Historical Approaches: Kuhn, Lakatos, and Feyerabend PDF

CHAPTER 6 HISTORICAL APPROACHES:...

CHAPTER 6 HISTORICAL APPROACHES: KUHN, LAKATOS AND FEYERABEND Martin Carrier All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted under U.S. or applicable copyright law. Historical approaches dominated philosophy of science as a whole between approximately 1965 and 1985. Thomas Kuhn’s Structure of Scientific Revolutions appeared in 1962 and was first broadly discussed as a challenge to Karl Popper’s falsificationist methodology at a confer- ence held in London in 1965. The ensuing controversy between Kuhn and Popper and the later contributions from Imre Lakatos, Paul Feyerabend and others did much to fuel the debate and lifted theory change to the top of the agenda of philosophy of science. Two characteristics are essential for the historical approaches in question: first, the emphasis on the practice of science in its historical development (accompanied by a more detailed consideration of case studies from the history of science) and, second, the focus on theory evaluation and, in particular, theory comparison. Historical approaches chiefly aimed to uncover the reasons for theory change or the nature of scientific progress. Theory evolution and theory evaluation were the focal points of the historical approaches under consideration. The background: cumulative history and falsificationism The controversy about historical approaches unfolded against a back- Copyright 2012. Continuum. drop that had shaped philosophical thoughts on scientific progress for EBSCO Publishing : eBook Collection (EBSCOhost) - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY AN: 842508 ; James Robert Brown.; Philosophy of Science: The Key Thinkers Account: towson.main.eds  Historical Approaches 133 quite some time. This cumulative view of progress suggests that theory change amounts to piling up truths (mostly about laws of nature). That is, what science has once recognized as true remains part of the system of knowledge forever. Science needs to retract theoretical claims to be sure, but such withdrawal is confined to illustrations of the laws or to intuitive models, for one, and to restrictions of the scope of laws, for another. The first item is highlighted by Moritz Schlick, among others. He claimed that no theory that has been verified will ever be jettisoned in its entirety. Rather, its key elements will be preserved by its succes- sor theory. Only two sorts of changes are typically introduced by the successor: first, adding new details that serve to increase the match with the data and, second, removing misleading illustrations that had been introduced hastily so as to facilitate understanding and use of the theory (Schlick 1938, 203). Consequently, what has been established by scientific research is essentially left unaffected by subsequent scientific progress. However, the domain of application of the laws of nature, as accepted by science, may turn out to be smaller than anticipated at their discovery. This second item was underscored by Pierre Duhem. Duhem argued that knowledge of the laws of physics is always tentative and incomplete and that, correspondingly, exceptions occur in experience. As a result, the pertinent law needs to be restricted to particular cir- cumstances. Such restrictions in scope are increasingly recognized in the course of scientific progress (Duhem 1906, 141–4). For instance, special relativity and quantum theory have made us realize that the domain of Newtonian mechanics is bound to small velocities (as compared to the speed of light) and great lengths (as compared to the atomic scale), but the Newtonian theory is nevertheless valid within this reduced realm. Underlying the cumulative view is the assumption that the scientific method is suitable for uncovering true results. Scientific tests are taken to be so challenging that all accounts that pass them can be relied upon. The tough examination process and the demanding standards of accept- ance are the reason why there are no substantial retractions in science. At the surface, a much more agitated picture of scientific progress emerges from Popperian critical rationalism or falsificationism. Popper took seriously the deep ruptures that affected fundamental physics in the first three decades of the twentieth century. In his view, the acceptance of relativity and quantum theories involved abandoning the entrenched EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  134 Philosophy of Science: The Key Thinkers system of classical physics. Such episodes show that even venerable theoretical traditions may turn out to be ill- conceived. There is more to scientific progress than revealing unexpected limits of scope. In contrast to the impoverished standard image of falsification- ism, Popper was well aware of the flexibility that theoretical systems contain for coping with recalcitrant data. He distinguishes between a logical and a pragmatic sense of falsification. Falsification in the logical sense refers to the contradiction within a set of theoretical claims and observation statements. Falsifiability in this logical sense is a property of a theoretical system as a whole; it says that if certain observation statements were adopted, the theoretical system would be untenable (Popper 1935, 52–3). However, Popper also acknowledges that ‘conven- tionalist ploys’, later called ‘immunization strategies’, could be invoked so as to avoid the refutation of particular assumptions. By appending unsupported auxiliary hypotheses or by adjusting the definitions of the pertinent concepts, specific theoretical claims could be rescued from the threat of contrary experience (Popper 1935, 15–6, 45–50). Accordingly, Popper remains in broad agreement with a celebrated thesis of Duhem to the effect that the bearing of an anomaly on a particular part of a network of hypotheses cannot be assessed by rely- ing on logic and experience alone. Duhem’s thesis says that different theoretical hypotheses can be held responsible for a given empirical problem, and different measures can be taken to fix the flaw. Additional considerations, subsumed by Duhem under the heading of ‘bon sens’, guide scientists in picking a culprit.1 Popper acknowledges that in many cases theoretical principles can be neither proved nor disproved by experience. Seemingly anomalous data leave loopholes for scientists intent on saving such a principle from prima facie contrary evidence. However, as Popper argues, this strategy of rescuing theories from refutation fails to capture an important fea- ture of theory change. The advancement of science is characterized by the occasional abandoning of theoretical accounts. The latter are not defended come hell or high water. The phlogiston theory, the caloric theory and the ether theory, all well- entrenched in their time, were given up in response to empirical criticism. They could have been preserved but instead were dropped. In Popper’s view, the underlying rationale is that continuing to adjust a troubled theory until it finally accords with recalcitrant data deprives us of the opportunity to learn from our EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  Historical Approaches 135 mistakes. We can only realize that something is fundamentally wrong if we accept the verdict of experience and treat a theory as refuted although it is not, logically speaking. Falsifiability in this pragmatic sense is an attitude of scientists toward theories in the first place, not primarily a property of theories themselves (although resoluteness is of no avail if a theory fails to entail any specific consequences). Popperian falsifiabil- ity is hence chiefly produced by a methodological convention (Popper 1935, 22–3, 26, 49; 1963, 323–4; Lakatos 1970, 23). Popper sees his methodology expressed in features of the growth of knowledge and patterns of theory change. Such patterns involve ven- turing bold conjectures and trying to refute them in serious and imagi- native attempts. Science is at its best when expectations are thwarted. As a result, Popper rejects the Baconian methodological maxim that scientists should be free of prejudices in approaching nature. On the contrary, premature anticipations are essential for scientific progress, provided that such prejudices are severely tested. The key to scientific method is not waiving prejudices but controlling them by examining their empirical consequences (Popper 1935, 223–5). Regarding theory change, Popper argues, on the one hand, that a theory replaces a refuted predecessor, which means that the two are incompatible; on the other hand, he also insists that the succes- sor approximately preserves the earlier account if the circumstances are suitably restricted (Popper 1935, 199, 221–2). This reveals that Popper rejects the cumulative view in letter but retains it in spirit. It was only Kuhn and Feyerabend who abandoned this view in word and deed. Kuhnian scientific revolutions are characterized by a succession of theo- ries that cannot be reconstructed as preservation in the limit (Kuhn 1962, 2–3, 92–8). Similarly, Feyerabend claims that scientific change does not proceed by embedding a theory in a more comprehensive one but by a complete replacement of the predecessor (Feyerabend 1962, 44–5). As a result, scientific progress does not merely extend the system of knowl- edge to new ground but also encompasses collapse and retreat. Kuhn’s paradigm theory Kuhn opens his Structure of Scientific Revolutions with the commit- ment to link up philosophy of science with the history of science more EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  136 Philosophy of Science: The Key Thinkers intimately than before. If history is taken seriously it can ‘produce a decisive transformation in the image of science by which we are now possessed’ (Kuhn 1962, 1). Although Popper was not Kuhn’s primary target of his reproach of historical superficiality, the ensuing Popper- Kuhn controversy served to make the relevant fault lines salient. Kuhn introduced the distinction between a comprehensive theoretical framework and its more specific articulation. He labelled the framework a ‘paradigm’ and defined it to include theoretical principles, method- ological or metaphysical commitments and a collection of exemplary solutions to problems (whence derives the designation ‘paradigm’). For example, the paradigm of nineteenth- century wave optics proceeded from the assumption that light is to be conceived as transverse oscilla- tions of a pervasive medium. Specific versions of the paradigm consisted in more concrete accounts of optical phenomena such as refraction, diffraction and interference. A scientific discipline that is dominated by one particular paradigm has entered the stage of ‘normal science.’ The shared commitment to an overarching framework makes it unnecessary for scientists to enter in-principle debates about the appropriateness of certain approaches and thus allows them to focus on more productive, technical work.2 In normal science, a paradigm has reached a monopoly and rules unquestioned. Its claims are not subjected to empirical examination. They are compared with experience, to be sure, but if the two don’t match, the anomaly is not viewed as a shortcoming of the paradigm. The blame is rather attributed either to additional unrecognized influences or to a lack of creativity and skill on the part of the scientists. As a result, paradigms are immune to empirical counterinstances in normal science (Kuhn 1962, 77–80; 1970a, 6–7; Hoyningen-Huene 1989, 218–19). Kuhn’s picture of normal science stands in stark contrast with the Popperian maxim of testing theories severely. Popper’s methodological advice to scientists is to take pains to actively produce counterinstances to a theory and to take them seriously in any event as potential refuta- tions of the theory. By contrast, Kuhn’s normal scientists are licensed to shelve unsolved problems and to go ahead undauntedly. The dif- ference between Popper and Kuhn is not to be understood as the gap between lofty normative principles and a sloppy practice. Indeed, Kuhn gives epistemological reasons for the lenience toward anomalies he assumes to be characteristic of normal science. His claim is that every EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  Historical Approaches 137 theory is fraught with anomalies, so that taking each such anomaly as a potential refutation would put an end to science (Kuhn 1962, 79–82). By contrast, the immunity paradigms enjoy in normal science provides a basis for the tenacious pursuit of theories which is in turn a necessary precondition for overcoming enduring challenges. In striking opposition to Popper, Kuhn considers it a characteristic of mature science that the corresponding scientific community abandons the critical discussion of the fundamentals of a discipline (Kuhn 1970a, 6). Mature science is dis- tinguished by waiving Popperian severe tests. However, when anomalies continue to pile up, confidence in the paradigmatic principles is weakened and finally lost. A ‘crisis’ emerges in whose course alternative theoretical options are considered and pursued. Such a crisis frequently gives rise to a ‘paradigm shift’ that is characteristic of a ‘scientific revolution’. One of Kuhn’s pivotal historical claims is that a theory is never dropped without a promising successor candidate. Abandoning a paradigm is tantamount to adopting a new one (Kuhn 1962, 77, 145–7). In contrast to the smooth development of normal science, Kuhnian revolutions amount to the wholesale replace- ment of a conceptual framework; they are non- cumulative in that they involve taking back achievements that were accepted as part of the system of knowledge before.3 The non- cumulative character of scientific revolutions becomes con- spicuous in two features, namely, in changes of problems considered relevant and in the emergence of so- called Kuhn-losses.4 The change of problems is unsurprising at first sight. After all, it conforms well to the traditional picture of scientific progress that problems are solved and new problems are thereby generated. Kuhn admits that problem changes of this kind appear in a revolution, but he stresses that an additional pattern of ‘problem dissolution’ emerges. This occurs when a problem is dismissed as ill-posed and misleading (Kuhn 1962, 103). For instance, one of the challenges of optical theory in the latter part of the nineteenth century was to determine the mechanical properties of the ether such that the known laws of light propagation would follow. After the acceptance of special relativity, the ether was dropped. Accordingly, the question as to the mechanical properties of the ether was not answered, but rather rejected as bereft of scientific significance. Second, scientific revolutions frequently involve what is now called Kuhn-losses. A new paradigm may be accepted in spite of the fact that EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  138 Philosophy of Science: The Key Thinkers some of the former explanatory achievements are thereby lost. More specifically, some of the empirical problems solved before are reopened again. In this way, anomalies are produced by the paradigm shift itself. To be sure, Kuhn-losses are accepted by the scientific community only in small measure, but their mere existence vitiates any claim to the effect that the new paradigm reproduces all the explanatory achievements of its predecessor. One of Kuhn’s favourite examples is taken from the Chemical Revolution. In the framework of the so- called phlogiston the- ory, a metal was regarded as a compound of a specific component (the ‘calx’) and phlogiston. Since phlogiston was assumed to be present in all metals, the theory could explain why they resembled one another to a much greater extent than the corresponding calces (oxides in modern terminology). Lavoisier’s oxygen theory, by contrast, considered metals to be elementary and thus could not account for their similarity. The adoption of the oxygen theory made an empirical problem reappear that was considered settled before (Kuhn 1962, 132, 157, 170; 1970a, 20; 1977, 323). As a result of these and other discrepancies, the old and the new paradigm are separated by a yawning chasm. Revolutions are non- cumulative; they involve a fundamental transformation in which the earlier account is not even approximately preserved by its successor. The contrast between theories separated by a revolution is far-reaching and unbridgeable (Kuhn 1962, 5–6, 97–110). At first glance, Kuhnian crises and revolutionary ruptures fit well with the Popperian image of science; after all, in these periods, theories are empirically tested and evaluated. In this vein, the Popperian view was advertised as the methodology for innovative science (Watkins 1970, 32, 36–7). Kuhn demurred and claimed rather that even scientific revolutions do not involve falsifica- tions. No theory is rejected for the reason alone that it contradicts the evidence; instead, two or more theoretical options compete with one another, and the rejection of one is tantamount to the adoption of another one (Kuhn 1962, 145–7). In sum, the most important historical claims entertained by Kuhn are that periods of normal science or paradigm monopoly are followed by revolutions or periods of theoretical pluralism. In the former periods, no significant theoretical innovation occurs; alternatives are only con- sidered during the latter. A revolution involves the substitution, not the mere abandonment, of a paradigm. And the new paradigm cannot be EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  Historical Approaches 139 reconstructed as containing the claims of its predecessor ‘in the limit’; revolutions proceed in a thoroughly non-cumulative fashion.5 Lakatos’s methodology of scientific research programmes Lakatos attempted to save Popper’s general approach by adjusting it in light of what Lakatos took to be Kuhn’s insights into patterns of theory change, for one, and Kuhn’s assaults on the rationality of science, for another. Lakatos’s aim was to show that theory choice can be recon- structed as a process governed by methodological rules and that, corre- spondingly, theory change is much more rational and much less subject to merely subjective factors than Kuhn assumed. Lakatos’s basic unit of scientific progress is the ‘research program’, that is, a series of theories that share a ‘hard core’ and a ‘positive heuristic’. The hard core contains the fundamental postulates of the program; it is retained throughout the program’s period of active pur- suit. The positive heuristic encompasses guidelines for articulating the program; it singles out significant problems and offers tools to their solution. One of Lakatos’s historical assertions is that the development of an excellent program follows its positive heuristic and does not merely respond to conceptual and empirical difficulties (Lakatos 1970, 47–52, 68–9). The hard- core postulates are irrefutable within the pertinent pro- gram; dropping them means abandoning the program. However, their irrefutability is not based on the firm ground on which they rest but rather on the commitment of the program’s proponents to maintain them. Instead, anomalies are accommodated by adapting a ‘protective belt’ of auxiliary hypotheses. This protective belt is expanded, modified, restructured or replaced entirely according to the empirical needs set by the anomalies. Duhem’s thesis (see sect. 1) entails that an empiri- cal problem never bears directly on a specific theoretical hypothesis. Accordingly, there is always room left for the scientists to adhere to a hypothesis by redirecting the refuting force of an anomaly to assump- tions from the protective belt, where required adjustments are made (Lakatos 1970, 48–50). EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  140 Philosophy of Science: The Key Thinkers Modifications within the protective belt may concern, first, observa- tion theories, second, initial and boundary conditions and, third, addi- tional assumptions within the respective theory itself. Take the solar neutrino problem that plagued nuclear physicists and theoretical astron- omers during the last decades of the twentieth century. The pertinent hard core is a theoretical model of the sun that contains accounts of the relevant fusion processes, the elements produced, the energy converted and so on. This model also predicts that a certain amount of neutrinos is emitted by the sun. The flux of solar neutrinos that reach the earth can be measured, and the result was a third of the expected value. One could argue, first, that the procedures for detecting neutrinos are less efficient than assumed. This line entails that the right amount of neutrinos is emitted and that the fault lies with the detection process or the pertinent observation theories. Second, modifications at the level of initial and boundary conditions could be made by assuming a lower temperature in the interior parts of the sun. In such a case, a reduced amount of neutrinos would be emitted. A third option is to assume that the different types of neutrinos can be converted into one another. The measuring procedures were only able to detect electron neutrinos, that is, one of the three relevant types, so that a sort of detection leakage would be responsible for the missing neutrinos. The issue was resolved in 2001 in favour of the third option of neutrino oscillations. Until that point scientists had ample room for shifting the blame for the anomaly across the protective belt. The hard core and the positive heuristic constitute the program’s identity. A specific version of a program encompasses these invariant ele- ments together with the changing protective belt. A research program unfolds as a chain of subsequent versions, each of which arises from its predecessor by some theoretical modification, usually performed within the protective belt. Lakatos appeals to methodological standards to distinguish between acceptable and inappropriate theoretical changes. These standards primarily stipulate which theoretical changes within a program are methodologically sound.6 An acceptable successor version within a program is required, first, to remain in agreement with the positive heuristic of the program, second, to account for all those phe- nomena that are successfully explained by its predecessor – albeit possi- bly in a different fashion – and, third, to successfully predict some novel, hitherto unexpected facts. Lakatos demands the reproduction of the EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  Historical Approaches 141 previous empirical achievements and the anticipation of observational regularities that were unknown to science before. The successful predic- tion of novel facts is the chief display of scientific quality. Rival research programs are to be evaluated analogously. Programs that satisfy these program-internal standards are compared by applying the conditions of reproduction and anticipation to the competitors. A superior pro- gram has to reproduce the explanatory successes of the rival program and predict effects unexpected within the framework of the latter. The empirical confirmation of theoretical anticipations is the stuff of which methodological superiority and scientific progress is made (Lakatos 1970, 33–6, 47–52, 68–9; see also Popper 1963, 326–9, 334–5; Carrier 2002, 59–63; 2008a, 279). The demand to reproduce the explanatory achievements of another theory (a preceding version or an alternative program) does not amount to the requirement to reproduce the explanations themselves. Retention of the principles and the modes of derivation is not necessary; the rel- evant phenomena may well be accounted for in a disparate fashion. For this reason Lakatos’s condition of progress is compatible with non- cumulative change in the Kuhnian sense of a complete replacement, rather than approximate retention, of a theory by another one. Lakatos is chiefly concerned with the methodological explanation of theory choice. Kuhn had argued that during revolutionary periods the competitors advance different standards for judging the appro- priateness of problem solutions. As a result, each competitor appreci- ates its own assets and its rivals’ liabilities drawing on its own specific measures of adequacy. Naturally enough, the partisans of contrasting approaches fail to convince one another.7 One of Kuhn’s examples is again taken from the Chemical Revolution. Within the phlogistic frame- work, it was considered the chief task of chemical theory to account for the properties of chemical substances (such as hardness, combus- tibility, volatility and the like) along with their changes during chemi- cal reactions. Consequently, chemical explanations are to be judged according to their ability to provide such an account. As a result of the switch to the oxygen theory, these problems moved into the back- ground, whereas the challenge to accommodate reaction weights was considered pivotal. Consequently, the standards for judging the achievements of chemical theories were altered as well (Kuhn 1962, 107; 1977, 335–6).8 EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  142 Philosophy of Science: The Key Thinkers Lakatos shifts Kuhn’s theory-specific standards into the positive heu- ristic. Since the heuristic determines which kinds of theoretical means and procedures are acceptable within a program or which theoretical aims are to be followed, theory-specific standards have a legitimate place there. In any event, the dependence of such standards on and their variation with particular theories square well with the heuristic nature of the standards. However, in contrast to Kuhn, no theory- specific standard is granted any influence on the comparative evalu- ation of programs (Lakatos 1970, 68). After all, compliance with the positive heuristic is only required within a program, and the judgement of entire programs relies exclusively on overarching, or ‘transparadig- matic,’ criteria. Lakatos’s methodological criteria specify requirements for the facts a theory can rightly count in its favour. His criteria operate by singling out facts explained in a particularly demanding manner, namely, by success- fully predicting a novel fact at the same time. The appraisal of a theory is based exclusively on these successful predictions, not on all of its correct empirical consequences. If consideration is restricted to such a selected group of outstanding explanations, it is much easier to reach an agree- ment on how well the theory is doing empirically (Lakatos 1970, 36; 1971, 112). Lakatos’s methodology was formulated as a response to Kuhn’s, and it leaps to the eye that some of Lakatos’s concepts are modelled on Kuhn’s notions (Kuhn 1970b, 256). A research program roughly cor- responds to a paradigm, and a program change resembles a scientific revolution. The retention of the hard core and the positive heuristic in pursuing a program reproduces the continuity of normal science. Finally, the methodological license to give up accepted explanations and to accommodate the relevant phenomena in a different way captures the non- cumulative character of revolutions. However, Lakatos’s methodology does not simply reproduce Kuhn’s notions in a different conceptual framework. It aims rather at meeting methodological challenges raised by the paradigm theory. A pertinent issue is Kuhn’s claim about the immunity of paradigms to anomalies. This immunity looks like an irrational trait from a Popperian angle. Its analogue for research programs can be derived as follows. Supporting facts are constituted by correctly predicted empirical regularities. Such facts have not been accounted for by the competing program; they EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  Historical Approaches 143 would otherwise not be novel. Thus, Lakatos’s condition of progress suggests that only those facts which cannot be explained by its rival are suitable for buttressing a program. Conversely, only those facts militate against a program that favours its competitor. It follows that only those anomalies which can be solved by the rival in a qualified fashion (i.e., by predicting unknown phenomena at the same time) will count as failures (Lakatos 1970, 37). This entails that the mere inability to accommodate an observation does not bring a program into trouble (Lakatos 1970, 92). This conclusion agrees with the Kuhnian immunity claim so that the latter follows from Lakatos’s requirements for suc- cessful explanations. Research programs are rightly immune to mere anomalies (Carrier 2002, 63–5). Another historical pattern stressed by Kuhn is that paradigms are never dropped unless a potential successor is waiting in the wings. Scientific revolutions proceed by theory substitutions. Its Lakatosian analogue is No program abandonment without program replacement. A program is only discredited methodologically if a superior competitor is available. This condition can be derived from a corollary to the afore- mentioned immunity argument. This argument says that the liabilities of one theory are the assets of the other. There are no significant failures without an alternative solution. And obviously enough, if a theory is not in trouble, it should not be given up. It follows that a program can never be rated as deficient unless there is a contender attacking it with some success. Disqualification of a program is brought about by a cor- roborated rival (Lakatos 1970, 35; 1971, 112–13). Kuhn levels his claims of paradigm immunity and paradigm substi- tution as descriptional objections to Popper’s methodological require- ments. Within the framework of Lakatos’s methodology, by contrast, the two features of immunity and substitution constitute theorems rather than objections. They follow from Lakatos’s conception of how theories are to be evaluated. Accordingly, if theories develop in the Kuhnian way, no conflict with methodological rules arises. Instead, Lakatos’s concep- tion provides a methodological explanation of these patterns of theory change (Carrier 2002, 65). However, Lakatos’s portrayal of scientific change does not simply reproduce Kuhn’s picture but deviates from the latter in some respects. For instance, Lakatos’s methodology entails that science exhibits a thor- ough theoretical pluralism rather than a monopolistic rule of paradigms EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  144 Philosophy of Science: The Key Thinkers punctuated by occasional cataclysms. Further, Lakatos’s methodology rules out Kuhn losses since Lakatos demands that theories are not super- seded until all the phenomena they had explained are also accounted for by the successor (see above). Regarding pluralism, the verdict of the history of science points rather in favour of Lakatos, while regarding Kuhn losses, it rather agrees with Kuhn. Feyerabend’s methodological anarchism Paul Feyerabend’s position was formulated mostly in response to Lakatos, who Feyerabend took to be the most sophisticated rationalist in matters of methodology and whose position Feyerabend intended to contrast with his relativistic stance.9 Regarding science, Feyerabend’s ‘methodological anarchism’ entails that there is no way to discern epistemic achievements of different theories. No account is justifiably superior to any other. The only defensible principle regarding theory assessment is ‘anything goes’ (Feyerabend 1975a, 27–8). Theories are to be welcomed as long as a plurality of them compete with each other and none gains a dominant position (Feyerabend 1975b, 365). Pluralism is threatened if cognitive positions are given up so that the spectrum of approaches is narrowed. Consequently, Feyerabend’s objective is to show that rejecting such approaches is never justified. I focus on two of Feyerabend’s arguments regarding science. It is unwar- ranted to exclude a theory from further consideration on the ground of, first, external incoherence or its contradiction with other accepted theo- ries and, second, discrepancies between theory and evidence. External coherence demands that different theories should be in agreement with one another and thus give rise to a unified picture of a realm of facts. However, as Feyerabend claims, there is no good reason to drop a the- ory because it contradicts other, entrenched accounts. The argument is epistemological and says that error is discovered by comparison, that is, by using a contrasting approach, not by analysis, that is, by confin- ing scrutiny to a given approach. Weak spots are more easily revealed by bringing to bear a discrepant framework, whereas internal criticism always accepts some of the premises at hand. As a result, seemingly inferior ideas should be retained as a challenge for the dominant view. Deficient accounts should not be given up but rather be strengthened. EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  Historical Approaches 145 A theory grows and matures by mastering such challenges (Feyerabend 1975a, 29–32; 1975b, 361–2). All advocates of historical approaches support pluralism (although, perhaps, restricted to particular stages of theory development), but most regard it as a heuristic tool for promoting the growth of knowl- edge. Pluralism is seen as a temporary feature that is apt to produce a more sophisticated consensus than would have been accessible by pursuing just one approach. Distributing epistemic resources across a variety of pathways is the best strategy for arriving at the optimum solu- tion. Feyerabend dissents and requires pluralism to be permanent. The dominance of any theory is pernicious to science – even if the theory should be true. If the truth need not face contenders, the reasons for its superiority increasingly fall into oblivion. The truth is of no avail if it is mindlessly reproduced and eventually no longer understood. Pluralism is not a means for propelling the growth of knowledge; it rather rep- resents the ideal state of knowledge. As a result, external incoherence is a methodological asset rather than a liability (Feyerabend 1975a, 47, 52; 1975b, 365). A pivotal standard for judging theories is empirical adequacy. If the facts militate against a theory, it deserves to be given up. However, as Feyerabend puts forward, the facts may be dependent on theory. What he has in mind here is the theory-ladenness of observations: measuring procedures serve to link up indicator phenomena with theoretical states, and these links are established and accounted for by theory. However, if such observation theories are mistaken, a systematic observational error may arise (Carrier 1994, 10–19; 2008b, 69–77). Anomalies may be produced by erroneous observation theories. Again, as Feyerabend emphasizes, such mistakes are best disclosed by approaching the matter from a variety of viewpoints, in this case by employing different obser- vation theories. Feyerabend’s favourite example concerns Galileo and the motion of the earth. Within the framework of the then-accepted impetus phys- ics, the motion of the earth should have had observable effects on the terrestrial motion of bodies. The absence of such effects seem- ingly ruled out any diurnal and annual motion of the earth. However, Galileo adopted a different stance. He recognized that the apparent counterevidence was produced by appeal to the impetus principle: that all motion comes to rest unless maintained by a force. The phenomena EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  146 Philosophy of Science: The Key Thinkers were interpreted by using this principle as observation theory. Yet if the principle of inertia was invoked for this purpose, the anomalous evidence disappeared. This example shows that recalcitrant data are no good reason for dropping a theory. Rather, the trouble may be due to a mistaken observation theory (Feyerabend 1970, 204–5; 1975a, 64–7, 74, 98). Feyerabend subsumes the lesson to be drawn from these considera- tions under two principles. The principle of tenacity licenses continued adherence to a theory in spite of known difficulties and anomalies. The principle of proliferation recommends addressing such problems by elaborating alternative theories. Feyerabend acknowledges that both principles are part of Kuhn’s paradigm theory. However, Kuhn restricts the appropriateness of each principle to particular stages of the development of science. Tenacity matches the complacency of normal science; proliferation is the order of the day when winds of change are blowing. In contradistinction, Feyerabend would have it that the two principles represent different aspects of scientific method and are employed jointly (Feyerabend 1970, 203–8, 211). Feyerabend’s criticism is mostly directed at a ruthless falsifica- tionism that is, moreover, blind to the room left for fixing failures of a theory (that is expounded by Duhem’s thesis). No proponent of the theory- change tradition, and no philosopher of science in the twentieth century at that, has ever advocated such a position. The conclusion is, accordingly, that Feyerabend overstates his case. It is true, theories can never be judged unambiguously, but they may pass or fail demanding empirical tests and thus exhibit strongly unequal degrees of epistemic virtues like explanatory and predictive power. Feyerabend’s considerations fail to convincingly support the relativis- tic stance he adopts. Essentials of historical approaches Although the differences in detail among the various advocates of his- torical approaches to methodology are conspicuous, they also exhibit a large common ground. First, they share the appreciation of historical detail. The idea is that placing science in its actual complexity at the EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  Historical Approaches 147 focus of analysis is apt to give rise to a richer and more adequate account of methodology than the preceding practice of alluding superficially to schematic examples. Second, methodology is supposed to elucidate the process of comparative evaluation of theories and the subsequent selec- tion of one account at the expense of alternatives. The way of choosing between theories in a certain problem-situation is the key to methodol- ogy and to the rationality of science as a whole (Popper 1963, 335). The adoption of this framework leads to an emphasis on theory change and to the acceptance of pluralism within science. This approach goes along with abandoning the cumulative view of history. A prima facie distinction among the leading advocates concerns the bearing of methodological rules and criteria. On the face of it, Popper and Lakatos opt for the rule of law regarding the assessment of the epis- temic achievements of theories, whereas Kuhn and Feyerabend want to leave this assessment to the discretion of individual scientists.10 However, the boundaries between the two camps become fuzzy on closer scru- tiny. Popper emphasizes that judgement is necessary for resolving issues of refutation (Popper 1935, 16; 1963, 64–5), and Lakatos stresses that his tough standards are more appropriate for rationally reconstructing theory change than for directing it (Lakatos 1971, 117, 131–4). When it comes to the nitty-gritty, all parties to the dispute agree that the import of general methodological rules on the development of science is limited. Another point of contact among the proponents of the theory- change tradition is the adoption of a pronounced normative stance. Even Kuhn, who insisted on the descriptive adequacy of methodologi- cal accounts, also claimed that the pattern of theory change he sug- gested is epistemically appropriate. That is, research pursued along Kuhn’s lines stands a good chance of contributing to the growth of knowledge (Kuhn 1962, 205–6; 1970b, 237). Popper’s normative approach is made most conspicuous by his condemnation of Kuhn’s normal scientist as a deplorable figure. Normal scientists exist, but for the betterment of science they ought not (Popper 1970, 53). Lakatos distances himself from what he considers Kuhn’s merely psychological account of theory change and insists on keeping epistemological ele- ments in the methodology (Lakatos 1970, 90–1). Finally, Feyerabend opposes Kuhn’s account because he takes normal science to be a dull and complacent enterprise that even exhibits anti-humanitarian EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  148 Philosophy of Science: The Key Thinkers tendencies (Feyerabend 1970, 210). The bottom line is that for all posi- tions under review here, theory change is characterized with reference to the epistemic aims of science; science contributes first and foremost to the growth of knowledge. It is not alone the driving forces of theory change that are intended to be illuminated, but the good reasons that govern the process. Finally, the mainstream of the theory- change tradition attempted to steer a middle course between realism and relativism.11 Epistemic progress in science becomes manifest in objective achievements such as broad scope of theories and the increased precision of explanations based on them. But there are no convincing reasons for believing that science eventually will manage to represent the blueprint of the universe (Kuhn 1962, 205–6; Lakatos 1970, 99–101; Laudan 1984). Underlying this instrumentalist restraint is the commitment to the underdetermina- tion of theories by evidence and the emphasis on the role of imagina- tion in science. The facts always leave room for different theoretical accounts, and the pathways of science are paved by human creativity rather than dictates of nature. Science is and remains a human creation. The more recent debates about methodological matters tend to nar- row the issue to the dichotomy of realism and relativism. It might help to recognize that the historical approaches had elaborated a position in between, namely, the notion of objective but non-representational epistemic merit. The hallmark of the historical approaches is the connection estab- lished among theory choice, scientific progress and the rationality of science. Theory choice involves the adoption of a cognitively superior account whose credentials are specified or made explicit by methodo- logical approaches. These criteria refer to the relations between the the- oretical claims entertained and the observations made. The sequence of the theories adopted by the scientific community constitutes scientific progress, and the methodological criteria suggested are judged by their ability to capture essential elements of this development. The ability to reconstruct theory change as evolution in accordance with meth- odological judgement buttresses the claim that theory change proceeds rationally. Certain patterns of change constitute scientific rationality and scientific progress. Historical approaches set a counterpoint to abstract reasoning in matters of confirmation that cranks concrete evaluations of theories out EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  Historical Approaches 149 of overarching goals of science, such as high probability, and becomes manifest, in particular, in all sorts of inductive logic (such as Carnap’s confirmation theory or latter-day Bayesianism). Historical approaches attempt to include to a much greater extent the variegated details of the actual practice of science as it unfolds in history. Historical approaches agree with social constructivism in taking social factors seriously, but unlike the latter, they emphasize that science advances in epistemic respects. The primary aim is the elucidation of the process of the growth of knowledge in the sense of propositions of superior cognitive quality and empirical reliability. Notes 1 Duhem 1906, 152–4, 175–6. However, Popper, disagreeing with Duhem, claims that ‘in quite a few cases’ the mistaken hypothesis can be identified as the one that was necessary for deriving the refuted prediction (Popper 1963, 324). 2 See Hoyningen-Huene 1989, 133–62 for Kuhn’s notion of ‘paradigm’. 3 Bird 2000, 30–48, gives a succint overview of Kuhn’s notions of normal and revolutionary science. 4 See Carrier 2002, 55–57 for additional relevant features. 5 For a critical discussions of Kuhn’s historical claims see Bird 2000, 49–63. 6 Lakatos oscillated between different versions of his criteria. I confine myself to what I take to be the original version. For alternative readings see Carrier 2002, 69. 7 Kuhn 1962, 109–110. Such differences in the standards of evaluation have come to be called ‘methodological incommensurability,’ in contradistinction to ‘semantic incommensurability,’ that is, non-translatability due to deep theo- retical contrasts involved. Neither Kuhn nor Feyerabend initially distinguished between the two sorts of incommensurability (see Carrier 2001). 8 Kuhn later emphasized the importance of comprehensive standards that transcend single paradigms, but claimed that they are unable to guide methodo- logical judgement unambiguously. This ‘Kuhn-underdetermination’ plays no role in the debate under consideration (Carrier 2008a, 274–8). 9 In fact, Feyerabend attempts to show that Lakatos’s criteria are vacuous and that Lakatos is as relativistic implicitly as Feyerabend is explicitly (Feyerabend 1975b, 215–19). Feyerabend’s Against Method is dedicated to ‘Imre Lakatos. Friend and fellow anarchist.’ 10 For Kuhn, the pertinent leeway exists only in revolutionary periods, but outside of such periods no theory choice is made anyway. 11 The two exceptions are Popper, the realist, and Feyerabend, the relativist. EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  150 Philosophy of Science: The Key Thinkers References Bird, Alexander (2000), Thomas Kuhn. Princeton, NJ: Princeton University Press. Carrier, Martin (1994), The Completeness of Scientific Theories. On the Derivation of Empirical Indicators Within a Theoretical Framework: The Case of Physical Geometry. Dordrecht: Kluwer. — (2001), ‘Changing Laws and Shifting Concepts: On the Nature and Impact of Incommensurability’, in P. Hoyningen- Huene and H. Sankey (eds), Incommensurability and Related Matters. Dordrecht: Kluwer, pp. 65–90. — (2002), ‘Explaining Scientific Progress. Lakatos’s Methodological Account of Kuhnian Patterns of Theory Change’, in G. Kampis, L. Kvasz and M. Stöltzner (eds), Appraising Lakatos: Mathematics, Methodology, and the Man. Dordrecht: Kluwer, pp. 53–71. — (2008a), ‘The Aim and Structure of Methodological Theory’, in L. Soler, H. Sankey and P. Hoyningen- Huene (eds), Rethinking Scientific Change and Theory Comparison: Stabilities, Ruptures, Incommensurabilities? Dordrecht: Springer, pp. 273–90. — (2008b), Wissenschaftstheorie: Zur Einführung. Hamburg: Junius. Duhem, Pierre (1906/1981), La théorie physique – son objet, sa structure. Paris: Librairie philosophique J. Vrin. Feyerabend, Paul K. (1962/1981), ‘Explanation, Reduction and Empiricism’, in Realism, Rationalism and Scientific Method (Philosophical Papers 1). Cambridge: Cambridge University Press, pp. 44–96. — (1970), ‘Consolations for the Specialist’, in Imre Lakatos and Alan Musgrave (eds), Criticism and the Growth of Knowledge. Cambridge: Cambridge University Press, 197–230. — (1975a/1987), Against Method. Outline of an Anarchistic Theory of Knowledge. London: Verso. — (1975b/2006), ‘How to Defend Society Against Science’, in E. Selinger and R. P. Crease (eds), The Philosophy of Expertise. New York: Columbia University Press, pp. 358–69. Hoyningen- Huene, Paul (1989), Die Wisenschaftsphilosophie Thomas S. Kuhns. Brunswick: Vieweg. Kuhn, Thomas S. (1962/1996), The Structure of Scientific Revolutions. Chicago: University of Chicago Press. — (1970a), ‘Logic of Discovery or Psychology of Research?’, in Imre Lakatos and Musgrave (eds), Criticism and the Growth of Knowledge. Cambridge: Cambridge University Press, pp. pp. 1–23. — (1970b), ‘Reflections on My Critics’, in Imre Lakatos and Alan Musgrave (eds), Criticism and the Growth of Knowledge. Cambridge: Cambridge University Press, pp. 231–78. EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use  Historical Approaches 151 — (1977), ‘Objectivity, Value Judgment, and Theory Choice’, in The Essential Tension. Selected Studies in Scientific Tradition and Change. Chicago: University of Chicago Press, pp. 320–39. Lakatos, Imre (1970/1978), ‘Falsification and the Methodology of Scientific Research Programmes’, in John Worrall and G. Currie (eds), Imre Lakatos. The Methodology of Scientific Research Programmes (Philosophical Papers I). Cambridge: Cambridge University Press, pp. 8–101. — (1971/1978), ‘History of Science and Its Rational Reconstruction,’ in John Worrall and G. Currie (eds), Imre Lakatos. The Methodology of Scientific Research Programmes (Philosophical Papers I). Cambridge: Cambridge University Press, pp. 102–38. Lakatos, Imre and Alan Musgrave (eds) (1970/1974), Criticism and the Growth of Knowledge, Cambridge: Cambridge University Press. Laudan, Larry (1984), ‘Explaining the Succes of Science: Beyond Epistemic Realism and Relativism’, in J. T. Cushing, C. F. Delaney and G. M. Gutting (eds), Science and Reality. Recent Work in the Philosophy of Science. Essays in Honor of Ernan McMullin. South Bend, IN.: University of Notre Dame Press, pp. 83–105. Popper, Karl R. (1935), Logik der Forschung, Tübingen: Mohr, 1976. — (1963), Conjectures and Refutations. The Growth of Scientific Knowledge. London: Routledge, 2005. — (1970), ‘Normal Science and Its Dangers’, in Imre Lakatos and Alan Musgrave (eds), Criticism and the Growth of Knowledge. Cambridge: Cambridge University Press, pp. 51–58. Schlick, Moritz (1938), ‘Form und Inhalt. Eine Einführung in philosophisches Denken’, in Philosophische Logik, ed. B. Philippi, Frankfurt: Suhrkamp, 1986, pp. 110–222. Watkins, John (1970), ‘Against Normal Science’, in Imre Lakatos and Alan Musgrave (eds), Criticism and the Growth of Knowledge, Cambridge: Cambridge Univesity Press, pp. 25–37. Worrall, John and G. Currie (eds) (1978), Imre Lakatos. The Methodology of Scientific Research Programmes (Philosophical Papers I). Cambridge: Cambridge University Press. EBSCOhost - printed on 8/1/2022 1:11 PM via TOWSON UNIVERSITY. All use subject to https://www.ebsco.com/terms-of-use