Science as Successful Prediction
The following excerpt was originally published in Criticism and the Growth of Knowledge (1970).
by Imre Lakatos
Science: Reason or Religion?
For centuries knowledge meant proven knowledgeproven either by the power of the intellect or by the evidence of the senses. Wisdom and intellectual integrity demanded that one must desist from unproven utterances and minimize, even in thought, the gap between speculation and established knowledge. The proving power of the intellect or the senses was questioned by the skeptics more than two thousand years ago; but they were browbeaten into confusion by the glory of Newtonian physics. Einstein's results again turned the tables and now very few philosophers or scientists still think that scientific knowledge is, or can be, proven knowledge. But few realize that with this the whole classical structure of intellectual values falls in ruins and has to be replaced: one cannot simply water down the ideal of proving truthas some empiricists doto the ideal of 'probable truth' oras some sociologists of knowledge doto 'truth by changing consensus.'
Popper's distinction lies primarily in his having grasped the full implications of the collapse of the best-corroborated scientific theory of all times: Newtonian mechanics and the Newtonian theory of gravitation. In his view virtue lies not in caution in avoiding errors, but in ruthlessness in eliminating them. Boldness in conjectures on the one hand and austerity in refutations on the other: this is Popper's recipe. Intellectual honesty does not consist in trying to entrench, or establish one's position by proving (or 'probabilifying') itintellectual honesty consists rather in specifying precisely the conditions under which one is willing to give up one's position. Committed Marxists and Freudians refuse to specify such conditions: this is a hallmark of their intellectual dishonesty. Belief may be a regrettably unavoidable biological weakness to be kept under the control of criticism: but commitment is for Popper an outright crime.
Kuhn thinks otherwise. He too rejects the idea that science grows by accumulation of eternal truths. He too takes his main inspiration from Einstein's overthrow of Newtonian physics. His main problem too is scientific revolution. But while according to Popper science is "revolution in permanence," and criticism the heart of the scientific enterprise, according to Kuhn revolution is exceptional and, indeed, extra-scientific, and criticism is, in "normal" times, anathema. Indeed for Kuhn the transition from criticism to commitment marks the point where progressand "normal" sciencebegins. For him the idea that on "refutation" one can demand the rejection, the elimination of a theory, is "naive" falsificationism. Criticism of the dominant theory and proposals of new theories are only allowed in the rare moments of "crisis." This last Kuhnian thesis has been widely criticized and I shall not discuss it. My concern is rather that Kuhn, having recognized the failure both of justificationism and falsificationism in providing rational accounts of scientific growth, seems now to fall back on irrationalism.
For Popper scientific change is rational or at least rationally reconstructible and falls in the realm of the logic of discovery. For Kuhn scientific changefrom one "paradigm" to anotheris a mystical conversion which cannot be governed by rules of reason and which falls totally within the realm of (social) psychology of discovery. Scientific change is a kind of religious change.
The clash between Popper and Kuhn is not about a mere technical point in epistemology. It concerns our central intellectual values, and has implications not only for theoretical physics but also for the underdeveloped social sciences and even for moral and political philosophy. If even in science there is no a way of judging a theory but by assessing the number, faith and vocal energy of its supporters, then this must be even more so in the social sciences: truth lies in power. Thus Kuhn's position would vindicate, no doubt, unintentionally, the basic political credo of contemporary religious maniacs ("student revolutionaries").
In this paper I shall first show that in Popper's logic of scientific discovery two different positions are conflated. Kuhn understands only one of these, "naive falsificationism" (I prefer the term "naive methodological falsificationism"); I think that his criticism of it is correct, and I shall even strengthen it. But Kuhn does not understand a more sophisticated position the rationality of which is not based on "naive" falsificationism. I shall try to explainand further strengthenthis stronger Popperian position which, I think, may escape Kuhn's strictures and present scientific revolutions as constituting rational progress rather than as religious conversions.
Fallibilism versus Falsification
Sophisticated verses naive methodological falsification.
Progressive and degenerating problemshifts.
Sophisticated falsificationism differs from naive falsificationism both in its rules of acceptance (or "demarcation criterion") and its rules of falsification or elimination. For the naive falsificationist any theory which can be interpreted as experimentally falsifiable, is "acceptable" or "scientific." For the sophisticated falsificationist a theory is "acceptable" or "scientific" only if it has corroborated excess empirical content over its predecessor (or rival), that is, only if it leads to the discovery of novel facts. This condition can be analyzed into two clauses: that the new theory has excess empirical content ("acceptability"1) and that some of this excess empirical content is verified ("acceptability"2). The first clause can be checked instantly by a priori logical analysis; the second can be checked only empirically and this may take an indefinite time.
Again, for the naive falsificationist a theory is falsified by a "(fortified) observational" statement which conflicts with it (or rather, which he decides to interpret as conflicting with it). The sophisticated falsificationist regards a scientific theory T as falsified if and only if another theory T' has been proposed with the following characteristics: (1) T' has excess empirical content over T: that is, it predicts novel facts, that is, facts improbable in the light of, or even forbidden, by T, (2) T' explains the previous success of T, that is, all the unrefuted content of T is contained (within the limits of observational error) in the content of T'; and (3) some of the excess content of T' is corroborated.
Let us take a series of theories, T1, T2, T3, . . . where each subsequent theory results from adding auxiliary causes to (or from semantical reinterpretations of) the previous theory in order to accommodate some anomaly, each theory having at least as much content as the unrefuted content of its predecessor. Let us say that such a series of theories is theoretically progressive (or "constitutes a theoretically progressive problemshift") if each new theory has some excess empirical content over its predecessor, that is, if it predicts some novel, hitherto unexpected fact. Let us say that such a series of theories is also empirically progressive (or "constitutes an empirically progressive problemshift") if some of this excess empirical content is also corroborated, that is, if each new theory leads us to the actual discovery of some new fact. Finally, let us call a problemshift progressive if it is both theoretically and empirically progressive, and degenerating if it is not. We "accept" problemshifts as "scientific" only if they are at least theoretically progressive; if they are not, we "reject" them as "pseudoscientific." Progress is measured by the degree to which a problemshift is progressive, by the degree to which the series of theories leads us to the discovery of novel facts. We regard a theory in the series "falsified" when it is superseded by a theory with higher corroborated content.
This demarcation between progressive and degenerating problemsshifts sheds new light on the appraisal of scientificor, rather, progressiveexplanations. If we put forward a theory to resolve a contradiction between a previous theory and a counterexample in such a way that the new theory, instead of offering a content-increasing (scientific) explanation, only offers a content-decreasing (linguistic) reinterpretation, the contradiction is resolved in a merely symmetrical, unscientific way. A given fact is explained scientifically only if a new fact is also explained with it.
Sophisticated falsificationism thus shifts the problem of how to appraise theories to the problem of how to appraise series of theories. Not an isolated theory, but only a series of theories can be said to be scientific or unscientific: to apply them the term "scientific" to one single theory is a category mistake.
The time-honored empirical criterion for a satisfactory theory was agreement with the observed facts. Our empirical criterion for a series of theories is that it should produce new facts. The idea of growth and the concept of empirical character are soldered into one.
This revised form of methodological falsificationism has many new features. First, it denies that "in the case of a scientific theory, our decision depends upon the result of experiments. If these confirm this theory, we may except it until we find a better one. If they contradict the theory, we reject it." It denies that "what ultimately decides the fate of a theory is the result of the test, i.e., an agreement about basic statements." Contrary to naive falsificationism, no experiment, experimental report, observation statement or well-corroborated low-level falsifying hypothesis alone can lead to falsification. There is no falsification before the emergence of a better theory. But then the distinctively negative character of naive falsificationism vanishes; criticism becomes more difficult, and also positive, constructive. But, of course, if falsification depends on the emergence of better theories, on the invention of theories which anticipate new facts, then falsification is not simply a relation between a theory and the empirical basis, but a multiple relation between competing theories, the original "empirical basis," and the empirical growth resulting from the competition. Falsification can thus be said to have a "historical character." Moreover, some of the theories which bring about falsification are frequently proposed after the "counterevidence" This may sound paradoxical for people indoctrinated with the naive falsificationism. Indeed, this epistemological theory of the relation between theory and experiment differs sharply from the epistemological theory of naive falsificationism. The very term "counterevidence" has to be abandoned in the sense that no experimental result must be interpreted directly as "counterevidence." If we still want to retain this time-honored term, we have to redefined it like this: "counterevidence" to T1' is a corroborating instance to T2 which is either inconsistent with or independent of T1 (with the proviso that T2 is a theory which satisfactorily explains the empirical success of T1). This shows that "crucial counterevidence"or "crucial experiments"can be recognized as such among the scores of anomalies only with hindsight, in the light of some superseding theory.
Thus the crucial element in falsificationism is whether the new theory offers any novel, excess information compared with its predecessor and whether some of this excess information is corroborated. Justificationists valued 'confirming' instances of a theory; naive falsificationists stressed 'refuting' instances; for the methodological falsificationists it is therather rarecorroborating instances of the excess information which are the crucial ones; these receive all the attention.
The main contemporary protagonist of the ideal of "truth by consensus" are Polanyi and Kuhn. For the historical background and criticism of this position, cf. Musgrave [1969a], Musgrave [1969b] and Lakatos .
Indeed he introduces his  by arguing against the "development-by-accumulation" idea of scientific growth. But his intellectual debt is to Koyre rather than Popper. Koyre showed that positivism gives bad guidance to the historian of science, for the history of physics can only be understood in the context of a succession of "metaphysical" research programs. Thus scientific changes are connected with vast cataclysmic metaphysical revolutions. Kuhn develops this message of Burtt and Koyre and the vast success of his book was partly due to his hard-hitting, direct criticism of justificationist historiography which created a sensation among ordinary scientists and historians of science whom Burtt's, Koyre's (or Popper's) message has not yet reached. But, unfortunately, his message had some authoritarian and rationalist overtones.
If I already know P1: "Swan A is white," Pw: "All swans are white" represents no progress, because it may only lead to the discovery of such further similar facts as P2: "Swan B is white." So-called "empirical generalizations" constitute no progress. A new fact must be improbable or even impossible in the light of previous knowledge.
The appropriateness of the term "problemshift" for a series of theories rather than of problems may be questioned. I choose it partly because I have not found a more appropriate alternative "theoryshift" sounds dreadful partly because theories are always problematical, they never solve all the problems they have set out to solve.
Indeed, in the original manuscript of my [1968a] I wrote: "A theory without excess corroboration has no excess explanatory power; therefore, according to Popper, it does not represent growth and therefore it is not 'scientific'; therefore we should say, it has no explanatory power" (p. 386). I cut out the italicized half of the sentence under pressure from my colleagues who thought it sounded too eccentric. I regret it now.
Poppers's conflation of "theories" and "series of theories" prevented him from getting the basic idea of sophisticated falsificationism across more successfully. His ambiguous usage lead to such confusing formulations as "Marxism [as the core of a series of theories or of a 'research program'] is irrefutably" and, at the same time, "Marxism [as a particular conjunction of this core and some specified auxiliary hypotheses, initial conditions and a ceteris paribus clause] has been refuted." (Cf. Popper .)
Of course, there is nothing wrong in saying that an isolated, single theory is "scientific" if it represents an advance on its predecessor, as long as one clearly realizes that in this formulation we appraise the theory as the outcome ofand in the context ofa certain historical development.
"In most cases we have, before falsifying a hypothesis, another one up our sleeves" (Popper [1959a], p. 87, footnote 1). But, as our argument shows, we must have one. Or, as Feyerabend put it: "The best criticism is provided by those theories which can replace the rivals they have removed" (, p. 227). He notes that in some cases "alternatives will be quite indispensable for the purpose of refutation" (ibid. p. 254). But according to our argument refutation without an alternative shows nothing but the property of our imagination in providing a rescue hypothesis.
In the distorting mirror of naive falsificationism, new theories which replace old refuted ones, are themselves born unrefuted. Therefore they do not believe that there is a relevant difference between anomalies and crucial counterevidence. For them, anomaly is a dishonest euphemism for counterevidence. But in actual history new theories are born refuted: they inherit many anomalies of the old theory. Moreover, frequently it is only the new theory which dramatically predicts that fact which will function as crucial counterevidence against its predecessor, while the "old" anomalies may well stay on as "new" anomalies.
( Imre Lakatos, Criticism and the Growth of Knowledge, New York: Cambridge University Press, 1970, pp. 91-195; from Theodore Schick, Jr., ed., Readings in the Philosophy of Science, Mountain View, CA: Mayfield Publishing Company, 2000, pp. 20-23. )
Home Page | Further Reading | Site Map | Send Feedback