Blite and Whack

A year or two ago my eye was caught by the cover of a magazine on an American news-stand. It was a magazine for the working woman, and its title, in the best traditions of the me-generation, was Self. The cover advertised articles with titles like ‘The Problems of the Kept Man’ and ‘What if He Says No?’ But what attracted my attention was the rather Californian injunction flashed in bold letters across the top: ‘Let’s Be Real!’

Of such stirring raw material is philosophical reflection made. But it can be difficult for philosophers to persuade others that worrying about such hardy perennials as realism and idealism is any more than neurotic. It is often difficult for them even to show their colleagues where the nub of the problem lies. Sir Karl Popper, who was 80 in 1982, is probably the philosopher outstandingly associated in the public mind with showing that such disputes do matter, that the enterprise of science, which aims to discover the truth about the universe we live in, is continually under threat from quacks and the pedlars of false certainty. He has shown that what satisfies our curiosity may not be true. His attacks on Marxism, astrology and psychoanalysis have been widely influential. Many working scientists testify to the effect he has had on the actual practice of their art. And although there is a suspicion of redundancy about the title In Pursuit of Truth on a philosopher’s festschrift (rather as though one discovered the title Against Sin on a festschrift for the Pope), it is Popper above all who is associated with demonstrating that if we care about knowing the truth, it matters very much how we set about it.

A Pocket Popper (no hyphen) is a selection of 30 readings from his work, in four sections on epistemology, the philosophy of science, metaphysics and social philosophy. The selection admirably shows the breadth of Popper’s concerns, and the reason it is a deeply unsatisfactory book is no reflection on the job the editor has done. It is that Popper is particularly badly misrepresented by being carved into bite-size chunks. The book gives one the unfair impression of Popper as moving briskly down an agenda of problems, ticking them off as he solves them in a mere five or six pages apiece. This not only tends to obscure the links between the different areas of his thought: it gives a taste of the energy of his argumentative style but barely a hint of its subtlety. Since Popper’s public reputation is as something of a blunderbuss, some of whose arguments devastate their target while others miss by many yards, this impression is an unfortunate one.

Any reader new to Popper who notes the confident, almost dismissive tones of the extract in which he ‘solves’ Hume’s problem of induction in a page and a half would be unlikely to guess that Popper thinks induction so difficult and complex a problem that he returns to it again and again in the main body of The Logic of Scientific Discovery and in the first volume of the Postscript under review here. Die Logik der Forschung was published in 1934 but was not translated into English until 1959. The Postscript has been circulated like samizdat since 1957 but is only now available in published form. Of its three volumes, the second has been discussed in these columns already (Vol. 5, No 15), while the third is on quantum physics.

The first volume, Realism and the Aim of Science, is the most directly continuous with the original work. Like the original, it is concerned with what Popper sees as the two central problems of the theory of knowledge: namely, the problem of induction and the problem of demarcation (of what distinguishes science from non-science). Underlying both of these is the problem of reconciling our sense that there exists a world independent of our theories about it, with the sceptical doubts of Hume concerning our justification for drawing general conclusions about the world on the basis of particular and finite observations. Many philosophers have resolved the tension simply by declaring that Hume was wrong, that the fact that we cannot make a deductively valid inference from the particular to the general does not mean we cannot make an inductively valid one. That, it is said, is what the ‘logic of inference’ is all about.

Popper’s claim – and if true, it is a remarkable one – is to have solved the problem of induction while accepting that in the essentials Hume was right, and that therefore we are not justified in reasoning from individual instances to general laws. Thus no number of observations of white swans can justify our assertion that all swans everywhere are white. They cannot even make it probable, he insists, that all swans everywhere are white. There is only one way of reasoning from the general to the particular, and that is by falsification, not verification. A single instance of a black swan does justify our assertion that it is false that all swans are white. Science proceeds not by trying to justify true theories but by seeking to eliminate false ones. What is distinctive about science, then – and this is where demarcation comes in – is that it makes possible the pursuit, and thus the elimination, of falsehood. Justification, and with it subjective certainty, are chimerical: if you care more about certainty than about truth you will never go beyond what the observations warrant, so you will go nowhere. It matters not a jot how our theories are reached: we may dream them, invent them as a joke or discern them in the tea-leaves. What does matter is the criticism to which we subject them, and the possibility of rejecting them as false. If there is no such possibility, these theories are no part of the enterprise of science. For Popper – unlike the logical positivists – this does not make them nonsense: they may be respectable metaphysics (like the principle of causality), or meaningful but bogus (like astrology).

The matter does not, of course, end there – as the four hundred-odd pages of this volume weightily testify. For clearly we care about falsehood, not intrinsically, but as a means to truth. Does Popper’s programme make the truth any less elusive? Arguments have raged about this for many years, and much of the book is devoted to a defence of his programme against attacks which, he acknowledges, come in varied and subtle forms. They fall into two main kinds. First, there is the contention that falsification is no more justifiable than verification, since the falsifying observations themselves are not ‘given’ but are implicitly theory-laden (and thus inductive in nature after all). Secondly, since there is an infinity of potential theories of the world, a finite number of observations will falsify very few of these. There will still be an infinity unscathed, all but one of which will be false. Falsification can guide us towards the truth, it is claimed, only if we covertly appeal to a principle of induction to help us choose between so far unfalsified competing theories.

Both of these objections are given considerable space in T.E. Burke’s The Philosophy of Popper, which might more accurately be described as meditations on Popperian themes than as a proper critical assessment of Popper’s work as a whole. In some respects, this is a strength: Dr Burke writes well and with clarity about scepticism in general, points to some illuminating (and surprising) affinities between Popper and the later Wittgenstein, and makes a sympathetic case for scepticism as a challenge that seriously threatens Popper’s realism. He stresses the logical gap between the criterion for the correctness of assertions (‘the set of conditions under which we licence ourselves to use them thus’) and what we actually claim to be the case – the commitment we undertake in making them. ‘It is in this gap that scepticism takes root.’ Burke argues that to accept the gap (as Popper and Hume do) is to cave in to scepticism: but, he says, the theory-ladenness of observations is in the end surmountable, and this provides a way to close the gap. We can not accept, he says, Popper’s ‘unrestricted fallibilism’, the view that ‘every statement we make is overshadowed by the possibility of error, and the best we can do is to agree, in some cases, to ignore this possibility and treat certain sentences as true.’ On the contrary, ‘we cannot suggest that (even possibly) we always go wrong, or may go wrong in any given case, however favourable the conditions, because to do so would deprive us of the means to make any such assessments at all.’ The way-out is to assimilate commitments to criteria: ‘to say how things are is of necessity to say how, under certain specific circumstances, we find them’ – it is ‘a statement of the result of some finding-out operation’. This verificationist approach circumvents, he claims, the theory-ladenness of observation and in the process makes falsificationism unnecessary: some general truths are ultimately derivable from finite numbers of particular observations. Hume’s insidious arguments are the undoing of Popper’s claim to be a realist.

Burke’s version of verificationism is a rather literal one that has come under considerable fire – for example, from Quine and Dummett. To say that making a statement is simply saying what we would observe in verifying circumstances is just implausible: statements about, say, intermediate vector bosons are not just statements about the behaviour of instruments at the big new underground particle accelerator at CERN: we certainly don’t go to the vast expense of building an accelerator just to see what its instruments say – we think that such experiments tell us something about the world outside the laboratory.

But more important than this, Burke’s book gives disappointingly little sense of the variety and ingenuity of the responses Popper has made to the many charges of scepticism that have been levelled at him. That observations even in natural science are theory-laden is now a fashionable doctrine, but it was an integral part of The Logic of Scientific Discovery, not a subsequent reluctant concession. Accepting the truth of falsifying observations is a decision, not an unquestionable given. The fact that we may all decide not to question the truth of an observation does not settle the matter, and later generations may do the questioning for us. The Copernican system depended on challenging the decision of Aristotelian cosmologists not to question the obvious truth that the earth does not move. Of course, to falsify one theory we must accept another, so we cannot make sense of what it would be to question all our theories at once. But that does not mean that any of them are immune from being criticised one at a time. In baulking at Popper’s fallibilism, Burke does not make clear how naturally it springs from Popper’s repudiation of justification, and his rejection of the quest for certainty as utterly distinct from, and inimical to, the quest for truth.

What of the second main objection to Popper’s programme – that based on the infinity of so-far unfalsified theories? Here the verdict must be mixed. A.J. Ayer once objected: ‘Why should a hypothesis which has failed the test be discarded unless this shows it to be unreliable: that is, except on the assumption that having failed once it is likely to fail again?’ This would smuggle in the principle of induction under another guise. Popper’s reply to this is straightforward: having failed once, it is false. A theory like ‘all swans are white’ needs only the discovery that a black swan was born in 1983 to be false, even if all swans ever born before and thereafter are white. However, Ayer’s objection can be reformulated: even if a black swan is born in 1983, why should we not adopt a theory which says that all swans except those born at a certain place and time in 1983 are white? To Popper, such a theory would be messy and ad hoc, but is it any the less likely to be true?

Of the remaining infinity of unfalsified theories, Popper prescribes the selection of those that are the simplest, the boldest, the most sweeping, the least ad hoc. What all these virtues have in common is that they involve a theory in laying itself as open as possible to falsification: that is, they help to speed up the rate at which we discard false theories. Now one doubt which might arise is this: speeding up the rejection rate may perhaps be desirable for some reason, but it cannot be because it makes us more likely to hit on the truth, since however fast we reject theories we shall always be left with an infinite number. So if we presuppose that simpler theories are more likely to be true, or, in Popper’s later formulation, are more like the truth, are we not appealing to induction? Popper’s answer is no: the claim that in a particular case the simpler or bolder theory is true is not an untestable presupposition: it is a testable conjecture. We do not assume that all simpler unfalsified theories are closer to the truth: we pick simpler theories on other grounds, and conjecture that they are true – and we may be wrong. As David Miller puts it in his interesting, if brisk article in the festschrift, Popper’s view that ‘reality, though unknown, is in some respects similar to what science tells us’ is ‘a consequence of what science tells us, not an assumption science has to make’.

But there is a second doubt that is harder to dismiss. It is the doubt raised by Nelson Goodman’s famous paradox, and we can express it by saying that the ability even to pick out some theories as simpler (never mind truer) than others itself rests on an inductive principle. In the case of our swans, someone might have a theory that all swans were ‘blite’, where something is blite if it is black and born at a certain time and place in 1983, and white in all other circumstances. Likewise, a swan is ‘whack’ if it is white and born at that time and place, and black otherwise. The telling point is that our theory that ‘all swans are white’ looks thoroughly ad hoc to someone who perceives objects as blite and whack: to him, we hold the extraordinary theory that all swans are blite, except for some unexplained reason at a certain time and place in 1983, when they are whack. Goodman’s paradox runs very deep: there does seem to be something strangely arbitrary about our perceiving the properties of the world in the way that we do. To my mind, Popper nowhere adequately answers it, perhaps because his critics have usually seen it as vitiating his alleged claim that simpler theories are more likely to be true – in fact, Popper makes no such claim. What Goodman’s paradox really vitiates is Popper’s view that we are able to perceive simplicity in the first place, independently of the principle of induction.

It seems reasonable to conclude, then, that reports of the death of the problem of induction are an exaggeration – but that Popper’s own ‘solution’ also has a great deal more life in it than most of his critics have been willing to concede. Some of this vigour is well represented in the essays by David Miller and Gunnar Andersson in In Pursuit of Truth. They are the best pieces in an otherwise fairly dismal collection, some parts of which are embarrassingly rambling and ill-conceived. It seems to have been designed on the principle that an 80th-birthday present to a revered colleague should be based on contributions from a large number of people at minimal inconvenience to themselves, should contain many fulsome expressions of appreciation, and above all should be scandalously expensive – forgetting that if it is a book it should also be worth reading. At £25 this is a pocket-popper in quite the wrong sense.

A major part of the Postscript is devoted to a defence of no less than three kinds of indeterminism. Space forbids a proper discussion of them here, but since Roger Sperry’s book, Science and Moral Priority, is relevant to one of them they must be distinguished. The first is Popper’s theory of the objective nature of probability: probability statements, he argues in the first and third volumes, are statements about objective propensities inherent in the world, not expressions merely of our subjective uncertainty. His arguments are complex, powerful, and in the end may be found entirely convincing. But he does make the disposal of the subjective approach look suspiciously easy. For example, he argues that the subjective approach cannot make sense of the notion of probability via the limit of a sequence of independent experiments, since every time we perform an experiment our information changes, so that the experiments are not independent. There are two problems with this argument: the first being that the subjective approach defines probability relative to some given body of knowledge (not necessarily all the knowledge we possess), so that what we learn as we perform the successive experiments does not change their independence. Secondly, the subjective approach does not claim to define probability via the limit of a sequence of experiments, since it attempts (with debatable success) to define it, by the method of Ramsey, Savage and others, via individuals’ preferences for different actions (such as gambles).

The second argument for indeterminism is the centrepiece of the second volume, and is an attack on the possibility of Laplace’s demon, a creature who, if furnished with all the laws of nature and able to measure initial conditions to any desired degree of accuracy, could predict events indefinitely far in the future to any degree of accuracy. Physicists are currently becoming very interested in ‘chaotic’ or ‘turbulent’ systems, ones which are in the normal sense deterministic, but in which disturbances, however tiny, in the initial conditions lead eventually to indefinitely large disturbances in the behaviour of components of the system. The existence of such systems decisively refutes the possibility of Laplace’s demon. What Popper’s argument shows, though, is not that there cannot be some function which determines behaviour, but simply that we could never know it well enough to use it. This is a special case of the undoubted truth that we shall never know everything, and some who are disturbed by determinism may not find that it completely lays their fears to rest.

Neither of these arguments, powerful and important though they are, has anything very much to say about human freedom in the traditional sense, and certainly has no connection with political liberalism. I suspect this may be because on these issues metaphysics is deservedly silent, but Popper thinks it important to establish indeterminism in a third sense. This is intimately connected with his ‘three-worlds’ metaphysics, in which World I comprises the world of physics, World 2 the world of subjective consciousness, and World 3 is ‘the world of the products of the human mind’. He argues that in addition to the kinds of indeterminism already advanced, it is necessary for human freedom and creativity that World I not be ‘causally closed’ but ‘causally open towards Worlds 2 and 3’: ‘If nature were fully deterministic then so would be the realm of human actions; in fact there would be no actions, but only the appearance of actions.’

If for the moment we restrict attention to the relation between Worlds 1 and 2, one of the problems with Popper’s view is that he does not consider the possibility that there is any alternative to the two extremes of reductionism (the view that a complete and exhaustive account of mental properties can be given in physical terms, which Popper rightly considers untenable) and a full-blooded Cartesian dualism. A great deal of contemporary philosophy of mind is devoted to exploring just such alternatives. Roger Sperry, a neuro-physiologist and winner of the 1982 Nobel Prize for Medicine for his work on brain bisection, takes a determined stand against reductionism by urging a variety of functionalism in which conscious phenomena emerge as holistic properties of neural organisation, and themselves play a causal role in brain function. This, he argues, is exactly similar to the way in which chemical properties of molecules are holistic or organisational properties not wholly reducible to sub-atomic properties. And, we might add, it is exactly similar to the way in which statements about the behaviour of football teams are holistic and not wholly reducible to statements about individual footballers, even if every action of a football team is also an action of some individual players. But, he insists, ‘our proposed mind-brain model also is non-dualistic in that it makes mind and brain inseparable parts of the same continuous hierarchy, the great bulk of which, by common agreement, is not dualistic. It becomes illogical to make a special exception of the principle at one level of mind and not at those above and below.’ In regard to Popper, to whom he devotes a chapter, he writes that ‘the current promotion of the separate worlds with a capital “W” in a true dualistic sense seems fundamentally inaccurate and misleading.’ To be fair, Sperry concedes that it has not always been clear how completely dualistic Popper’s view is intended to be. If it isn’t dualistic though, it is doubtful whether it succeeds in establishing his third variety of indeterminism in the sense in which he intends.

Though its arguments are not helped by being cradled in a sometimes sedative prose style, Sperry’s book is often interesting, particularly for those of us prone to pontificate about scientific practice from the sidelines. It is concerned with ethics as well as the philosophy of mind, for in the kind of anti-reductive neurophysiology Sperry proposes ‘human values can also be viewed objectively in causal control terms as universal determinants of human decision-making.’ As a result, ‘science becomes the most effective and reliable means available for determining valid criteria for moral value and meaning.’ Such an approach naturally prompts the objection: it is all very well to discover via neurophysiology what human value systems are and how they affect our behaviour – quite another to say what they ought to be. Insisting dogmatically on a fact-value gap is no answer either – but the ease with which Sperry travels from one to the other takes a bit too much for granted, especially in view of the vexed state of the sociobiology debate.

Nonetheless, his strictures against the glib separation of the concerns of science from those of the humanities (C.P. Snow’s complaint applied specifically to subject-matters rather than generally to cultures) are apt and important. What a tonic, then, to pick up Art, Science and Human Progress, a volume of lectures sponsored by the Richard Bradford Trust, which was founded in 1969 ‘to explore the relationships between the methods of scientific investigation and of artistic creation in literature, the visual arts, and other art forms’. The Popperian process of conjecture and refutation in science is a most fruitful vehicle for the exploration of parallels with the humanities, and provides a valuable focus in helping lecturers keep to a difficult brief. The temptation to be generally expansive is hard to resist in this context. The late Lord Clark, for instance, in an interesting lecture on the growth of television, provides some moments of unintentional comedy when he agrees morosely with Burns about ‘this unlimited promiscuity’ (whether in life or on television it is hard to tell) that ‘ach, it hardens a’ within, and petrifies the senses,’ and proposes a hypothesis that would doubtless intrigue Sir Keith Joseph: the ‘spiritual and intellectual decline which has overtaken us in the last thirty years’ may, he suggests, be due to ‘diversion of all the best brains on technology’.

There are two splendid lectures in the volume, from E.H. Gombrich and the late H.D.F. Kitto. Kitto discusses Sophocles in terms of Arnold’s line: ‘He saw life steadily and saw it whole.’ Gombrich explores in a sparkling way the analogy between evolutionary processes in which variants continually spring up and the process of criticism which shapes the evolution of aesthetic movements (though he stresses that the analogy must not be pushed too far). Criticism is integral to art as it is in Popper’s evolutionary view of knowledge: theories of the avant-garde which loftily ignore criticism as reactionary ‘can lead to a crippling elimination of all negative feedback, and with its disappearance the notion of an artistic experiment also loses its meaning.’ Popper’s name is not always in the foreground in this volume (except in Peter Medawar’s clear and straightforward account of his philosophy): but the spirit and influence of Popper make it the most fitting tribute I have seen to him on his 80th birthday.