The Myth of the Framework

In Defence of Science and Rationality

Karl Popper, M.A. Notturno, M.A. Notturno

• Publisher: Routledge
• Year: 2014
• ISBN:
  9781135974800
• Edition: 1
• Topic: Filosofia
• Subtopic: Storia e teoria della filosofia

In a career spanning sixty years, Sir Karl Popper has made some of the most important contributions to the twentieth century discussion of science and rationality. The Myth of the Framework is a new collection of some of Popper's most important material on this subject.
Sir Karl discusses such issues as the aims of science, the role that it plays in our civilization, the moral responsibility of the scientist, the structure of history, and the perennial choice between reason and revolution. In doing so, he attacks intellectual fashions (like positivism) that exagerrate what science and rationality have done, as well as intellectual fashions (like relativism) that denigrate what science and rationality can do. Scientific knowledge, according to Popper, is one of the most rational and creative of human achievements, but it is also inherently fallible and subject to revision.
In place of intellectual fashions, Popper offers his own critical rationalism - a view that he regards both as a theory of knowlege and as an attitude towards human life, human morals and democracy.
Published in cooperation with the Central European University.

1

THE RATIONALITY OF SCIENTIFIC REVOLUTIONS

Selection versus instruction

The title of this series of Spencer lectures, ‘Progress and Obstacles to Progress in the Sciences’, was chosen by the organizers of the series. The title seems to imply that progress in science is a good thing, and that an obstacle to progress is a bad thing – a position held by almost everybody until quite recently. Perhaps I should make it clear at once that I accept this position, although with some slight and faintly obvious reservations to which I shall briefly allude later. Of course, obstacles which are due to the inherent difficulty of the problems tackled are welcome challenges. (Indeed, many scientists were greatly disappointed when it turned out that the problem of tapping nuclear energy was comparatively trivial, involving no new revolutionary change of theory.) But stagnation in science would be a curse. Still, I agree with Professor Bodmer’s suggestion that scientific advance is only a mixed blessing¹. Let’s face it: blessings are mixed, though there may be some exceedingly rare exceptions.

My talk will be divided into two parts. The first part (sections I–VIII) is devoted to progress in science, and the second part (sections IX–XIV) to some of the social obstacles to progress.

Remembering Herbert Spencer, I shall discuss progress in science largely from an evolutionary point of view – more precisely, from the point of view of the theory of natural selection. Only the end of the first part (that is, section VIII) will be spent in discussing the progress of science from a logical point of view, and in proposing two rational criteria of progress in science, which will be needed in the second part of my talk.

In the second part I shall discuss a few obstacles to progress in science, more especially ideological obstacles. And I shall end (sections XI–XIV) by discussing the distinctions between, on the one hand, scientific revolutions, which are subject to rational criteria of progress, and, on the other hand, ideological revolutions, which are only rarely rationally defensible. It appeared to me that this distinction was sufficiently interesting to call my lecture The Rationality of Scientific Revolutions’. The emphasis here must be, of course, on the word ‘scientific’.

I

I now turn to progress in science. I will be looking at progress in science from a biological or evolutionary point of view. I am far from suggesting that this is the most important point of view for examining progress in science. But the biological approach offers a convenient way of introducing the two leading ideas of the first half of my talk. They are the ideas of instruction and selection.

From a biological or evolutionary point of view, science, or progress in science, may be regarded as a means used by the human species to adapt itself to the environment: to invade new environmental niches, and even to invent new environmental niches.² This leads to the following problem.

We can distinguish between three levels of adaptation: genetic adaptation, adaptive behavioural learning, and scientific discovery (which is a special case of adaptive behavioural learning). My main problem in this part of my talk will be to enquire into the similarities and dissimilarities between the strategies of progress or adaptation on the scientific level and on those two other levels: the genetic level and the behavioural level. And I will compare the three levels of adaptation by investigating the role played on each level by instruction and by selection.

II

In order not to lead you blindfolded to the result of this comparison I will annnounce at once my main thesis. It is a thesis asserting the fundamental similarity of the three levels, as follows.

On all three levels – genetic adaptation, adaptive behaviour, and scientific discovery – the mechanism of adaptation is fundamentally the same.

This can be explained in some detail.

Adaptation starts from an inherited structure which is basic for all three levels: the gene structure of the organism. To it corresponds, on the behavioural level, the innate repertoire of the types of behaviour that are available to the organism and, on the scientific level, the dominant scientific conjectures or theories. These structures are always transmitted by instruction, on all three levels: by the replication of the coded genetic instruction on the genetic and the behavioural levels, and by social tradition and imitation on the behavioural and the scientific levels. On all three levels, the instruction comes from within the structure. If mutations, or variations, or errors occur, then these are new instructions, which also arise from within the structure, rather than from without, from the environment.

These inherited structures are exposed to certain pressures, or challenges, or problems: to selection pressures, to environmental challenges, to theoretical problems. In response, variations of the genetically or traditionally inherited instructions are produced³ by methods which are at least partly random. On the genetic level, these are mutations and recombinations⁴ of the coded instruction. On the behavioural level, they are tentative variations and recombinations within the repertoire. On the scientific level, they are new and revolutionary tentative theories. On all three levels we get new tentative trial instructions – or, briefly, tentative trials.

It is important that these tentative trials are changes that originate within the individual structure in a more or less random fashion – on all three levels. The view that they are not due to instruction from without, from the environment, is supported(if only weakly) by the fact that very similar organisms may sometimes respond in very different ways to the same new environmental challenge.

The next stage is that of selecting from the available mutations and variations: those of the new tentative trials which are badly adapted are eliminated. This is the stage of the elimination of error. Only the more or less well adapted trial instructions survive and are inherited in their turn. Thus we may speak of adaptation by ‘the method of trial and error’ – or better, by ‘the method of trial and the elimination of error’. The elimination of error, or of badly adapted trial instructions, is also called ‘natural selection’. It is a kind of ‘negative feedback’ that operates on all three levels.

It is to be noted that in general no equilibrium state of adaptation is reached by any one application of the method of trial and the elimination of error, or by natural selection. First, because no perfect or optimal trial solutions to the problem are likely to be offered. Secondly – and this is more important – because the emergence of new structures, or of new instructions, involves a change in the environmental situation. New elements of the environment may become relevant. And in consequence, new pressures, new challenges and new problems may arise as a result of the structural changes which have arisen from within the organism.

On the genetic level the change may be a mutation of a gene, with a consequent change of an enzyme. Now the network of enzymes forms the more intimate environment of the gene structure. Accordingly, there will be a change in this intimate environment. And with it, new relationships between the organism and the more remote environment may arise – and further, new selection pressures.

The same happens on the behavioural level. For the adoption of a new kind of behaviour can be equated in most cases with the adoption of a new ecological niche. As a consequence, new selection pressures will arise, and new genetic changes.

On the scientific level, the tentative adoption of a new conjecture or theory may solve one or two problems. But it invariably opens up many new problems, for a new revolutionary theory functions exactly like a new and powerful sense organ. If the progress is significant then the new problems will differ from the old problems: the new problems will be on a radically different level of depth. This happened, for example, in relativity. It happened in quantum mechanics. And it is happening right now, most dramatically, in molecular biology. In each of these cases, new horizons of unexpected problems were opened up by the new theory.

This, I suggest, is the way in which science progresses. And our progress can best be gauged by comparing our old problems with our new ones. If the progress that has been made is great, then the new problems will be of a character undreamt-of before. There will be deeper problems, and there will be more of them. The further we progress in knowledge, the more clearly we can discern the vastness of our ignorance.⁵

I will now sum up my thesis.

On all the three levels that I am considering, the genetic, the behavioural, and the scientific levels, we are operating with inherited structures which are passed on by instruction – either through the genetic code or through tradition. On all the three levels, new structures and new instructions arise by trial changes from within the structure: by tentative trials, which are subject to natural selection or the elimination of error.

III

So far I have stressed the similarities in the working of the adaptive mechanism on the three levels. This raises an obvious problem: what about the differences?

The main difference between the genetic and the behavioural level is this. Mutations on the genetic level are not only random but completely ‘blind’ in two senses.⁶ First, they are in no way goal directed. Secondly, the survival of a mutation cannot influence the further mutations, not even the frequencies or probabilities of their occurrence (though admittedly, the survival of a mutation may sometimes determine what kind of mutations may possibly survive in future cases). On the behavioural level, trials are also more or less random. But they are no longer completely ‘blind’ in either of the two senses mentioned. First, they are goal directed. And secondly, animals may learn from the outcome of a trial: they may learn to avoid the type of trial behaviour that has led to a failure. (They may even avoid it in cases in which it could have succeeded.) Similarly, they may also learn from success. And successful behaviour may be repeated, even in cases in which it is not adequate. However, a certain degree of ‘blindness’ is inherent in all trials.⁷

Behavioural adaptation is usually an intensely active process: the animal – especially the young animal at play – and even the plant are actively investigating the environment.⁸

This activity, which is largely genetically programmed, seems to me to mark an important difference between the genetic level and the behavioural level. I may here refer to the experience which the Gestalt psychologists call ‘insight’, an experience that accompanies many behavioural discoveries.⁹ But it must not be overlooked that even a discovery accompanied by ‘insight’ may be mistaken: every trial, even one with ‘insight’, is of the nature of a conjecture or a hypothesis. Köhler’s apes, it will be remembered, sometimes hit with ‘insight’ on what turns out to be a mistaken attempt to solve their problem. And even great mathematicians are sometimes misled by intuition. Thus animals and men have to try out their hypotheses. They have to use the method of trial and of error elimination.

On the other hand I agree with Köhler and Thorpe¹⁰ that the trials of problem-solving animals are in general not completely blind. Only in extreme cases, when the problem which confronts the animal does not yield to the making of hypotheses, will the animal resort to more or less blind and random attempts in order to get out of a disconcerting situation. Yet even in these attempts, goal-directedness is usually discernible, in sharp contrast to the blind randomness of genetic mutations and recombinations.

Another difference between genetic change and adaptive behavioural change is that the former always establishes a rigid and almost invariable genetic structure. The latter, admittedly, leads sometimes also to a fairly rigid behaviour pattern which is dogmatically adhered to – radically so in the case of ‘imprinting’ (Konrad Lorenz) – but in other cases it leads to a flexible pattern which allows for differentiation or modification. For example, it may lead to exploratory behaviour, or to what Pavlov called the ‘freedom reflex’.¹¹

On the scientific level, discoveries are revolutionary and creative. Indeed, a certain creativity may be attributed to all levels, even to the genetic level: new trials, leading to new environments and thus to new selection pressures, create new and revolutionary results on all levels, even though there are strong conservative tendencies built into the various mechanisms of instruction.

Genetic adaptation can of course operate only within the time span of a few generations – at the very least, say, one or two generations. In organisms that replicate very quickly this may be a short time span, and there may be simply no room for behavioural adaptation. More slowly reproducing organisms are compelled to invent behavioural adaptation in order to adjust themselves to quick environmental changes. They thus need a behavioural repertoire, with types of behaviour of greater or lesser latitude or range. The repertoire, and the latitude of the available types of behaviour, can be assumed to be genetically programmed. And since, as indicated, a new type of behaviour may be said to involve the choice of a new environmental niche, new types of behaviour may indeed be genetically creative. For they may in their turn determine new selection pressures, and thereby indirectly decide upon the future evolution of the genetic structure.¹²

On the level of scientific discovery two new aspects emerge. The most important one is that scientific theories can be formulated linguistically, and that they can even be published. Thus they become objects outside of ourselves: objects open to investigation. As a consequence, they are now open to criticism. Thus we can get rid of a badly fitting theory before the adoption of the theory makes us unfit to survive. By criticizing our theories we can let our theories die in our stead. This is of course immensely important.

The...