PART I
THE CRITICAL APPROACH CHAPTER I
INDUCTION
But as for certain truth, no man has known it,
Nor will he know it; neither of the gods,
Nor yet of all the things of which I speak.
And even if by chance he were to utter
The perfect truth, he would himself not know it:
For all is but a woven web of guesses.
XENOPHANES
IN THIS introductory chapter, the problem of induction is treated more fully and in a wider setting than in my Logic of Scientific Discovery (L.Sc.D., for short) to which the present work is a sequel: here I shall discuss all its more interesting ramifications of which I am aware.
In section 2, I try to give an outline of my theory of knowledge, to re-formulate the problem of induction, and to re-state its solution. The discussion of the viewsâlargely metaphysical viewsâwhich tend to prevent this solution from being accepted is carried on to section 16, entitled âDifficulties of Metaphysical Realism. By a Metaphysical Realistâ.
Chapters Two and Three (sections 17 to 26) are concerned with the problem of demarcationâthe demarcation between science and metaphysics. (I do not attempt to demarcate between sense and nonsense.) There I try to show that the problem of demarcation, and its solution by a testability criterion of demarcation, have a significance which reaches far beyond the borders of philosophy.
Chapter Three (sections 27 to 32) is concerned with the problem of corroboration, and of introducing the technical term âdegree of corroborationâ. I try to show that this problem is of some interest, in view of the problem situation existing in the philosophy of science; for its solution corrects the widespread but mistaken belief that scientific induction can help us to assess the probability of a hypothesis.
1. A Puzzled Philosopher Abroad.
Not long after I first came to London, in the autumn of 1935, about a year after the publication of my Logik der Forschung (L.d.F., for short), I was taken to a meeting of the Aristotelian Society. Bertrand Russell, whom I had long admired as the greatest philosopher since Kant, read a paper on âThe Limits of Empiricismâ.1 Believing that our empirical knowledge was obtained by induction, and deeply impressed by the force of Humeâs criticism, Russell suggested that we had to assume some principle of induction which could not be based upon induction in its turn; a principle whose adoption marked the limits of empiricism. Russellâs position was almost the same as one which, rightly or wrongly, I had ascribed to Kant in the first section of my L.d.F.
Having been invited to participate in the discussion, I said that I did not believe in induction at all, although I did believe in empiricismâan empiricism which did not impose upon itself those Kantian limits which Russell was prepared to accept. This statement (which I formulated as pointedly as I could manage with the little English at my disposal) was taken for a joke by the audience, who graciously laughed and clapped. I then suggested that the whole trouble was due to the mistaken belief that scientific knowledge was an especially strict or certain or august kind of knowledge. This statement met with the same reception as the first. I concluded with an attempt to explain that, in the usual sense of âknowâ, whenever I know that it is raining, it must be true that it is raining; for if it is not true, then I simply cannot know that it is raining, however sincerely I may believe that I know it. In this sense of the word, âknowledgeâ always means âtrue and certain knowledgeâ; and âto knowâ means, in addition, to be in possession of sufficient reason for holding that our knowledge is true and certain. But, I said, there was no such thing as scientific knowledge in this sense. If, nonetheless, we chose to label the results of our scientific endeavours with the customary name âscientific knowledgeâ, then we ought to be clear that scientific knowledge was not a species of knowledge; least of all, a species distinguished by a high degree of solidity or certainty. On the contrary, measured by the high standards of scientific criticism, âscientific knowledgeâ always remained sheer guessworkâalthough guesswork controlled by criticism and experiment. (It could not even attain any positive degree of âprobabilityâ if this term was used in the sense of the probability calculus, for example, in the form given to it by Keynes or Jeffreys.) I ended by saying that merely by recognizing that scientific knowledge consists of guesses or hypotheses, we can solve the problem of induction without having to assume a principle of induction, or any limits to empiricism.
My little speech was well received, which was gratifying, but for the wrong reasons, which was puzzling. For it had been taken as an attack upon science, and perhaps even as an expression of a somewhat superior attitude towards it.
Admittedly, I had attacked, by implication, Science with a capital âSâ, and those of its devotees who were ready to take its pronouncements as gospel truth. But I knew, of course, that Russell, with his deep and critical understanding of science, and his love of truth, was no such devotee. Thus the dismissal of Science with a capital âSâ, although implicit in what I had said, had not been my main point at all. Rather, what I had hoped to convey to the audience was this: if we assume that what is called âscientific knowledgeâ consists only of guesses or conjectures, then this assumption is sufficient for solving the problem of inductionâcalled by Kant âthe problem of Humeââwithout sacrificing empiricism; that is to say, without adopting a principle of induction and ascribing to it a priori validity. For guesses are not âinduced from observationsâ (although they may, of course, be suggested to us by observations). This fact allows us to accept without reservation (and without Russellâs limits of empiricism) Humeâs logical criticism of induction and to give up the search for an inductive logic, for certainty, and even for probability, while continuing in our scientific search for truth.
This, my main point, was lost. And I realized that it could hardly have been otherwise. For if people think on inductive linesâand who does not?âthen a remark like âI do not believe in inductionâ can hardly be interpreted in any other sense than âI do not believe in scienceâ. Nor do I think that I should have conveyed my meaning better had I begun, say, with the words, âI believe in the greatness of science, but I do not believe that the methods or procedures of science are inductive in any senseâ. Had I said this, people would no doubt have heard âScienceâ in place of âscienceâ, and they might have concluded that I wanted to uphold some doctrine of intuition or intuitionism, or perhaps some form of scientific authoritarianism.
Having just used two âismsâ, I may perhaps mention that I have often been reproached, even by some of my most sympathetic philosophic friends, for my bad habit of discussing philosophy in terms of âismsâ; and I am quite ready to admit that it might be more straightforward if, without any reference to âismsâ, one could just explain oneâs tenets, state oneâs arguments, and be done with it. But my puzzling experience illustrates why this method does not always work. We never address ourselves to completely open minds. However open minded an audience may be, they cannot help harbouring, if only subconsciously, connected theories, views, and expectations about the world, and even about the ways in which we learn to know it. They have adopted positions; usually typical positions: âismsâ.
Most of us, especially most philosophers, hold a great number of theories consciously, and after critical examination; and we may be prepared both to defend these by argument and to give them up when good arguments are brought against them. But we all also hold theories which we take for granted more or less unconsciously and therefore uncritically; and these uncritically held theories often contain the strongest reason for continuing to hold those other theories consciously. That this is so has been known for a long time: Bacon described such unconscious assumptions as idols and as prejudices. In Plat0âs Dialogues, Socrates frequently makes his partners realize that certain positions taken up by them imply that they hold theories or views of which they are not fully aware and which sometimes are even mutually conflicting. Before him, Parmenides speaks of delusive opinions uncritically held by âthe mortalsâ.
One of the oldest, more interesting and perhaps more important tasks of philosophy is the critical examination of such âpositionsâ and the theories or views they involveâespecially those which are uncritically taken for granted. In doing so it is often found that there are clusters of related views related by common assumptions, by common preferences, or by common dislikes. Obviously it is often convenient and even necessary to give names to these positions, or views, or clusters of views. Hence the âismsâ.
That âismsâ have gone out of fashion in modern philosophy, and that using them or similar names is felt to be a sign of bad taste, is due to the fact that the critical discussion of theories or positions or clusters of views has gone out of fashion. But fashions, especially in philosophy, should not be accepted. They should be examined critically because they are themselves nothing but âismsâââismsâ adopted uncritically.
All this bears on induction. Many philosophers and even some scientists believe that induction is an undeniable fact of common sense: that the actual use of what is now often called âinductive proceduresâ cannot be seriously denied. This may or may not be so (see section 3 below). But we should at any rate learn to listen to those who deny facts of common sense. Philosophers have been extremely patient in listening to people who asserted and others who denied the existence of tables and chairs, or meteorites, or ghosts (both in and out of machines), or analytic statements. But the discussions of all these assertions and âismsâ are not nearly so fundamental for philosophy as the one about the existence or nonexistence of inductive procedures. So let us discuss the matter critically. Perhaps the assumption that inductive proceduresâan âinductive logicââexist is, after all, a prejudice, so that all that does exist is merely a myth, a mistaken âismâ (âinductivismâ).
If there is such a thing as inductivism, then this helps explain why âismsâ are unfashionable. For such things as uncritically held âismsâ are a danger to inductivism. Bacon saw this. But his remedyâpurge your mindâwas naive. So it has seemed better to inductivists to look away, or else to study âismsâ inductively.
These theories which, if held unconsciously, are obviously held uncritically, are often incorporated in our language; and not only in its vocabulary, but also in its grammatical structure. This was first seen, to my knowledge, by Bertrand Russell, when he pointed out that many philosophical theories depend on the mistaken assumption that âall propositions are reducible to subject-predicate formâ, an assumption which is closely connected with the grammatical structure of Indo-European languages.2 Later, a similar doctrine was developed by Benjamin Lee Whorf who stressed, more especially, the dependence of our idea of time upon our language.3
These facts about language are sometimes used to defend the following radical conclusion. We are, intellectually, it may be said, the prisoners of our language: we cannot think except in terms of theories (of substance, or of space and time, for example) which, unknown to us, are incorporated in our language; and we cannot escape by our own effortsâfor example by means of a critical discussionâfrom our prison, for the critical discussion would have to be conducted with the help of our language; and it would therefore remain within itâwithin the prison. Only by learning a new language of a different structureâone which is essentially not fully translatable into our old languageâand therefore only through the clash w...