7.

Science: Conjectures
and Refutations

Karl R. Popper

Together with Rudolf Carnap and Thomas Kuhn, Karl Popper

( I 902- I994) is one of the three most influential philosophers
of science of the twentieth century. Schooled in Vienna when the
Vienna Circle was dominant, Popper rejected both their preoccupa-
tion with a meaning criterion and their attempt to formulate a
logic of confirmation, substituting instead a criterion of demarcation
between science and non-science. His criterion concerns, not a
theory's confirmability, but instead its falsifiability, that is, its
susceptibility to be proven incorrect. That criterion, and his
falsificationist account of scientific methodology - the method of
conjecture and refutation - is sketched in the selection below.

Mr Turnbull had predicted evil consequences, problems which interest me most, and about
... and was now doing the best in his power to those developments with which I am most inti-
bring about the verification ofhis own prophecies. mately acquainted. I therefore decided to do
Anthony Trollope what I have never done before: to give you a report
on my own work in the philosophy of science,
since the autumn of 1919 when I first began to
I grapple with the problem, "When should a theory
be ranked as sdentific?" or "ls there a criterion for
When I received the list of participants in this the sdentific character or status of a theory?"
course and realized that I had been asked to The problem which troubled me at the time
speak to philosophical colleagues I thought, was neither, "When is a theory true?" nor, "When
after some hesitation and consultation, that you is a theory acceptable?" My problem was differ-
would probably prefer me to speak about those ent. I wished to distinguish between sdence and

A lecture givm at Peterhouse, Cambridge, in Summer 1953, originally published by the British Council under the title
"Philosophy of Science: a Personal Report" in British Philosophy in Mid-Century, ed. C. A. Mace, 1957 (George Allen
& Unwin, 1957).

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
472 KARL R. POPPER

pseudo-science; knowing very well that science It was during the summer of 1919 that I began
often errs, and that pseudo-science may happen to feel more and more dissatisfied with these
to stumble on the truth. three theories - the Marxist theory of history.
I knew, of course, the most widely accepted psycho-analysis, and individual psychology; and
answer to my problem: that science is dis- I began to feel dubious about their claims to
tinguished from pseudo-science - or from scientific status. My problem perhaps first took
"metaphysics" - by its empirical method, which the simple form, "What is wrong with ~1arxism,
is essentially inductive, proceeding from observa- psycho-analysis, and individual psychology? \\·b y
tion or experiment. But this did not satisfy me. are they so different from physical theories, from
On the contrary, I often formulated my prob- Newton's theory, and especially from the theorv
lem as one of distinguishing between a genuinely of relativity?"
empirical method and a non-empirical or even To make this contrast clear I should explain
a pseudo-empirical method - that is to say, a that few of us at the time would have said that
method which, although it appeals to observation we believed in the truth of Einstein's theory of gray-
and experiment, nevertheless does not come up itation. This shows that it was not my doubting
to scientific standards. The latter method may the truth of those other three theories \\'hich
be exemplified by astrology, with its stupendous bothered me, but something else. Yet neither
mass of empirical evidence based on observation was it that I merely felt mathematical physics
- on horoscopes and on biographies. to be more exact than the sociological or psycho-
But as it was not the example of astrology logical type of theory. Thus what worried me was
which led me to my problem I should perhaps neither the problem of truth, at that stage at least.
briefly describe the atmosphere in which my nor the problem of exactness or measurabilit)·.
problem arose and the examples by which it was It was rather that I felt that these other three
stimulated. After the collapse of the Austrian theories, though posing as sciences, had in fact
Empire there had been a revolution in Austria: the more in common with primitive myths than with
air was full of revolutionary slogans and ideas, science; that they resembled astrology rather than
and new and often wild theories. Among the astronomy.
theories which interested me Einstein's theory of I found that those of my friends who v.·ere
relativity was no doubt by far the most import- admirers of Marx, Freud, and Adler, were
ant. Three others were Marx's theory of history, impressed by a number of points common to
Freud's psycho-analysis, and Alfred Adler's so- these theories, and especially by their apparent
called "individual psychology." explanatory power. These theories appeared to Ix
There was a lot of popular nonsense talked able to explain practically everything that hap·
about these theories, and especially about relativity pened within the fields to which they referred.
(as still happens even today), but I was fortunate The study of any of them seemed to have the effect
in those who introduced me to the study of of an intellectual conversion or revelation, open-
this theory. We all - the small circle of students ing your eyes to a new truth hidden from those
to which l belonged - were thrilled with the not yet initiated. Once your eyes were thus opened
result of Eddington's eclipse observations which you saw confirming instances everywhere: the
in 1919 brought the first important confirmation world was full of verifications of the theon·.
of Einstein's theory of gravitation. It was a great Whatever happened always confirmed it. Thus its
experience for us, and one which had a lasting truth appeared manifest; and unbelievers \\·en:
influence on my intellectual development. clearly people who did not want to see the mani-
The three other theories · I have mentioned fest truth; who refused to see it, either because
were also widely discussed among students at it was against their class interest, or because of
that time. I myself happened to come into per- their repressions which were still "un-analy-sed~
sonal contact with Alfred Adler, and even to and crying aloud for treatment.
co-operate \vith him in his social work among The most characteristic element in this situ-
the children and young people in the working- ation seemed to me the incessant strean1 of
class districts of Vienna where he had established confirn1ations, of observations which "verified"
social guidance clinics. the theories in question; and this point \vas

Digitized .by Google Original from

UNIVERSITY OF MICHIGAN
 SCIENCE: CONJECTURES AND REFUTATIONS 473

constantly emphasized by their adherents. A favour of these theories. It began to dawn on

Marxist could not open a newspaper without me that this apparent strength was in fact their
finding on every page confirming evidence for his weakness.
interpretation of history; not only in the news, With Einstein's theory the situation was strik-
but also in its presentation - which revealed ingly different. Take one typical instance -
the class bias of the paper - and especially of Einstein's prediction, just then confirmed by the
course in what the paper did not say. The Freudian findings of Eddington's expedition. Einstein's
analysts emphasized that their theories were con- gravitational theory had led to the result that
stantly verified by their "clinical observations." light must be attracted by heavy bodies (such
As for Adler, I was much impressed by a personal as the sun), precisely as material bodies were
experience. Once, in 1919, I reported to him a attracted. As a consequence it could be calculated
case which to me did not seem particularly that light from a distant fixed star whose appar-
Adlerian, but which he found no difficulty in ent position was close to the sun would reach the
analysing in terms of his theory of inferiority earth from such a direction that the star would
feelings, although he had not even seen the child. seem to be slightly shifted away from the sun;
Slightly shocked, I asked him how he could be so or, in other words, that stars dose to the sun would
sure. "Because of my thousandfold experience," look as if they had moved a little away from the
he replied; whereupon I could not help saying: sun, and from one another. This is a thing which
..And with this new case, I suppose, your experi- cannot normally be observed since such stars
ence has become thousand-and-one-fold." are rendered invisible in daytime by the sun's
What I had in mind was that his previous overwhelming brightness; but during an eclipse
observations may not have been much sounder it is possible to take photographs of them. If the
than this new one; that each in its turn had been same constellation is photographed at night one
interpreted in the light of "previous experience," can measure the distances on the two photo-
and at the same time counted as additional con- graphs, and check the predicted effect.
firmation. What, I asked myself, did it confirm? Now the impressive thing about this case is
No more than that a case could be interpreted in the risk involved in a prediction of this kind.
the light of the theory. But this meant very little, If observation shows that the predicted effect
I reflected, since every conceivable case could is definitely absent, then the theory is simply
be interpreted in the light of Adler's theory, or refuted. The theory is incompatible with certain
equally of Freud's. I may illustrate this by two very possible results ofobservation - in fact with results
different examples of human behaviour: that of which everybody before Einstein would have
a man who pushes a child into the water with expected. 1 This is quite different from the situ-
the intention of drowning it; and that of a man ation I have previously described, when it turned
who sacrifices his life in an attempt to save the out that the theories in question were compat-
child. Each of these two cases can be explained ible with the most divergent human behaviour,
with equal ease in Freudian and in Adlerian so that it was practically impossible to describe
terms. According to Freud the first man suffered any human behaviour that might not be claimed
from repression (say, of some component of his to be a verification of these theories.
Oedipus complex), while the second man had These considerations led me in the winter
achieved sublimation. According to Adler the of 1919- 20 to conclusions which I may now
first man suffered from feelings of inferiority reformulate as follows.
(producing perhaps the need to prove to himself
that he dared to commit some crime), and so did (I) It is easy to obtain confirmations, or verifi-
the second man (whose need was to prove to him- cations, for nearly every theory - if we look
self that he dared to rescue the child). I could not for confirmations.
think of any human behaviour which could not (2) Confirmations should count only if they
be interpreted in terms of either theory. It was pre- are the result of risky predictions; that is
cisely this fact - that they always fitted, that they to say, if, unenlightened by the theory in
were always confirmed - which in the eyes of their question, we should have expected an event
admirers constituted the strongest argument in which was incompatible with the theory -

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
474 KARL R. POPPER

an event which would have refuted the that they were quite unimpressed by any unfav-
theory. ourable evidence. Moreover, by malcing their
(3) Every "good" scientific theory is a prohibi- interpretations and prophecies sufficiently vague
tion: it forbids certain things to happen. they were able to explain away anything that
The more a theory forbids, the better it is. might have been a refutation of the theory had
( 4) A theory which is not refutable by any the theory and the prophecies been more precise.
conceivable event is nonscientific. Irrefut- In order to escape falsification they destroyed
ability is not a virtue of a theory (as people the testability of their theory. It is a typical
often think) but a vice. soothsayer's trick to predict things so vague!,·
(S) Every genuine test of a theory is an attempt that the predictions can hardly fail: that th~·
to falsify it, or to refute it. Testability is become irrefutable.
falsifiability; but there are degrees of test- The Marxist theory of history, in spite of
ability: some theories are more testable, the serious efforts of some of its founders and
more exposed to refutation, than others; followers, ultimately adopted this soothsaying
they take, as it were, greater risks. practice. In some of its earlier formulations (for
( 6) Confirming evidence should not count except example in Marx's analysis of the character of
when it is the result of a genuine test of the the "coming social revolution") their predictions
theory; and this means that it can be pre- were testable, and in fact falsified. 2 Yet instead of
sented as a serious but unsuccessful attempt accepting the refutations the followers of Marx
to falsify the theory. (I now speak in such re-interpreted both the theory and the evidence
cases of "corroborating evidence.") in order to make them agree. In this way th~·
(7) Some genuinely testable theories, when found rescued the theory from refutation; but they did
to be false, are still upheld by their admirers so at the price of adopting a device which made
- for example by introducing ad hoc some it irrefutable. They thus gave a "conventionalist
auxiliary assumption, or by reinterpreting the twist" to the theory; and by this stratagem ~
theory ad hoc in such a way that it escapes destroyed its much advertised claim to scientific
refutation. Such a procedure is always pos- status.
sible, but it rescues the theory from refuta- The two psycho-analytic theories were in a
tion only at the price of destroying, or at different class. They were simply non-testable,
least lowering, its scientific status. (I later irrefutable. There was no conceivable human
described such a rescuing operation as a behaviour which could contradict them. This does
"conventionalist twist" or a "conventionalist not mean that Freud and Adler were not seeing
stratagem.") certain things correctly: I personally do not
doubt that much of what they say is of consider-
One can sum up all this by saying that the able importance, and may well play its part one
criterion of the scientific status of a theory is its day in a psychological science which is testable.
falsifiability, or refutability, or testability. But it does mean that those 'clinical observa-
tions' which analysts natvely believe confirm their
theory cannot do this any more than the daily
II confirmations which astrologers find in their
practice.3 And as for Freud's epic of the Ego, the
I may perhaps exemplify this with the help of Super-ego, and the Id, no substantially stronger
the various theories so far mentioned. Einstein's claim to scientific status can be made for it than
theory of gravitation clearly satisfied the criterion for Homer's collected stories from Olympus. These
of falsifiability. Even if our measuring instru- theories describe some facts, but in the manner
ments at the time did not allow us to pronounce of myths. They contain most interesting psycho-
on the results of the tests with complete assurance, logical suggestions, but not in a testable form.
there was clearly a possibility of refuting the theory. At the same time I realized that such myths may
Astrology did not pass the test. Astrologers be developed, and become testable; that histor-
were greatly impressed, and misled, by what they ically speaking all - or very nearly all - scientific
believed to be confirming evidence - so much so theories originate from myths, and that a myth

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
 SCIENCE: CONJECTURES AND REFUTATIONS 475

may contain important anticipations of scientific which also had obvious practical consequences
theories. Examples are Empedocles' theory of (for example, political ones). But I did not yet
evolution by trial and error, or Parmenides' myth realize its full implications, or its philosophical
of the unchanging block universe in which noth- significance. When I explained it to a fellow stu-
ing ever happens and which, if we add another dent of the Mathematics Department (now a
dimension, becomes Einstein's block universe (in distinguished mathematician in Great Britain), he
which, too, nothing ever happens, since everything suggested that I should publish it. At the time
is, four-dimensionally speaking, determined and I thought this absurd for I was convinced that my
laid down from the beginning). I thus felt that if problem, since it was so important for me, must
a theory is found to be non-scientific, or "meta- have agitated many scientists and philosophers who
physical" (as we might say), it is not thereby found would surely have reached my rather obvious
to be unimportant, or insignificant, or "meaning- solution. That this was not the case I learnt from
less," or "nonsensical."4 But it cannot claim to be Wittgenstein's work, and from its reception; and
backed by empirical evidence in the scientific so I published my results thirteen years later in
sense - although it may easily be, in some genetic the form of a criticism of Wittgenstein's criterion
sense, the "result of observation." of meaningfulness.
(There were a great many other theories of Wittgenstein, as you all know, tried to show in
this pre-scientific or pseudo-scientific character, the Tractatus (see for example his propositions
some of them, unfortunately, as influential as the 6.53; 6.54; and S) that all so-called philosophical
Marxist interpretation of history; for example, or metaphysical propositions were actually non-
the racialist interpretation of history - another propositions or pseudo-propositions: that they
of those impressive and all-explanatory theories were senseless or meaningless. All genuine (or
which act upon weak minds like revelations.) meaningful) propositions were truth functions
Thus the problem which I tried to solve by of the elementary or atomic propositions which
proposing the criterion of falsifiability was neither described "atomic facts," i.e. - facts which can
a problem of meaningfulness or significance, nor in principle be ascertained by observation. In
a problem of truth or acceptability. It was the other words, meaningful propositions were fully
problem of drawing a line (as well as this can reducible to elementary or atomic propositions
be done) between the statements, or systems which were simple statements describing pos-
of statements, of the empirical sciences, and all sible states of affairs, and which could in principle
other statements - whether they are of a religious be established or rejected by observation. If we
or of a metaphysical character, or simply pseudo- call a statement an "observation statement" not
scientific. Years later - it must have been in only if it states an actual observation but also
1928 or 1929 - I called this first problem of mine if it states anything that may be observed, we
the "problem of demarcation." The criterion of shall have to say (according to the Tractatus, S and
falsifiability is a solution to this problem of 4.52) that every genuine proposition must be a
demarcation, for it says that statements or systems truth-function of, and therefore deducible from,
of statements, in order to be ranked as scientific, observation statements. All other apparent proposi-
must be capable of conflicting with possible, or tions will be meaningless pseudo-propositions;
conceivable, observations. in fact they will be nothing but nonsensical
gibberish.
This idea was used by Wittgenstein for a charac-
III terization of science, as opposed to philosophy.
We read (for example in 4.11, where natural
Today I know, of course, that this criterion science is taken to stand in opposition to philo-
of demarcation - the criterion of testability, or sophy): "The totality of true propositions is the
falsifiability, or refutability - is far from obvious; total natural science (or the totality of the nat-
for even now its significance is seldom realized. ural sciences)." This means that the propositions
At that time, in 1920, it seemed to me almost which belong to science are those deducible from
trivial, although it solved for me an intellectual true observation statements; they are those proposi-
problem which had worried me deeply, and one tions which can be verified by true observation

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
476 KARL R. POPPER

statements. Could we know all true observation made nonsense of my views.0 My protests that I
statements, we should also know all that may be was trying to solve, not their pseudo-problem o f
asserted by natural science. meaning, but the problem of demarcation, were
This amounts to a crude verifiability criterion of no avail.
of demarcation. To make it slightly less crude, My attacks upon verification had some effect,
it could be amended thus: "The statements however. They soon led to complete confusion
which may possibly fall within the province of in the camp of the verificationist philosophers
science are those which may possibly be verified of sense and nonsense. The original proposal of
by observation statements; and these statements, verifiability as the criterion of meaning was at lea.st
again, coincide with the class of all genuine or clear, simple, and forceful . The modifications
meaningful statements." For this approach, then, and shifts which were now introduced were the
verifiability, meaningfulness, and scientific character very opposite. 7 This, I should say, is now seen
all coincide. even by the participants. But since I am usually
I personally was never interested in the so-called quoted as one of them I wish to repeat that
problem of meaning; on the contrary, it appeared although I created this confusion I never par-
to me a verbal problem, a typical pseudo-problem. ticipated in it. Neither falsifiability nor testabilitv
I was interested only in the problem of demarca- were proposed by me as criteria of meaning: and
tion, i.e. in finding a criterion of the scientific although I may plead guilty to having introduc~
character of theories. It was just this interest both terms into the discussion, it was not I who
which made me see at once that Wittgenstein's introduced them into the theory of meaning.
verifiability criterion of meaning was intended Criticism of my alleged views was widespread
to play the part of a criterion of demarcation and highly successful. I have yet to meet a criti-
as well; and which made me see that, as such, cism of my views. 8 Meanwhile, testability is being
it was totally inadequate, even if all misgivings widely accepted as a criterion of demarcation.
about the dubious concept of meaning were set
aside. For Wittgenstein's criterion of demarcation
- to use my own terminology in this context - IV
is verifiability, or deducibility from observation
statements. But this criterion is too narrow (and I have discussed the problem of demarcation in
too wide): it excludes from science practically some detail because I believe that its solution is
everything that is, in fact, characteristic of it the key to most of the fundamental problems of
(while failing in effect to exclude astrology). No the philosophy of science. I am going to give you
scientific theory can ever be deduced from observa- later a list of some of these other problems, but
tion statements, or be described as a truth-function only one of them - the problem of induction - can
of observation statements. be discussed here at any length.
All this I pointed out on various occasions I had become interested in the problem oi
to Wittgensteinians and members of the Vienna induction in 1923. Although this problem is very
Circle. In 1931-2 I summarized my ideas in a closely connected with the problem of demarca-
largish book (read by several members of the tion, I did not fully appreciate the connection for
Circle but never published; although part of it was about five years.
incorporated in my Logic of Scientific Discovery); I approached the problem of induction through
and in 1933 I published a letter to the Editor of Hume. Hume, I felt, was perfectly right in point-
Erkenntnis in which I tried to compress into two ing out that induction cannot be logically justified.
pages my ideas on the problems of demarcation He held that there can be no valid logical~ argu-
and induction.s In this letter and elsewhere I ments allowing us to establish "that those instance..
described the problem of meaning as a pseudo- of which we have had no experience, resen,ble rhos..·.
problem, in contrast to the problem of demarca- of which we have had experience." Consequent!,·
tion. But my contribution was classified by "even cifter the observation of the frequent or con-
members of the C ircle as a proposal to replace stant conjunction of objects, we have no recison to
the verifiability criterion of 1nea11i11g by a fal sifi - draw any inference concerning any object beyond
ability criterion of r11ea11i11g - \Vhich effectively those of which we have had experienct·." For

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
 SCIENCE: CONJECTURES AND REFUTATIONS 477

"shou'd it be said that we have experience" 10 - Hume, I felt, had never accepted the full force
experience teaching us that objects constantly of his own logical analysis. Having refuted the
conjoined with certain other objects continue to logical idea of induction he was faced with the
be so conjoined- then, Hwne says, "I wou'd renew following problem: how do we actually obtain
my question, why from this experience we form any our knowledge, as a matter of psychological fact,
conclusion beyond those past instances, of which we if induction is a procedure which is logically
have had experience." In other words, an attempt invalid and rationally unjustifiable? There are
to justify the practice of induction by an appeal two possible answers: ( l) We obtain our know-
to experience must lead to an infinite regress. ledge by a non-inductive procedure. This answer
As a result we can say that theories can never be would have allowed Hume to retain a form of
inferred from observation statements, or rationally rationalism. (2) We obtain our knowledge by
justified by them. repetition and induction, and therefore by a
I found Hume's refutation of inductive infer- logically invalid and rationally unjustifiable pro-
ence clear and conclusive. But I felt completely cedure, so that all apparent knowledge is merely
dissatisfied with his psychological explanation of a kind of belief - belief based on habit. This
induction in terms of custom or habit. answer would imply that even scientific know-
ledge is irrational, so that rationalism is absurd,
!... I and must be given up. (I shall not discuss here
the age-old attempts, now again fashionable,
We must thus replace, for the purposes of a psy- to get out of the difficulty by asserting that
chological theory of the origin of our beliefs, the though induction is of course logically invalid
na"ive idea of events which are similar by the idea if we mean by "logic" the same as "deductive
of events to which we react by interpreting them logic," it is not irrational by its own standards, as
as being similar. But if this is so (and I can see may be seen from the fact that every reasonable
no escape from it) then Hume's psychological man applies it as a matter of fact: it was Hume's
theory of induction leads to an infinite regress, great achievement to break this uncritical identi-
precisely analogous to that other infinite regress fication of the question of fact - quid facti? - and
which was discovered by Hume himself, and the question of justification or validity - quid
. . ?
used by him to explode the logical theory of 1uns . ...
induction. For what do we wish to explain? In It seems that Hume never seriously considered
the example of the puppies we wish to explain the first alternative. Having cast out the logical
behaviour which may be described as recognizing theory of induction by repetition he struck a
or interpreting a situation as a repetition of bargain with common sense, meekly allowing the
another. Clearly, we cannot hope to explain this re-entry of induction by repetition, in the guise of
by an appeal to earlier repetitions, once we realize a psychological theory. I proposed to tum the
that the earlier repetitions must also have been tables upon this theory of Hume's. Instead of
repetitions-for-them, so that precisely the same explaining our propensity to expect regularities
problem arises again: that of recognizing or inter- as the result of repetition, I proposed to explain
preting a situation as a repetition of another. repetition-for-us as the result of our propensity
To put it more concisely, similarity-for-us is the to expect regularities and to search for them.
product of a response involving interpretations Thus I was led by purely logical considerations
(which may be inadequate) and anticipations to replace the psychological theory of induc-
or expectations (which may never be fulfilled). It tion by the following view. Without waiting,
is therefore impossible to explain anticipations, passively, for repetitions to impress or impose
or expectations, as resulting from many repeti- regularities upon us, we actively try to impose
tions, as suggested by Hume. For even the first regularities upon the world. We try to discover
repetition-for-us must be based upon similarity- similarities in it, and to interpret it in terms
for-us, and therefore upon expectations - precisely of laws invented by us. Without waiting for prem-
the kind of thing we wished to explain. ises we jump to conclusions. These may have to
This shows that there is an infinite regress be discarded later, should observation show that
involved in Hume's psychological theory. they are wrong.

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
478 KARL R. POPPER

This was a theory of trial and error - of Katz, 12 "divides the environment into edible and
conjectures and refutations. It made it possible to inedible things. An animal in flight stts roads
understand why our attempts to force inter- to escape and hiding places.... Generally speak-
pretations upon the world were logically prior to ing, objects change ... according to the needs
the observation of similarities. Since there were of the animal." We may add that objects can Ix
logical reasons behind this procedure, I thought classified, and can become similar or dissimilar,
that it would apply in the field of science also; that only in this way - by being related to needs and
scientific theories were not the digest of observa- interests. This rule applies not only to animals but
tions, but that they were inventions - conjectures also to scientists. For the animal a point of "iew
boldly put forward for trial, to be eliminated is provided by its needs, the task of the moment.
if they clashed with observations; with observa- and its expectations; for the scientist by his
tions which were rarely accidental but as a rule theoretical interests, the special problem under
undertaken with the definite intention of test- investigation, his conjectures and anticipations.
ing a theory by obtaining, if possible, a decisive and the theories which he accepts as a kind of
refutation. background: his frame of reference, his "horizon
of expectations."
The problem "Which comes first, the hypo-
V thesis (H) or the observation (0)," is soluble; as
is the problem, "Which comes first, the hen (HJ
The belief that science proceeds from observation or the egg ( 0)." The reply to the latter is, ~ An
to theory is still so widely and so firmly held earlier kind of egg"; to the former, "An earlier
that my denial of it is often met with incredulity. kind of hypothesis." It is quite true that an"·
I have even been suspected of being insincere particular hypothesis we choose will have been
- of denying what nobody in his senses can preceded by observations - the observations.
doubt. for example, which it is designed to explain. But
But in fact the belief that we can start with these observations, in their tum, presupposed
pure observations alone, without anything in the the adoption of a frame of reference: a frame
nature of a theory, is absurd; as may be illustrated of expectations: a frame of theories. If they were
by the story of the man who dedicated his life to significant, if they created a need for explanation
natural science, wrote down everything he could and thus gave rise to the invention of a hypo-
observe, and bequeathed his priceless collection thesis, it was because they could not be explained
of observations to the Royal Society to be used within the old theoretical framework, the old
as inductive evidence. This story should show us horizon of expectations. There is no danger here
that though beetles may profitably be collected, of an infinite regress. Going back to more and
observations may not. more primitive theories and myths we shall in the
Twenty-five years ago I tried to bring home end find unconscious, inborn expectations.
the same point to a group of physics students The theory of inborn ideas is absurd, I think;
in Vienna by beginning a lecture with the follow- but every organism has inborn reactions or
ing instructions: "Take pencil and paper, carefully responses; and among them, responses adapted
observe, and write down what you have observed" to impending events. These responses we may
They asked, of course, what I wanted them to describe as "expectations" without implying that
observe. Clearly the instruction, "Observe!" is these "expectations" are conscious. The newborn
absurd. 11 (It is not even idiomatic, unless the baby "expects," in this sense, to be fed (and, one
object of the transitive verb can be taken as could even argue, to be protected and loved).
understood.) Observation is always selective. It In view of the close relation between expecta-
needs a chosen object, a definite task, an interest, tion and knowledge we may even speak in quite
a point of view, a problem. And its description a reasonable sense of "inborn knowledge." This
presupposes a descriptive language, with property "knowledge" is not, however, valid a priori; an
words; it presupposes similarity and classification, inborn expectation, no matter how strong and
which in its turn presupposes interests, points of specific, may be mistaken. (The newborn child
view, and problen1s. "A hungry animal," writes may be abandoned, and starve.)

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
 SCIENCE: CONJECTURES AND REFUTATIONS 479

Thus we are born with expectations; with To sum up this logical criticism of Hume's psy-
"knowledge" which, although not valid a priori, chology of induction we may consider the idea of
is psychologically or genetically a priori, i.e. prior building an induction machine. Placed in a simp-
to all observational experience. One of the most lified "world" (for example, one of sequences of
important of these expectations is the expectation coloured counters) such a machine may through
of finding a regularity. It is connected with an repetition "learn," or even "formulate," laws of
inborn propensity to look out for regularities, or succession which hold in its "world." If such a
with a need to find regularities, as we may see from machine can be constructed (and I have no doubt
the pleasure of the child who satisfies this need. that it can) then, it might be argued, my theory
This "instinctive" expectation of finding regu- must be wrong; for if a machine is capable of
larities, which is psychologically a priori, cor- performing inductions on the basis of repetition,
responds very closely to the "law of causality" there can be no logical reasons preventing us
which Kant believed to be part of our mental from doing the same.
outfit and to be a priori valid. One might thus be The argument sounds convincing, but it is
inclined to say that Kant failed to distinguish mistaken. In constructing an induction machine
between psychologically a priori ways of think- we, the architects of the machine, must decide a
ing or responding and a priori valid beliefs. priori what constitutes its "world"; what things
But I do not think that his mistake was quite as are to be taken as similar or equal; and what kind
crude as that. For the expectation of finding of "laws" we wish the machine to be able to "dis-
regularities is not only psychologically a priori, cover" in its "world." In other words we must
but also logically a priori: it is logically prior to build into the machine a framework determin-
all observational experience, for it is prior to any ing what is relevant or interesting in its world:
recognition of similarities, as we have seen; and the machine will have its "inborn" selection
all observation involves the recognition of sim- principles. The problems of similarity will have
ilarities (or dissimilarities). But in spite of being been solved for it by its makers who thus have
logically a priori in this sense the expectation is interpreted the "world" for the machine.
not valid a priori. For it may fail: we can easily
construct an environment (it would be a lethal
one) which, compared with our ordinary environ- VI
ment, is so chaotic that we completely fail to
find regularities. (All natural laws could remain Our propensity to look out for regularities, and
valid: environments of this kind have been used to impose laws upon nature, leads to the psy-
in the animal experiments mentioned in the chological phenomenon of dogmatic thinking or,
next section.) more generally, dogmatic behaviour: we expect
Thus Kant's reply to Hume came near to regularities everywhere and attempt to find them
being right; for the distinction between an a even where there are none; events which do not
priori valid expectation and one which is both yield to these attempts we are inclined to treat
genetically and logically prior to observation, as a kind of "background noise"; and we stick to
but not a priori valid, is really somewhat subtle. our expectations even when they are inadequate
But Kant proved too much. In trying to show and we ought to accept defeat. This dogmatism
how knowledge is possible, he proposed a theory is to some extent necessary. It is demanded by a
which had the unavoidable consequence that our situation which can only be dealt with by forcing
quest for knowledge must necessarily succeed, our conjectures upon the world. Moreover, this
which is clearly mistaken. When Kant said, "Our dogmatism allows us to approach a good theory
intellect does not draw its laws from nature but in stages, by way of approximations: if we accept
imposes its laws upon nature," he was right. But defeat too easily, we may prevent ourselves from
in thinking that these laws are necessarily true, finding that we were very nearly right.
or that we necessarily succeed in imposing them It is clear that this dogmatic attitude, which
upon nature, he was wrong. ' 3 Nature very often makes us stick to our first impressions, is indic-
resists quite successfully, forcing us to discard our ative of a strong belief; while a critical attitude,
laws as refuted; but if we live we may try again. which is ready to modify its tenets, which admits

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
480 KARL R. POPPER

doubt and demands tests, is indicative of a weaker I found many other links between the ~ --
belief. Now according to Hume's theory, and chology of knowledge and psychological fields
to the popular theory, the strength of a belief which are often considered remote from it - for
should be a product of repetition; thus it should example the psychology of art and music; in fact.
always grow with experience, and always be greater my ideas about induction originated in a conjecture
in less primitive persons. But dogmatic thinking, about the evolution of Western polyphony. But
an uncontrolled wish to impose regularities, a you will be spared this story.
manifest pleasure in rites and in repetition as
such, are characteristic of primitives and children;
and increasing experience and maturity sometimes VII
create an attitude of caution and criticism rather
than of dogmatism. My logical criticism of Hume's psychological
I may perhaps mention here a point of theory, and the considerations connected \\'ith
agreement with psycho-analysis. Psycho-analysts it (most of which I elaborated in 1926-7, in a
assert that neurotics and others interpret the thesis entitled "On Habit and Belief in l,ay.•sft'' )
world in accordance with a persona! set pattern may seem a little removed from the field of the
which is not easily given up, and which can often philosophy of science. But the distinction between
be traced back to early childhood. A pattern or dogmatic and critical thinking, or the dogmatic
scheme which was adopted very early in life is and the critical attitude, brings us right back to
maintained throughout, and every new experience our central problem. For the dogmatic attitude
is interpreted in terms of it; verifying it, as it were, is clearly related to the tendency to verif>· our
and contributing to its rigidity. This is a descrip- laws and schemata by seeking to apply them and
tion of what I have called the dogmatic attitude, to confirm them, even to the point of neglecting
as distinct from the critical attitude, which shares refutations, whereas the critical attitude is one
with the dogmatic attitude the quick adoption of readiness to change them - to test them; to
of a schema of expectations - a myth, perhaps, refute them; to falsify them, if possible. This
or a conjecture or hypothesis - but which is suggests that we may identify the critical attitude
ready to modify it, to correct it, and even to give with the scientific attitude, and the dogmatic
it up. I am inclined to suggest that most neuroses attitude with the one which we have described as
may be due to a partially arrested development pseudo-scientific.
of the critical attitude; to an arrested rather than It further suggests that genetically speaking
a natural dogmatism; to resistance to demands the pseudo-scientific attitude is more primitive
for the modification and adjustment of certain than, and prior to, the scientific attitude: that
schematic interpretations and responses. This it is a prescientific attitude. And this primitivi~·
resistance in its turn may perhaps be explained, or priority also has its logical aspect. For the
in some cases, as due to an injury or shock, critical attitude is not so much opposed to the dog-
resulting in fear and in an increased need for assur- matic attitude as super-imposed upon it: criticism
ance or certainty analogous to the way in which · must be directed against existing and influential
an injury to a limb makes us afraid to move beliefs in need of critical revision - in other
it, so that it becomes stiff. (It might even be words, dogmatic beliefs. A critical attitude needs
argued that the case of the limb is not merely for its raw material, as it were, theories or beliefs
analogous to the dogmatic response, but an which are held more or less dogmatically.
instance of it.) The explanati_o n of any concrete Thus science must begin with myths, and with
case will have to take into account the weight of the criticism of myths; neither with the colle\."tion
the difficulties involved in making the necessary of observations, nor with the invention of experi·
adjustments - difficulties which may be consider- ments, but with the critical discussion of m\1hs. ,
able, especially in a complex and changing and of magical techniques and practices. The
world: we know fro1n experiments on animals scientific tradition is distinguished from the pre-
that varying degrees of neurotic behaviour may scientific tradition in having two layers. Like the
be produced at will by correspondingly varying latter, it passes on its theories; but it also passl>s
difficulties. on a critical attitude towards them. The theories

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
 SCIENCE: CONJECTURES AND REFUTATIONS 481

are passed on, not as dogmas, but rather with an eagerness to revise the theory if we succeed
the challenge to discuss them and improve upon in designing a test which it cannot pass - then
them. This tradition is Hellenic: it may be traced Hume was wrong. In such an acceptance of
back to Thales, founder of the first school (l do theories there is nothing irrational. There is not
not mean "of the first philosophical school," but even anything irrational in relying for practical
simply "of the first school") which was not mainly purposes upon well-tested theories, for no more
concerned with the preservation of a dogma. rational course of action is open to us.
The critical attitude, the tradition of free dis- Assume that we have deliberately made it our
cussion of theories with the aim of discovering task to live in this unknown world of ours; to
their weak spots so that they may be improved adjust ourselves to it as well as we can; to take
upon, is the attitude of reasonableness, of ration- advantage of the opportunities we can find in it;
ality. It makes far-reaching use of both verbal and to explain it, if possible ( we need not assume
argument and observation - of observation in that it is), and as far as possible, with the help of
the interest of argument, however. The Greeks' laws and explanatory theories. If we have made this
discovery of the critical method gave rise at first our task, then there is no more rational procedure
to the mistaken hope that it would lead to the than the method of trial and error - of conjecture
solution of all the great old problems; that it and refutation: of boldly proposing theories; of
would establish certainty; that it would help to trying our best to show that these are erroneous;
prove our theories, to justify them. But this hope and of accepting them tentatively if our critical
was a residue of the dogmatic way of thinking; efforts are unsuccessful.
in fact nothing can be justified or proved (out- From the point of view here developed all
side of mathematics and logic). The demand for laws, all theories, r:emain essentially tentative, or
rational proofs in science indicates a failure to conjectural, or hypothetical, even when we feel
keep distinct the broad realm of rationality and unable to doubt them any longer. Before a the-
the narrow realm of rational certainty: it is an ory has been refuted we can never know in what
untenable, an unreasonable demand. way it may have to be modified. That the sun will
Nevertheless, the role of logical argument, of always rise and set within twenty-four hours is
deductive logical reasoning, remains all-important still proverbial as a law "established by induc-
for the critical approach; not because it allows tion beyond reasonable doubt." It is odd that this
us to prove our theories, or to infer them from example is still in use, though it may have served
observation statements, but because only by purely well enough in the days of Aristotle and Pytheas
deductive reasoning is it possible for us to discover of Massalia - the great traveller who for centuries
what our theories imply, and thus to criticize was called a liar because of his tales of Thule, the
them effectively. Criticism, I said, is an attempt land of the frozen sea and the midnight sun.
to find the weak spots in a theory, and these, as The method of trial and error is not, of course,
a rule, can be found only in the more remote simply identical with the scientific or critical
logical consequences which can be derived from approach - with the method of conjecture and
it. It is here that purely logical reasoning plays refutation. The method of trial and error is
an important part in science. applied not only by Einstein but, in a more
Hume was right in stressing that our theories dogmatic fashion, by the amoeba also. The differ-
cannot be validly inferred from what we can ence lies not so much in the trials as in a critical
know to be true - neither from observations and constructive attitude towards errors; errors
nor from anything else. He concluded from this which the scientist consciously and cautiously
that our belief in them was irrational. If "belief" tries to uncover in order to refute his theories with
means here our inability to doubt our natural laws, searching arguments, including appeals to the
and the constancy of natural regularities, then most severe experimental tests which his theories
Hume is again right: this kind of dogmatic belief and his ingenuity permit him to design.
has, one might say, a physiological rather than a The critical attitude may be described as the
rational basis. If, however, the term "belief" is conscious attempt to make our theories, our
taken to cover our critical acceptance of scientific conjectures, suffer in our stead in the struggle
theories - a tentative acceptance combined with for the survival of the fittest. It gives us a chance

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
482 KARL R. POPPER

to survive the elimination of an inadequate I recently came across an interesting formula-

hypothesis -when a more dogmatic attitude would tion of this belief in a remarkable philosophical
eliminate it by eliminating us. (There is a touching book by a great physicist - Max Born's Natural
story of an Indian community which disappeared Philosophy of Cause and Chana. •s He writes:
because of its belief in the holiness of life, "Induction allows us to generalize a number
including that of tigers.) We thus obtain the of observations into a general rule: that night
fittest theory within our reach by the elimina- follows day and day follows night ... But \\'hile
tion of those which are less fit. (By "fitness" I do everyday life has no definite criterion for the
not mean merely "usefulness" but truth; .... ) validity of an induction, ... science has worked
I do not think that this procedure is irrational out a code, or rule of craft, for its application..,
or in need of any further rational justification. Born nowhere reveals the contents of this induc-
tive code (which, as his wording shows, contains
a "definite criterion for the validity of an induc-
VIII tion"); but he stresses that "there is no logical
argument" for its acceptance: "it is a question of
Let us now tum from our logical criticism of the faith"; and he is therefore "willing to call induc-
psychology ofexperience to our real problem - the tion a metaphysical principle." But why d<XS
problem of the logic of science. Although some of he believe that such a code of valid inductive rules
the things I have said may help us here, in so far must exist? This becomes clear when he speaks
as they may have eliminated certain psycholo- of the "vast communities of people ignorant
gical prejudices in favour of induction, my treat- of, or rejecting, the rule of science, among them
ment of the logical problem of induction is com- the members of anti-vaccination societies and
pletely independent of this criticism, and of all psy- believers in astrology. It is useless to argue with
chological considerations. Provided you do not them; I cannot compel them to accept the same
dogmatically believe in the alleged psychological criteria of valid induction in which I belieYe:
fact that we make inductions, you may now the code of scientific rules." This makes it quite
forget my whole story with the exception of two clear that "valid induction" was here meant to sen,r
logical points: my logical remarks on testability as a criterion of demarcation between science and
or falsifiability as the criterion of demarcation; and pseudo-science.
Hume's logical criticism of induction. But it is obvious that this rule or craft of u,'lllid
From what I have said it is obvious that there induction" is not even metaphysical: it simply
was a close link between the two problems which does not exist. No rule can ever guarantee that a
interested me at that time: demarcation, and generalization inferred from true observations.
induction or scientific method. It was easy to however often repeated, is true. (Born himself does
see that the method of science is criticism, i.e. not believe in the truth of Newtonian physics.
attempted falsifications. Yet it took me a few years in spite of its success, although he believes that
to notjce that the two problems - of demarcation it is based on induction.) And the success of
and of induction - were in a sense one. science is not based upon rules of induction,
Why, I asked, do so many scientists believe but depends upon luck, ingenuity, and the purely
in induction? I found they did so because they deductive rules of critical argument.
believed natural science to be characterized by I may summarize some of my conclusions as
the inductive method - by a method starting follows:
from, and relying upon, long ~quences of obser-
vations and experiments. They believed that the ( l ) Induction, i.e. inference based on manv
djfference between genuine science and meta- observations, is a myth. It is neither a psy-
physical or pseudo-scientific speculation de- chological fact, nor a fact of ordinary life,
pended solely upon whether or not the inductive nor one of scientific procedure.
method was employed. They believed (to put it (2) The actual procedure of science is to operate
in my own terminology) that only the inductive with conjectures: to jump to conclusions -
method could provide a satisfactory criterion of often after one single observation (as noticed
demarcation. for example by Hume and Born).

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
 SCIENCE: CONJECTURES AND REFUTATIONS 483

( 3) Repeated observations and experiments func- including Bertrand Russell, have done before him),
tion in science as tests of our conjectures or in favour of what he calls a "metaphysical prin-
hypotheses, i.e. as attempted refutations. ciple"; a metaphysical principle which he does
( 4) The mistaken belief in induction is fortified not even attempt to formulate; which he vaguely
by the need for a criterion of demarcation describes as a "code or rule of craft"; and of which
which, it is traditionally but wrongly believed, I have never seen any formulation which even
only the inductive method can provide. looked promising and was not clearly untenable.
( 5) The conception of such an inductive method, But in fact the principles (a) to (c) do not clash.
like the criterion of verifiability, implies a We can see this the moment we realize that the
faulty demarcation. acceptance by science of a law or of a theory is
( 6) None of this is altered in the least if we say tentative only; which is to say that all laws
that induction makes theories only probable and theories are conjectures, or tentative hypo-
rather than certain. theses (a position which I have sometimes called
"hypotheticism"); and that we may reject a law
or theory on the basis of new evidence, without
IX necessarily discarding the old evidence which
originally led us to accept it. 17
If, as I have suggested, the problem of induction The principle of empiricism (c) can be fully
is only an instance or facet of the problem of preserved, since the fate of a theory, its accept-
demarcation, then the solution to the problem ance or rejection, is decided by observation and
of demarcation must provide us with a solution experiment - by the result of tests. So long as
to the problem of induction. This is indeed the a theory stands up to the severest tests we can
case, I believe, although it is perhaps not imme- design, it is accepted; if it does not, it is rejected.
diately obvious. But it is never inferred, in any sense, from the
For a brief formulation of the problem of empirical evidence. There is neither a psycho-
induction we can turn again to Born, who writes: logical nor a logical induction. Only the falsity of
" ... no observation or experiment, however ex- the theory can be inferred from empirical evidence,
tended, can give more than a finite number of and this inference is a purely deductive one.
repetitions"; therefore, "the statement of a law - Hume showed that it is not possible to infer
B depends on A - always transcends experience. a theory from observation statements; but this
Yet this kind of statement is made everywhere does not affect the possibility of refuting a theory
and all the time, and sometimes from scanty by observation statements. The full appreciation
material." 16 of this possibility makes the relation between
In other words, the logical problem of induc- theories and observations perfectly clear.
tion arises from (a) Hume's discovery (so well This solves the problem of the alleged clash
expressed by Born) that it is impossible to justify between the principles (a), (b), and (c), and with
a law by observation or experiment, since it it Hume's problem of induction.
"transcends experience"; (b) the fact that science
proposes and uses laws "everywhere and all the
time." (Like Hume, Born is struck by the "scanty X
material," i.e. the few observed instances upon
which the law may be based.) To this we have to Thus the problem of induction is solved. But
add (c) the principle of empiricism which asserts nothing seems less wanted than a simple solution
that in science, only observation and experiment to an age-old philosophical problem. Wittgenstein
may decide upon the acceptance or rejection of and his school hold that genuine philosophical
scientific statements, including laws and theories. problems do not exist; 18 from which it clearly fol-
These three principles, (a), (b), and (c), appear lows that they cannot be solved. Others among
at first sight to clash; and this apparent clash my contemporaries do believe that there are
constitutes the logical problem of induction. philosophical problems, and respect them; but they
Faced with this clash, Born gives us (c), the prin- seem to respect them too much; they seem to
ciple of empiricism (as Kant and many others, believe that they are insoluble, if not taboo; and

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
484 KARL R. POPPER

they are shocked and horrified by the claim that clearly not justifiable, the following new problem
there is a simple, neat, and lucid, solution to any must arise: how do you justify your method oi
of them. If there is a solution it must be deep, they trial and error? Reply: the method of trial and error
feel, or at least complicated. is a method ofeliminating false theories by obsen·a-
However this may be, I am still waiting for a tion statements; and the justification for this is the
simple, neat and lucid criticism of the solution purely logical relationship of deducibility which
which I published first in 1933 in my letter to the allows us to assert the falsity of universal state-
Editor of Erkenntnis,1 9 and later in The Logic of ments if we accept the truth of singular ones.
Scientific Discovery. Another question sometimes asked is this: "'·hy
Of course, one can invent new problems of is it reasonable to prefer non-falsified state-
induction, different from the one I have formu- ments to falsified ones? To this question some
lated and solved. (Its formulation was half its involved answers have been produced, for ex-
solution.) But I have yet to see any reformulation ample pragmatic answers. But from a pragmati..:
of the problem whose solution cannot be easily point of view the question does not arise. since
obtained from my old solution. I am now going false theories often serve well enough: most
to discuss some of these re-formulations. formulae used in engineering or navigation are
One question which may be asked is this: how known to be false, although they may be excellent
do we really jump from an observation statement approximations and easy to handle; and they are
to a theory? used with confidence by people who know them
Although this question appears to be psycho- to be false.
logical rather than philosophical, one can say The only correct answer is the straightforward
something positive about it without invoking one: because we search for truth (even though "''e
psychology. One can say first that the jump is can never be sure we have found it), and because
not from an observation statement, but from the falsified theories are known or believed to
a problem-situation, and that the theory must be false, while the non-falsified theories ma,·• still
allow us to explain the observations which created be true. Besides, we do not prefer every non-
the problem (that is, to deduce them from the falsified theory - only one which, in the light
theory strengthened by other accepted theories of criticism, appears to be better than its com-
and by other observation statements, the so- petitors: which solves our problems, which is
called initial conditions). This leaves, of course, well tested, and of which we think, or rather con-
an immense number of possible theories, good jecture or hope (considering other provisionally
and bad; and it thus appears that our question accepted theories), that it will stand up to further
has not been answered. tests.
But this makes it fairly clear that when we It has also been said that the problem of
asked our question we had more in mind than, induction is, "Why is it reasonable to believe
"How do we jump from an observation statement that the future will be like the past?," and that a
to a theory?" The question we had in mind was, satisfactory answer to this question should make
it now appears, "How do we jump from an it plain that such a belief is, in fact, reasonable.
observation statement to a good theory?" But to My reply is that it is reasonable to believe that
this the answer is: by jumping first to any theory the future will be very different from the past in
and then testing it, to find whether it is good many vitally important respects. Admittedly it is
or not; i.e. by repeatedly applying the critical perfectly reasonable to act on the assumption
method, eliminating many .bad theories, and that it will, in many respects, be like the past. and
inventing many new ones. Not everybody is able that well-tested laws will continue to hold (since
to do this; but there is no other way. we can have no better assumption to act upon);
Other questions have sometimes been asked. but it is also reasonable to believe that such a
The original problem of induction, it was said, course of action will lead us at times into severe
is the problem of justifying induction, i.e. of trouble, since some of the laws upon which ,ve
justifying inductive inference. If you answer this now heavily rely may easily prove unreliable.
problem by saying that what is called an "induct- (Remember the midnight sun!) One might even
ive inference'' is always invalid and therefore say that to judge from past experience, and from

. Digitized by Google Original from

UNIVERSITY OF MICHIGAN
 SCIENCE: CONJECTURES AND REFUTATIONS 485

our general scientific knowledge, the future will long history of human endeavours to know more
not be like the past, in perhaps most of the ways about our world.
which those have in mind who say that it will. Yet another way of putting the problem of
Water will sometimes not quench thirst, and air induction is in terms of probability. Let t be the
will choke those who breathe it. An apparent theory and e the evidence: we can ask for P(t,e),
way out is to say that the future will be like the that is to say, the probability of t, given e. The
past in the sense that the laws of nature will not problem of induction, it is often believed, can
change, but this is begging the question. We then be put thus: construct a calculus of probability
speak of a "law of nature" only if we think that which allows us to work out for any theory t what
we have before us a regularity which does not its probability is, relative to any given empirical
change; and if we find that it changes then we shall evidence e; and show that P( t,e) increases with the
not continue to call it a "law of nature." Of accumulation of supporting evidence, and reaches
course our search for natural laws indicates that high values - at any rate values greater than 1/2.
we hope to find them, and that we believe that In The Logic of Scientific Discovery l explained
there are natural laws; but our belief in any par- why I think that this approach to the problem is
ticular natural law cannot have a safer basis than fundamentally mistaken. 20 To make this clear,
our unsuccessful critical attempts to refute it. I introduced there the distinction between prob-
I think that those who put the problem of ability and degree ofcorroboration or confirmation.
induction in terms of the reasonableness of our (The term "confirmation" has lately been so
beliefs are perfectly right if they are dissatisfied with much used and misused that I have decided to
a Humean, or post-Humean, sceptical despair of surrender it to the verificationists and to use
reason. We must indeed reject the view that a belief for my own purposes "corroboration" only. The
in science is as irrational as a belief in primitive term "probability" is best used in some of the many
magical practices - that both are a matter of senses which satisfy the well-known calculus of
accepting a "total ideology," a convention or a tra- probability, axiomatized, for example, by Keynes,
dition based on faith. But we must be cautious Jeffreys, and myself; but nothing of course de-
if we formulate our problem, with Hume, as one pends on the choice of words, as long as we do
of the reasonableness of our beliefs. We should not assume, uncritically, that degree of corrobora-
split this problem into three - our old problem tion must also be a probability - that is to say,
of demarcation, or of how to distinguish between that it must satisfy the calculus of probability.)
science and primitive magic; the problem of the I explained in my book why we are interested
rationality of the scientific or critical procedure, in theories with a high degree of corroboration,
and of the role of observation within it; and And I explained why it is a mistake to conclude
lastly the problem of the rationality of our accept- from this that we are interested in highly prob-
ance of theories for scientific and for practical able theories. I pointed out that the probability
purposes. To all these three problems solutions of a statement (or set of statements) is always the
have been offered here. greater the less the statement says: it is inverse
One should also be careful not to confuse the to the content or the deductive power of the
problem of the reasonableness of the scientific statement, and thus to its explanatory power.
procedure and the (tentative) acceptance of the Accordingly every interesting and powerful state-
results of this procedure - i.e. the scientific ment must have a low probability; and vice
theories - with the problem of the rationality versa: a statement with a high probability will be
or otherwise of the belief that this procedure will scientifically uninteresting, because it says little
succeed. In practice, in practical scientific research, and has no explanatory power. Although we
this belief is no doubt unavoidable and reason- seek theories with a high degree of corroboration,
able, there being no better alternative. But the belief as scientists we do not seek highly probable theories
is certainly unjustifiable in a theoretical sense, as but explanations; that is to say, powerful and im-
I have argued. Moreover, if we could show, on probable theories. 21 The opposite view- that science
general logical grounds, that the scientific quest aims at high probability - is a characteristic
is likely to succeed, one could not understand development of verificationism: if you find that
why anything like success has been so rare in the you cannot verify a theory, or make it certain by

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
486 KARL R. POPPER

induction, you may turn to probability as a kind situations, if actually observed, mean that the the-
of" Ersatz" for certainty, in the hope that induc- ory is refuted. But what kind of clinical respon~
tion may yield at least that much. would refute to the satisfaction of the analyst not
1 have discussed the two problems of demar- merely a particular analytic diagnosis but p;y(ho--
analysis itself? And have such criteria ever ~n
cation and induction at some length. Yet since I
discussed or agreed upon by analysts? ls there
set out to give you in this lecture a kind of report
not, on the contrary, a whole family of analytK
on the work I have done in this field I shall concepts, such as "ambivalence" (I do not sugge,t
have to add, in the form of an Appendix, a few that there is no such thing as ambivalence>, " ·hi.:h
words about some other problems on which I have would make it difficult, if not impossible. 10 d!?J'~
been working, between 1934 and 1953. I was upon such criteria? Moreover, how much head"·a,·
led to most of these problems by trying to think has been made in investigating the question of th<'
out the consequences of the solutions to the extent to which the (conscious or uncons.:ious 1
two problems of demarcation and induction. But expectations and theories held by the analvs1
time does not allow me to continue my narrative, influence the "clinical responses" of the patient?
and to tell you how my new problems arose out (To say nothing about the conscious attempu to
influence the patient by proposing interpr<'ta-
of my old ones. Since I cannot even start a dis-
tions to him, etc.) Years ago I introduced the
cussion of these further problems now, I shall
term "Oedipus effect" to describe the influence of
have to confine myself to giving you a bare list a theory or expectation or prediction upon rhe ''"'-mr
of them, with a few explanatory words here and which it predicts or describes: it will be remembered
there. But even a bare list may be useful, I think. that the causal chain leading to Oedipus· parricide
It may serve to give an idea of the fertility of was started by the oracle's prediction of this f''ent
the approach. It may help to illustrate what our This is a characteristic and recurrent theme oi
problems look like; and it may show how many such myths, but one which seems to have iailed
there are, and so convince you that there is no to attract the interest of the analysts, perhaps
need whatever to worry over the question whether not accidentally. (The problem of confinnato~·
philosophical problems exist, or what philosophy dreams suggested by the analyst is discu~ h"
Freud, for example in Ge.sammelrt: S<hri~en. Ill,
is really about. So this list contains, by implica-
1925, where he says on p. 314: ulf anybody as~n~
tion, an apology for my unwillingness to break that most of the dreams which can be utilized in
with the old tradition of trying to solve problems an analysis ... owe their origin to [the analyst· s:
with the help of rational argument, and thus for suggestion, then no objection can be made from
my unwillingness to participate wholeheartedly the point of view of analytic theory. Yet there is
in the developments, trends, and drifts, of con- nothing in this fact," he surprisingly adds. "~·hich
temporary philosophy. would detract from the reliability of our results. - i
4 The case of astrology, nowadays a typical pseuJo-
science, may illustrate this point. It was attacked
by Aristotelians and other rationalists, down to
Notes Newton's day, for the wrong reason - for it.s
now accepted assertion that the planets had an
I This is a slight oversimplification, for about half "influence" upon terrestrial ("sublunar") e.,.ents.
of the Einstein effect n1ay be derived from the In fact Newton's theory of gravity, and especi.ilh·
classical theory, provided we assume a ballistic the lunar theory of the tides, was historicalh·
theory of light. speaking an offspring of astrological lore. NC\\1on.
2 Sec, for exan1ple, my Open Society and Its Ene111ies, it seems, was most reluctant to adopt a thror.,.
ch. 15, section iii, and notes.13-14. which came from the same stable as for example
3 "(:linical observations," like all other observa- the theory that "influenza" epidemics are due to
tions, are interpretations in tire light of theories . . . ; an astral "influence." And Galileo, no doubt for
and for this reason alone they are apt to seem the same reason, actually rejected the lunar the-
to support those theories in the light of which ory of the tides; and his misgivings about Kepler
they were interpreted. But real support can be may easily be explained by his misgivings about
obtained onlv, from observations undert.iken as astrology.
tests (by "attempted refutations"); and for this 5 My Logic of Scientific Discovery ( 1959, 1960, I 96 I ).
purpose critrri11 of rcf111a1io11 have to be laid down here usually referred to as L.5<.D., is the translation
beforehand: it must be agreed which observable of l.ogik der Forschung ( 1934 ), with a number of

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
 SCIENCE: CONJECTURES AND REFUTATIONS 487

additional notes and appendices, including (on it quite certain that we have not overlooked
pp. 312-14) the letter to the Editor of Erkenntnis a mistake). On the first level, there is a logical
mentioned here in the text which was first pub- asymmetry: one singular statement - say about
lished in Erkenntnis, 3, 1933, pp. 426 f. the perihelion of Mercury - can formally falsify
Concerning my never published book men- Kepler's laws; but these cannot be formally verified
tioned here in the text, see R. Carnap's paper by any number of singular statements. The attempt
"Ueber Protokollstdze" (On Protocol-Sentences), to minimize this asymmetry can only lead to
Erkenntnis, 3, 1932, pp. 215-28 where he gives an confusion. On another level, we may hesitate to
outline of my theory on pp. 223-8, and accepts accept any statement, even the simplest observa-
it. He calls my theory "procedure B," and says tion statement; and we may point out that every
( p. 224, top): "Starting from a point of view dif- statement involves interpretation in the light of
ferent from Neurath's" (who developed what theories, and that it is therefore uncertain. This
Carnap calls on p. 223 "procedure A"), "Popper does not affect the fundamental asymmetry, but
developed procedure B as part of his system." it is important: most dissectors of the heart before
And after describing in detail my theory of tests, Harvey observed the wrong things - those, which
Carnap sums up his views as follows (p. 228): "After they expected to see. There can never be anything
weighing the various arguments here discussed, it like a completely safe observation, free from the
appears to me that the second language form with dangers of misinterpretation. (This is one of the
procedure B - that is in the form here described reasons why the theory of induction does not
- is the most adequate among the forms of scientific work.) The "empirical basis" consists largely of a
language at present advocated . .. in the ... theory mixture of theories of lower degree of universality
of knowledge." This paper of Carnap's contained (of"reproducible effects"). But the fact remains that,
the first published report of my theory of critical relative to whatever basis the investigator may
testing. (See also my critical remarks in L.Sc.D., note accept (at his peril), he can test his theory only by
I to section 29, p. 104, where the date "1933" should trying to refute it.
read "1932".) 9 Hume does not say "logical" but "demonstrative,"
6 Wittgenstein's example of a nonsensical pseudo- a terminology which, I think, is a little mislead-
proposition is: "Socrates is identical." Obviously, ing. The following two quotations are from the
"Socrates is not identical" must also be nonsense. Treatise of Human Nature, Book I, Part III,
Thus the negation of any nonsense will be non- sections vi and xii. (The italics are all Hume's.)
sense, and that of a meaningful statement will IO This and the next quotation are from loc. cit.,
be meaningful. But the negation of a testable (or section vi. See also Hume's Enquiry Concerning
falsifiable) statement need not be testable, as was Human Understanding, section IV, Part II, and
pointed out, first in my L.Sc.D., (e.g. pp. 38 f.) his Abstract, edited 1938 by J. M. Keynes and
and later by my critics. The confusion caused by P. Sraffa, p. 15, and quoted in L.Sc.D., new
taking testability as a criterion of meaning rather appendix •v11. text to note 6.
than of demarcation can easily be imagined. 11 See section 30 of L.Sc.D.
7 The most recent example of the way in which the 12 Katz, loc. cit.
history of this problem is misunderstood is A. R. 13 Kant believed that Newton's dynamics was a
White's "Note on Meaning and Verification," priori valid. (See his Metaphysical Foundations of
Mind, 63, 1954, pp. 66 ff. J. L Evans's article, Mind, Natural Scienc.e, published between the first and the
62, 1953, pp. I ff., which Mr White criticizes, is second editions of the Critique of Pure Reason.) But
excellent in my opinion, and unusually perceptive. if, as he thought, we can explain the validity of
Understandably enough, neither of the authors can Newton's theory by the fact that our intellect
quite reconstruct the story. (Some hints may be imposes its laws upon nature, it follows, I think,
found in my Open Society, notes 46, 51 and 52 to that our intellect must succeed in this; which makes
ch. 11.) it hard to understand why a priori knowledge
8 In L.Sc.D. I discussed, and replied to, some likely such as Newton's should be so hard to come by.
objections which afterwards were indeed raised 14 A thesis submitted under the title "Gewohnheit und
without reference to my replies. One of them is the Gesetzerlebnis" to the Institute of Education of the
contention that the falsification of a natural law is City of Vienna in 1927. (Unpublished.)
just as impossible as its verification. The answer 15 Max Born, Natural Philosophy ofCause and Chance,
is that this objection mixes two entirely different Oxford, 1949, p. 7.
levels of analysis (like the objection that mathem- I6 Natural Philosophy of Cause and Chance, p. 6.
atical demonstrations are impossible since checking, 17 I do not doubt that Born and many others would
no matter how often repeated, can never make agree that theories are accepted only tentatively.

Digitized by Google Original from

UNIVERSITY OF MICHIGAN
488 KARL R. POPPER

But the widespread belief in induction shows ability: it may have values between - I ( refutation
that the far-reaching implications of this view are oft bye) and C(r,r) :S: +l. Statements r which .ue
rarely seen. lawlike and thus non-verifiable cannot even rea.::h
18 Wittgenstein still held this belief in 1946. C(t,e) = C(t,t) upon empirical evidence e. CU.ti 1s
19 See note 5, above. the degree of corroborability of r, and is equal to the
20 LSc.D. (see note 5 above), ch. x, especially sections degree of testability oft, or to the contmt oft. Bcxawe
80 to 83, also section 34 ff. See also my note of the demands implied in point ( 6) at the end of
"A Set of Independent Axioms for Probability," section I above, I do not think, however, that it is
Mind, N.S. 47, 1938, p . 275. (This note has since possible to give a complete formalization of th< idea
been reprinted, with corrections, in the new of corroboration (or, as I previously used to s.av.
appendix •ii of L.Sc.D.) of confirmation).
21 A definition, in terms of probabilities (see the next (Added 1955 to the first proofs of this pa~r: :
note), of C(t,e), i.e. of the degree of corroboration See also my note "Degree of Confirmation.-
(of a theory t relative to the evidence e) satisfying British Journal for the Philosophy of Scio,,£. .5.
the demands indicated in my L.Sc.D., sections 82 1954, pp. 143 ff. (See also 5, p. 334.) I have s,n.:e
to 83, is the following: simplified this definition as follows (8 ./ P.S..
1955, 5, p. 359:)
C(t,e) = E(t,e)(I + P(t)P(t,e)),
C(r,e) = (P(e,t) - P(e))l(P(e,t) - P(et) + p, •. : ,
where E(t,e) = (P(e,t) - P(e))/(P(e,r) + P(e)) is a
(non-additive) measure of the explanatory power For a further improvement, see 8./.P.S. 6. I9:'-5.
oft with respect to e. Note that C( t,e) is not a prob- p. 56.

Digitized by Google Original from

UNIVERSITY OF MICHIGAN