POPPER’S VIEW OF SCIENCE

Bacon said, you cannot criticize a man and admire him. This view of Bacon was once very popular. In his obituary of David Hume, his friend Adam Smith says, I will not speak here of his philosophy since those who endorse it admire it and those who reject it despise it.

What is this point? The American philosopher, Max Otto, has referred (see note 2 above) to the fact that a working geologist, Lewis G. Westgate, had held this: there is value in publishing criticizable hypotheses in the hope of eliciting criticism, because that will be useful in the progress of science, as it opens the road to better hypotheses. Westgate was even aware of fighting a very popular prejudice. As late as in 1940 he had to state:

It may seem ungracious to pick on the mistakes of our predecessors and to drag them out for public discussion. But the theory of scientific method is one thing and its practice is often quite different. And it may be that by the examination of its faulty working and the errors which follow, we can best avoid errors ourselves. The progress of scientific discovery is more important than the concern of the individuals. The earlier investigators [Agassiz] were pioneers. We honour them as the founders of the science [of geology].⁵

This passage is very interesting. It clearly identifies the popular view that public criticism is disrespectful and useless. It opposes the popular view by showing what the use is of public criticism and by paying homage to those criticized.

Westgate’s view is influenced by that of T. C. Chamberlain, who, in turn, was influenced by William Whewell, the little known but highly influential Cambridge methodologist and historian of science of the second quarter of the last century. Whewell considered experimentation a method of criticism, and he considered that failed criticism led to the con­viction that the object of the failed criticism is true.

Popper agrees with Whewell that experiment is criticism, but he refuses to see any validation in the failure of criticism. Most philosophers who study Popper’s view consider it to be a modification of Whewell’s. Where­as Whewell viewed validation as conclusive, they think that Popper has claimed such validation to be tentative. To correct this impression we must return to the problem which Whewell’s theory of validation has come to solve and see whether Popper solves it in a similar manner or not. What problem, then, does the theory of validation come to solve? The problem of induction, of course. What is the problem of induction?

The problem of induction is peculiar: solutions to it are offered and hotly debated, but it is seldom stated and still less discussed. There is no easy way of learning from the vast literature on the problem what the problem is; one may even claim that it is not possible to arrive at a concise statement of the problem, because in different periods one may have to state it differently. It seems strange that there is a problem so very important that dozens of books and papers in the learned periodicals appear yearly in attempts to offer new solutions to it and to debate existing solutions to it, yet there is practically no discussion of the problem itself, its formulation, its significance, etc. Alfred North Whitehead has said this problem is a scandal in philosophy - a remark often quoted in the literature.

Here is another novelty in Popper’s philosophy. The novelty may be stated in a popular saying: a problem well-stated is a problem half-solved. Old as the saying is, it has not led a single philosopher to present fully the problem of induction - a task which could easily fill a volume or two of very interesting material. On this point, I regret to stress, Popper has failed too. In his classical work, The Logic of Scientific Discovery, Popper briefly states the problem and he does not discuss it at all. His statement is technical and unusual (though it was usual in the eighteenth century), with the result that quite a number of philosophers have misread his solution as a solution to the problem of induction as they see it, not as he sees it.⁷

The problem of induction is not a problem at all; it is a collection of problems with varying intellectual backgrounds; in different contexts different parts of the collection gain significance. Let me illustrate this with one example.

The way most philosophers of science would - and do - flippantly present the problem of induction is, how does experience validate a hypothesis? This may be reformulated as, how does a scientist choose one hypothesis out of a few competing hypotheses in the light of experience? To claim that the two above formulations are of one and the same problem amounts to two highly questionable assumptions which may, in the manner indicated here, be smuggled into philosophy before the debate even starts. The first is that scientists choose validated hypotheses. This amounts to saying science contains no error; even if the validated hypo­thesis turns out to be not so good, at the time it was the best validated it was rational to choose it. The second smuggled assumption is that validation is done through competition - an assumption rejected even by some who implicitly accept it!

Suppose a hypothesis can never be validated in any way.

The importance, we may say, is this. Let us assume that scientists do choose to endorse new hypotheses to believe in and to act upon. Now, if the choice is rational it must be justified; hence, no validation of new hypotheses, no rationality but mere alteration of fashion. This view of the importance of induction amounts to equating rationality with learn­ing - an appealing proposal, but surely not one to smuggle through by implicitly identifying the two formulations (in the previous paragraph) of the problem of induction. Yet this, I contend, is precisely how the proposal is made in a number of significant volumes on the theory of induction.

It is an amusing fact that Popper’s explicit stress on the theory of rationality as the theory of the growth of knowledge has impressed his fellow-philosophers; it has impressed them, I suppose, precisely because on this very point they are in profound agreement with him. On this point they do not say, we have hinted at this idea before, we can quote William Whewell on it (assuming they know who Whewell is), and such deprecatory remarks. Here they feel free to sympathize with Popper. But what if scientific hypotheses cannot be validated? If they can be validated we have (let us say) learning and rationality - but if we do not have valida­tion is it necessarily the case that we do not have learning?

There is an optical illusion of this kind in many domains. For instance, one may need money for bail, and very badly; but if one simply cannot mobilize it, one may, but often fails to, think of other ways to get bail, e.g. bonds. Whenever a problem turns out to be too difficult, one may dig deeper and find the problem giving rise to it and try to solve it dif­ferently.

The problem of induction, on a deeper level, is the problem of learning from experience, and the problem of rational choice of hypotheses; but these cannot be solved by validation - at least not according to Popper who agrees with Hume and others on this.

At last we have come to the problem of induction - or half of it (we have dropped rationality for a while) - in a formulation favourable to Popper.

The problem of induction is, how do we gain theoretical knowledge from experience? The usual view of this problem, we have seen, is that it is the same as, how does experience tell us which theory to endorse? How does experience validate a theory? Whewell accepts all this, and tells us to examine carefully any theory by testing severely its corollaries, and to believe that theory which stands up to criticism. Whewell adds that such a theory will remain validated forever and this part of his theory we may simply ignore. After ignoring this part, does Popper agree with this paragraph?

Not at all. Popper denies, first and foremost, that the problem of in­duction (how do we gain theoretical knowledge from experience?) is the same as the problem of choice of hypothesis (which theory should I choose in the light of experience?) or that these are the same as the problem of validation (how does experience validate the hypothesis we choose?). For the last problem (how does experience validate a hypothesis?) Popper’s answer is totally negative; experience never validates a hypo­thesis. As to the problem of choice of a hypothesis, Popper says, experience never forces us to choose a hypothesis, but at most permits the choice of a hypothesis; we are not allowed to choose a refuted hypothesis, and if we choose an unrefuted hypothesis but refuse to test it for fear that it be refuted we are hardly better off; but we are only allowed, not forced, to choose a hypothesis which we test but fail to refute.

In brief, Popper’s solution to the problem of induction is totally revolutionary as a solution to that problem of induction*, in other contexts it is all too often stated, even as sheer commonsense: we learn from experience by criticism, by learning from mistakes. Both Whewell and Popper recognize the importance of attempts at criticism and the fact that they are important whether or not they are successful. Yet for Whe­well the successful criticisms are preludes to failed criticisms which validate a theory which thus gains supreme importance. For Popper successful criticisms are the junctures at which theoretical knowledge is gained. Popper’s theory of the importance of failed attempts to criticize does not come to solve the problem of induction.

We see here the marked novelty. Both Whewell and Popper have the same problem; both speak of unsuccessful and successful criticism. To solve the same problem, Whewell uses the former, Popper uses the latter.

The theory of development of thought through criticism may sound strange, but it is itself not novel at all. It is Socratic, and at least in one field of inquiry it was made by one school of thought the central philos­ophy of that field. The field is theology and the school is that of negative theology or theologia negativa. Popper has now applied it to science and so it is, just like Herzl’s solution, a new solution, as a solution to a new problem, but otherwise old. The novelty of it is, indeed, quite obvious: it sounds very strange to the modern ear. It sounds so strange to the modern ear because of the widespread of Baconion ideas. It is the prerogative of God alone to remain hidden forever, said Bacon, promising his reader that Mother Nature must sooner or later reveal her secrets to us if we go about our scientific business the right way, i.e. not offend Her by putting Her in the chains of our preconceived ideas, but flattering Her by taking serious account of Her smallest move. It is well known that in seventeenth century literature Nature often stands for God, even though only Spinoza had the courage to say it out loud. It is surprising that this did not lead at once to a full fledged scientia negativa', that such a theory was adumbrated, by Pascal, Boyle, and others, only makes matters even more puzzling.

Take away Bacon’s promise that Nature must reveal herself, and you do not as yet get Popper’s philosophy: conventionalist philosophers already did that, yet they insisted that science does not err. Take away both Bacon’s promise and his contempt for all error, and you can easily render theologia negativa into scientia negativa. But why should you do that? What problem should this come to solve? The problem of induction.

This can be seen only if the problem of induction is severed from the problems of choice and of validation of science, which the conventionalist failed to do. Consequently, the conventionalists declared that we simply cannot gain theoretical knowledge. Perhaps; but Popper’s view may be true, and then perhaps we can.

If you take Popper’s view as a serious candidate then you have im­mediately one serious problem on your hands; what criticism is tribute to the criticized and what not? It is a problem which Popper has studied only by implication. He has repeatedly stated that criticism is a tribute and hence we should choose the best opponent to criticize and criticize him respectfully. But which opponent is the best? Take a significant problem, take a list of its known solutions and try to criticize them explicitly. Any other solution that is a serious candidate has to be im­mune to these criticisms and then we may consider it new and under­take to examine it critically.

Apply now this criterion to Popper’s solution of the problem of in­duction. All past solutions are theories of validation, and traditional criticisms of the theory of induction apply to all theories of validation as such (any validation is in need of validation). Popper’s theory, not being a theory of validation, is immune to past criticism. Hence it is new. Is it also true? My answer to that must await another occasion.

This offers us a solution to the problem, which criticism is respectful, which is not? Any criticism is respectful if it contains recognition of the value of the criticized theory; a hitherto uncriticized theory can be ap­preciated because it is hitherto uncriticized; this enhances the value of criticism as well. We have here also a novel criterion of novelty as was promised earlier in this discussion. A solution is novel if it is immune to criticisms of older alternatives to it: a criticism is novel if it hits on as yet uncriticized solutions. All this is so strikingly simple it needs no elabora­tion and looks deceptively familiar. In a sense it is; as a solution it is new, and as a new solution it has strikingly new and interesting applications.

NOTES

¹ B. Russell, ‘Had Newton Never Lived’, Radio Times 15 (1927), 49-50. See also be­ginning of Ch. 2 of his Scientific Outlook.

² Max Otto, Science and the Moral Life, New York, 1949, p. 156.

³ Popper tends to view even idealism as a version of realism: in Berkeley’s view the thing-in-itself is God, in Mach’s view it is elements of sensation. This is intriguing but rather problematic both in its paradoxicality and vis-a-vis its historical context.

⁴ Representation and portrayal are two very different matters, but we shall not enter this point here.

⁵ Scientific Monthly (October 1940), 309.

⁶ Ibid., p. 301.

⁷ But let it also be said that Popper returns to the problem again and again, and that parts of his Conjectures and Refutations contain more extensive discussions of the same problem. He has finally tried to do justice to the problem in his Objective Know­ledge, Oxford 1972, Chapter 1.