.......

CONJECTURES

sense similar to that in which processes or things may be said to be parts of

the world; that the world consists of facts in a sense in which it may be
said to consist of (four dimensional) processes or of (three dimensional)
things. They believe that, just as certain nouns are names of things, sentences
are names of facts. And they sometimes even believe that sentences are some-
thing like pictures of facts, or that they are projections of facts.7 But all this is
mistaken. The fact that there is no elephant in this room is not one of the
processes or parts ofthe world; nor is the fact that a hailstorm in Newfound-
10
land occurred exactIy I I I years after a tree collapsed in the New Zealand bush.
Facts are something like a common product oflanguage and reality; they are
reality pinned down by descriptive statements. They are like abstracts from TRUTH, RATIONALITY, AND THE
a book, made in a language which is different from that of the original, and
determined not only by the original book but nearly as much by the principles
GROWTH OF SCIENTIFIC KNOWLEDGE
of selection and by other methods of abstracting, and by the means of which
the new language disposes. New linguistic means not only help us to describe 1. THE GROWTH OF KNOWLEDGE: THEORIES AND PROBLEMS
new kinds of facts; in a way, they even create new kinds of facts. In a certain
sense, these facts obviously existed before the new means were created which I
were indispensable for their description; I say, 'obviously' because a calcula-
tion, for example, ofthe movements of the planet Mercury of 100 years ago, MY aim in this lecture is to stress the significance of one particular aspect of
carried out today with the help of the calculus of the theory of relativity, may science-its need to grow, or, if you like, its need to progress. I do not have
certainly be a true description of the facts concerned, even though the theory in mind here the practical or social significance of this need. What I wish to
was not yet invented when these facts occurred. But in another sense we might discuss is rather its intellectual significance. I assert that continued growth is
say that these facts do not exist as facts before they are singled out from the essential to the rational and empirical character of scientific knowledge; that
continuum of events and pinned down by statements-the theories which if science ceases to grow it must lose that character. It is the way of its growth
describe them. These questions, however, although closely connected with our which makes science rational and empirical; the way, that is, in which
scientists discriminate between available theories and choose the better one
problem, must be left for another discussion. I have mentioned them only in
order to make clear that even should the solutions I have proposed be more or or (in the absence of a satisfactory theory) the way they give reasons for
less correct, there would still be open problems left in this field. rejecting all the available theories, thereby suggesting some of the conditions
with which a satisfactory theory should comply.
7I had in mindWittgensteinin the Tractatus.Note that thispaperwaswrittenin 1946. You will have noticed from this formulation that it is not the accumulation
of observations which I have in mind when I speak of the growth of scientific
knowledge, but the repeated overthrow of scientific theories and their replace-
ment by better or more satisfactory ones. This, incidentally, is a procedure
which might be found worthy of attention even by those who see the most
important aspect of the growth of scientific knowledge in new experiments
and in new observations. For our critical examination of our theories leads
us to attempts to test and to overthrow them; and these lead us further to
experiments and observations of a kind which nobody would ever have
dreamt of without the stimulus and guidance both of our theories and of our

This lecture was never delivered, or published before. It was prepared for the International
Congress for the Philosophy of Science in Stanford, August 1960, but because of its length only
a small part of it could be presented there. Another part formed my Presidential Address to
the British Society for the Philosophy of Science, delivered in January 1961. I believe that the
lecture contains (especially in parts 3 to 5) some essential further developments of the ideas of
my Logic of Scientific Discovery.
214 215

"""'-- -
 CONJECTURES
10 TRUTH, RATIONALITY, AND THE GROWTH OF KNOWLEDGE
criticism of them. For indeed, the most interesting experiments and observa-
tions were carefully designed by us in order to test our theories, especially progress (unless we adopt a very narrow view of our possible aims in life);
our new theories. for almost every gain is balanced, or more than balanced, by some loss. And
In this paper, then, I wish to stress the significance of this aspect of science in most fields we do not even know how to evaluate change.
and to solve some of the problems, old as well as new, which are connected Within the field of science we have, however, a criterion of progress: even
with the notions of scientific progress and of discrimination among com- before a theory has ever undergone an empirical test we may be able to
peting theories. The new problems I wish to discuss are mainly those con- say whether, provided it passes certain specified tests, it would be an im-
nected with the notions of objective truth, and of getting nearer to the provement on other theories with which we are acquainted. This is my first
thesis.
truth-notions which seem to me of great help in analysing the growth of
knowledge. To put it a little differently, I assert that we know what a good scientific
theory should be like, and-even before it has been tested-what kind of
Although I shall confine my discussion to the growth of knowledge in
theory would be better still, provided it passes certain crucial tests. And it is
science, my remarks are applicable without much change, I believe, to the
growth of pre-scientific knowledge also-that is to say, to the general way in this (meta-scientific) knowledge which makes it possible to speak of progress
in science, and of a rational choice between theories.
which men, and even animals, acquire new factual knowledge about the
world. The method of learning by trial and error-of learning from our II
mistakes-seems to be fundamentally the same whether it is practised by
Thus it is my first thesis that we can know of a theory, even before it has been
lower or by higher animals, by chimpanzees or by men of science. My interest
tested, that ifit passes certain tests it will be better than some other theory.
is not merely in the theory of scientific knowledge, but rather in the theory of
My first thesis implies that we have a criterion of relative potential satis-
knowledge in general. Yet the study of the growth of scientific knowledge is,
factoriness, or of potential progressiveness, which can be applied to a theory
I believe, the most fruitful way of studying the growth of knowledge in general.
even before we know whether or not it will turn out, by the passing of some
For the growth of scientific knowledge may be said to be the growth of crucial tests, to be satisfactory in fact.
ordinary human knowledge writ large (as I have pointed out in the 1958
This criterion of relative potential satisfactoriness (which I formulated
Preface to my Logic of Scientific Discovery).
some time ago,2 and which, incidentally, allows us to grade theories accord-
But is there any danger that our need to progress will go unsatisfied,
ing to their degree of relative potential satisfactoriness) is extremely simple
and that the growth of scientific knowledge will come to an end? In par-
and intuitive. It characterizes as preferable the theory which tells us more;
ticular, is there any danger that the advance of science will come to an end
because science has completed its task? I hardly think so, thanks to the in- that is to say, the theory which contains the greater amount of empirical
information or content; which is logically stronger; which has the greater
finity of our ignorance. Among the real dangers to the progress of science is
explanatory and predictive power; and which can therefore be more severely
not the likelihood of its being completed, but such things as lack of imagina-
tested by comparing predicted facts with observations. In short, we prefer an
tion (sometimes a consequence oflack of real interest); or a misplaced faith
interesting, daring, and highly informative theory to a trivial one.
in formalization and precision (which will be discussed below in section v);
All these properties which, it thus appears, we desire in a theory can be
or authoritarianism in one or another of its many forms.
shown to amount to one and the same thing: to a higher degree of empirical
Since I have used the word 'progress' several times, I had better make quite content or of testability.
sure, at this point, that I am not mistaken for a believer in a historical law of
progress. Indeed I have before now struck various blows against the belief in ill
a law of progress, 1and I hold that even science is not subject to the operation My study of the content of a theory (or of any statement whatsoever) was
of anything resembling such a law. The history of science, like the history of based on the simple and obvious idea that the informative content of the
all human ideas, is a history of irresponsible dreams, of obstinacy, and of conjunction, ab, of any two statements, a, and b, will always be greater than,
error. But science is one of the very few human activities-perhaps the only or at least equal to, that of any of its components.
one-in which errors are systematically criticized and fairly often, in time, Let a be the statement 'It will rain on Friday'; b the statement 'It will be
corrected. This is why we can say that, in science, we often learn from our
mistakes, and why we can speak clearly and sensibly about making progress 2 See the discussion of degrees of testability, empirical content, corroborability, and
corroboration in my L.Sc.D., especiallysections 31 to 46; 82 to 85; new appendix .ix; also
there. In most other fields of human endeavour there is change, but rarely the discussion of degrees of explanatory power in this appendix, and especially the com-
1 See especially my Poverty of Historicism (2nd edn., 1960), and ch. 16 of the present
parison of Einstein's and Newton's theories (in note 7 on p. 401). In what follows, I shall
volume. sometimes refer to testability, etc., as the 'criterion of progress', without going into the
more detailed distinctions discussed in my book.
216 217

"""'-- -
 CONJECTURES 10 TRUTH. RATIONALITY, AND THE GROWTH OF KNOWLEDGE

fine on Saturday'; and ab the statement 'It will rain on Friday and it will be deeply ingrained probabilistic prejudice seems unavoidable if the matter is
fine on Saturday': it is then obvious that the informative content of this last really to be cleared up. Even if, as would be easy enough, I were to base my
statement, the conjunction ab, will exceed that of its component a and also own theory upon the calculus of content, or of logical strength, it would still
that ofits component b. And it will also be obvious that the probability of ab be necessary to explain that the probability calculus, in its ('logical') applica-
(or, what is the same, the probability that ab will be true) will be smaller than tion to propositions or statements, is nothing but a calculusof the logical weak-
that of either of its components. ness or lack of content of these statements (either of absolute logical weakness
Writing Ct(a) for 'the content of the statement a', and Ct(ab) for 'the or of relative logical weakness). Perhaps a head-on collision would be avoid-
content of the conjunction a and b', we have able if people were not so generally inclined to assume uncritically that a high
(1) Ct(a) < Ct(ab) > Ct(b) probability must be an aim of science, and that, therefore, the theory of
induction must explain to us how we can attain a high degree of probability
This contrasts with the corresponding law of the calculus of probability,
for our theories. (And it then becomes necessary to point out that there is
(2) p(a) > p(ab) <p(b) something else-a 'truthlikeness' or 'verisimilitude'-with a calculus totally
where the inequality signs of (1) are inverted. Together these two laws, (1) and different from the calculus of probability with which it seems to have been
(2), state that with increasing content, probability decreases, and vice versa; confused.)
or in other words, that content increases with increasing improbability. (This To avoid these simple results, all kinds of more or less sophisticated
analysis is of course in full agreement with the general idea of the logical theories have been designed. I believe I have shown that none of them is
content of a statement as the class of all those statements which are logically successful. But what is more important, they are quite unnecessary. One
entailed by it. We may also say that a statement a is logically stronger than merely has to recognize that the property which we cherish in theories and
a statement b if its content is greater than that of b-that is to say, if it entails which we may perhaps call 'verisimilitude' or 'truthlikeness' (see section xi
more than b.) below) is not a probability in the sense of the calculus of probability of which
This trivial fact has the following inescapable consequences: if growth or (2) is an inescapable theorem.
knowledge means that we operate with theories of increasing content, it must It should be noted that the problem before us is not a problem of words. I
also mean that we operate with theories of decreasing probability (in the do not mind what you call 'probability', and I do not mind if you call those
sense of the calculus of probability). Thus if our aim is the advancement or degrees for which the so-called 'calculus of probability' holds by any other
growth of knowledge, then a high probability (in the sense of the calculus of name. I personally think that it is most convenient to reserve the term 'prob-
probability) cannot possibly be our aim as well: these two aims are in- ability' for whatever may satisfy the well-known rules of this calculus (which
compatible. Laplace, Keynes, Jeffreys and many others have formulated, and for which I
I found this trivial though fundamental result about thirty years ago, and have given various formal axiom systems). If (and only if) we accept this
I have been preaching it ever since. Yet the prejudice that a high probability terminology, then there can be no doubt that the absolute probability of a
must be something highly desirable is so deeply ingrained that my trivial statement a is simply the degree of its logical weakness, or lack of informative
result is still held by many to be 'paradoxical'. 3 Despite this simple result content, and that the relative probability of a statement a, given a statement b,
the idea that a high degree of probability (in the sense of the calculus of is simply the degree of the relative weakness, or the relative lack of new in-
probability) must be something highly desirable seems to be so obvious to formative content in statement a, assuming that we are already in possession
most peop.le that they are not prepared to consider it critically. Dr Bruce of the information b.
Brooke-Wavell has therefore suggested to me that I should stop talking in Thus if we aim, in science, at a high informative content-if the growth of
this context of 'probability' and should base my arguments on a 'calculus of knowledge means that we know more, that we know a and b, rather than a
content' and of 'relative content'; or in other words, that I should not speak alone, and that the content of our theories thus increases-then we have to
about science aiming at improbability; but merely say that it aims at maxi- admit that we also aim at a low probability, in the sense of the calculus of
mum content. I have given much thought to this suggestion, but I do not probability.
think that it would help: a head-on collision with the widely accepted and And since a low probability means a high probability of being falsified, it
follows that a high degree of falsifiability, or refutability, or testability, is one
3 See for example J. C. Harsanyi, 'Popper's Improbability Criterion for the Choice of of the aims of science-in fact, precisely the same aim as a high informative
ScientificHypotheses', Philosophy, 35, 1960, pp. 332 ff. Incidentally, I do not propose any content.
criterion' for the choice of scientific hypotheses: every choice remains a risky guess.
Moreover, the theoretician's choice is the hypothesis most worthy of further critical dis- The criterion of potential satisfactoriness is thus testability, or improb-
cussion(rather than of acceptance). ability: only a highly testable or improbable theory is worth testing, and is
218 219

'III~I~
.
........
 CONJECTURES 10 TRUTH. RATIONALITY,ANDTHE GROWTHOF KNOWLEDGE

actually (and not merely potentially) satisfactory if it withstands severe tests even though they had accidental components: every one of these men was
searching for an effect of the kind he found.
-especially those tests to which we could point as crucial for the theory III
We can even say that some discoveries, such as Columbus' discovery of
before they were ever undertaken.
It is possible in many cases to compare the severity oftests objectively. It is America, corroborate one theory (of the spherical earth) while refuting at
the same time another (the theory of the size of the earth, and with it, of the
even possible, if we find it worth while, to define a measure ofthe severity of
tests. (See the Addenda to this volume.) By the same method we can define
nearest way to India); and that they were chance-discoveries to the extent to II
the explanatory power and the degree of corroboration of a theory.4 which they contradicted all expectations, and were not consciously under-
taken as tests of those theories which they refuted.
IV
v
The thesis that the criterion here proposed actually dominates the progress of
science can easily be illustrated with the help of historical examples. The The stress I am laying upon change in scientific knowledge, upon its growth,
or its progressiveness, may to some extent be contrasted with the current ideal jlll
theories of Kepler and Galileo were unified and superseded by Newton's
of science as an axiomatized deductive system. This ideal has been dominant
logically stronger and better testable theory, and similarly Fresnel's and
Faraday's by Maxwell's. Newton's theory, and Maxwell's, in their turn, were in European epistemology from Euclid's Platonizing cosmology (for this is,
unified and superseded by Einstein's. In each such case the progress was I believe, what Euclid's Elements were really intended to be) to that of Newton,
towards a more informative and therefore logically less probable theory: and further to the systems of Boscovic, Maxwell, Einstein, Bohr, Schriidinger,
towards a theory which was more severely testable because it made pre- and Dirac. It is an epistemology that sees the final task and end of scientific
dictions which, in a purely logical sense, were more easily refutable. activity in the construction of an axiomatized deductive system. "
A theory which is not in fact refuted by testing those new and bold and As opposed to this, I now believe that these most admirable deductive
improbable predictions to which it gives rise can be said to be corroborated systems should be regarded as stepping stones rather than as ends:s as im-
by these severe tests. I may remind you in this connection of Galle's discovery portant stages on our way to richer, and better testable, scientific knowledge.
of Neptune, of Hertz's discovery of electromagnetic waves, of Eddington's Regarded thus as means or stepping stones, they are certainly quite in-
eclipse observations, of Elsasser's interpretation of Davisson's maxima as dispensable, for we are bound to develop our theories in the form of deductive
interference fringes of de Broglie waves, and of Powell's observations of the systems. This is made unavoidable by the logical strength, by the great in-
first Yukawa mesons. formative content, which we have to demand of our theories if they are to be
All these discoveries represent corroborations by severe tests-by predic- better and better testable. The wealth of their consequences has to be un-
tions which were highly improbable in the light of our previous knowledge folded deductively; for as a rule, a theory cannot be tested except by testing,
(previous to the theory which was tested and corroborated). Other important one by one, some of its more remote consequences; consequences, that is,
discoveries have also been made while testing a theory, though they did not which cannot immediately be seen upon inspecting it intuitively.
lead to its corroboration but to its refutation. A recent and important case is Yet it is not the marvellous deductive unfolding of the system which makes
the refutation of parity. But Lavoisier's classical experiments which show that a theory rational or empirical but the fact that we can examine it critically;
the volume of air decreases while a candle burns in a closed space, or that the that is to say, subject it to attempted refutations, including observational
weight of burning iron-filings increases, do not establish the oxygen theory of tests; and the fact that, in certain cases, a theory may be able to withstand
combustion; yet they tend to refute the phlogiston theory. those criticisms and those tests-among them tests under which its pre-
Lavoisier's experiments were carefully thought out; but even most so- decessors broke down, and sometimes even further and more severe tests. It
called 'cbance-discoveries' are fundamentally of the same logical structure. is in the rational choice of the new theory that the rationality of science lies,
For these so-called 'chance-discoveries' are as a rule refutations of theories rather than in the deductive development of the theory.
which were consciously or unconsciously held: they are made when some of Consequently there is little merit in formalizing and elaborating a deductive
our expectations (based upon these theories) are unexpectedly disappointed. non-conventional system beyond the requirements of the task of criticizing
Thus the catalytic property of mercury was discovered when it was accident- and testing it, and of comparing it critically with competitors. This critical
ally found that in its presence a chemical reaction had been speeded up which
SI have been influenced in adopting this view by Dr J. Agassi who, in a discussion in
had not been expected to be influenced by mercury. But neither Oersted's nor 1956,convinced me that the attitude of looking upon the finished deductive systems as an
Riintgen's nor Becquerel's nor Fleming's discoveries were really accidental, end is a relic of the long domination of Newtonian ideas (and thus, I may add, of the II
Platonic, and Euclidean, tradition). For an even more radical view of Dr Agassi's see the
4 See especially appendix *ix to my L.Se.D. last footnote to this chapter.
220 221
 CON1ECTURES 10 TRUTH, RATIONALITY, AND THE OROWTH OF KNOWLEDOE II~I
I
comparison, though it has, admittedly, some minor conventional and arbi- 2. THB THBORY OF OB1BCTIVBTRUTH: CORRBSPONDBNCB I

trary aspects, is largely non-conventional, thanks to the criterion of progress. TO THB FACTS
It is this critical procedure which contains both the rational and the empirical VII
elements of science. It contains those choices, those rejections, and those
decisions, which show that we have learnt from our mistakes, and thereby So far I have spoken about science, its progress, and its criterion of progress I
added to our scientific knowledge. without even mentioning truth. Perhaps surprisingly, this can be done without
IIII!
VI falling into pragmatism or inStrumentalism. Indeed, it is even possible to
argue in favour of the intuitive satisfactoriness of the criterion of progress in
Yet perhaps even this picture of science-as a procedure whose rationality science without ever speaking about the truth of its theories. In fact, before I
consists in the fact that we learn from our mistakes-is not quite good enough. became acquainted with Tarski's theory oftruth,7 it appeared to me safer and
It may still suggest that science progresses from theory to theory and that it more economical to discuss the criterion of progress without getting too
consists of a sequence of better and better deductive systems. Yet what I deeply involved in the highly controversial problem connected with the use
really wish to suggest is that science should be visualized as progressingfrom of the word 'true'.
problems to problems-to problems of ever increasing depth. My attitude at the time was this: although I accepted, as almost every-
For a scientific theory-an explanatory theory-is, if anything, an attempt body does, the objective or absolute or correspondence theory of truth-
to solve a scientific problem, that is to say, a problem concerned or connected truth as correspondence with the facts-I preferred to avoid the topic. For it
with the discovery of an explanation.6 appeared to me hopeless to try to understand clearly this strangely elusive
Admittedly, our expectations, and thus our theories, may precede, his- idea of a correspondence between a statement and a fact.
torically, even our problems. Yet science starts only with problems. Problems In order to recall why the situation appeared so hopeless we only have to
crop up especially when we are disappointed in our expectations, or when our remember, as one example among many, Wittgenstein's Tractatus with its
theories involve us in difficulties, in contradictions; and these may arise surprisingly naive picture theory, or projection theory, of truth. In this book
either within a theory, or between two different theories, or as the result of a a proposition was conceived as a picture or projection of the fact which it was
clash between our theories and our observations. Moreover, it is only through intended to describe and as having the same structure (or 'form') as that
a problem that we become conscious of holding a theory. It is the problem fact; just as a gramophone record is indeed a picture or a projection of a
which challenges us to learn; to advance our knowledge; to experiment; and sound, and shares some of its structural properties.8
to observe. Another of these unavailing attempts to explain this correspondence was
Thus science starts from problems, and not from observations; though due to Schlick, who gave a beautifully clear and truly devastating criticism9
observations may give rise to a problem, especially if they are unexpected; of various correspondence theories-including the picture or projection
that is to say, if they clash with our expectations or theories. The conscious theory-but who unfortunately produced in his turn another one which was
task before the scientist is always the solution of a problem through the con- no better. He interpreted the correspondence in question as a one-one cor-
struction of a theory which solves the problem; for example, by explaining respondence between our designations and the designated objects, although
unexpected and unexplained observations. Yet every worthwhile new theory counter examples abound (designations applying to many objects, objects
raises new problems; problems of reconciliation, problems of how to conduct designated by many designations) which show that this interpretation is
new and previously unthought-of observational tests. And it is mainly untenable.
through tIie new problems which it raises that it is fruitful. All this was changed by Tarski's theory of truth and of the correspondence
Thus we may say that the most lasting contribution to the growth of of a statement with the facts. Tarski's greatest achievement, and the real
scientific knowledge that a theory can make are the new problems which it significance of his theory for the philosophy of the empirical sciences lies, I
raises, so that we are led back to the view of science and of the growth of believe, in the fact that he re-established a correspondence theory of absolute
knowledge as always starting from, and always ending with, problems- or objective truth which showed that we are free to use the intuitive idea of
problems of an ever increasing depth, and an ever increasing fertility in truth as correspondence with the facts. (The view that his theory is applicable Ilil
suggesting new problems. only to formalized languages is, I think, mistaken. It is applicable to any !Ii
consistent and-more or less-'natural' language. So we must try to learn
6 Compare this and the following two paragraphs with my Poverty of Historicism, section !'
28, pp. 121 ff., and chs. 1 and 16 of this volume. 7 See my L.Se.D., especially section 84.
8Cp. Wittgenstein's Tractatus, especially4.0141; also 2.161; 2.17; 2.223; 3.11. lull
9 See especiallypp. 56-7 of his remarkable Erkenntnislehre,2nd edn., 1925.
222 223
1111111

IIII11
 CONJECTURES 10 TRUTH,RATIONALITY,ANDTHE GROWTHOF KNOWLEDGE
from Tarski's analysis how to dodge its inconsistencies; which means, for truth-these are all subjective (or 'epistemic') theories oftruth, in contra-
admittedly, the introduction of a certain amount of 'artificiality'-or caution distinction to Tarski's objective (or 'metalogical') theory. They are subjective
-into its use.) in the sense that they all stem from the fwuJamental subjectivist position
Although I may assume in this assembly some familiarity with Tarski's which can conceive of knowledge only as a special kind of mental state, or as a
theory of truth, I may perhaps explain the way in which it can be regarded, disposition, or as a special kind of belief, characterized, for example, by its
from an intuitive point of view, as a simple elucidation of the idea of corres- history or by its relation to other beliefs.
pondence with the facts. I shall have to stress this almost trivial point be- If we start from our subjective experience of believing, and thus look upon
cause, in spite of its triviality, it will be crucial for my argument. knowledge as a special kind of belief, then we may indeed have to look upon
The highly intuitive character of Tarski's ideas seems to become more truth-that is, true knowledge-as some even more special kind of belief: as
evident (as I have found in teaching) if we first decide explicitly to take 'truth' one that is well-founded or justified. This would mean that there should be
as a synonym for 'correspondence with the facts', and then (forgetting all some more or less effectivecriterion, if only a partial one, of well-foundedness;
about 'truth') proceed to define the idea of' correspondence with the facts'. some symptom by which to differentiate the experience of a well.founded
Thus we shall first consider the following two formulations, each of which belief from other experiences of belief. It can be shown that all subjective
states very simply (in a metalanguage) under what conditions a certain theories of truth aim at such a criterion: they try to define truth in terms of
assertion (in an object language) corresponds to the facts. the sources or origins of our beliefs,10 or in terms of our operations of
(1) The statement, or the assertion, 'Snow is white' corresponds to the verification, or of some set of rules of acceptance, or simply in terms of the
facts if, and only if, snow is, indeed, white. quality of our subjective convictions. They all say, more or less, that truth is
(2) The statement, or the assertion, 'Grass is red' corresponds to the facts what we are justified in believing or in accepting, in accordance with certain
if, and only if, grass is, indeed, red. rules or criteria, of origins or sources of our knowledge, or of reliability, or
These formulations (in which the word 'indeed' is only inserted for ease, stability, or biological success, or strength of conviction, or inability to think
and may be omitted) sound, of course, quite trivial. But it was left to Tarski otherwise.
to discover that, in spite of their apparent triviality, they contained the The objective theory of truth leads to a very different attitude. This may be
solution of the problem of explaining correspondence with the facts and, seen from the fact that it allows us to make assertions such as the following:
with it, truth. a theory may be true even though nobody believes it, and even though we
I have said that Schlick's theory was mistaken, yet I think that certain have no reason for accepting it, or for believing that it is true; and another
comments he made (loc. cit.) about his own theory throw some light on theory may be false, although we have comparatively good reasons for
Tarski's. For Schlick says that the problem of truth shared the fate of some accepting it.
others whose solutions were not easily seen because they were mistakenly sup- Clearly, these assertions would appear to be self-contradictory from the
posed to lie on a very deep level, while actually they were fairly plain and, at point of view of any subjective or epistemic theory of truth. But within
first sight, unimpressive. Tarski's solution may well appear unimpressive at the objective theory, they are not only consistent, but quite obviously true.
first sight. Yet its fertility and its power are impressive indeed. This, however, A similar assertion which the objective correspondence theory would make
is not my topic here. quite natural is this: even if we hit upon a true theory, we shall as a rule be
vm merely guessing, and it may well be impossible for us to know that it is true.
An assertion like this was made, apparently for the first time, by Xeno.
Thanks to Tarski's work, the idea of objective or absolute truth-that is phanesll who lived 2,500 years ago; which shows that the objective theory of
truth as correspondence with the facts-appears to be accepted today with truth is very old indeed-antedating Aristotle, who also held it. But only with
confidence by all who understand it. The difficulties in understanding it seem Tarski's work has the suspicion been removed that the objective theory of
to have two sources: first, the combination of an extremely simple intuitive truth as correspondence with the facts may be either self-contradictory
idea with a certain amount of complexity in the execution of the technical (because of the paradox of the liar), or empty (as Ramsey suggested), or
programme to which it gives rise; secondly, the widespread but mistaken barren, or at the very least redundant, in the sense that we can do without it
dogma that a satisfactory theory of truth would have to be a theory of true (as I once thought myself).
belief-of well-founded, or rational belief. Indeed, the three rivals of the In my theory of scientific progress I might perhaps do without it, up to a
correspondence theory of truth-the coherence theory which mistakes con- point. Since Tarski, however, I no longer see any reason why I should try to
sistency for truth, the evidence theory which mistakes 'known to be true' for 10See my Introduction to this volume, 'On the Sources of Knowledge and of Ignorance'.
'true', and the pragmatic or instrumentalist theory which mistakes usefulness 11 Seethe Introduction,p. 26,and ch. 5, p. 152r., above.
224 225
 CONJECTURES 10 TRUTH,RATIONALITY,ANDTHE GROWTHOF KNOWLEDGE
avoid it. And if we wish to elucidate the difference between pure and applied Though things have improved since then, subjectivism is still rampant in
science, between the search for knowledge and the search for power or for the philosophy of science, and especially in the field of probability theory.
powerful instruments, then we cannot do without it. For the difference is that, The subjectivist theory of probability, which interprets degrees of prob-
in the search for knowledge, we are out to find true theories, or at least ability as degrees of rational belief, stems directly from the subjectivist
theories which are nearer than others to the truth-which correspond better approach to truth-especially from the coherence theory. Yet it is still em-
to the facts; whereas in the search for theories that are merely powerful braced by philosophers who have accepted Tarski's theory of truth. At least
instruments for certain purposes, we are, in many cases, quite well served by some of them, I suspect, have turned to probability theory in the hope that it
theories which are known to be false.12 would give them what they had originally expected from a subjectivist or
So one great advantage of the theory of objective or absolute truth is that epistemological theory of the attainment of truth through verification; that is,
it allows us to say-with Xenophanes-that we search for truth, but may not a theory of rational and justifiable belief, based upon observed instances'!s
know when we have found it; that we have no criterion of truth, but are It is an awkward point in all these subjective theories that they are irrefu-
nevertheless guided by the idea of truth as a regulative principle (as Kant or table (in the sense that they can too easily evade any criticism). For it is
Peirce might have said); and that, though there are no general criteria by always possible to uphold the view that everything we say about the world, or
which we can recognize truth-except perhaps tautological truth-there are everything we print about logarithms, should be replaced by a belief-state-
something like criteria of progress towards the truth (as I shall explain ment. Thus we may replace the statement 'Snow is white' by 'I believe that
presently). snow is white' or perhaps even by 'In the light of all the available evidence I
The status of truth in the objective sense, as correspondence to the facts, believe that it is rational to believe that snow is white'. The possibility of re-
and its role as a regulative principle, may be compared to that of a mountain placing any assertion about the objective world by one of these subjectivist
peak which is permanently, or almost permanently, wrapped in clouds. The circumlocutions is trivial, though in the case of the assertions expressed in
climber may not merely have difficulties in getting there-he may not know logarithm tables-which might well be produced by machines-somewhat
when he gets there, because he may be unable to distinguish, in the clouds, unconvincing. (It may be mentioned in passing that the subjective interpreta-
between the main summit and some subsidiary peak. Yet this does not affect tion of logical probability links these subjectivist replacements, exactly as in
the objective existence of the summit, and if the climber tells us 'I have some the case of the coherence theory of truth, with an approach which, on closer
doubts whether I reached the actual summit', then he does, by implication, analysis, turns out to be essentially 'syntactic' rather than 'semantic'-
recognize the objective existence of the summit. The very idea of error, or of although it can of course always be presented within the framework of a
doubt (in its normal straightforward sense) implies the idea of an objective 'semantical system'.)
truth which we may fail to reach. It may be useful to sum up the relationships between the objective and
Though it may be impossible for the climber ever to make sure that he has subjective theories of scientific knowledge with the help of a little table:
reached the summit, it will often be easy for him to realize that he has not OBJBCTIVEOR LOGICALOR SUBJECTIVBOR PSYCHOLOGICALOR
reached it (or not yet reached it); for example, when he is turned back by an ONTOLOGICAL THEORIES BPISTEMOLOGICAL THEORIES
overhanging wall. Similarly, there will be cases when we are quite sure that truth as correspondence truth as property of our state
we have not reached the truth. Thus while coherence, or consistency, is no with the facts of mind-or knowledge or belief
criterion of truth, simply because even demonstrably consistent systems may objective probability subjective probability
be false in fact, incoherence or inconsistency do establish falsity; so, if we are (inherent in the situation, and (degree of rational belief based
lucky, we may discover inconsistencies and use them to establish the falsity testable by statistical tests) upon our total knowledge)
of some of our theories.13 objective randomness lack of knowledge
In 1944, when Tarski published the first English outline of his investiga- (statistically testable)
tions into the theory of truth (which he had published in Poland in 1933), equiprobability lack of knowledge
few philosophers would have dared to make assertions like those of Xeno- (physical or situational symmetry)
phanes; and it is interesting that the volume in which Tarski's paper was In all these cases I am inclined to say not only that these two approaches
published also contained two subjectivist papers on truth.14 should be distinguished, but also that the subjectivist approach should be
discarded as a lapse, as based on a mistake-though perhaps a tempting
12See the discussion of the 'second view' (called 'instrumentalism') in ch. 3, above.
13See Alfred Tarslei's paper, 'The Semantic Conception of Truth', in Philosophy and I' Cpo Carnap, Logical Foundations o/Probability, 1950, p. 177. Cp. my L.Sc.D., especially
Phenom. Research, 4, 1943-4, pp. 341 ff. (Cp. especially section 21.) section 84.
14See the volume referred to in the preceding note, especiallypp. 279 ;wd 336. 227
226
 illl
CONJECTURES 10 TRUTH,RATIONALITY,AND THE GROWTHOF KNOWLEDGE 1111

mistake. There is, however, a similar table in which the epistemological its habit of appealing to empirical evidence in support of its dogmas-
(right hand) side is not based on a mistake. astrologers do so too-but solely in the critical approach-in an attitude I111

III~
which, of course, involves the critical use, among other arguments, of em-
truth conjecture
pirical evidence (especially in refutations). For us, therefore, science has
testability empirical test nothing to do with the quest for certainty or probability or reliability. We are
explanatory or predictive power degree of corroboration not interested in establishing scientific theories as secure, or certain, or prob- 'il
II
(that is, report of the results able. Conscious of our fallibility we are only interested in criticizing them II
.verisimilitude' of tests) and testing them, in the hope of finding out where we are mistaken; of
learning from our mistakes; and, if we are lucky, of proceeding to better
3. TRUTH AND CONTENT: VERISIMILITUDE VERSUS PROBABILITY theories.
Considering their views about the positive or negative function of argu-
IX
ment in science, the first group-the justificationists-may be also nicknamed
Like many other philosophers I am at times inclined to classify philosophers the 'positivists' and the second-the group to which I belong-the critics or
as belonging to two main groups-those with whom I disagree, and those who the 'negativists'. These are, of course, mere nicknames. Yet they may perhaps
agree with me. I also call them the verificationists or the justificationist suggest some of the reasons why some people believe that only the positivists
philosophers of knowledge (or of belief), and the falsificationists or fallibilists or verificationists are seriously interested in truth and in the search for truth,
or critical philosophers of knowledge (or of conjectures). I may mention in while we, the critics or negativists, are flippant about the search for truth, and
passing a third group with whom I also disagree. They may be cal1ed the addicted to barren and destructive criticism and to the propounding of
disappointed justificationists-the irrationalists and sceptics. views which are clearly paradoxical.
The members of the first group-the verificationists or justificationists- This mistaken picture of our views seems to result largely from the adop-
hold, roughly speaking, that whatever cannot be supported by positive tion of a justificationist programme, and of the mistaken subjectivist approach
reasons is unworthy of being believed, or even of being taken into serious to truth which I have described.
consideration. For the fact is that we too see science as the search for truth, and that, at
On the other hand, the members of the second group-the falsificationists least since Tarski, we are no longer afraid to say so. Indeed, it is only with
or fallibilists-say, roughly speaking, that what cannot (at present) in prin- respect to this aim, the discovery of truth, that we can say that though we are
ciple be overthrown by criticism is (at present) unworthy of being seriously fallible, we hope to learn from our mistakes. It is only the idea of truth
considered; while what can in principle be so overthrown and yet resists all which al10ws us to speak sensibly of mistakes and of rational criticism, and
our critical efforts to do so may quite possibly be false, but is at any rate which makes rational discussion possible-that is to say, critical discussion in
not unworthy of being seriously considered and perhaps even of being search of mistakes with the serious purpose of eliminating as many of these
believed-though only tentatively. mistakes as we can, in order to get nearer to the truth. Thus the very idea of
Verificationists, I admit, are eager to uphold that most important tradition error-and of fallibility-involves the idea of an objective truth as the
of rationalism-the fight of reason against superstition and arbitrary author- standard of which we may fall short. (It is in this sense that the idea of truth
ity. For they demand that we should accept a belief only ifit can be justified is a regulative idea.)
by positive ev.idence;that is to say, shown to be true, or, at least, to be highly Thus we accept the idea that the task of science is the search for truth, that
probable. In other words, they demand that we should accept a belief only is, for true theories (even though as Xenophanes pointed out we may never
if it can be verified, or probabilistically confirmed. get them, or know them as true if we get them). Yet we also stress that truth 1111111,
Falsificationists (the group of fallibilists to which I belong) believe-as is not the only aim of science. We want more than mere truth: what we look
most irrationalists also believe-that they have discovered logical arguments for is interesting truth-truth which is hard to come by. And in the natural
which show that the programme of the first group cannot be carried out: sciences (as distinct from mathematics) what we look for is truth which has
that we can never give positive reasons which justify the belief that a theory is a high degree of explanatory power, in a sense which implies that it is logically
true. But, unlike irrationalists, we falsificationists believe that we have also improbable truth.
discovered a way to realize the old ideal of distinguishing rational science For it is clear, first of all, that we do not merely want truth-we want more
from various forms of superstition, in spite of the breakdown of the original truth, and new truth. We are not content with 'twice two equals four', even
inductivist or justificationist programme. We hold that this ideal can be though it is true: we do not resort to reciting the multiplication table if we
realized, very simply, by recognizing that the rationality of science lies not in are faced with a difficult problem in topology or in physics. Mere truth is not
228 229

'('-".'-';c
~" "".: "'F"'~'I':'
 ........

CONJECTURES 10 TRUTH. RATIONALITY.ANDTHE GROWTHOF KNOWLEDGE

enough; what we look for are answers to our problems. The point has been even if these should soon turn out to be false, then this interest is due to our
well put by the German humorist and poet Busch, of Max-and-Moritz fame, methodological conviction that only with the help of such bold conjectures
in a little nursery rhyme-I mean a rhyme for the epistemological nursery: 16 can we hope to discover interesting and relevant truth.
Twice two equals four: 'tis true, There is a point here which, I suggest, it is the particular task of the logician
But too empty, and too trite. to analyse. 'Interest', or 'relevance', in the sense here intended, can be
What I look for is a clue objectively analysed; it is relative to our problems; and it depends on the
To some matters not so light. explanatory power, and thus on the content or improbability, of the informa-
Only if it is an answer to a problem-a difficult, a fertile problem, a prob- tion. The measures alluded to earlier (and developed in the Addenda to this
lem of some depth-does a truth, or a conjecture about the truth, become volume) are precisely such measures as take account of some relative content
relevant to science. This is so in pure mathematics, and it is so in the natural of the information-its content relative to a hypothesis or to a problem.
sciences. And in the latter, we have something like a logical measure of the I can therefore gladly admit that falsificationists like myself much prefer
depth or significance of the problem in the increase of logical improbability an attempt to solve an interesting problem by a bold conjecture, even (and
or explanatory power of the proposed new answer, as compared with the best especially) if it soon turns out to be false, to any recital of a sequence of
theory or conjecture previously proposed in the field. This logical measure is irrelevant truisms. We prefer this because we believe that this is the way in
essentially the same thing which I have described above as the logical criterion which we can learn from our mistakes; and that in finding that our con-
of potential satisfactoriness and of progress. jecture was false, we shall have learnt much about the truth, and shall have
My description of this situation might tempt some people to say that truth got nearer to the truth.
does not, after all, playa very big role with us negativists even as a regulative I therefore hold that both ideas-the idea of truth, in the sense of corres-
principle. There can be no doubt, they will say, that negativists (like myself) pondence with the facts, and the idea of content (which may be measured by
much prefer an attempHo solve an interesting problem by a bold conjecture, the same measure as testability)-play about equally important roles in our
even if it soon turns out to befalse, to any recital of a sequence of true but un- considerations, and that both can shed much light on the idea of progress in
science.
interesting assertions. Thus it does not seem, after all, as if we negativists
had much use for the idea of truth. Our ideas of scientific progress and of x
attempted problem-solving do not seem very closely related to it.
This, I believe, would give quite a mistaken impression of the attitude of Looking at the progress of scientific knowledge, many people have been
moved to say that even though we do not know how near or how far we are
our group. Call us negativists, or what you like: but you should realize that
from the truth, we can, and often do, approach more and more closely to the
we are as much interested in truth as anybody-for example, as the members
of a court of justice. When the judge tells a witness that he should speak 'The truth. I myself have sometimes said such things in the past, but always with
truth, the whole truth, and nothing but the truth', then what he looks for is as a twinge of bad conscience. Not that I believe in being over-fussy about what
we say: as long as we speak as clearly as we can, yet do not pretend that what
much of the relevant truth as the witness may be able to offer. A witness who
likes to wander off into irrelevancies is unsatisfactory as a witness, even we are saying is clearer than it is, and as long as we do not try to derive
though these irrelevancies may be truisms, and thus part of 'the whole truth'. apparently exact consequences from dubious or vague premises, there is no
harm whatever in occasional vagueness, or in voicing every now and then our
It is quite obvious that what the judge-or anybody else-wants when he
asks for 'the whole truth' is as much interesting and relevant true information feelings and general intuitive impressions about things. Yet whenever I used
to write, or to say, something about science as getting nearer to the truth,
as can be got; and many perfectly candid witnesses have failed to disclose
some important information simply because they were unaware of its rele- or as a kind of approach to truth, I felt that I really ought to be writing
vance to the case. 'Truth', with a capital 'T', in order to make quite clear that a vague and higWy
metaphysical notion was involved here, in contradistinction to Tarski's
Thus when we stress, with Busch, that we are not interested in mere truth
'truth' which we can with a clear conscience write in the ordinary way with
but in interesting and relevant truth, then, I contend, we only emphasize a smallletters.17
point which everybody accepts. And if we are interested in bold conjectures, It was only quite recently that I set myself to consider whether the idea of
16From W. Busch, Schein und Sein (first published posthumously in 1909; p. 28 of the truth involved here was really so dangerously vague and metaphysical after
[nseledition. 1952). My attention has been drawn to this rhyme by an essay on Busch as a
philosopher which my late friend Julius Kraft contributed to the volume Erziehung und 17Similarmisgivingsare expressedby Quinewhenhe criticizesPeircefor operatingwith
Politik (Essays for Minna Specht, 1960);see p. 262. My translation makes it perhaps more the ideaof approachingto truth. SeeW. V. Quine,Wordand Object,New York. 1960,
like a nursery rhyme than Busch intended. p.23.
230 231
 "
CONJECTURES
10 TRUTH, RATIONALITY, AND THE GROWTH OF KNOWLEDGE ~
all. Almost at once I found that it was not, and that there was no particular
difficulty in applying Tarski's fundamental idea to it. one-the idea of a degree of better (or worse) correspondence to truth or of
For there is no reason whatever why we should not say that one theory greater (or less) likeness or similarity to truth; or to use a term already
corresponds better to the facts than another. This simple initial step makes mentioned above (in contradistinction to probability) the idea of (degrees of)
verisimilitude.
everything clear: there really is no barrier here between what at first sight
appeared to be Truth with a capital 'T' and truth in a Tarskian sense. It should be noted that the idea that every statement or theory is not only
But can we really speak about better correspondence'1 Are there such either true or false but has, independently of its truth value, some degree of
things as degrees of truth'1 Is it not dangerously misleading to talk as if verisimilitude, does not give rise to any multi-valued logic-that is, to a
Tarskian truth were located somewhere in a kind of metrical or at least logical system with more than two truth values, true and false; though some
topological space so that we can sensibly say of two theories-sayan earlier of the things the defenders of multi-valued logic are hankering after seem to
theory tl and a later theory t2, that t2 has superseded t1>or progressed be realized by the theory of verisimilitude (and related theories alluded to in
beyond t1>by approaching more closely to the truth than tl '1 section 3 of the Addenda to this volume).
I do not think that this kind of talk is at all misleading. On the contrary, I XI
believe that we simply cannot do without something like this idea of a better
or worse approximation to truth. For there is no doubt whatever that we can Once I had seen the problem it did not take me long to get to this point. But
say, and often want to say, of a theory t2 that it corresponds better to the strangely enough, it took me a long time to put two and two together, and
facts, or that as far as we know it seems to correspond better to the facts, to proceed from here to a very simple definition of verisimilitude in terms of
than another theory tl' truth and of content. (We can use either logical or empirical content, and
I shall give here a somewhat unsystematic list of six types of case in which thus obtain two closely related ideas of verisimilitude which however merge
we should be inclined to say of a theory tl that it is superseded by t2 in the into one if we consider here only empirical theories, or empirical aspects of
theories.)
sense that t2 seems-as far as we know-to correspond better to the facts than
t1>in some sense or other. Let us consider the content of a statement a; that is, the class of all the
(1) t2 makes more precise assertions than t1>and these more precise logical consequences of a. If a is true, then this class can consist only of true
assertions stand up to more precise tests. statements, because truth is always transmitted from a premise to all its con-
clusions. But if a is false, then its content will always consist of both true and
(2) t2 takes account of, and explains, more facts than tl (which will include
for example the above case that, other things being equal, t2's assertions are false conclusions. (Example: 'It always rains on Sundays' is false, but its
more precise). conclusion that it rained last Sunday happens to be true.) Thus whether a
(3) t2 describes, or explains, the facts in more detail than tl' statement is true or false, there may be more truth, or less truth, in what il
(4) t2 has passed tests which tl has failed to pass. says, according to whether its content consists of a greater or a lesser number
of true statements.
(5) t2 has suggested new experimental tests, not considered before t2 was
Let us ca1lthe class of the true logical consequences of a the 'truth-content'
designed (and not suggested by t1>and perhaps not even applicable to tl);
and t2 has passed these tests. of a (a German term 'Wahrheitsgehalt'-reminiscent of the phrase 'there
is truth in what you say'-ofwhich 'truth-content' may be said to be a trans-
(6) t2 has unified or connected various hitherto unrelated problems.
lation, has been intuitively used for a long time); and let us call the class of
If we reflect upon this list, then we can see that the contents of the theories
tl and t2 play an important role in it. (It will be remembered that the logical the false consequences of a-but only these-the 'falsity-content' of a. (The
'falsity-content' is not, strictly speaking, a 'content', because it does not con-
content of a statement or a theory a is the class of all statements which follow tain any of the true conclusions of the false statements which form its ele-
logically from a, while I have defined the empirical content of a as the class of ments. Yet it is possible-see the Addenda-to define its measure with the
all basic statements which contradict a.18) For in our list of six cases, the help of two contents.) These terms are precisely as objective as the terms
empirical content of theory t2 exceeds that of theory tl. 'true' or 'false' and 'content' themselves. Now we can say:
This suggests that we combine here the ideas of truth and of content into
Assuming that the truth-content and thefalsity-content of two theories t1and
tz are comparable, we can say that t2 is more closely similar to the truth, or
18This definition is logicallyjustified by the theorem that, so far as the 'empirical part'
of the logicalcontent is concerned, comparison of empirical contents and of logical contents corresponds better to the facts, than t1>if and only if either
always yield the same results; and it is intuitivelyjustified by the consideration that a state- (a) the truth-content but not the falsity-content oft2 exceeds that oft1>
ment a tells the more about our world of experience the more possible experiences it (b) the falsity-content oft1> but not its truth-content, exceeds that oftz.
excludes (or forbids). About basic statements see also the Addenda to this volume.
232 Ifwe now work with the (perhaps fictitious) assumption that the content and
233
 ---

[:
CONJECTURES 10 TRUTH, RATIONALITY, AND THE GROWTH OF KNOWLEDGE

truth-content of a theory a are in principle measurable, then we can go slightly This leads me to my third point. Let me first say that I do not suggest that I ~I,
beyond this definition and can define Vs(a), that is to say, a measure of the the explicit introduction of the idea of verisimilitude will lead to any changes 1111'

verisimilitude or truthlikeness of a. The simplest definition will be in the theory of method. On the contrary, I think that my theory of test-
ability or corroboration by empirical tests is the proper methodological
Vs(a) = CtT(a) - CtF(a) counterpart to this new metalogical idea. The only improvement is one of
where CtT(a) is a measure of the truth-content of a, and CfF(a) is a measure clarification. Thus I have often said that we prefer the theory t2 which has ~I

of the falsity-content of a. A slightly more complicated but in some respects passed certain severe tests to the theory t1 which has failed these tests,
because a false theory is certainly worse than one which, for all we know, may lu
preferable definition will be found in section 3 of the Addenda to the present
volume. be true.
It is obvious that Vs(a) satisfies our two demands, according to which To this we can now add that even after 12has been refuted in its turn, we
Vs(a) should increase can still say that it is better than t10for although both have been shown to be
false, the fact that t2 has withstood tests which t1 did not pass may be a good '1
(a) if CtT(a) increases while CtAa) does not, and
indication that the falsity-content of t1 exceeds that of t2 while its truth- "
(b) if CtF(a) decreases while CtT(a) does not. "
Some further considerations of a slightly technical nature and the defini- content does not. Thus we may still give preference to t2, even after its
tions of CtT(a) and especially CtAa) and Vs(a) will be found in the Addenda. falsification, because we have reason to think that it agrees better with the
Here I want only to discuss three non-technical points. facts than did fl'
XII All cases where we accept t2 because of experiments which were crucial
between t2 and tl seem to be of this kind, and especially all cases where the
The first point is this. Our idea of approximation to truth, or of verisimilitude, experiments were found by trying to think out, with the help of t2, cases where
has the same objective character and the same ideal or regulative character as t2 leads to other results than did t1' Thus Newton's theory allowed us to
the idea of objective or absolute truth. It is not an epistemological or an predict some deviations from Kepler's laws. Its success in this field estab-
epistemic idea-no more than truth or content. (In Tarski's terminology, it is lished that it did not fail in cases which refuted Kepler's: at least the now
obviously a 'semantic' idea, like truth, or like logical consequence, and, known falsity-content of Kepler's theory was not part of Newton's, while it
therefore, content.) Accordingly, we have here again to distinguish between was pretty clear that the truth-content could not have shrunk, since Kepler's
the question 'What do you intend to say if you say that the theory t2 has a theory followed from Newton's as a 'first approximation'.
higher degree of verisimilitude than the theory t11', and the question 'How do Similarly, a theory t2 which is more precise than tl can now be shown to
you know that the theory t2 has a higher degree of verisimilitude than the have-always provided its falsity content does not exceed that of t1-a
theory t1l' higher degree of verisimilitude than t1' The same will hold for t2 whose
We have so far answered only the first of these questions. The answer to numerical assertions, though false, come nearer to the true numerical values
the second question depends on it, and it is exactly analogous to the following than those of t1'
(absolute rather than comparative) question about truth: 'I do not know-I Ultimately, the idea of verisimilitude is most important in cases where we
only guess. But I can examine my guess critically, and if it withstands severe know that we have to work with theories which are at best approximations-
criticism, then this fact may be taken as a good critical reason in favour ofit.' that is to say, theories of which we actually know that they cannot be true.
My second point is this. Verisimilitude is so defined that maximum veri- (This is often the case in the social sciences.) In these cases we can still speak
similitude would be achieved only by a theory which is not only true, but of better or worse approximations to the truth (and we therefore do not need
completely comprehensively true: if it corresponds to all facts, as it were, and, to interpret these cases in an instrumentalist sense). ~!
of course, only to real facts. This is of course a much more remote and un- XIII
attainable ideal than a mere correspondence with some facts (as in, say, II
'Snow is usually white'). It always remains possible, of course, that we shall make mistakes in our
But all this holds only for the maximum degree of verisimilitude, and not relative appraisal of two theories, and the appraisal will often be a con-
for the comparison of theories with respect to their degree of verisimilitude. troversial matter. This point can hardly be over-emphasized. Yet it is also
This comparative use of the idea is its main point; and the idea of a higher or important that in principle, and as long as there are no revolutionary changes
lower degree of verisimilitude seems less remote and more applicable and in our background knowledge, the relative appraisal of our two theories, t1
therefore perhaps more important for the analysis of scientific methods than and t2, will remain stable. More particularly, our preferences need not t
the-in itself much more fundamental-idea of absolute truth itself. change, as we have seen, if we eventually refute the better of the two theories.
234 235 I~
i
 --
10 TRUTH, RATIONALITY, AND THE GROWTH OF KNOWLEDGE
CONJECTURES
'Now of this world thus arranged to seem wholly like truth I shall tell
Newton's dynamics, for example, even though we may regard it as refuted,
has of course maintained its superiority over Kepler's and Galileo's theories.
you. . .'
Yet already in the same generation or the next, Epicharmus, in a criticism
The reason is its greater content or explanatory power. Newton's theory of Xenophanes, seems to have used the word 'eikotos' in the sense of
continues to explain more facts than did the others; to explain them with
'plausible', or something like it (DK,21 A 15); though the possibility cannot be
greater precision; and to unify the previously unconnected problems of
excluded that he may have used it in the sense of 'like the truth', and that it
celestial and terrestrial mechanics. The reason for the stability of relative
was Aristotle (our sour~e is Met., IOlOa4) who read it in the sense of 'plaus-
appraisals such as these is quite simple: the logical relation between the ible' or 'likely'. Some three generations later, however, 'eikos' is used quite
theories is of such a character that, first of all, there exist with respect to them
unambiguously in the sense of 'likely' or 'probable' (or perhaps even of 'more
those crucial experiments, and these, when carried out, went against Newton's
frequently than not') by the sophist Antiphon when he writes (DK,B 60): 'If
predecessors. And secondly, it is of such a character that the later refutations one begins a thing well it is likely to end well.'
of Newton's theory could not support the older theories: they either did not
All this suggests that the confusion between verisimilitude and probability
affect them, or (as with the perihelion motion of Mercury) they could be
goes back almost to the beginning of Westem philosophy: and this is under-
claimed to refute the predecessors also.
standable if we consider that Xenophanes stressed the fallibility of our
I hope that I have explained the idea of better agreement with the facts, or
knowledge which he described as uncertain guesswork and at best 'like the
of degrees of verisimilitude, sufficiently clearly for the purpose of this brief
truth'. This phrase, it seems, lent itself to misinterpretation as 'uncertain and
survey.
XIV at best of some fair degree of certainty'-that is, 'probable'.
Xenophanes himself seems to have distinguished clearly between degrees
A brief remark on the early history of the confusion between verisimilitude of certainty and degrees of truthlikeness. This emerges from another frag-
and probability may perhaps be appropriate here. ment (quoted above towards the end of chapter 5, p. 153) which says that
As we have seen, progress in science means progress towards more interest- even if by chance we were to hit upon, and pronounce, the final truth (that is,
ing, less trivial, and therefore less 'probable' theories (where 'probable' is we may add, perfect truthlikeness), we should not know it. Thus great un-
taken in any sense, such as lack of content, or statistical frequency, that certainty is compatible with greatest truthlikeness.
satisfies the calculus of probability) and this means, as a rule, progress I suggest that we return to Xenophanes and re-introduce a clear distinction
towards less fanilliar and less comfortable or plausible theories. Yet the idea between verisimilitude and probability (using this latter term in the sense laid
of greater verisimilitude, of a better approximation to the truth, is usually down by the calculus of probability).
confused, intuitively, with the totaI1y different idea of probability (in its The differentiation between these two ideas is the more important as they
various senses of 'more likely than not', 'more often than not', 'seems likely have become confused; because both are closely related to the idea of truth,
to be true', 'sounds plausible', 'sounds convincing'). The confusion is a very and both introduce the idea of an approach to truth by degrees. Logical
old one. We have only to remember some of the other words for 'probable', probability (we do not discuss here physical probability) represents the idea
such as 'likely' which comes originally from 'like the truth' or 'verisimilar' of approaching logical certainty, or tautological truth, through a gradual
('eoikotos', 'eikotos', 'eikos,' etc., in Greek; 'verisimilis' in Latin; 'wahrschein- diminution of informative content. Verisimilitude, on the other hand,
lich' in German) in order to see some of the traces, and perhaps some of represents the idea of approaching comprehensive truth. It thus combines
the sources, of this confusion. truth and content while probability combines truth with lack of content.20
Two at least of the earliest of the Presocratic philosophers used 'eoikota' The feeling that it is absurd to deny that science aims at probability stems,
in the sense of 'like the truth' or 'similar to the truth'. Thus we read in Xeno- I suggest, from a misguided 'intuition'-from the intuitive confusion between
phanes (DK,B 35): 'These things, let us suppose, are like the truth.' the two notions of verisimilitude and of probability which, as it now turns
It is fairly clear that verisimilitude or truthlikeness is meant here, rather out, are utterly different.
than probability or degree of incomplete certainty. (Otherwise the words 'let
us suppose' or 'let it be conjectured' or 'let it be imagined' would be re- passage thus: 'This world-arrangement (or world-order) I shall expound to you in all its
parts as something probable and plausible.' In translating '(wholly) like truth' or '(wholly)
dundant, and Xenophanes would have written something like, 'These things, like the truth', I am somewhat influenced by the line (DK,B 35) quoted above from Xeno-
let it be said, are probable'.) phanes (and also by K. Reinhardt's Parmenides,p. 5 f., where Wilamowitz is referred to).
Using the same word ('eoikota'), Parmenides wrote (DK, B 8, 60):19 See also section vii of the Introduction to the present volume; the quotation from Osiander
in section i of ch. 3; section xii of ch. 5, above; and Addendum 6, below.
11In this fragment'eoikota'has beenmostfrequentlytranslatedas 'probable'or'plaus- 20This, incidentally, holds for both, absolute probability, p(a), and relative probability,
ible'. For exampleW. Kranz, in Diels-Kranz,Fragmenteder Vorsokratiker,6th edn., p(a, b); and there are corresponding absolute and relative concepts ofverisimi1itude.
translatesit 'wahrschein/ich-einleuchtend'
that is, 'probable and plausible';he reads the 237
236

""""--
-
 ---
CONJECTURES 10 TRUTH, RATIONALITY,AND THE GROWTHOF KNOWLEDGE
4. BACKGROUND KNOWLEDGE AND SCIENTIFIC GROWTH not individually but only as a corporate body.21 Now it has to be admitted
xv that we can often test only a large chunk of a theoretical system, and some-
times perhaps only the whole system, and that, in these cases, it is sheer guess-
People involved in a fruitful critical discussion of a problem often rely, if only
work which of its ingredients should be held responsible for any falsification;
unconsciously, upon two things: the acceptance by all parties of the common
a point which I have tried to emphasize-also with reference to Duhem-for
aim of getting at the truth, or at least nearer to the truth, and a considerable
a long time past.22 Though this argument may turn a verificationist into a
amount of common background knowledge. This does not mean that either
sceptic, it does not affect those who hold that all our theories are guesses
of these two things is an indispensible basis of every discussion, or that these
anyway.
two things are themselves' a priori', and .cannot be critically discussed in their
This shows that the holistic view of tests, even if it were true, would not
turn. It only means that criticism never starts from nothing, even though
create a serious difficulty for the fallibilist and falsificationist. On the other
every one of its staiting points may be challenged, one at a time, in the course
of the critical debate. hand, it should be said that the holistic argument goes much too far. It is
possible in quite a few cases to find which hypothesis is responsible for the
Yet though every one of our assumptions may be challenged, it is quite
impracticable to challenge all of them at the same time. Thus all criticism refutation; or in other words, which part, or group of hypotheses, was
necessary for the derivation of the refuted prediction. The fact that such
must be piecemeal (as against the holistic view of Duhem and of Quine);
logical dependencies may be discovered is established by the practice of
which is only another way of saying that the fundamental maxim of every
independenceproofs ofaxiomatized systems; proofs which show that certain
critical discussion is that we should stick to our problem, and that we should
axioms of an axiomatic system cannot be derived from the rest. The more
subdivide it, if practicable, and try to solve no more than one problem at a
simple of these proofs consist in the construction, or rather in the discovery,
time, although we may, of course, always proceed to a subsidiary problem,
or replace our problem by a better one. of a model-a set of things, relations, operations, or functions-which satis-
fies all the axioms except the one whose independence is to be shown: for this
While discussing a problem we always accept (if only temporarily) all
one axiom-and therefore for the theory as a whole-the model constitutes a
kinds ofthings as unproblematic: they constitute for the time being, and for
counter example.
the discussion of this particular problem, what I call our background know
Now let us say that we have an axiomatized theoretical system, for example
ledge. Few parts of this background knowledge will appear to us in all con-
of physics, which allows us to predict that certain things do not happen, and
texts as absolutely unproblematic, and any particular part of it may be
that we discover a counter example. There is no reason whatever why this
challenged at any time, especially if we suspect that its uncritical acceptance
may be responsible for some of our difficulties. But almost all of the vast counter example may not be found to satisfy most of our axioms or even all
our axioms except one whose independence would be thus established. This
amount of background knowledge which we constantly use in any informal
shows that the holistic dogma of the 'global' character of all tests or counter
discussion will, for practical reasons, necessarily remain unquestioned; and
examples is untenable. And it explains why, even without axiomatizing our
the misguided attempt to question it all-that is to say, to start from scratch-
physical theory, we may well have an inkling of what has gone wrong with
can easily lead to the breakdown of a critical debate. (Were we to start the
our system.
race where Adam started, I know of no reason why we should get any further
than Adam did.) This, incidentally, speaks in favour of operating, in physics, with highly
XVI analysed theoretical systems-that is, with systems which, even though they
may fuse all the hypotheses into one, allow us to separate various groups of
The fact that, as a rule, we are at any given moment taking a vast amount of hypotheses, each of which may become an object of refutation by counter
traditional knowledge for granted (for almost all our knowledge is traditional) examples. (An excellent recent example is the rejection, in atomic theory, of
creates no difficulty for the falsificationist or fallibilist. For he does not
the law of parity; another is the rejection of the law of commutation for
accept this background knowledge; neither as established nor as fairly certain, conjugate variables, prior to their interpretation as matrices, and to the
nor yet as probable. He knows that even its tentative acceptance is risky, and statistical interpretation of these matrices.)
stresses that every bit of it is open to criticism, even though only in a piece-
meal way. We can never be certain that we shall challenge the right bit; but XVII
since our quest is not for certainty, this does not matter. It will be noticed that One fact which is characteristic of the situation in which the scientist finds
this remark contains my answer to Quine's holistic view of empirical tests; himself is that we constantly add to our background knowledge. If we discard
a view which Quine formulates (with reference to Duhem), by asserting that
21See W. V. Quine, From a LogiealPoint of View, 1953.p. 41.
our statements about the external world face the tribunal of sense experience 22See my L.Se.D., especially sections 19 to 22; and this volume, ch. 3, text to note 28.
238 239

""'- ~
 ---
CONJECTURES 10 TRUTH. RATIONALITY,AND THE GROWTHOF KNOWLEDGB
some parts of it, others, closely related to them, will remain. For example, What is the general problem situation in which the scientist finds himself?
even though we may regard Newton's theory as refuted-that is, his system He has before him a scientific problem: he wants to find a new theory capable I
of ideas, and the formal deductive system which derives from it-we may of explaining certain experimental facts; facts which the earlier theories I
still assume, as part of our background knowledge, the approximate truth, successfully explained; others which they could not explain; and some by
within limits, of its quantitative formulae. which they were actually falsified. The new theory should also resolve, if
The existence ofthis background knowledge plays an important role in one possible, some theoretical difficulties (such as how to dispense with certain
of the arguments which support (I believe) my thesis that the rational and ad hoc hypotheses, or how to unify two theories). Now if he manages to
empirical character of science would vanish if it ceased to progress. I can produce a theory which is a solution to all these problems, his achievement
sketch this argument here only in its barest outline. will be very great. II
A serious empirical test always consists in the attempt to find a refutation, Yet it is not enough. I have been asked, 'What more do you want?' My
a counter example. In the search for a counter example, we have to use our answer is that there are many more things which I want; or rather, which I
background knowledge; for we always try to refute first the most risky pre- think are required by the logic of the general problem situation in which the
dictions, the 'most unlikely. . . consequences' (as Peirce already saw23);which scientist finds himself; by the task of getting nearer to the truth. I shall con-
means that we always look in the most probable kinds of places for the most fine myself here to the discussion of three such requirements.
probable kinds of counter examples-most probable in the sense that we The first requirement is this. The new theory should proceed from some
should expect to find them in the light of our background knowledge. Now simple, new, and powerful, unifying idea about some connection or relation
if a theory stands up to many such tests, then, owing to the incorporation of (such as gravitational attraction) between hitherto unconnected things (such
the results of our tests into our background knowledge, there may be, after a as planets and apples) or facts (such as inertial and gravitational mass) or
time, no places left where (in the light of our new background knowledge) new 'theoretical entities' (such as field and particles). This requirement of
counter examples can with a high probability be expected to occur. But this simplicity is a bit vague, and it seems difficult to formulate it very clearly. It
means that the degree of severity of our test declines. This is also the reason seems to be intimately connected with the idea that our theories should des.
why an often repeated test will no longer be considered as significant or as cribe the structural properties of the world-an idea which it is hard to think
severe: there is something like a law of diminishing returns from repeated out fully without getting involved in an infinite regress. (This is so because any
tests (as opposed to tests which, in the light of our background knowledge, idea of a particular structure of the world-unless, indeed, we think of a
are of a new kind, and which therefore may still be felt to be significant). These purely mathematical structure-already presupposes a universal theory; for
are facts which are inherent in the knowledge-situation; and they have often example, explaining the laws of chemistry by interpreting molecules as struc-
been described-especially by John Maynard Keynes and by Ernest Nagel- tures of atoms, or of subatomic particles, presupposes the idea of universal
as difficult to explain by an inductivist theory of science. But for us it is all laws that regulate the properties and the behaviour of the atoms, or of the
very easy. And we can even explain, by a similar analysis of the knowledge- particles. ) Yet one important ingredient in the idea of simplicity can be logic-
situation, why the empirical character of a very successful theory always ally analysed. It is the idea of testability.24 This leads us immediately to our
grows stale, after a time. We may then feel (as Poincare did with respect to second requirement.
Newton's theory) that the theory is nothing but a set of implicit definitions or For, secondly,we requirethat the newtheory shouldbe independently
conventions-until we progress again and, by refuting it, incidentally re- testable.25Thatis to say,apartfromexplaining all the explicandawhichthe
establish.itslost empiricalcharacter. (De mortuis nil nisi bene: once a theory new theory was designed to explain,it must have new and testable con-
i
is refuted, its empirical character is secure and shines without blemish.) sequences(preferablyconsequencesof a newkind); it must lead to the pre.
diction of phenomena which have not so far been observed.
This requirement seems to me indispensable since without it our new
S. THREE REQUIREMENTSFOR THE GROWTH OF KNOWLEDGE theory might be ad hoc; for it is always possible to produce a theory to fit any
XVIIT 24 See sections 31-46 of my L.Se.D. More recently I have stressed (in lectures) the need to
relativize comparisons of simplicity to those hypotheses which compete qua solutions of a
But let us return again to the idea of getting nearer to the truth-to the
certain problem, or set of problems. The idea of simplicity, though intuitively connected with
search for theories which agree better with the facts (as indicated by the list the idea of a unified or coherent system or a theory that springs from one intuitive picture of
of six comparisons in section x above). the facts, cannot be analysed in terms of numerical paucity of hypotheses. For every theory
can be formulated in one statement; and it seems that, for every theory and every n, there is
23Seethe CollectedPapers of c. S. Peirce, vol. YD,7.182and 7.206.I owe this reference to a set of n independent axioms (though not necessarily 'organic' axioms in the Warsaw sense). I
W. B. Gallie (cp. Philosophy,35, 1960, p. 67), and a similar one to David Rynin. 25 For the idea of an independent test see my paper 'The Aim of Science', Ratio, 1,1957.
240 241

""'-- -- u
 ---..--
CONJECTURES 10 TRUTH, RATIONALITY, AND THE GROWTH OF KNOWLEDGE

given set of explicanda. Thus our two first requirements are needed in order Refutations have often been regarded as establishing the failure of a
to restrict the range of our choice among the possible solutions (many of scientist, or at least of his theory. It should be stressed that this is an in-
them uninteresting) of the problem in hand. ductivist error. Every refutation should be regarded as a great success; not
If our second requirement is satisfied then our new theory will represent a merely a success of the scientist who refuted the theory, but also of the
potential step forward, whatever the outcome of the new tests may be. For it scientist who created the refuted theory and who thus in the first instance
will be better testable than the previous theory: the fact that it explains all the suggested, if only indirectly, the refuting experiment.
explicanda of the previous theory, and that, in addition, it gives rise to new Even if a new theory (such as the theory of Bohr, Kramers, and Slater)
tests, suffices to ensure this. should meet an early death, it should not be forgotten; rather its beauty
Moreover, the second requirement also ensures that our new theory wiH,to should be remembered, and history should record our gratitude to it-for
some extent, be fruitful as an instrument of exploration. That is to say, it will bequeathing to us new and perhaps still unexplained experimental facts and,
suggest to us new experiments, and even if these should at once lead to the with them, new problems; and for the services it has thus rendered to the
refutation of the theory, our factual knowledge will have grown through the progress of science during its successful but short life.
unexpected results of the new experiments. Moreover, they will confront us AIl this indicates clearly that our third requirement is not indispensable:
with new problems to be solved by new explanatory theories. even a theory which fails to meet it can make an important contribution to
Yet I believe that there must be a third requirement for a good theory. It is science. Yet in a different sense, I hold, it is indispensable none the less.
this. We require that the theory should pass some new, and severe, tests. (Bohr, Kramers and Slater rightly aimed at more than making an important
XIX contribution to science.)
In the first place, I contend that further progress in science would become
Clearly, this requirement is totally different in character from the previous impossible if we did not reasonably often manage to meet the third require-
two. These could be seen to be fulfilled, or not fulfilled, largely by analysing , ment; thus if the progress of science is to continue, and its rationality not to
the old and the new theories logically. (They are 'formal requirements'.) The decline, we need not only successful refutations, but also positive successes.
third requirement, on the other hand, can be found to be fulfilled, or not ful- \
f
We must, that is, manage reasonably often to produce theories that entail
filled, only by testing the new theory empirically. (It is a 'material require- new predictions, especially predictions of new effects, new testable con-
ment', a requirement of empirical success.) sequences, suggested by the new theory and never thought of before.28 Such
Moreover, the third requirement clearly cannot be indispensable in the a new predict~on was that planets would under certain circumstances deviate
same sense as are the two previous ones. For these two are indispensable for from Kepler's laws; or that light, in spite of its zero mass, would prove to be
deciding whether the theory in question should be at all accepted as a serious subject to gravitational attraction (that is, Einstein's eclipse-effect). Another
candidate for examination by empirical tests; or in other words, whether it is example is Dirac's prediction that there will be an anti-particle for every
an interesting and promising theory. Yet on the other hand, some of the most elementary particle. New predictions of these kinds must not only be pro-
interesting and most admirable theories ever conceived were refuted at the duced, but they must also be reasonably often corroborated by experimental
very first test. And why not? The most promising theory may fail if it makes evidence, I contend, if scientific progress is to continue.
predictions of a new kind. An example is the marvellous theory of Bohr, We do need this kind of success; it is not for nothing that the great theories
Kramers and SIater26 of 1924 which, as an intellectual achievement, might of science have all meant a new conquest of the unknown, a new success in
almost ra~ with Bohr's quantum theory of the hydrogen atom of 1913. Yet predicting what had never been thought of before. We need successes such
unfortunately it was almost at once refuted by the facts-by the coincidence as that of Dirac (whose anti-particles have survived the abandonment of
experiments of Bothe and Geiger.27 This shows that not even the greatest some other parts of his theories), or that ofYukawa's meson theory. We need
physicist can anticipate the secrets of nature: his inspirations can only be the success, the empirical corroboration, of some of our theories, if only in
guesses, and it is no fault of his, or of his theory, if it is refuted. Even Newton's order to appreciate the significance of successful and stirring refutations (like
theory was in the end refuted; and indeed, we hope that we shall in this way that of parity). It seems to me quite clear that it is only through these tempor-
succeed in refuting, and improving upon, every new theory. And if it is ary successes of our theories that we can be reasonably successful in attribut-
refuted in the end, why not in the beginning? One might well say that it is ing our refutations to definite portions of the theoretical maze. (For we are
merely a historical accident if a theory is refuted after six months rather than reasonably successful in this-a fact which must remain inexplicable for. one
after six years, or six hundred years. who adopts Duhem's and Quine's views on the matter.) An unbroken
26Phil.Mag.,47, 1924,pp. 785If. 28I have drawn attention to 'new' predictions of this kind and to their philosophical
27Zeitschr.f. Phys.,32, 1925,pp. 63If. significance in ch. 3. See especially pp. 117f.
242 243

"'- -
 --
CONJECTURES 10 TRUTH, RATIONALITY, AND THE GROWTH OF KNOWLEDGE

sequence of refuted theories would soon leave us bewildered and helpless: we we may at the same time feel the need for encouragement in our belief that
should have no clue about the parts of each of these theories-or of our we are making progress). Yet this should certainly not induce in us the atti-
background knowledge-to which we might, tentatively, attribute the failure tude of merely producing theories so that they can be superseded.
of that theory. For our aim as scientists is to discover the truth about our problem; and
we must look at our theories as serious attempts to find the truth. If they are
xx not true, they may be, admittedly, important stepping stones towards the
truth, instruments for further discoveries. But this does not mean that we can
Earlier I suggested that science would stagnate, and lose its empirical char- ever be content to look at them as being nothing but stepping stones, nothing
acter, if we should fail to obtain refutations. We can now see that for very but instruments; for this would involve giving up even the view that they are
similar reasons science would stagnate, and lose its empirical character, if we instruments of theoretical discoveries; it would commit us to looking upon
should fail to obtain verifications of new predictions; that is, if we should them as mere instruments for some observational or pragmatic purpose. And
only manage to produce theories that satisfy our first two requirements but this approach would not, I suspect, be very successful, even from a pragmatic
not the third. For suppose we were to produce an unbroken sequence of point of view: if we are content to look at our theories as mere stepping
explanatory theories each of which would explain all the explicanda in its stones, then most of them will not even be good stepping stones. Thus we
field, including the experiments which refuted its predecessors; each would ought not to aim at theories which are mere instruments for the exploration
also be independently testable by'predicted new effects; yet each would be of facts, but we ought to try to find genuine explanatory theories: we should
at once refuted when these predictions were put to the test. Thus each make genuine guesses about the structure of the world. In brief, we should
would satisfy our first two requirements, but all would fail to satisfy the not be satisfied with the first two requirements.
third.
Of course, the fulfilment of our third requirement is not in our own hands.
I assert that, in this case, we should feel that we were producing a sequence No amount of ingenuity can ensure the construction of a successful theory.
of theories which, in spite of their increasing degree of testability, were ad hoc, We also need luck; and we also need a world whose mathematical structure
and that we were not getting any nearer to the truth. And indeed, this feeling is not so intricate as to make progress impossible. For indeed, if we should
may well be justified: this whole sequence of theories might easily be ad hoc. cease to progress in the sense of our third requirement-if we should only
For if it is admitted that a theory may be ad hoc if it is not independently test- succeed in refuting our theories but not in obtaining some verifications of
able by experiments of a new kind but merely explains all the explicanda, predictions of a new kind-we might well decide that our scientific problems
including the experiments which refuted its predecessors, then it is clear that have become too difficult for us because the structure (if any) of the world is
the mere fact that the theory is also independently testable cannot as such beyond our powers of comprehension. Even in this case we might proceed,
ensure that it is not ad hoc. This becomes clear if we consider that it is always for a time, with theory construction, criticism, and falsification: the rational
possible, by a trivial stratagem, to make an ad hoc theory independently test- side of the method of science might, for a time, continue to function. Yet I
able, if we do not also require that it should pass the independent tests in believe that we should feel that, especially for the functioning of its empirical
question: we merely have to connect it (conjunctively) in some way or other side, both kinds of successes are essential: success in refuting our theories,
with any testable but not yet tested fantastic ad hoc prediction which may and success on the part of some of our theories in resisting at least some of our
occur to us (or to some science fiction writer). most determined attempts to refute them.
Thus our third requirement, like the second, is needed in order to eliminate
trivial and other ad hoc theories.29 But it is needed also for what seem to me XXI
even more serious reasons. It may be objected that this is merely good psychological advice about the
I think that we are quite right to expect, and perhaps even to hope, that attitude which scientists ought to adopt-a matter which, after all, is their
even our best theories will be superseded and replaced by better ones (though private affair-and that a theory of scientific method worthy of its name
29Dr Jeny Giedymin (in a paper 'A Generalization of the Refutability Postulate', should be able to produce logical or methodological arguments in support of
Studia Logica, 10, 1960,see especiallypp. 103if.) has fonnulated a general methodological our third requirement. Instead of appealing to the attitude or the psychology
principle of empiricism which says that our various rules of scientific method must not of the scientist, our theory of science should even be able to explain his atti-
permit what he calIs a 'dictatorial strategy'; that is they must exclude the possibility that we tude, and his psychology, by an analysis of the logic of the situation in which
shall always win the game played in accordance with these rules: Nature must be able to
defeat us at least sometimes.If we drop our third requirement, then we can always win, and he finds himself. There is a problem here for our theory of method.
need not consider Nature at all, as far as the construction of 'good' theories is concerned: I accept this challenge, and I shall produce three reasons: the first from
speculations about answers which Nature may give to our questions will play no role in our the idea of truth; the second from the idea of getting nearer to the truth
problem situation which will always be fully determined by our past failures alone.
244 245

..... -
 ,..-

CONJECTURES 10 TRUTH, RATIONALITY,ANDTHE GROWTHOF KNOWLEDGE

(verisimilitude); and the third from our old idea of independent tests and energy equation) as superior to one which can only save the known pheno-
of crucial tests. mena (Lorentz-Fitzgerald).
(1) The first reason why our third requirement is so important is this. We (3) The same point-the importance of crucial tests-can be made without
know that if we had an independently testable theory which was, moreover, appealing to the aim of increasing the verisimilitude of a theory, by using an
true, then it wouldprovide us with successful predictions (and only with success- old argument of mine-the need to make the tests of our explanations
ful ones). Successful predictions-though they are not, of course, sufficient independent.30 This need is a result of the growth of knowledge-of the in-
conditions for the truth of a theory-are therefore at least necessary con- corporation of what was new and problematic knowledge into background
ditions for the truth of an independently testable theory. In this sense-and knowledge, with a consequent loss of explanatory power to our theories.
only in this sense-our third requirement may even be said to be 'necessary', These are my main arguments.
if we seriously accept truth as a regulative idea.
XXII
(2) The second reason is this. If it is our aim to strengthen the verisimilitude
of our theories, or to get nearer to the truth, then we should be anxious not Our third requirement may be divided into two parts: first we require of a
only to reduce the falsity content of our theories but also to strengthen their good theory that it should be successful in some of its new predictions;
truth content. secondly we require that it is not refuted too soon-that is, before it has been
Admittedly this may be done in certain cases simply by constructing the strikingly successful. Both requirements sound strange. The first because the
new theory in such a way that the refutations of the old theory are explained logical relationship between a theory and any corroborating evidence cannot,
('saving the phenomena', in this case the refutations). But there are other it seems, be affected by the question whether the theory is temporally prior
cases of scientific progress-cases whose existence shows that this way of to the evidence. The second because if the theory is doomed to be refuted, its
increasing the truth content is not the only possible one. intrinsic value can hardly depend upon delaying the refutation.
The cases I have in mind are cases in which there was no refutation. Our explanation of this slightly puzzling difficulty is simple enough: the
Neither Galileo's nor Kepler's theories were refuted before Newton: what successful new predictions which we require the new theory to produce are
Newton tried to do was to explain them from more general assumptions, and identical with the crucial tests which it must pass in order to become suf-
thus to unify two hitherto unrelated fields of inqniry. The same may be said ficiently interesting to be accepted as an advance upon its predecessor, and
of many other theories: Ptolemy's system was not refuted when Copernicus to be considered worthy of further experimental examination which may
produced his. And though there was, before Einstein, the puzzling experiment eventually lead to its refutation.
of Michelson and Morley, this had been successfully explained by Lorentz Yet the difficulty can hardly be resolved by an inductivist methodology. It
and Fitzgerald. is therefore not surprising that inductivists such as John Maynard Keynes
It is in cases like these that crucial experiments become decisively important. have asserted that the value of predictions (in the sense of facts derived from
We have no reason to regard the new theory as better than the old theory- the theory but previously not known) was imaginary; and indeed if the value
to believe that it is nearer to the truth-until we have derived from the new of a theory would lie merely in its relation to its evidential basis, then it would
theory new predictions which were unobtainable from the old theory (the be logically irrelevant whether the supporting evidence precedes or follows in
phases of Venus, the perturbations, the mass-energy equation) and until we time the invention of the theory. Similarly the great founders of the hypo-
have found that these new predictions were successful. For it is only this thetical method used to stress the 'saving of the phenomena', that is to say,
success whic~ shows that the new theory had true consequences (that is, a the demand that the theory should explain known experience. Successful new
truth content) where the old theories had false consequences (that is, a falsity prediction-of new effects-seems to be a late idea, for obvious reasons;
content). perhaps it was first mentioned by some pragmatist, although the distinction
Had the new theory been refuted in any of these crucial experiments then between the prediction of known effects and the prediction of new effects was
we should have had no reason to abandon the old one in its favour-even if hardly ever made explicitly. But it seems to me quite indispensable as a part
the old theory was not wholly satisfactory. (This was the fate of the Bohr- of an epistemology which views science as progressing to better and better
Kramers-Slater theory.) explanatory theories, that is, not merely to instruments of exploration, but
In all these important cases we need the new theory in order to find out to genuine explanations,
where the old theory was deficient. Admittedly, the situation is different if the Keynes' objection (that it is an historical accident whether this support was
deficiency of the old theory is already known before the new theory is in- known before the theory was proposed, or only afterwards so that it could
vented; but logically the case has enough similarity with the other cases to attain the status of a prediction) overlooks the all-important fact that it is
regard a new theory which leads to new crucial experiments (Einstein's mass- 30Seeespeciallymy paper'TheAimof Science',Ratio,1, 1957.
246 247

""'-- -
 -~
,I
CONJECTURES 10 TRUTH, RATIONALITY, AND THE GROWTH OF KNOWLEDGE

through our theories that we learn to observe, that is to say, to ask questions I always thought that the comparison of these degrees leads to a criterion
which lead to observations and to their interpretations. This is the way which is more important and more realistic than the simpler criterion of
our observational knowledge grows. And the questions asked are, as a rule, falsifiability which I proposed at the same time, and which has been widely
crucial questions which may lead to answers that decide between competing discussed. But this simpler criterion is also needed. In order to show the need
theories. It is my thesis that it is the growth of our knowledge, our way of for the falsifiability or testability criterion as a criterion of the empirical
choosing between theories, in a certain problem situation, which makes character of scientific theories, I will discuss, as an example, a simple, purely
science rationa1. Now both the idea of the growth of knowledge and that of existential statement which is formulated in purely empirical terms. I hope
a problem situation are, at least partly, historical ideas. This explains why this example will also provide a reply to the often repeated criticism that it is
another partly historical idea-that of a genuine prediction of evidence (it perverse to exclude purely existential statements from empirical science and
may be about past facts) not known when the theory was first proposed- to classify them as metaphysica1.
may play an important role here, and why the apparently irrelevant time My example consists of the following purely existential theory:
element may become relevant)1 'There exists a finite sequence of Latin elegiac couplets such that, if it is
I shall now briefly sum up our results with respect to the epistemologies of pronounced in an appropriate manner at a certain time and place, this is
the two groups of philosophers I have dealt with, the verificationists and the '\ immediately followed by the appearance of the Devil-that is to say, of a
rI
falsificationists. man-like creature with two small horns and one cloven hoof.'
I
While the verificationists or inductivists in vain try to show that scientific I Clearly, this untestable theory is, in principle, verifiable. Though according
beliefs can be justified or, at least, established as probable (and so encourage, to my criterion of demarcation it is excluded as non-empirical and non-
by their failure, the retreat into irrationalism), we of the other group have scientific or, if you like, metaphysical, it is not so excluded by those positivists
found that we do not even want a highly probable theory. Equating ration- who consider all well-formed statements and especially all verifiable ones as
ality with the critical attitude, we look for theories which, however fallible, empirical and scientific.
progress beyond their predecessors; which means that they can be more ft Some of my positivist friends have indeed assured me that they consider
severely tested, and stand up to some of the new tests. And while the verifica-
tionists laboured in vain to discover valid positive arguments in support of
\ my existential statement about the Devil to be empirical. It is empirical
though false, they said. And they indicated that I was mistaking false em-
their beliefs, we for our part are satisfied that the rationality of a theory lies pirical statements for non-empirical ones.
in the fact that we choose it because it is better than its predecessors; because However, I think that the confusion, if any, is not mine. I too believe that
it can be put to more severe tests; because it may even have passed them, if the existential statement is false: but I believe that it is a false metaphysical
we are fortunate; and because it may, therefore, approach nearer to the truth. statement. And why, I ask, should anybody who takes it for empirical think
that it is false'! Empirically, it is irrefutable. No observation in the world
can establish its falsity. There can be no empirical grounds for its falsity.
APPENDIX: A PRESUMABLY FALSE YET FORMALLY Moreover, it can be easily shown to be highly probable: like all existential
HIGHLY PROBABLE NON-EMPIRICAL STATEMENT statements, it is in an infinite (or sufficiently large) universe almost logically
In the text of this chapter I have drawn attention to the criterion of progress true, to use an expression of Carnap's. Thus, if we take it to be empirical, we
and of rationality based on the comparison of degrees of testability or degrees
have no reason to reject it, and every reason to accept it and to believe in it-
of the empirical content or explanatory power of theories. I did so because especially upon a subjective theory of probable belief.
11
these degrees have been little discussed so far. )
Probability theory tells us even more: it can be easily proved not only that
I) empirical evidence can never refute an almost logically true existential state-
31Verificationists may think that the preceding discussion of what I have called here the ment, but that it can never reduce its probability.a2(Its probability could be
third requirement quite unnecessarilyelaborates what nobody contests. Falsificationistsmay 4
think otherwise; and personally I feel greatly indebted to Dr Agassi for drawing myatten- reduced only by some information which is at least 'almost logically false',
tion to the fact that I have previouslynever explained clearlythe distinction betweenwhat are and therefore not by an observational evidence statement.) So the empirical
called here the second and third requirements. He thus induced me to state it here in some t probability or degree of empirical confirmation (in Carnap's sense) of our
detail. I should mention, however, that he disagrees with me about the third requirement
which, as he explained to me, he cannot accept because he can regard it only as a residue of ( statement about the devil-summoning spell must for ever remain equal to
verificationist modes of thought. (See also his paper in the Australasit.nJournal of Philo- I unity, whatever the facts may be.
sophy, 39, 1961, where he expresses his disagreement on p. 90.) I admit that there may
be a whiff of verificalionismhere; but this seems to me a case where we have to put up 32This is a consequence of the 'principle of stability' of the probability calculus; see
with it, if we do not want a whiff of some form of instrumentalism that takes theories to theorem (26), section V, of my paper 'Creative and Non-Creative Definitions in the Cal-
be mere instruments of exploration. culus of Probability', Synthese, 15,1963, No.2, pp. 167ff.
248 249
 .........-.

CONJECTURES

It would of course be easy enough for me to amend my criterion of

demarcation so as to include such purely existential statements among the
empirical statements. I merely should have to admit not only testable or
falsifiable statements among the empirical ones, but also statements which
may, in principle, be empirically 'verified'.
But I believe that it is better not to amend my original falsifiability criterion.
For our example shows that, if we do not wish to accept my existential
statement about the spell that summons the devil, we must deny its empirical
character (notwithstanding the fact that it can easily be formalized in any
model language sufficient for the expression of even the most primitive
scientific assertions). By denying the empirical character of my existential
statement, I make it possible to reject it on grounds other than observational
evidence. (See chapter 8, section 2, for a discussion of such grounds; and see
chapter 11, especially pp. 275-277, for a discussion and a formalization of a
similar argument.)
This shows that it is preferable, as I have been trying to make clear for
some considerable time, not to assume uncritically that the terms 'empirical'
and 'well-formed' (or 'meaningful') must coincide-and that the situation is
hardly improved if we assume, uncritically, that probability or probabilistic
REFUTATIONS
'confirmability' may be used as a criterion of the empirical character of
statements or theories. For a non-empirical and presumably false statement
may have a high degree of probability, as has been shown here.

250

""-
.-