5

Towards an Evolutionary Theory of Knowledge*

My dear Director, Ladies and Gentlemen,

In 1944 I was travelling with my wife in a bitterly cold bus, returning from a skiing holiday on Mount Cook. The bus stopped in the middle of nowhere, at a snowed-in rural New Zealand post office. To my surprise, I heard my name called, and someone handed me a telegram – the telegram that changed our lives. It was signed F.A. Hayek, and it offered me a Readership at the London School of Economics. The appointment followed in 1945, and in 1949 I was given the title of Professor of Logic and Scientific Method.

Today’s lecture to the Alumni of the School, for which you, Dr Patel, so kindly invited me, is the first public lecture I have ever been asked to give at the L.S.E. I hope, Dr Patel, that you will allow me to regard it, quite informally, as my slightly belated Inaugural Address. It is an occasion to which I have been looking forward eagerly for the last 40 years.

My second request to you, dear Dr Patel, is to allow me to change the wording of the title of this address. When I was urged by the School to produce a title, I had little time to think. I now feel that ‘Evolutionary Epistemology’ sounds pretentious, especially since there exists a less pretentious equivalent. So, please let me change my title to its equivalent and let me call this Inaugural Address ‘Towards an Evolutionary Theory of Knowledge’.

It is my aim, and my problem, at this Inaugural to interest you in work done and, even more, in work yet to be done in the theory of knowledge, by placing it in the wide and exciting context of biological evolution, and to show you that we can learn something new from such an exercise.

I do not start by asking a question such as ‘What is knowledge?’ and, even less, ‘What does “knowledge” mean?’. Instead, my starting point is a very simple proposition – indeed, an almost trivial one – the proposition that animals can know something: that they can have knowledge. For example, a dog may know that his master returns home, on working days, at about 6 p.m., and the behaviour of the dog may give many indications, clear to his friends, that he expects the return of his master at that time. I shall show that, in spite of its triviality, the proposition that animals can know something completely revolutionizes the theory of knowledge as it is still widely taught.

There are, of course, some people who would deny my trivial proposition. They might say, perhaps, that in attributing knowledge to the dog I am merely using a metaphor, and a blatant anthropomorphism. Even biologists interested in the theory of evolution have said such things. My reply is: yes, it is a blatant anthropomorphism, but it is not merely a metaphor. And the anthropomorphism is a very useful one: one that is indispensable in any theory of evolution. If you speak of the nose of the dog, or of his legs, then these are also anthropomorphisms, even though we take for granted that it is simply true that the dog has a nose – if somewhat different from our human nose.

Now, if you are interested in the theory of evolution, you will find that part of it is the important theory of homology, and that the dog’s nose and my nose are homologous, which means that they are both inherited from a distant common ancestor. Without this hypothetical theory of homology, evolutionary theory could not exist. Obviously, the theory of homology is a highly speculative and very successful hypothesis, and one that all evolutionists adopt. My attribution of knowledge to the dog is therefore an anthropomorphism, but it is not just a metaphor. Rather, it implies the hypothesis that some organ of the dog, in this case presumably the brain, has a function that corresponds not only in some vague sense to the biological function of human knowledge, but is homologous with it.

Please note that the things that may be homologous are, in the original sense, organs. But they may also be functions of organs. And they may also be procedures. Even behaviour may be hypothesized to be homologous in the evolutionary sense; for example, courtship behaviour, especially ritualized courtship behaviour. That this is indeed homologous in the hereditary or genetic sense between, say, different but closely related species of birds, is very convincing. That it is homologous between ourselves and some species of fish seems highly dubious, yet it remains a serious hypothesis. Of course, the possession of a mouth or of a brain in fish is most convincingly homologous with our possession of the corresponding organs: it is quite convincing that they are genetically derived from the organs of a common ancestor.

I hope the central importance of the theory of homology for the theory of evolution has become sufficiently clear for my purpose; that is, for defending the existence of animal knowledge, not as a mere metaphor, but as a serious evolutionary hypothesis.

This hypothesis does in no way imply that animals will be aware of their knowledge; and it thereby draws attention to the fact that we ourselves possess knowledge of which we are not aware, not conscious.

Our own unconscious knowledge has often the character of unconscious expectations, and sometimes we may become conscious of having had an expectation of this kind when it turns out to have been mistaken.

An example of this is an experience that I had several times in my long career: in going down some stairs and reaching the last step, I almost fell, and became aware of the fact that I had unconsciously expected that there was one more step, or one fewer step, than actually existed.

This led me to the following formulation: when we are surprised by some happening, the surprise is usually due to an unconscious expectation that something else would happen.

I shall now try to give you a list of 19 interesting conclusions that we can draw, and partly have drawn (although so far unconsciously) from our trivial proposition that animals can know something.

  1. Knowledge has often the character of expectations.
  2. Expectations have usually the character of hypotheses, of conjectural or hypothetical knowledge: they are uncertain. And those who expect, or who know, may be quite unaware of this uncertainty. In the example of the dog, the dog may die without ever having been disappointed in this expectation of his master’s timely return: but we know that the timely return was never certain, and that the dog’s expectation was a very risky hypothesis. (After all, there might have been a railway strike.) So we can say:
  3. Most kinds of knowledge, whether of men or animals, are hypothetical or conjectural; especially the very ordinary kind of knowledge, just described as having the character of an expectation; say, the expectation, supported by a printed official timetable, that at 5.48 p.m. there will arrive a train from London. (In some libraries, embittered or merely discerning readers returned timetables to the shelves headed ‘Fiction’.)
  4. In spite of its uncertainty, of its hypothetical character, much of our knowledge will be objectively true: it will correspond to the objective facts. Otherwise we could hardly have survived as a species.
  5. We therefore must clearly distinguish between the truth of an expectation or a hypothesis and its certainty; and therefore between two ideas: the idea of truth and the idea of certainty; or, as we may also say, between truth and certain truth – for example, a mathematically demonstrable truth.
  6. There is much truth in much of our knowledge, but little certainty. We must approach our hypotheses critically; we must test them as severely as we can, in order to find out whether they cannot be shown to be false after all.
  7. Truth is objective: it is correspondence to the facts.
  8. Certainty is rarely objective: it is usually no more than a strong feeling of trust, of conviction, although based on insufficient knowledge. Such feelings are dangerous since they are seldom well-founded. Strong feelings of conviction make dogmatists of us. They may even turn us into hysterical fanatics who try to convince themselves of a certainty which they unconsciously know is not available.

Before proceeding with this list to the next point (to point 9), I wish to digress for a moment. For I want to say a few words against the widespread doctrine of sociological relativism, often unconsciously held, especially by sociologists who study the ways of scientists and who think that they thereby study science and scientific knowledge. Many of these sociologists do not believe in objective truth, but think of truth as a sociological concept. Even a former scientist such as the late Michael Polanyi thought that truth was what the experts believe to be true – or, at least, the great majority of the experts. But in all sciences, the experts are sometimes mistaken. Whenever there is a breakthrough, a really important new discovery, this means that the experts have been proved wrong, and that the facts, the objective facts, were different from what the experts expected them to be. (Admittedly, a breakthrough is not a frequent event.)

I do not know of any creative scientist who has made no mistakes; and here I am thinking of the very greatest – Galileo, Kepler, Newton, Einstein, Darwin, Mendel, Pasteur, Koch, Crick, and even Hilbert and Gödel. Not only are all animals fallible, but also all men. So there are experts, but no authorities – a fact that has not yet established itself sufficiently. Of course, we are all very conscious of the fact that we ought not to make mistakes, and we all try very hard. (Perhaps Gödel tried harder than anyone else.) But still, we are fallible animals – fallible mortals, as the early Greek philosophers would have it: only the Gods can know, we mortals can only opine and guess.

I guess, indeed, that it is the suppressed sense of our own fallibility that is responsible for our despicable tendency to form cliques and to go along with whatever seems to be fashionable: that makes so many of us howl with the wolves. All this is human weakness, which means it ought not to exist. But it does exist, of course; it is even to be found among some scientists. And as it exists, we ought to combat it; first in ourselves, and then, perhaps, in others. For I hold that science ought to strive for objective truth, for truth that depends only on the facts; on truth that is above human authority and above arbitration, and certainly above scientific fashions. Some sociologists fail to understand that this objectivity is a possibility towards which science (and therefore scientists) should aim. Yet science has aimed at truth for at least 2,500 years.

But let us now return to our evolutionary theory of knowledge, to our trivial starting proposition that animals can know something, and to our list of results obtained from, or suggested by, this trivial proposition.

9.  Can only animals know? Why not plants? Obviously, in the biological and evolutionary sense in which I speak of knowledge, not only animals and men have expectations and therefore (unconscious) knowledge, but also plants; and, indeed, all organisms.

10.  Trees know that they may find much-needed water by pushing their roots into deeper layers of the earth; and they know (or the tall ones do) how to grow up vertically. Flowering plants know that warmer days are about to arrive, and they know how and when to open their flowers, and to close them – according to sensed changes in radiation intensity or in temperature. Thus they have something like sensations or perceptions to which they respond, and something like sense organs. And they know, for example, how to attract bees and other insects.

11.  An apple tree that sheds its fruit or its leaves offers a beautiful example for one of the central points of our investigations. The tree is adapted to the seasonal changes of the year. Its structure of inbuilt biochemical processes keeps them in step with these law-like and long-term environmental changes. It expects these changes: it is attuned to them, it has foreknowledge of them. (Trees, especially tall trees, are also finely adjusted to such invariants as gravitational forces.) But the tree responds, in an appropriate and well-adapted manner, also to short-term changes and forces, and even to momentary events in the environment. Chemical changes in the stems of the apples and of the leaves prepare them for falling; but they usually fall in response to the momentary pull of the wind: the ability to respond appropriately to short-term or even momentary events or changes in the environment is closely analogous to the ability of an animal to respond to short-term perceptions, to sense experiences.

12.  The distinction between adaptation to, or (unconscious) knowledge of, law-like and long-term environmental conditions, such as gravity and the cycle of the changing seasons, on the one side, and adaptation to, or knowledge of, environmental short-term changes and events, on the other side, is of the greatest interest. While short-term events occur in the life of the individual organisms, the long-term and law-like environmental conditions are such that adaptation to them must have been at work throughout the evolution of countless generations. And if we look more closely at short-term adaptation, at the knowledge of, and the response to, environmental short-term events, then we see that the ability of the individual organism to respond appropriately to short-term events (such as a particular pull of the wind or, in the animal kingdom, the appearance of a foe) is also a long-term adaptation, and also the work of adaptation going on through countless generations.

13.  A grazing flock of wild geese is approached by a fox. One of them sees the fox and gives the alarm. It is precisely a situation like this – a short-term event – in which the eyes of an animal can save its life. The animal’s ability to respond appropriately depends on its possession of eyes – of sense organs – adapted to an environment in which daylight is periodically available (analogous to the change of the seasons or to the constant availability of the directional pull of gravity, used by the tree to find the direction of its growth); in which deadly foes are threatening (that is, in which crucially significant objects exist for visual identification); and in which escape is possible if these foes are identified at a sufficient distance.

14.  All this adaptation is of the nature of long-term knowledge about the environment. And a little thinking will make it clear that without this kind of adaptation, without this kind of knowledge of law-like regularities, sense organs like the eyes would be use less. Thus we must conclude that the eyes could never have evolved without an unconscious and very rich knowledge about long-term environmental conditions. This knowledge, no doubt, evolved together with the eyes and with their use. Yet at every step it must have somehow preceded the evolution of the sense organ, and of its use. For the knowledge of the pre-conditions of its use are built into the organ.

15.  Philosophers and even scientists often assume that all our know ledge stems from our senses, the ‘sense data’ which our senses deliver to us. They believe (as did, for example, the famous theorist of knowledge, Rudolf Carnap) that the question ‘How do you know?’ is in every case equivalent to the question ‘What are the observations that entitle you to your assertion?’. But seen from a biological point of view, this kind of approach is a colossal mistake. For our senses to tell us anything, we must have prior knowledge. In order to be able to see a thing, we must know what ‘things’ are: that they can be located within some space; that some of them can move while others cannot; that some of them are of immediate importance to us, and therefore are noticeable and will be noticed, while others, less important, will never penetrate into our consciousness: they may not even be unconsciously noticed, and they may simply leave no trace whatever upon our biological apparatus. For this apparatus is highly active and selective, and it actively selects only what is at the moment of biological importance. But in order to do so, it must be able to use adaptation, expectation: prior knowledge of the situation must be available, including its possibly significant elements. This prior knowledge cannot, in turn, be the result of observation; it must, rather, be the result of an evolution by trial and error. Thus the eye itself is not the result of observation, but the result of evolution by trial and error, of adaptation, of non-observational long-term knowledge. And it is the result of such knowledge, derived not from short-term observation, but from adaptation to the environment and to such situations as constitute the problems to be solved in the task of living; situations that make our organs, among them our sense organs, significant instruments in the moment-by-moment task of living.

16.  I hope I have been able to give you some idea of the importance of the distinction between long-term and short-term adaptation, long-term and short-term knowledge, and of the fundamental character of long-term knowledge: of the fact that it must always precede short-term or observational knowledge, and of the impossibility that long-term knowledge can be obtained from short-term knowledge alone. Also, I hope that I have been able to show you that both kinds of knowledge are hypothetical: both are conjectural knowledge, although in a different way. (Our know ledge, or a tree’s knowledge, of gravity will turn out to be gravely mistaken if we, or the tree, are placed in a no-longer accelerated rocket or ballistic missile.) Long-term conditions (and the knowledge of them) may be subject to revision; and an instance of short-term knowledge may turn out to be a misinterpretation.

And so we come to the decisive and perhaps most general proposition, valid for all organisms including man, even though it may perhaps not cover all forms of human knowledge.

17.  All adaptations to environmental and to internal regularities, to long-term situations and to short-term situations, are kinds of knowledge – kinds of knowledge whose great importance we can learn from evolutionary biology. There are, perhaps, some forms of human knowledge that are not, or not obviously, forms of adaptations, or of attempted adaptations. But, roughly speaking, almost all forms of knowledge of an organism, from the unicellular amoeba to Albert Einstein, serve the organism to adapt itself to its actual tasks, or to tasks that may turn up in the future.

18.  Life can only exist, and can only survive, if it is in some degree adapted to its environment. We can thus say that knowledge – primitive knowledge, of course – is as old as life. It originated together with the origin of pre-cellular life, more than three thousand eight hundred million years ago. (Unicellular life came into existence not much later.) This happened soon after the earth cooled down sufficiently to allow some of the water in its atmosphere to liquify. Until then, water had existed only in the form of steam or clouds, but now hot liquid water began to collect in rocky basins, big or small, forming the first rivers, lakes, and seas.

19.  Thus, the origin and the evolution of knowledge may be said to coincide with the origin and the evolution of life, and to be closely linked with the origin and evolution of our planet earth. Evolutionary theory links knowledge, and with it ourselves, with the cosmos; and so the problem of knowledge becomes a problem of cosmology.

Here I end my list of some of the conclusions that can be drawn from the proposition that animals can have knowledge.

I may perhaps very briefly refer to my book, The Logic of Scientific Discovery, first published in German in 1934, and published in English for the first time 25 years later, in 1959. In the Preface to the first English edition I wrote of the fascination of the problem of cosmology, and I said of it: ‘It is the problem of understanding the world – including ourselves, and our knowledge, as part of the world.’

This is how I still see the setting of the evolutionary theory of knowledge.

When our solar system evolved and the earth had cooled sufficiently, there must have developed conditions in some place on earth that were favourable to the origin and to the evolution of life. Unicellular bacterial life quickly spread all over the earth. But those originally so very favourable local conditions could hardly have prevailed over many different geographical regions; so it seems that life must have had a struggle. Yet, in a comparatively short time, many different bacterial forms evolved that were adapted to very different environmental conditions.

Such, it appears, are the facts. Of course, they are far from certain: they are hypothetical interpretations of some geological findings. But if they are even approximately correct, they refute, for two reasons, the at present most widely accepted theory of the origin of life: the so-called ‘soup theory’ or ‘broth theory’.

First reason: as the leading defenders of the soup theory assert, this theory demands a low temperature for the soup or broth in which the macromolecules develop, and later join up to form the first organism. The reason for their assertion is that, if the temperature is not very low (the broth must be considerably supercooled below 0°C), the macro-molecules quickly decompose, instead of joining up.

But what we know of the earth in those days indicates that no such cool places existed. The surface of the earth, and even more the seas, were much hotter than today; and even today a watering place supercooled below 0°C would not easily be available, except perhaps near the North Pole or within a refrigerating plant.

Second reason: the theory that the macromolecules in the soup have joined up, and so have organized themselves into a living organism, is improbable in the extreme. The improbability is so great that one would have to assume an extremely long time-span in order to make the event a little less improbable; a time-span far longer, indeed, than the calculated time for which the cosmos has existed. So say even some of the most prominent defenders of the soup theory.

This constitutes a sound refutation of the theory in question, for as geologists found, the time-span between the formation of (boiling hot) liquid water and the origin of life is surprisingly short, and far too short to allow for an event of such extreme improbability to happen – even if the high temperature were acceptable to the soup theorists.

These two reasons constitute two refutations of the at present ruling soup theory of the origin of life. (There are many other refutations.) It is therefore fortunate that an alternative theory was published in 1988, a theory which is not beset by these or by similar difficulties. It assumes only the existence of such simple inorganic micromolecules as those of water, iron, carbon dioxide and hydrosulphide. No organic macromolecules are assumed to be present before the first metabolic cycles start and, with them, the chemical self-organizing of life. The new theory shows in detail how organic molecules (such as sugar) may evolve in time, perhaps deep down in the sea, bound to the surface of pyrite crystals, rather than in a solution. The anaerobic formation of the pyrite crystal creates the free chemical energy needed for the chemical processes – especially carbon fixation – that constitute the earliest form of pre-cellular life.

This new theory of the origin of life has been developed by its author in considerable detail, and it seems to be very successful: it explains many biochemical pathways. It is readily testable by experiments. But its greatest strength is that it can explain many biochemical facts that were unexplained before.

Günter Wächtershäuser, the author of this new biochemical theory, has also provided another biochemical theory – one that is of still greater relevance to the evolutionary theory of knowledge and to the problems we are discussing here. He has produced a biochemical theory of the origin of the first light-sensitive organ; that is, of the earliest evolutionary predecessor of our eyes. Since our eyes are our most important sense organs, this result is of great interest to our discussion.

The main result is this. It is known that some early unicellular micro-organism, presumably a bacterium, invented a revolutionary electro-chemical method of transforming sunlight into chemical energy: a method of using sunlight as a food, a method of feeding on sunlight. It was a bold and, indeed, a dangerous invention since, as we all know, too much sunlight – and especially the ultraviolet portion of sunlight – can kill. So with this invention, several problems arose for the cell (that previously may have been living deep down in the dark sea). They were pointed out by Wächtershäuser.

The first problem was to find out where sunlight is and, by using this information, to move towards it. This problem was solved by the first formation of a sense organ with the function of our eyes, a sense organ chemically linked to some already existing executive mechanism for the movement of the cell.

A second problem that arose was that of avoiding the danger of getting too much ultraviolet sunlight: of moving away in time, before suffering damage, towards some shade, presumably towards a deeper layer of sea water.

Thus in the evolution of the eye, even its earliest predecessor had to become a controller of the movement of the cell. It had to become part of the feeding mechanism of the cell, and part of its security movements: its mechanism for evading danger. The eye helped to avoid radiation damage to the cell – to anticipate danger. Even its very first function assumed prior knowledge of environmental states and possibilities.

Wächtershäuser pointed out that the revolutionary invention of feeding on sunlight would have been self-destructive without that other, that essentially protective invention of moving out of sunlight (and presumably also into it) becoming part of the invention of the early eye and its link to the apparatus of motion. And so the problem arises in his theory: how could these two great inventions come together?

If we take an interest in biological evolution, especially in early evolution, then we must be constantly aware of the fact that life is, basically, a chemical process. It was Heraclitus, half a millennium before the birth of Christ, who said that life was a process, like fire; and, indeed, our life is something like a complex process of chemical oxidation. In the earliest stages of its evolution, when free oxygen was not available, sulphur played its role instead. As you may know, it was the bacterial invention of using sunlight as food – which, incidentally, led later to the self-invention of the kingdom of plants – that produced the greatest of all life-induced revolutions in the history of our environment: it introduced oxygen into the atmosphere. And so it created the air that we know, that makes our life possible, life as we know it: our breathing, our lungs, our fire (within and without). Heraclitus was right: we are not things, but flames. Or a little more prosaically, we are, like all cells, processes of metabolism; nets of chemical processes, of highly active (energy-coupled) chemical pathways.

The great Belgian biochemist Marcel Florkin (1900–1979) was one of the first to see clearly that the evolution of life, or organisms, is an evolution of nets of chemical pathways. The net of pathways that constitutes a cell at some given period of time may make it possible for some new pathway, often just a slight variation, to graft itself upon the then extant system. The new pathway may have been impossible without some of the chemical compounds produced by the old system of pathways. As Florkin pointed out, the net of chemical pathways of an extant cell often still retains, as part of this net, the archaic pathways of some billion years ago that made the later graftings possible. This, as Florkin pointed out, is analogous to the way in which anatomic pathways of the anatomic construction of the developing embryo may still retain those of its archaic ancestors of, say, some hundred million years ago. Thus the extant pathways of the metabolism may reveal some of its evolutionary history; a situation analogous to the so-called ‘biogenetic law’ of Fritz Müller and Ernst Haeckel.

It is within this setting of Florkin’s ideas that Wächtershäuser was able to explain the riddle of the coincidence of the two great inventions: the invention of feeding on sunlight and the invention of light-sensitivity, of the archaic eye. The explanation is that both inventions are chemically very closely related: one of the pathways producing the machinery for feeding on sunlight and the pathway for producing the visual apparatus are structurally connected.

We may speculate that the invention resulted from the general tendency of organisms to explore their environment; in this case, by rising towards the surface layer of the sea. Presumably, the one or the other of these bacteria had, accidentally, evolved to a stage that made it possible to invent both of these new, chemically connected, grafts. Other organisms will also have boldly ventured near the surface, only to be destroyed by sunlight. But one (or perhaps a few) had the right chemical outfit, and survived. It was able to turn the surface layer of the sea into the richest feeding ground for its offspring; and its offspring exuded those huge amounts of oxygen that transformed the atmosphere.

We see that the Darwinian trial-and-error method turns out to be a method of the (partly accidental) variation and accretion of chemical pathways. In extant cells the pathways are controlled, step by chemical step, by enzymes which are highly specific chemical catalysts, that is, chemical means of speeding up specific chemical steps; and the enzymes are partly controlled by the genes. But a genetic mutation, and the synthesis of a new enzyme, will not lead to a new step in the net of pathways unless the new enzyme accidentally fits into the extant net; it is always the existing structure of the net of pathways that determines what new variations or accretions are possible. It is the existing net that contains the potentiality for new inventions; and a fitting enzyme, if not yet available, may become available soon. In some cases it may decide the future evolution of the species by determining which of the potential steps will be realized. (One step may lead to a slow evolution while another step may lead to a cascade of further steps. Both steps will be equally Darwinian, since they are subject to selection; their apparently different speeds are likely to be explicable in chemical terms.)

I will now try to list some of the lessons to be learnt for the theory of knowledge from all that has been said so far.

The main lesson to be drawn may be formulated, perhaps with some little exaggeration, as follows. Even in the most primitive organisms, and even in the most primitive cases of sensitivity, everything depends upon the organism itself: upon its structure, its state, its activity. More especially, even if we confine our discussion for the moment to the problem of obtaining some knowledge from the environment with the help of the organism’s sensitivity to the momentary state of its environment, even then everything will depend on the organism’s own state, its long-term structure, its state of preparation for solving its problems, its state of activity.

In order to develop more fully what I have just said only very roughly, it is useful to introduce here a variant of the Kantian terminology of a priori and a posteriori. In Kant, knowledge a priori means knowledge that we possess prior to sense-observation; and knowledge a posteriori means knowledge we possess posterior to sense-observation, or after observation; and I will use the terms ‘a priori’ and ‘a posteriori’ only in this temporal or historical sense. (Kant himself uses his term a priori to mean, in addition, knowledge that is not merely prior to observation but also ‘a priori valid’; by which he means necessarily or certainly true. Of course, I shall not follow him in this since I am stressing the uncertain and conjectural character of our knowledge.) So I shall use the term ‘a priori’ to characterize that kind of knowledge – of fallible or conjectural knowledge – which an organism has prior to sense experience; roughly speaking, it is inborn knowledge. And I shall use the term ‘a posteriori’ for knowledge that is obtained with the help of the sensitivity of the organism to momentary changes in the state of its environment.

Using this Kantian terminology with the modifications I have just indicated, we can now say that Kant’s own position – highly revolutionary at the time – is this.

(A)Most knowledge of detail, of the momentary state of our surroundings, is a posteriori.

(B)But such a posteriori knowledge is impossible without the a priori knowledge that we somehow must possess before we can acquire observational or a posteriori knowledge: without it, what our senses tell us can make no sense. We must first have established an overall frame of reference, or else there will be no context available to make sense of our sensations.

(C)This a priori knowledge contains, especially, knowledge of the structure of space and time (of space and time relations), and of causality (of causal relations).

I think that, in all these points, Kant is right. (Incidentally, I also think that he had hardly a real successor in this except perhaps Schopenhauer.) In my opinion, Kant anticipated the most important results of the evolutionary theory of knowledge.

But I am going much further than Kant. I think that, say, 99 per cent of the knowledge of all organisms is inborn and incorporated in our biochemical constitution. And I think that 99 per cent of the knowledge taken by Kant to be a posteriori and to be ‘data’ that are ‘given’ to us through our senses is, in fact, not a posteriori, but a priori. For our senses can serve us (as Kant himself saw) only with yes-and-no answers to our own questions; questions that we conceive, and ask, a priori; and questions that sometimes are very elaborate. Moreover, even the yes-and-no answers of the senses have to be interpreted by us – interpreted in the light of our a priori preconceived ideas. And, of course, they are often misinterpreted.

Thus, all our knowledge is hypothetical. It is an adaptation to a partly unknown environment. It is often successful and often unsuccessful, the result of anticipatory trials and of unavoidable errors, and of error elimination. Some of the errors that have entered the inheritable constitution of an organism are eliminated by eliminating their bearer; that is, the individual organism. But some errors escape, and this is one reason why we are all fallible: our adaptation to the environment is never optimal, and it is always imperfect. A frog is constituted a priori so that it can see its prey – a fly – only when the fly moves. When the fly sits still, the frog cannot see it, even if it is very close: the frog’s adaptation is imperfect.

Organisms and their organs incorporate expectations about their environment; and expectations – as we have seen – are homologous with our theories: as homologous as is the nose of my dog with my nose. So I suggest the hypothesis that adaptations and expectations are homologous even with scientific theories (and vice versa theories with adaptations and expectations). Theories may often contain evaluations. A unicellular organism’s sensitivity to light, to heat, and to acidity may help it to escape from too little or too much of any of these. The organism’s structure may incorporate the theory: ‘the surrounding water can be dangerous: it may be too hot or too cold, and it may be too much or too little acid.’ Clearly, such evaluations can evolve only if the organism is able to take action; for example, by moving away if it anticipates danger from these environmental states. Problems, values, and activity all evolve together.

I have said something about the origin of the archaic eye; and we can now say that its invention incorporates new discoveries, new theories, new knowledge about the environment and also the possibility of new values. For the first bacterium that not only achieved the new chemical synthesis, but went with it to a layer near the surface of the sea and survived, after millions of its brothers had succumbed, proved by its survival that it had solved a problem of adaptation; and in solving a problem, it introduced a new theory about new values. The invention was incorporated in the structure of the organism; in new, inheritable knowledge and therefore in new a priori knowledge.

Within this great revolution, the momentary signals conveyed by the eye to the organism were as such comparatively unimportant. They became important only together with the state of the organism; say, its need for food. The eye certainly helped the organism to feed on sunrays without destruction. But signals as such that, by homology, we might call the ‘data’ need not even be noticed. What leads to action are the interpreted signals (and interpretation is part of the action): signals plus the new theoretical evaluation of advantages and of danger; not objective ‘data’, but enticements and warnings acquired and interpreted with the help of the a priori structure of the organism.

We have seen that, even in bacteria, theories or hypotheses come before the signals, the ‘sensations’. I need hardly stress that, especially in science, hypotheses come before what some scientists still call the ‘data’; misleadingly, because they are not given to us, but actively (and sometimes at great peril) sought and acquired by us.

Observations (or ‘data’) may lead in science to the abandonment of a scientific theory and thereby induce some of us to think up a new tentative theory – a new trial. But the new theory is our product, our thought, our invention; and a new theory is only rarely thought up by more than a few people, even when there are many who agree on the refutation of the old theory. The few are those who see the new problem. Seeing a new problem may well be the most difficult step in creating a new theory.

The invention of the eye is thus an invention of new theoretical priori knowledge, of an adaptation to the environment. It was from the first an adaptation to a long-term environmental structure: to the existence of potentially edible sunlight; it thus incorporates knowledge of this environmental structure. It is theoretical knowledge of a high degree of universality, almost like Kantian knowledge of space and time. It creates the possibility of momentary ‘observation’ or, more precisely, of the adaptation to a momentary situation of the environment. It may induce in the organism states of enticement or of repulsion, and it may make possible the release of prepared actions upon the environment. Thus, the invention of a highly universal theory (in this case the invention of a sense organ) may come before the observation (the use of the sense organ): it makes observation possible and introduces it into the set of actions that are available to the organism. And so it is itself an adaptation, found by trial and error. Theories (scientific or otherwise) are trials, inventions; they are not the results of many observations; they are not derived from many data.

Clearly, the first invention of the eye was a great achievement. Much of it has been preserved, and much evolved further. And yet, we – in common with all animals – have forgotten the knowledge that sunlight is edible, and how to eat it. And to this day we have not fully regained this knowledge.

Ladies and Gentlemen, I am one of those who love science and who think that science is enlightened common sense. I even think it is not much more than enlightened bacterial common sense! This is a view that, admittedly, clashes with common sense; but I hope that I have shown in this lecture that it need not clash with enlightened common sense. I have, I believe, refuted classical empiricism – the bucket theory of the mind that says that we obtain knowledge just by opening our eyes and letting the sense-given or god-given ‘data’ stream into a brain that will digest them.

Christopher Isherwood expressed this view by the title of his book, I am a Camera. But when he chose this title he forgot that even a camera must have an a priori built-in constitution; that there are primitive cameras and surprisingly evolved ones; and that in a failing light in which a bad camera records nothing, a good camera may produce a perfect picture, giving us all we want of it. It is better adapted to the environment, and also to our needs, that is, to our problems: it incorporates certain values that we have evolved while working on the evolution of the camera. But a lot of things it cannot do; for example, it cannot improve itself; and it cannot invent a new important problem, or a new tentative solution.

All organisms are problem finders and problem solvers. And all problem solving involves evaluations and, with these, values. Only with life do problems and values enter the world. And I do not believe that computers will ever invent important new problems, or new values.

Of these new values that we have invented, two seem to me the most important for the evolution of knowledge: a self-critical attitude – a value that we should always teach ourselves to live up to; and truth – a value that we should always seek our theories to live up to.

The first of these values, a self-critical attitude, first enters the world with certain objective products of life, such as spiders’ webs, birds’ nests, or beaver dams: products that can be repaired or improved. The emergence of the self-critical attitude is the beginning of something even more important: of the critical approach, an approach that is critical in the interest of objective truth. (I hope that it was the critical approach that inspired the Founders of the London School of Economics to choose the dam-repairing beaver for its coat of arms.)

Both of these values, the critical approach and objective truth, enter our world only with the human language, the first and most important product of the human mind. Language makes it possible to consider our theories critically: to look at them as if they were external objects, as if they belonged to the world outside of ourselves which we share with others. Theories become objects of criticism, like the beaver dam. And we can try to repair them in the light of that most important value: correspondence to the facts – truth.

I have often said that from the amoeba to Einstein there is only one step. Both work with the method of trial and error. The amoeba must hate error, for it dies when it errs. But Einstein knows that we can learn only from our mistakes, and he spares no effort to make new trials in order to detect new errors, and to eliminate them from our theories. The step that the amoeba cannot take, but Einstein can, is to achieve a critical, a self-critical attitude, a critical approach. It is the greatest of the virtues that the invention of the human language puts within our grasp. I believe that it will make peace possible.

Let me end with a quotation from Albrecht Dürer, an artist and a scientist:

But I shall let the little I have learnt go forth into the day in order that someone better than I may guess the truth, and in his work may prove and rebuke my error. At this I shall rejoice that I was yet the means whereby this truth has come to light.

* Lecture delivered at the London School of Economics, 9 June 1989, before the Alumni of the School; the Director of the School, Dr I. Patel, was in the chair. Published previously by Thoemmes Press, Bristol, as part of A World of Propensities, and reproduced here with kind permission.