The Burden of Disproof

What makes you think that next time you plug in your brimming kettle boiling water will be produced? I ask, as the sceptic in philosophy always asks, not because I have any specific reason for thinking that there is something wrong with the kettle, its lead or the electricity supply in your kitchen; and not because I dispute the validity of certain scientific theories about the behaviour of heated liquids. Let’s assume all is well in these respects, that there really is a pattern in our past experience which correlates the application of heat to water with its reaching its boiling-point: I want to know why the existence of such a pattern in the past gives you any reason to believe it will continue to hold in the future. We have after all no guarantee that the future will resemble the past: the contrary of any matter of fact, however well-established, always remains possible – and this suggests we have no more reason to expect a continuation of an observed pattern than we do to expect its complete breakdown. We certainly cannot avoid this conclusion by arguing that its continuation, although not certain, is more likely than its breakdown, for that would presuppose that the frequency with which we have encountered something in the past can form the basis of a calculation of the probability of its future re-appearance, which is precisely what is in question.

This is how scepticism about induction – the practice of drawing conclusions about a wider, unobserved domain on the basis of observations of a limited sample – takes shape; and what makes it so intractable is that exactly the same question-begging circularity infects all of the most widely-canvassed responses to the sceptic’s request for a justification. If we defend our kettle-boiling plans by invoking a scientific theory about the behaviour of liquids, we presuppose that that hitherto-justified theory will continue to hold good in the future; talk of a Principle of the Uniformity of Nature on which all beliefs about physical reality rest simply conjures up a metaphysical fog to shroud the question of why we should believe that such a Uniformity will continue to hold; and pointing to the past success of the inductive method invites the sceptic to ask why we should believe in its future success. On the other hand, capitulation to the sceptic seems to amount to rejecting the basis of that knowledge of reality on which most of our everyday transactions with the world rely, and by means of which the edifice of the sciences is constructed. So philosophers have tended either to ignore the obvious inductive conclusion to be derived from their past failures and continue the search for a non-question-begging answer, or to look for some way of undermining the vantage-point from which the sceptic poses the question.

What distinguishes Sir Karl Popper as a philosopher of science is that he accepts both the cogency of the sceptical critique of induction and the legitimacy of the scientific endeavour, and accordingly devotes himself to the Herculean task of characterising the methodology of science in such as way as to demonstrate that its practitioners need not employ any variant of inductive reasoning. The pivot of his account is the logical asymmetry between verification and falsification. If the sceptic is right to assert that no body of positive evidence can verify a claim which reaches beyond that evidence, then no set of confirmatory scientific observations or experiments can verify a law-like scientific hypothesis; but a single conflicting observation can conclusively falsify it. The claim that ‘all swans are white’ is not provable by any number of observations of white swans, but it is definitively disproven by the observation of a single black swan. In short, scientific laws are testable in spite of being unprovable, because we can systematically attempt to refute them. From this logical observation, Popper has drawn the following conclusions: that scientists should aim at falsifying rather than verifying their best available theories, so that scientific progress is a matter of eliminating falsehood rather than establishing truth; that a good scientific theory is one formulated in such a way as to expose it as clearly as possible to such attempts at refutation; and that whenever these efforts succeed, the failed hypothesis should be replaced with one which provides as bold and imaginative a solution to the problem as possible, so that it, too, will then be clearly exposed to possible refutation.

This vision of science as an arena within which leaps of theoretical creativity drive a search for the facts that will reveal their inadequacy, and in which the best available theory is at most a provisional resting-place for thought and practice, manages to combine demystification and exhilaration in proportions that have proved inspiring to many of Popper’s readers; and its implications stretch far beyond the scientific realm. For Popper’s account appears not only to demonstrate that science can do without the inductive method, but also to provide an alternative criterion by means of which to demarcate science from non-science. Put briefly, the distinguishing mark of a genuinely scientific theory is its falsifiability: a theory is scientific only if it is testable; and it is testable only if some imaginable observation would refute it. It is on these grounds that Popper, in The Poverty of Historicism, launched his famous and influential attack on what he saw as the scientific pretensions of Marxism and psychoanalysis. Marxism failed the demarcation test because, although its proponents advanced many falsifiable predictions, they responded to their falsification by endlessly reformulating both their theory and their characterisations of the evidence so as to bring the two into a seamless accord. Psychoanalysis failed because no conceivable observations could contradict its theories; by attempting to account for everything, its practitioners succeeded in explaining nothing.

It is worth pointing out that, at least in Popper’s hands, falsifiability functioned solely as an attempt to define the bounds of science. The failure of Marxism and psychoanalysis to pass the demarcation test simply placed them outside the scientific realm; it did not entail that those theories were of no value whatever, or that they were entirely empty or nonsensical. Certain of Popper’s avowed followers have employed the falsifiability criterion as part of a crude polemic against any and all variants of Marxism and psychoanalysis; but in so doing, they merely repeat the intellectual error of those among their opponents who assume that defending the validity and value of the work of Marx and Freud means defending its scientific validity and value. Nonetheless, Popper’s conception of science has strongly influenced his view of the best approach to matters lying outside the domain of the natural sciences: his famous defence of the procedures and institutions of Western liberal democracy in The Open Society essentially depends on the claim that any other political arrangements will fail to acknowledge what his analysis of scientific method demonstrates – the provisional, fallible nature of human cognitive processes. Highly authoritarian political practices, for example, severely constrain a society’s ability to develop, implement and refine social policy imaginatively and critically, and so to exercise rational control over the structures which most profoundly influence the course of its citizens’ lives. In this sense, for Popper, liberal democracy is an indispensable framework for the rational practice of politics.

It is the development and defence of this distinctive combination of views on the nature of science and politics that has given Popper the reputation he now enjoys, and which made the idea of celebrating his 90th birthday a commendable one. However, In Search of a Better World offers little If any indication of the nature or ultimate worth of that intellectual project, and so constitutes a profoundly unsatisfying way of commemorating the work of its author. Indeed, if any theme predominates across this heterogeneous collection of occasional essays and lectures, it is Popper’s deep commitment to a theory that many of his otherwise-devoted supporters find difficult to swallow – the ‘three worlds’ hypothesis. According to Popper, in addition to the material world and the subjective world of minds, there is a third world: a world of objective structures originally produced by mind-endowed creatures but thereafter existing independently of them. This world encompasses ideas, art, science, language and institutions that are encoded and preserved in, but not reducible to, material objects such as brains, books and buildings. The most puzzling thing about this abstract realm is why Popper feels the need to hypothesise its existence. His main argument appears to be that, once created, its denizens have unexpected and unavoidable consequences: the example of mathematical truths is clearly at the forefront here. However, conceiving of mathematics simply as rule-governed human activity would allow us to accommodate such facts without invoking Platonic Forms. Since not every application of every rule can be foreseen, unexpected consequences might emerge, and our commitment to the rule would ensure that we regarded such consequences as unavoidable, but they would neither emerge nor continue to hold in the absence of human mathematical practices and dispositions; in short, the autonomy Popper highlights can be understood as real but relative. Since, therefore, the truth of his hypothesis is not established by the facts he cites, the most prominent theme in this collection serves primarily to detract from the respect in which his other work has been and continues to be held.

This is not, however, all that the book has to offer. It also contains musings on such matter as the nature of Beethoven’s creative self-criticism and speculation about the origin of commercial book-publishing in the Athens of Pisistratus; but such thoughts can hardly be developed within the compass of a short lecture in a way that will generate any conviction in their plausibility which is not itself derived from one’s respect for Popper’s abilities in entirely different intellectual fields. Some of Popper’s more recent pronouncements in those areas – on the nature of science and the social sciences, the development of human knowledge and the responsibility of intellectuals to defend and refine the idea of the open society – are reprinted here, and disciples eager to devour every last scrap from his filing-cabinet will no doubt be delighted to read them; but they are summaries of work long ago completed or manifestos for future research, and so rarely indicate even the structure of the arguments which buttress their conclusions.

In the years since their development it has become increasingly clear how far Popper’s views on politics and culture presuppose his views on science, and how far his views on science are open to fundamental criticism. Two main difficulties have emerged. First, if Popper’s account is to provide a viable methodology for science, he must explain how there can be any non-inductive way of ascertaining which of several competing theories is worthy of adoption. For, placed as it must be within an endless cycle of conjecture and refutation, any prevailing theory in any given science at any given time can neither be known nor justifiably believed to be true; at best, we can claim that it has so far survived our best attempts to refute it. But why should the fact of its present survival give us any reason to rely on it in the future? After all, both this theory and any of its less successful competitors generate an infinite number of testable propositions about reality: so the successful theory may nevertheless contain a huge quantity of undiscovered falsehoods, larger even than a theory that has hitherto failed most of our tests. It is thus impossible to see how a theory’s ability to stir vive our best efforts at refutation can establish that it contains more truth and no more falsity, or no less truth and less falsity, than any of its less successful rivals – unless we illicitly assume that its past success is a good indication of its future success, and so reintroduce inductive reasoning.

Determining when to reject any given theory is equally problematic. For the criterion of falsification operates within a much more complex practical arena than the logical asymmetry between verification and falsification might suggest. No scientific theory ever en counters reality ‘naked’: when a scientist attempts to refute a given hypothesis by carrying out an experiment, she must set up and employ complex equipment and observe the results, and both her construction and her observations will be theory-laden – ineliminably influenced by her understanding of how her equipment works and how reality works. An observation which might appear to conflict with the hypothesis under test can thus always be accommodated by assuming an equipment malfunction or by assuming an unknown factor in the experimental environment; in other words, it can be treated as an anomaly rather than a refutation. And this is not merely a logical point; every successful scientific theory has co-existed with a certain number of anomalous test results, and the refusal of scientists to treat all anomalies as refutations has been a precondition of progress – the most famous example concerns Uranus, whose orbit appeared eccentric on Newtonian principles, but was correctly explained by retaining those principles and positing the existence of another planet (Neptune). There cannot therefore be any such thing as an unquestionable procedure for rejecting scientific theories; we can certainly say that some ways of coping with anomalies will be more respectable than others, but we cannot provide an abstract recipe for determining whether any given theory should be jettisoned or maintained in the face of seemingly recalcitrant evidence.

Popper’s followers have attempted to modify his approach in the light of these difficulties – most famously, by shifting its focus from single theories and single experiments to sequences of such theories embodying rigorous stipulations regarding the treatment of anomalous evidence. Thus far, however, there is little reason to believe that exactly the same difficulties do not re-emerge with unaltered force in these new contexts; and such accounts continue to neglect what the history of science reveals: the relative rarity of determined attempts to refute prevailing theories and to construct radical alternatives to them, a corresponding tendency to treat potential refutations as anomalies, and the importance of socio-cultural factors in determining the focus and trajectory of experimentation. This neglect continues because Popperians fail to see that relating methodological abstractions to the particularities of practice does not reduce logic to sociology but rather restores it to the medium within which it has a function and so a meaning.

If, then, Popper has comprehensively misrepresented the nature of rationality in science, the idea of defending liberal democracy on the grounds that its institutions are best suited to the exercise of rationality (thus understood) loses much of its attraction. In truth, once the positive political vision that Popper offers is detached from his forceful and valuable critique of alternative historicist and authoritarian models, it has little to recommend it. For it amounts to presenting ideal citizens as problem-solvers, agreed on their ends and concerned only to identify the most effective means of achieving them, and focusing on concrete, manageable projects with short-term pay-offs rather than Utopian social blueprints. Such advice borders on the platitudinous (step forward those who prefer ineffective ways of carrying out unmanageable projects); and to conceive such tasks as the main business of politics is to ignore the reality of fundamental disagreements about what society’s most pressing problems might be, how far their amelioration might bring about greater overall suffering or require radical social reform, and how far the interests of some groups may be irreconcilable with those of others. As is often the case with attempts to look at politics through natural-scientific spectacles, the fundamental conflicts of value and power which make it so intractable are omitted rather than overcome. Popperian piecemeal social engineering is effectively politics depoliticised, a vision whose resources run out just when the going gets tough.

In philosophy, as in science, much can be learned from failure; in this case, Popper’s failure to construct a viable, induction-free scientific methodology teaches us just how deeply rooted in human reason and human life the concept of induction really is. To adapt one of his own favourite quotations from Xenophanes, his web of guesses turned out to be a snare – but a very well-woven one.