5. Rationality and theory-change
So much, then, for history of science as warrant. But this was by no means the principal message that the first readers of Kuhn’s work took away with them. They saw the book as historicist in a far more controversial way, as indeed challenging the very rationality of science itself on historicist grounds. The major theme of the book was, after all, the discontinuity of paradigm-change and the consequences thereof. Later stages of theory-development cannot properly be said to be any nearer the truth than earlier stages were. The metaphors of religious conversion and gestalt shift, the stress on incommensurability, caught immediate attention. All in all, the book seemed designed to challenge the rationality, the reason-giving, that had defined natural science since its earliest phases.
In a postscript to the second edition of Structure and in a number of essays in his later work, The Essential Tension, Kuhn labored to show that the implications of his account of paradigm-change for the rationality of that change were far less radical than his critics were assuming them to be. His disavowals were not entirely successful; his critics continued to use terms like ‘irrationalist’ or ‘relativist’, ‘subjectivist’, to describe his message. The rhetoric of the first edition of Structure evidently lingered in people’s minds. I would argue, however, that if his writings be taken as a whole, it seems fair to say in retrospect that he was trying to redefine the rationality of paradigm-change in science rather than to undermine it fatally (McMullin, 1993).
First, he always emphasized that there are “good reasons” for paradigm-change (Kuhn, 1971, 199), notably the perception of anomaly and the recognition of an alternative to the ruling paradigm. What he wanted to bring out was that these reasons are not coercive; they cannot compel assent and there is no precise point at which defense of the older paradigm becomes illogical. Second, though he emphasized that the factors governing theory-choice in science function as values to be maximized rather than as logical rules to be mechanically followed, he also maintained in his later work that these values or shared criteria carry through paradigm-change; they are, he boldly asserts, “permanent attributes of science” (Kuhn, 1977, 335), available to guide paradigm-choice. No matter how deep the incommensurability between two successive paradigm-languages may run, the values governing theory-choice are there to serve as criteria applicable to each paradigm alike. Proponents of the two paradigms, Kuhn goes on, will disagree as to the relative weight to be given to the criteria, but the fact is that they will have a stock of perfectly good, though admittedly not compelling, reasons to call in support of their own paradigms. Third, the challenge of incommensurability, which had seemed almost insurmountable in the first edition of Structure, seemed to dwindle in his later work, notably in his last treatment of that topic, in the final essay of the Festschrift in his honor, World Changes (Horwich, 1993), where he makes it clearer than he had originally done that to call two paradigms incommensurable is not to imply that they cannot be rationally compared. Since it is perfectly obvious that even in the case of the most profound paradigm-shift in the history of natural sciences, the Galilean one, Galileo himself could compare the Aristotelian paradigm of his own early work to the new one he was proposing for mechanics, this clarification on Kuhn’s part was welcome.
But if Kuhn’s views on the rationality of paradigm-change were in the end less radical than at first they seemed to be, he never retreated from the more radical claim that the history of science shows that science cannot be construed as converging on truth about the natural world; the discontinuities in the historical record are compatible with a modified notion of scientific rationality but not (he claimed) with the traditional view of science as progressing towards an ever-deeper grasp of real structure. There was no progression in that ontological sense from Aristotle to Newton to Einstein, only an improvement in problem-solving, in the accuracy and scope of prediction. In a famous sentence, he declared: “The notion of a match between the ontology of a theory and its ‘real’ counterpart in nature now seems to me illusive in principle” (Kuhn, 1971, 206). Though he made it clear that his motivation for this strongly anti-realist view of theory came in the first instance from the philosophy of language, he claimed that it was confirmed by his reading of history. He allowed that this would seem a relativist position to many, but insisted that this was as far as the history of science allowed one to go.
I will return in a moment to this challenge to the traditional realism of the sciences and the influence it has exerted in the intervening years. But first, a further possible intrusion of historicity into the one-time atemporal logic of science must be noted.
6. Historicity and rationality
Kuhn, as we just saw, responded to the charge of irrationalism lodged against him by claiming that he was perfectly prepared to allow on historical grounds that paradigm-change and, more broadly, theory-assessment in general are shaped by criteria that remain relatively unchanged over time. Later writers have tended to question this restriction of the historicist thesis, arguing that the criteria of theory-assessment, as for the acceptance of an observation-claim, have themselves changed substantially over the years. Shapere and Laudan, for example, have developed somewhat different accounts of how and why such change has taken place (Shapere, 1984; Laudan, 1984a). Like Kuhn himself, they are anxious to maintain the overall rationality of theory-assessment in science, and believe that this can be done by providing a second-order account of how the criteria themselves might alter, on grounds that can themselves be described without circularity as rational.
The key to this is to show, on historical or on logical grounds, how the chosen criteria are fitted to serve the long-range goals of science. Of course, this raises the further historicist query: might not the goals themselves alter, and would this change still count as rational? Laudan developed what he called a “reticulated” model of scientific inquiry, involving aims, methods for achieving these aims, and at the bottom level, factual claims. Each of these three elements, he argued, calibrates the others, allowing any one of them to change but keeping the whole in a sort of meta-rational balance. It seems clear that the goals of science have, in fact, changed over the course of history: Aristotle’s goal of demonstration, worthy though it was, proved impracticable, as did the seventeenth-century goal of universal mechanical explanation as that goal was then understood. The challenge, then, is to reconcile this story of maturing goals with the claim to a complex form of rationality, in which accuracy of prediction is likely to serve a key role at the meta-level (McMullin, 1989, Introduction).
I have spent a good deal of time at what I think to be the most characteristic emphasis in the philosophy of science of the last half-century, namely, the emphasis on historicity in one form or another. But other themes impose themselves too. In the logicist picture of science, the individual subject was almost invisible; the stress was on rule, universally applicable. Already among the positivists themselves, as we saw, there was a growing realization that human decision entered in at many of the key moments in scientific inquiry. When and how was an observation to be certified as basic? Which of two fairly effective theories was to be preferred? There was room for legitimate disagreement in response to questions like these. Popper and Carnap used the slippery term ‘convention’; others spoke of “decisional” elements. Of course, the notion that scientific claims are routinely underdetermined by the data offered in their support was in no sense new. It was a commonplace already in the seventeenth century, though somewhat obscured in the later Newtonian era.
We have already seen how Kuhn urged that the criteria involved in theory-assessment ought to be regarded as values to be maximized rather than as rules to be followed. Different scientists will understand values like coherence, simplicity, and so forth, differently and also weight their importance differently. In this way, the prevalence of controversy at the growing points of science becomes intelligible, a feature of science that resisted explanation in the hard-and-fast terms favored by strict positivism (Engelhardt and Caplan, 1987). One of the liveliest issues in post-positivist philosophy of science has been the propriety of allowing metaphysical considerations to affect theory-assessment. It is clear that they do, in fact, do so, sometimes in dramatic ways, as, for example, in the famous disagreement between Bohr and Einstein regarding the merits of the new quantum theory. But should they do so? Metaphysics, after all, is not outside the range of rational argument; it ought not reduce to a matter of taste or the contingency of personal background, as its positivist critics charged it did. There are, of course, difficult issues here, issues that are in no way new but that present themselves here in a new guise. It seems fair to say that they are not likely to be resolved soon!
8. The realism debate
The thrust of philosophy of science in recent decades has been, as we have seen, to call in question many of the idealizations on which the logical-empiricist program depended. Restoring the historical dimension to science, recognizing how far from simple rule are many of the decisions that scientists must make in the course of their work, led from logicism to what has at times amounted almost to an anti-logicism, seen in its most extreme form in Feyerabend’s later writings. No one has been inclined to deny the progressive character of science if it be regarded merely as an instrument of prediction, a set of problem-solving devices. It would surely be difficult in our technology-driven society, even for a philosopher, to query the fast-accelerating power of the natural sciences to mold the earth to human desires. Where question arises is in regard to ontology, to the ability of the sciences to reveal the underlying structures of the natural world. That the sciences can do this would seem beyond question to most people, scientists and non-scientists alike. How, then, explain the rise of anti-realism among philosophers of science in the ‘sixties and ‘seventies? Was it just the professional perversity of philosophers showing itself, their irrepressible inclination to call in question the obvious? Or was it the sort of consequence one might have anticipated from the decline of positivism?
A preliminary comment: Anti-realism, in its instrumentalist form, has been a recurrent feature of the history of science. It has usually been local, that is, affecting only a particular science or set of sciences, rather than global, that is, affecting theoretical natural science generally. Thus, an instrumentalist construal of mathematical astronomy was more or less standard in medieval and Renaissance times; a similar approach to atomic physics was not uncommon in the nineteenth century. There were specific reasons in each case for this sort of reservation in regard to a particular domain of science. Their empiricist account of meaning led the logical positivists, especially in their earlier stricter period, to be wary about theoretical language and to question, in consequence, the ontologies that go with it. Thus, there was an anti-realist cast to their thinking, this time of a global sort, that regarded the explanatory success of physical theory as inadequate testimony of the existence of the entities postulated by the theory. In an influential essay “The theoretician’s dilemma”, Hempel called this ontological skepticism in question and went on himself later to defend a strong form of realism (Hempel, 1958; 1966).
But Kuhn, as we have seen, defended anti-realism as a direct consequence of his account of meaning and of paradigm-change. Since each paradigm tends to sweep away its predecessor, there can, he claimed, be no confidence in the permanence of the theoretical entities associated with any particular paradigm, no matter how well-supported that paradigm may appear. Later writers, like Laudan, Rorty, and Fine, picked up on this historicist theme, Laudan in particular developing an effective polemic against what he called convergent realism (Laudan, 1984b). Global anti-realism of this sweeping sort has one nasty consequence: it calls into question any sort of reality-claim for entities the warrant for whose existence would be the explanatory success of the theories in which they appear. This would include, then, all historical theories, cosmology, evolutionary biology, palaeontology, all attempts to establish the sequence of events in the distant past as well as the sorts of entities that populated the universe long ago, or for that matter, not so long ago. Quod nimis probat, nihil probat; the medieval adage “what proves too much, proves nothing”, comes immediately to mind (McMullin, 1984).
Global forms of anti-realism, inspired by accounts of theoretical meaning or of theory-change, by attacks on the misnamed “inference to best explanation”, or by the more extreme forms of social constructivism, are all vulnerable to this kind of commonsense objection. Among anti-realisms, much more viable are local or selective anti-realisms directed at specific sciences for reasons peculiar to that science. It is no accident that the most effective critics of scientific realism in recent decades are philosophers of quantum physics, like van Fraassen, who are motivated primarily by the ontological puzzles of their own field (van Fraassen, 1980). Indeed, van Fraassen’s notion of empirical adequacy would implicitly commit him to some sort of realism for entities other than the sorts of unobservables with which quantum physics deals. It would be quite consistent to hesitate to assign ontological status to the categorically remote sorts of quantum and field entities invoked by the theoretical physicist and yet to maintain a realist approach more generally. Finding a persuasive philosophical warrant for scientific realism in the face of arguments from the historical instability of theory or from the apparent circularity of arguments relying on the best explanation of the explanatory success of specific theories, is at the heart of the current debate. Hacking, for example, in his Representing and Intervening has argued for a form of realism that does not rely on an appeal to the explanatory virtues of theory (Hacking, 1983). The anti-realist case seems often, indeed, to reduce to a critique of the arguments currently advanced for realism rather than providing a defence for, and a clarification of the practical implications of, anti-realism in its own right. I would argue that paying special attention to the diachronic criteria of theory, that is, to those that track specific theories over the course of time, criteria like fertility, for example, affords the most effective positive argument in support of a qualified form of realism (McMullin, 1996b). But to pursue this would lead too far afield.
9. The social turn
I have left till last what has been perhaps the most spectacular new development in the philosophy of science of the last thirty years: the rapid rise and diversification in, social approaches to the understanding of science (Mendelsohn et al., 1977; Barnes and Edge, 1982; Brown, 1984; McMullin, 1992). This key development has taken two very different forms. In keeping with Kuhn’s directing of attention to the practice of science in the broadest sense, there has been a transformation of the historiography of science to include topics and approaches that would previously have been thought marginal to the pursuit of “real” history of science, that is, to the history of scientific ideas and their validation. Merton and his students had been applying the techniques of sociology to the community of scientists ever since the ‘thirties, treating that community, as Kuhn would later do, as in principle a relatively insulated one. But this reverential attitude was not to last. The combined efforts of historians, sociologists, ethnographers, anthropologists, from the ‘seventies onwards began to present a much more complex picture of the actual practice of science, both contemporary and historical.
The topic of experiment, overly idealized in earlier philosophy of science, came in for particularly intensive and illuminating analysis (Franklin, 1986; Galison, 1987; Gooding, 1989). The different “styles” of different laboratories, for example, seemed to be a topic worthy of the sort of scrutiny that anthropologists have long given different tribal cultures (Traweek, 1988). In consequence of this continuing research, we now have a far fuller grasp of the complex reality of the natural sciences as involving quite different sorts of practices.
But right from the beginning there was another more ambitious project of a rather different character under way under the capacious banner of the “social”. Though it went under the name of sociology, its aims were overtly philosophical in character. How, its proponents asked, are we to understand the nature of “scientific” knowledge itself in the light of this new emphasis on the social aspects of science as a practice and not just as a disembodied set of propositions? The earliest exponents of what they called the “sociology of scientific knowledge” were associated with the University of Edinburgh and their so-called “strong program” was intended, not just to amend, but to replace philosophy of science as it had been practiced from Aristotle’s day up to our own. The traditional focus on the propositional would be replaced with a new and broader concern with the social, in all that elastic term’s many applications.
Bloor and Barnes, when announcing their vision for a new field, SSK, that in their view would be more sociology than philosophy, relied on the claim that social factors, particularly what they called “interests”, social, political, personal, played a significant role at every step in the sciences, notably in the shaping of experimental “data” and in the final stages of theory-assessment (Bloor, 1976; Barnes, 1976). So far, this might not seem so different from the doctrine of the Idols in Bacon’s New Organon, or the notion of “convention” we have seen in the work of Popper and Carnap. Where it diverges is that the proponents of SSK see these social influences as constitutive and as ineliminable, as a necessary part of scientific decision-making, which (they claim) is radically underdetermined from the side of the physical world. Scientific theories are, then, to a significant extent social constructions imposed upon reality. The question of their truth is thus beside the point. The entities they postulate lack a credible claim on reality; they are, in effect, no more than aids in prediction. Here social constructionist versions of anti-realism meet those that derive from historical accounts of theory-change.
Not all of the socially-inspired philosophies of science are so emphatic in their challenge to traditional objectivist commitments. And some, indeed, are more extreme in that regard (e.g. Woolgar, 1988). There is a fair amount of disagreement between the more philosophically-inclined exponents of SSK and the more descriptively-oriented defenders of ethnomethodological approaches. And there is little agreement about how best to respond to the reflexivity challenge (must not the claims of SSK themselves to be taken to be socially-constructed and thus to have, at best, no more than a tenuous hold on truth?) (Ashmore, 1989).
The social-constructionist challenge to the objectivity of scientific judgement faces one question, in particular (McMullin, 1984b). If theory-choice is as heavily influenced by socio-political and other, on the face of it non-epistemic, factors, why should it yield theories that continue to improve dramatically in the crucial matter of accurate prediction? One response has been to reject outright the distinction between epistemic and non-epistemic. Another has been to claim that certain sorts of social factor can also be epistemic, that is, truth-yielding. Some feminist philosophers of science, among them Helen Longino, argue that feminist values can lead to a more holistic approach to such areas as neuro-physiology, an approach that is more likely in the long run to yield good science (Longino, 1990). This is an interesting proposal, one that could in principle be tested. There is obviously a better chance that a philosophical world-view might turn out to be a properly epistemic factor in scientific theory-choice than, say, class-interest or the rivalry between competing research groups.
What is at issue in all this is not whether or not social, political, and personal interests play a role in purportedly cognitive decision-making in science. They obviously do, as case-studies of scientific controversy make clear (Englehardt and Caplan, 1987), and as is in any case evident from even the smallest familiarity with scientific research. Where the disagreement lies between the more subjectivist social accounts and the more traditional philosophies of science is as to whether the role of particular non-epistemic factors tends to be progressively limited by the normal operations of science over the course of time, and whether in consequence it can possibly be correct to claim that long-tested theory is more social construction than a fair representation of real extra-social structures in the natural world.
In this all-too-rapid review, I have tried to pick out the themes that have most actively engaged philosophers of science over the course of the last half-century. Inevitably much has gone without mention. I have emphasized the historicist turn of the ‘sixties and ‘seventies. But the pages of the main U.S. journal, Philosophy of Science, or the British Journal for the Philosophy of Science, remind one also of how much has remained the same. Many authors are still struggling with the same logical issues-reduction, probability theory, inductive validation…-as those that preoccupied the philosophers of science of the ‘thirties. Many more are turning to novel issues in the philosophy of the sciences, notably in the areas of evolutionary biology and quantum mechanics. The newest generation of philosophers of science appear to be attracted most to these latter fields, perhaps in part, at least, because the intellectual excitement generated by the historicist turn of nearly forty years ago has to some extent subsided.
Perhaps the most striking development of the last half-century, however, has been the growth of philosophy of science as a professional discipline. The first-ever LMPS Congress, which was held at Stanford University in the U.S. in 1960, was a small affair compared to the LMPS Congresses of more recent years. At the first-ever biennial meeting of the U.S. Philosophy of Science Association in Pittsburgh in 1968, a small meeting-room was more than sufficient for our numbers. Now there are hundreds of academic philosophers of science spread over many countries, and dozens of journals and conference-proceedings devoted to the field. I rather doubt that we have as many towering figures in the field as we had in those extraordinarily productive mid-century decades. But the quantity of really high-quality work is far greater today than before. And one feature of the historicist turn I have been chronicling is the impressive research that has been appearing in case-studies from the history of science as well as in the history of philosophy of science itself, published in consistently appealing new journals: Studies in the History and Philosophy of Science, Perspectives on Science, Science in Context. We understand a field best, whether it be science or the philosophy of science, they are saying by implication, by studying it not just as an atemporal propositional form but also as a living and changing organism.
It is this realization that has been the principal achievement of the tumultuous ‘sixties and ‘seventies. The excesses of those years and the years since are evident. But these should not blind us to the importance of the insights that were then gained, as well as of the issues that were then raised, many of them still only partially resolved. They will carry us well into the new millennium.