch early rather
than late. Here we have both decisions and prediction
with respect to one's own future voluntary action.

And one is not "really" the other because they are,
in this case contraries.l

On the other hand there are circumstances under which we

discover that decision and prediction need to be compatible.

 

15Paul Henle, Op, Cit.

1693. cit., p. 48.

202
My son would like bubble gum and a purple pony.

I wish to please him. I promise him the gum, but I

cannot promise him a purple pony. As an upright

father, I cannot decide to act in a way that I can-

not also predict. If, as Professor Henle suggests,

what I decide I cannot also predict, then it is

hard to see the basis upon which I could ever grant

my assurances of my own actions.17

Now the interesting cases for our purposes are

those in which predictions of our own behavior turn out
to be self-vitiating. I, for example, am one who is
adicted to nicotine and smoke two packs of cigarettes
each day. On the basis of evidence and certain laws of
human behavior I can quite accurately predict that tomor-
row I will smoke two packs of cigarettes. On the other
hand, I can also decide to quit smoking tomorrow. That
is, I can choose to vitiate my prediction. The question
then would be whether or not I could predict how I will
decide. Will I decide to stop smoking and hence to
vitiate my original prediction? Or will I simply follow
the course of least resistance? In order to answer this
question I would need a theory of self-vitiation which
would be able to determine whether my original fore-

diction was self-vitiating, self-fulfilling, or neither.18

And if I am extremely anxious to predict how I will act,

 

l7Ibid.

18We are assuming that an adequate theory of
self-vitiation would also be able to account for the
phenomena which we have labeled 'self-fulfillable'.

203
I may also wish to inquire whether this second fore-
diction is itself self-vitiating. I may then need to
resort to theory T}' assuming that theory T} was used in
the second prediction. It may also be the case that I

will make certain decisions on the basis of what I have

 

predicted from theory T}', i.e., I may choose to vitiate
what I have predicted or I may follow the line of least
resistance.

At this point an ancient philosophical contro-
versy comes to the fore: the controversy regarding the
primacy of the freedom of the will versus the primacy of
reason. And whether I choose to vitiate a fore-diction
on any "level" would presumably be a question of whether
I were more interested in knowing what.the future would
bring or whether I were more interested in being able Eg_
decide what the future would bring. And we may suppose
that an adequate theory of self-vitiation would, on the
basis of which of these dispositions was manifested by
a particular receiver of tokens, be able to determine on
which level an individual would stop deciding and Egg:
dicting.

From the analysis of self-vitiation we can see
that whether or not predictions of our own behavior are
successful will depend in large part on whether or not
the fore-dictions in question are self-vitiating. It is
certainly true that I make a number of successful pre-

dictions about my own behavior every day. These are not

204

simply decisions in which I choose one alternative over
another. I deduce on the basis of certain inductive
generalizations concerning my own and others' behavior
that I will act in such and such a way under certain
conditions. I predict that I will go to work tomorrow,
that I will eat at least three meals, that I will exhibit
certain moods, that if I hear a joke I will laugh, that
if I hear of the death of a friend I will be sad, that
I will continue to smoke two packs a day, that in the next
five-years I will probably spend at least five days in a
sick bed, that I probably will not die tomorrow, that
tomorrow I will perform certain bodily functions, etc.,
etc.

This is not to say that in all of these events
I may not choose to do otherwise. But in the case that
I do choose to do otherwise I can then raise the question
as to whether I can successfully predict how I will choose.
And it may turn out that certain kinds of personal pre-
dictions are unsuccessful and not empirically possible
due to the fact that they introduce the kinds of distur-
bance with which we are now quite familiar. In this
case we cannot retreat to an unseen observer in order to
resolve the problem. For here we are dealing with pre-
dictions of one's own behavior. But which of such pre-
dictions are or would be unsuccessful is itself an
empirical question which could be answered by an adequate

theory of self-vitiation. Again, we note however, that

205
this is not to say that any action is unpredictable in
principle; only that when certain predictors are in-
volved may the action prove to be unpredictable.

But even in such cases the individual predicting
his own behavior may choose one level of prediction over
another on empirical grounds. And just because a pre-
diction on one level contained a self—vitiating fore-
diction would not be adequate grounds for claiming that
fore-dictions on every level would be self-vitiating.

Again, 6.5 requires that successful predictions
must not contain self-vitiating fore-dictions.19 And
whether an individual can successfully predict his own
behavior in any set of specific circumstances will depend
to a large extent on whether his utterances of fore-

dictions are self-vitiating.

Summary

The sections of this chapter have been concerned
with arguments previously considered in Chapter II. We
have been principally concerned to show that the argu-
ments claiming that the phenomena of self-vitiation makes
successful prediction of the future behavior of human
beings impossible are not cogent. The difficulties in-

volved in the development of a theory of self-vitiation

 

19Although obviously they may contain self-
fulfilling fore-dictions.

206
are formidable indeed. But the difficulties involved
are not logical but empirical.

We went on to show that there are certain empirical
conditions which must be met if a particular prediction
is to be accepted as successful. The result of this
analysis has been to show that the success or failure of
a prediction may depend on the circumstances under which
a particular fore-diction is uttered (e.g. the relations
which obtain between the predictor and the receiver of
tokens).

The purposes for which this analysis was under-
taken have, we hope, been fulfilled. Whatever differences
do obtain between the social and physical sciences, it
seems clear that the phenomena of self—vitiation is not
one of them. It also seems clear that there are no
theoretical or logical obstacles in the way of developing
a theory of self-vitiation. Or, more modestly, we may
say that if there are logical difficulties involved, we

have been unable to uncover them.

CHAPTER VII

SUMMARY AND IMPLICATIONS

Summagy of the Anaiysis

In this essay we have attempted to call attention
to the great importance of the phenomena of self-vitiation
for the philosophy of science. What we can know about
the world is in some degree dependent on which of our
pretensions to knowledge have self-vitiating consequences.
In this chapter we shall review our argument and then
attempt to call attention, in somewhat more detail, to
some of the consequences of what we have presumably learned
from the analysis.

We first examined a number of instances of self-
vitiation. While these examples were highly selective,
they suggest that there appears to be little of human
concern which is immune to the possibility of being de—
scribed by self—vitiating utterances.

The arguments in the literature concerning the
impossibility of predicting the future history of mankind
we found to be incogent for a number of reasons. The
phenomena with which the arguments were concerned were
not clearly specified. Certain ambiguities in key terms
prevented the arguments from having the cogency usually
attributed to them.

These difficulties led us to undertake a more
formal and more rigorously precise analysis of the term
'self—vitiating'. we defined the term in such a manner
that the definiens enabled us to discriminate between

207

208
those utterances which are simply false and those which
we are warranted in accepting as true prior to the deter-
mination of their self-vitiating character.

We also called attention to the fact that this
class of utterances needs to be distinguished from the
traditional syntactical and semantical paradoxes. It is
no self-referential feature of the self-vitiating utter-
ance which creates the difficulty. Nor is it a syntactical
feature of these utterances which proves troublesome.

‘What is involved is that the fore-diction in question is
causally efficacious in bringing about a state of affairs
described of the negation of the utterance. That this
distinction has not been clearly drawn in the past may
account for the rather obscure fashion in which the
phenomena have been treated.

we found that the difficulties encountered with
certain malformed (from a syntactical standpoint) and
ambiguous or vague expressions could be overcome by
treating such utterances as members of the causal sequence
which leads to a self-vitiating consequence and reserving
the predicate 'self—vitiating' for well-formed, unambig-
uous and precise expressions. This restricted usage, we
assumed, will not handicap the social scientist in develop-
ing a theory of self-vitiation.

One of the consequences of the analysis is that
usually only certain members of a statement are self-

vitiating, If we say that a statement is self-vitiating,

209
this loose manner of speaking may be interpreted as
meaning that some particular member of that statement
is self-vitiating but not that every member is self-
vitiating. Hence if some member of A_is false, every
member of A_is false. But it is not the case that if
some member of ALis self-vitiating, every member is
self-vitiating.

We were fortunate in discovering a mode of analysis
for the predicate 'self—vitiating' which also proved to
be adequate to the analysis of the predicate 'self-
fulfilling'. The analysis, however, revealed an impor-
tant distinction between the two predicates. This
difference concerns the predicates as applied to general-
izations. we noted that, from a logical standpoint, one
disconfirming instance of a generalization which was
caused by an utterance of that generalizationxvould be
sufficient for asserting that the generalization is false.
But one confirming instance of a contingent generalization
which was caused by the utterance of that generalization
would not be sufficient grounds for holding that the
generalization is true.

Our analysis also showed certain important re-
lations which exist between the phenomena of self—
vitiation and the problem of distinguishing between law-
like and non-law-like generalizations. We were led to
suggest that an adequate treatment of the one would

contribute to a solution of the other. If the grounds

210
which are sufficient for determining that A_occurs are
also sufficient for determining the truth of A, then in
the event that A_is a generalization we could conclude
that A_is non—law-like.

An examination of the problem of boundary con-
ditions in scientific explanation revealed that self-
vitiating phenomena may be treated as a sub-class of that
set of problems encountered when the boundary conditions
are not made explicit in a particular explanation. The
difficulty in these cases is removed by making the
boundary conditions explicit. From the revised set of
premises the false explanandum is no longer deducible.

An adequate theory of self-vitiation would explain
why certain predictions fail; i.e. it would show that one
of the boundary conditions not considered in such cases
was the self-vitiating character of the fore-diction.
Even more important, perhaps, would be the fact that an
adequate theory of self-vitiation would enable us to
predict the self-vitiating consequences of certain ut-
terances. As part of a general theory of perturbance
such a theory of self-vitiation would be invaluable to
both the social and the physical sciences.

Prior to the development of a theory of self-
vitiation it would be required to examine a large number
of instances of self-vitiating hypotheses and then to
determine which features shared in common by all of these

utterances are relevant for defining the disposition

211
predicate 'self—vitiatable'. Once such a predicate is
available, it will be possible to begin formulating the
laws of self-vitiation, gathering evidence for the con-
firmation of the law-statements in question, and then
systematically ordering these statements.

The practical difficulties attending the develop-
ment of such a theory would be enormous. We have
attempted to show, however, that the arguments concerning
the supposed logical difficulties attending the develop-
ment of such a theory will not stand up under examination.
The lack of an adequate theory at present is a major
factor in our being unable to successfully predict the
future course of human history. But there are no apparent
theoretical reasons why such a theory could not be
developed. And if such theoretical reasons “exist" such
reasons have not yet been advanced in favor of the thesis
of the impossibility of predicting certain events. The
arguments which have been advanced do not adequately
support this thesis.

Our analysis has also attempted to show that the
claim that the phenomena of self-vitiation enable us to
distinguish between the social and non-social sciences
is not as cogent as itmay first appear. Any difference
which does emerge seems to be one of degree rather than
of kind. That is to say, we may expect, presystematically
at least, that we will discover more examples of this

type of phenomenon in the social sciences than in the

212
physical sciences. We noted, however, that as greater
use is made of machines such as electronic brains the
difficulty may become even more important for the
physical or non-social sciences as well.

In a positive vein our analysis attempted to in-
dicate some of the problems that the development of such
a theory could presumably resolve. Without an adequate
theory of self—vitiation certain advances in the sciences
do not appear to have much chance for success. With such
a theory, however, we may expect that our understanding

of the world in which we live will be greatly increased.

Some gmplications of the Preceding Analysis

We have suggested throughout the course of this
essay that the phenomena of self-vitiation is not simply
an interesting though somewhat irrelevant prdblem but
that it has great importance for the sciences and the
enterprise of human living generally. We shall in this
concluding section attempt to indicate in somewhat more
detail some of the practical implications of the develop—
ment of such a theory.

A theory of self-vitiation would presumably con-
tribute to a more adequate theory of knowledge. In
Chapter II we showed some of the difficulties attendant
upon our usage of the term 'to know'. (We did not pre-
tend to give an exhaustive account of the problem in

that section, nor can we attempt such an account here.

II! (II. ‘1' Ill! '3'. {I

213
we wish merely to indicate some of the other consequences
of our analysis.) When we say that we know that such-
and-such is the case, if an utterance of a statement
which describes what we claim to know is self-vitiatable,
how can our claim to know such—and-such be construed?
The preceding analysis indicates that such claims may be
elliptical ways of saying that such-and-such is the case,
provided that ggygggy§g_;§_not the case. In other words,
all absolute claims to knowledge (with the possible ex-
ception of such claims as "I know that the paper before
me appears to be white“) may be construed as provisional
claims. What is tacitly assumed is that certain boundary
conditions will hold. The interesting cases for a theory
of knowledge are just those cases in which the utterance
of what is construed to be knowledge is self-vitiating.
One boundary condition which is presumed to hold in this
case would be that the utterance not occur under circum-
stances g,

Presumably the fore-going remarks will only apply
to contingent or synthetic hypotheses and not to those
which are logically true or analytic. But the kinds of
distinctions which are required here are by no means
clear. The literature of the past few years is filled
with controversy concerning the distinction between for-
mal and non-formal, analytic and synthetic, contingent
and necessary statements. Our analysis may or may not

help to settle this controversy. But some of the

214
difficulty may be resolved by noting that should any
statement contain a self-vitiatable member we should
presumably not wish to count that statement as analytic.
And if a statement contain no self-vitiatable member we
should probably not wish to call that statement synthetic.l
Here we would have an empirical test available for deter-
mining whether to apply the predicate 'analytic' in any
given case. An analytic statement then would be one
which had no self-vitiatable members -- i.e. no members
such that if they occurred under circumstances g_would
be self-vitiating. An interesting consequence of this
treatment would be that to call a particular statement
'analytic' would be to claim that by law it was not
possible for the statement to be false. The possibility
here in question would be "empirical" as well as "logical"
or "mathematical". The laws in question would be the
laws of self-vitiation together with all the other laws
of the sciences. We are not attempting to defend the
distinction between analytic and synthetic as it has
traditionally been drawn. We are only attempting to show
that a theory of self-vitiation could presumably enable
one to mark the distinction on empirical grounds.‘ Whether
the distinction as thus drawn would prove useful is not

a question we need to decide in this essay.

 

l . .

Except perhaps 1n certain cases such as state-
ments in celestial mechanics of the kind indicated in
Chapter V.

215

The implications of our analysis for each of the
sciences cannot, it appears, be overestimated. One of
the requirements which the scientist imposes upon his
enterprise is that disturbing factors, perturbance effects,
and relevant differences between the subjects under con-
sideration be reduced to a minimum. But is there any way
open to the scientist to prevent the disturbance which is
the result of simply drawing a conclusion (i.e. uttering
a self-vitiating fore-diction) from a set of premises
describing the results of his experiment? .A theory of
self-vitiation presumably could answer this question for
the individual scientist in specific circumstances. What
would be available from such a theory would be a set of
requirements for uttering fore-dictions which would tend
to minimize the possibility of their being self-vitiating.
We have already suggested what at least one of these re-
quirements might be: a keeping of the results as secret
as possible, preventing them from being understood by
receivers who have the disposition to render them self-
vitiating.

And here we note some important value problems
which need to be considered. One problem is at least as
old as Plato's Republic and the question of whether or
not the beneficial lie is justified and the conditions
under which it would be justified. In a democratic
society the decision to keep certain results secret

appears to violate some of the values which a free society

_ ‘1']!- I 'If i I
I)..."

216
holds dear. We may even equate the kind of secrecy in
question with totalitarian and authoritarian regimes.
Once a theory of self-vitiation is available which can
enable us to predict successfully in a large number of
cases when certain utterances are or are not self-
vitiating, then as we noted in Chapter VI, part of the
value problems are solved. For if a scientist keeps
secret a conclusion which he knows will be self-vitiating
if uttered under certain circumstances, circumstances
which would be almost certain to obtain unless he kept
the conclusion secret, then he is not hiding the truth,
but refusing to publish what will in those circumstances
be false.

The major value problems are present ones. ”We do
not have a theory of self-vitiation; even an inadequate
theory is not available to us. Hence we are unable in
any given set of circumstances to determine with any
significant degree of assurance when a particular utter-
ance will be self-vitiating. Yet we may strongly suspect,
guess, or prophesy that certain utterances will have self-
vitiating consequences. What shall we do in such cases?
The answer to this question is by no means clear cut.

And until an adequate theory is forthcoming we shall
probably have to rely on guesses concerning probabilities,
weighing the costs of uttering over against the cost of

not uttering certain hypotheses.

217

The political and social consequences appear to
be enormous. If a political scientist or a sociologist
publishes certain findings concerning race relations, un-
employment, labor-management relations or any other of a
number of related issues, and these turn out to have self-
vitiating consequences which are disvaluable according to
some acceptable value schema, the entire society of which
the scientist is a part will presumably suffer the con—
sequences. How often such publications of this kind have
such effects is at present unknown to us. What is clear,
however, is that the individual scientist's dedication to
the truth is not always exhibited by his publishing what
purports to be the truth.

On the other hand suppose a social scientist
deduces a conclusion which if published would be self-
vitiating yet among the consequences of publication would
also be certain valuable results (e.g. the amelioration
of race relations, the decrease in unemployment, etc.).
Suppose further that the scientist has evidence that his
hypothesis is self-vitiatable and would be self-vitiating
if published. Shall he publish a lie? Plato's question
has not yet been satisfactorily answered; and presumably
it cannot be given a blanket answer.

The present day scientist, however, does not have
the kind of evidence which we would require to support an
assertion that a particular scientific hypothesis is self-

vitiating or is not self-vitiating. And whether a given

218
scientist will publish results of his particular inves—
tigations may in part depend on how concerned he is about
the possibility that his publication of results may have
self-vitiating consequences.

An even more troublesome problem arises in regard
to confirmation theory. It is likely that a theory of
self-vitiation will give us a set of probability state-
ments concerning the phenomena of self-vitiation. Suppose
that a scientist, on the basis of such a theory determines
that a particular conclusion, if published, has a prob—
ability of .7 of being self-vitiating. Shall he publish
the fore-diction? Or suppose that instead the probability
in question is .15. Shall he publish the fore-diction?
Whether or not he will publish may depend on the conse—
quences of publication. If the fore-diction in question
has consequences upon publication which are relatively
unimportant in the "total scheme of things" a probability
of .7 may not dissuade him from publication. But a fore-
diction whose publication may tend to result in atomic war
if it is self-vitiating, may have a probability as low as
.01 of being self-vitiating and yet the scientist may still

be hesitant to publish.2 Publishers also may be called

 

2These comments are suggested by and are related
to Professor Richard S. Rudner's comments in "The Scientist
Qgg_Scientist Makes Value Judgments", ghilosophy g§_Science,
20:1-6, January, 1953. One of Professor Rudner's main
points is that "how sure we must be before we accept a
hypothesis depends on how serious a mistake would be.“

219
upon to make certain value judgments in this general
area. Interesting questions of a legal nature may arise
once an adequate theory of self-vitiation is available.
Suppose that the publication of a particular hypothesis
had a presumed probability of .7 that it was self-
vitiating. Suppose further that the author and publisher
know (in one of the pertinent sensesdiscussed earlier)
that if the hypothesis is self-vitiating it will have con—
sequences which are extremely disvaluable to a certain
segment of the population. If author and publisher
decide to go through with publication will they then be
liable to lawsuit for damages? If not, would the same
hold true if the probability were .99? If so, would the
same hold true if the probability were .01?

These remarks indicate that certain delicate
problems of censorship may be given a slightly different
"twist" once an adequate theory of self-vitiation is
available to us. We may suspect that certain legal re-
strictions on publications will be urged by those Who
emphasize the consequences of certain kinds of publication.
Others may hold that the decision to publish or not to
publish must rest with the individual author and publisher
and that the freedom of the press must not suffer any more
inroads than it does at present. How this particular
argument will be resolved is by no means clear. We are

concerned at this point only to call attention to one

220
possible consequence which a theory of self-vitiation may
have once it has proven adequate for other purposes.

The problems which an adequate theory of self—
vitiation will presumably pose are by no means new. And
we have not pretended to do more than to suggest a few
of what appear to be some of the major difficulties we
shall encounter once such a theory is available to us.

Our task of analyzing the terms 'self-vitiating',
'self-vitiatable' and 'self—fulfilling' is for our pur-
poses complete. The taSk of developing an adequate
theory of self-vitiation remains. And if this essay
should prove to be a help in the development of such a

theory our hopes shall have been fulfilled.

BIBLIOGRAPHY

Books

Berlin, Isaiah. Historical Inevitability. London and
New York: Oxford university Press, 1954.

Braithwaite, Richard Bevan. Scientific Explanation.
New York: Harper and Brothers, 1960.

Carroll, Lewis. Through the Looking Glass. New York:
Random House, special edition, 1946.

Danto, Arthur and Sidney Morgenbesser, editors. Philosophy
of Science. New York: Meridian Books, Inc., 1960.

 

Dray, William. Laws and Explanation in History. London:
Oxford university Press, 1957.

Goodman, Nelson. Fact, Fiction and Forecast. Cambridge,
Massachusetts: Harvard university Press, 1955.

. The Structure of Appearance. Cambridge, Massachusetts:
Harvard University Press, 1951.

Leonard, Henry S. Principles of Right Reason. New York:
Henry Holt and Company, c.d., 1957.

. ggngular Terms. typescript, Widener Library,
1930.

Luce, R. D. and H. Raiffa.. Games and Decisions. New York:
Wiley and Sons, Inc., 1957.

McKinsey, John C. C. Introduction to the Theory of Games.
New York: McGraw-Hill Book Company, Inc., 1952.

Madden, Edward W. The Structure of Scientific Thought.
Boston: Houghton.Mifflin Company, 1960.

Merton, RObert K. Social Theory and Social Structure.
Glencoe, Illinois: The Free Press, 1957, revized and
enlarged edition.

Mill, John Stuart. A_System of Logic. New York: Harper
and Brothers, 1847.

Nagel, Ernest. The Structure of Science. New York:
Harcourt, Brace, and WOrld, c.d., 1961.

Popper, Karl Raymond. The Logic of Scientific Discovery.
London: Hutchinson, 1959.

. The Open Society and its Enemies. London:
Routledge and Kegan Paul, Ltd., 1945.

221

222

Popper, Karl Raymond. The Poverty of Historicism.
Boston: The Beacon Press, c.d., 1957.

Quine, Willard Van Orman. werd and Object. New York and
London: John Wiley and Sons, Inc., c.d., 1960.

 

Ryle, Gilbert. The Concept of Mind. New York: Barnes
and Ndble, Inc., c.d., 1949.

Venn, John. Principles of Empirical or Inductive Logic.
London and New York: Macmillan and Company, 1889.

Ziff, Paul. Semantic Analysis. Ithaca, New Ybrk: Cornell
University Press, c.d., 1960.

 

Journals

Abel, Theodore. "The Operation Called Verstehen," American
Journal gf_Sociology, 54:211-218, November, 1948.

Albery, Ethel M. "Causality in the Social Sciences,“
JOurnal g§_Philosophy, 51:695-706, November 11, 1954.

Bunge, Mario. "Do Computers Think?", Part I, The British
Journal for the philosopAy,g£_Science, 7:139-148,
August, 1956.

. "Do Computers Think?", Part II, The British
Journal for the Philosophy g§_Science, 7:212-219,
November, 1956.

Feuer, Lewis S. "Causality in the Social Sciences,"
JOurnal gf_Philosgphy, 51:681-695, November 11, 1954.

. "Indeterminacy and Economic Development,"
Philosgahx p_r; Science, 15:234, July, 1948.

Gewirth, Alan. "Can Men Change the Laws of Social Science,"
Philosophy g§_Science, 21:229-241, July, 1954.

Goodman, Nelson and Henry Leonard. "The Calculus of
Individuals and its Uses," Journal gf_Philosophy,
5:45-55, June, 1940.

 

Grunbaum, Adolf. "Historical Determinism, Social Activism,
and Prediction in the Social Sciences," The British

Journal for the PhilosopAyygf_Science, 7:236-240,
November, 1956.

223

Grfinbaum, Adolf. "Temporally-Asymmetric Principles, Parity
between Explanation and Prediction, Mechanism versus
Teleology," Philosophy pf.Science, 29:146-174, April,
1962.

 

Hartshorne, Charles. ”Freedom Requires Indeterminacy and
Freedom," Journal g£_Philospphy, 55: 793-811,
September 11, 1958.

Hempel, Carl G. and Paul Oppenheim. "Studies in the Logic
of Scientific Explanation," Philosophy g£_Science,
15: 135-175, April, 1948.

Henle, Paul. "Do We Discover our USe of Words?" Journal
g; Philosophy, 54: 750-758, November 7, 1957.

Leonard, Henry S. "The Logic of Existence," Philospphical
Studies, 7: 49-64, June, 1956.

Miles, T. R. "On the Difference between Men and Machines,"
The British Journal for the Philosophy g§_Science,
7: 277-292, February, 1957.

Quine, Willard Van Orman. "The Scope and Language of
Science,“ The British Journal for the Philosophy g§_
Science, 8: 1-17, May, 1957.

Rescher, Nicholas. "Belief-contravening Suppositions,"
Philosophical Review, 70: 176-196, April, 1961.

Richfield, Jerome and Irving M. Copi. "Deciding and
Predicting,“ Philosophy g£_Science, 28: 47-52,
January, 1961.

Roshwald, M. ”The Case for Indeterminism," The British
Journal for the PhilosopAyyg§_Science, 7: 240-242,
November, 1956.

Rudner, Richard S. "The Scientist Qua Scientist Makes
Value Judgments," Philosophy g£_Science, 20: l-6,
January, 1953.

Scheffler, Israel. "Explanation, Prediction, and Ab-
straction," The British JOurnal for the Philosophy
g§_Science, 7: 293—309, February, 1957.

Encyclopedias

Hempel, Carl G. "Fundamentals of Concept Formation in
Empirical Science," International Encyclopedia 2:
Unified Science, Vol. II, No. 7, Chicago: University
of Chicago Press, c.d., 1952.

‘314.‘"

 

ml“ mlulellfmlllill “Willi“ Hm IAIRI'ES
31293 03175 3613