You are here

Part IV. A Rationalist's Outlook

Chapter XIII: Cosmology

1 In the first two books of this study we examined the views of reason and revelation held by Catholics and by one important line of Protestant theologians. Neither of these traditions could be described as passing the examination with honours. In each of them much error and incoherence came to light.

Why have they persisted so long in spite of Such defects? In the last two chapters we have had a partial answer to this question. They have persisted because religion is a fundamental activity or attitude, always present in some form and degree, and these two great traditions gave it an expression which, in view of man's limited knowledge, was highly plausible. They persisted, again, because both developed an ethics of belief which made doubt or denial, even when reached by disinterested inquiry, a sin. They persisted, once more, because in both there was a large freightage of the sort of myth that was congenial to men's feeling. With the growth of knowledge, the relaxation of the ethics of belief, and the increasing freedom of intellect in religion, the massive edifice of dogma in which men had lived for centuries crumbled slowly away.

Unfortunately, the dissecting intellect is a better critic than architect. Though most philosophers would probably agree that the traditional theologies are no longer credible, they disagree on what should replace them. This is not, as many would like to think, an indictment of reason itself. Reason, in the present use of the term, means simply the disinterested application of our cognitive faculties to the problems of belief, and though this application has been protested in both the Christian traditions, the protest, as we have had full occasion to see, is not very cogent. If ‘reason’ pronounces diverse judgements, it is not because the ideal of validity differs from man to man; if it were, they could not be sure there was any point even in arguing with each other, for one man's proof might be another man's irrelevancy. The diversity of men's conclusions is due rather to the fact, admitted by all the great rationalists from Plato down, that man is brokenly and imperfectly rational. His thought is seldom picked up by the logic inherent in the nature of things and carried straight to a conclusion; his nature has long roots in an animal past which distorts his thought continually through his non-rational impulses and desires. How deeply his thought is affected by these sub-rational pushes and pulls is clear from the millennia of struggle it has taken for his reason to break free from myth into some measure of autonomy. The struggle, of course, still goes on.

2 When the proponent of revelation turns on the proponent of reason and says, ‘You have done your best to drive me from my ancestral house; now show me the more stately mansion your “reason” has prepared for me,’ the rationalist is likely to be embarrassed. There are all too many mansions in his province, many of them obviously disreputable and none of them bearing the plaque of official approval. The revelationist will complain that he is being invited to exchange certainty for confessed uncertainty, and a clearly structured universe, in which his place was clear and important, for a bewildering twilight world in which he is secure of little except his own insecurity. But the rationalist need not apologise. He is not inviting his neighbour to give up any certainty to which either logic or ethics entitles him. The rationalist has seen, or thinks he has seen, that the system of dogma, if accepted as revealed and certain, leads straight to self-contradiction, and therefore cannot be what it claims to be. And he is not offering his twilight world as one that is more comfortable or pleasing to live in than others, though he is personally grateful for having escaped the horror of the Lutheran world, in which most men are predestined to unremitting torment for what they are predestined to do. Whether a world view is pleasing or not has nothing to do with its truth, and it is truth that the rationalist above all wants. He wants to see things as they are. Whoever attempts this must admit that even the men of highest ability who have sought to make sense of the world have differed in their conclusions about it, and also that, considering the complex and shifting evidence, it would be astonishing if they did not. The chance that any one of them, even one as able and as devoted to truth as Spinoza or Hume, should arrive at any ultimate truth about the world which could stand in the form he gives to it is very small, and the chance of a less able inquirer smaller still.

But every negation has a positive base. The long series of negations with which historic dogmas have been greeted in this book could only have sprung from positive convictions of some sort, whether or not they were realised at the time; and it is only fair to opponents that, before taking leave of them, the writer should lay his own cards on the table. To set out these convictions with a proper defence would be to write a system of metaphysics, which is no task for the last pages of an overlong book. But a sketch may be better than nothing. It will at least give to outraged persons who have been involuntary targets a target to aim at in return; and it may give a glimpse to the curious of the sort of world one rationalist lives in.


3 I start with the law of contradiction: a statement and its contradictory cannot both be true. Here at the first step I hold we are committed to a world that is rational to this extent, that nothing in it is either self-contradictory or contradictory of anything else. The real world is a coherent world, in the sense that it is at least self-consistent.

It may be objected, of course, that the world is full of contradictions. Einstein contradicted Newton, Newton contradicted Leibniz, Leibniz contradicted Spinoza, and Spinoza contradicted Descartes. What the Kremlin wants in the Near East is a contradiction of what the White House wants. Even self-contradiction is notoriously common. Did not Russell hold that moral judgements fell outside the province of reason, and at the same time appeal to reason in condemning President Johnson? According to Marxism, progress consists in the constant overcoming of contradictions; the contradiction between the workers and the bourgeoisie, for example, was overcome by the October revolution.

Objections of this kind are mere misunderstandings. The law of contradiction does not say anything so absurd as that no statements are ever contradictory of others. It says only that if they are thus contradictory, they cannot both be true. And for two persons or nations or classes to have conflicting desires or interests is no contradiction. The fact that John wants Susan is not a denial of the fact that James wants her also, and is quite consistent with it. Conflicts of bodies or wills are common; inconsistencies between assertions or beliefs are common; inconsistencies between truths or facts, we should hold, do not and cannot occur.

Among those who have freed themselves from such mistakes, however, there are persons who would interpret the laws of logic in a way that would make a rationalist metaphysic impossible. Something must therefore be said about them, although it may involve a little more technicality than will be to everyone's taste.

4(1) First, there are those who say that the law of contradiction is merely a linguistic rule, a statement that we propose to use language in one way rather than another. When we say, for example, that the same surface cannot at once be black and non-black, what we are saying is that we propose not to use the term ‘non-black’ of any surface we call black. I agree that such a proposal may be implied, and that no sane person would want to use these words otherwise. But why? If we are free to use them otherwise, why do we never take advantage of that freedom and use ‘black’ and ‘non-black’ of the same surface? The answer is clear enough. If we did, our language would not answer to the facts; we should be applying two adjectives to the same thing when it was plain that if one applied the other would not. Linguisticism has recently gone so far as to suggest that language makes the world in its own image. This has things the wrong way about. The truth is rather that language is an attempt to reflect the world. Instead of black being made exclusive of non-black by our usage, the exclusion is there in nature, and our usage, if it is sensible, will respect and reflect this fact.

(2) Sometimes the interpretation takes a slightly different turn. The law of contradiction is not now a convention of language but a convention of thought. It is true that we normally choose to conform to it, but our conformity is a matter of choice, not of necessity; the rules of logic, Carnap once maintained, could be chosen ‘quite arbitrarily’.1 But those who take this position never hold to it in practice; nor can they. When they have shown by what they take to be sound reasoning that these rules are arbitrary, they consider that they have made out their case, and that anyone who, accepting their evidence, rejects their conclusion is unreasonable. But if the rules of inference are really arbitrary, why should I not accept the evidence and reject the conclusion? Indeed, why should I not say, since no logical rule now binds me, that I both accept the conclusion and reject it? Since the law of contradiction is a convention only, I choose to say that the statement ‘the law is conventional’ does not exclude the statement ‘the law is not conventional’, and hence that my opponent and I, though contradicting each other, are also in agreement. Of course no conventionalist would accept this nonsense. He never means that his statement of the arbitrariness of logic, or his argument for it, is itself merely arbitrary. They at least are really valid, and anyone who rejects them is missing the truth about logic. The theory that logic is conventional is not itself a mere convention. But if so, the theory has been abandoned.

(3) Sometimes still a different tack is taken. The laws of logic are laws of thought, but not laws of things. They are necessary in the sense that we cannot escape from them; we cannot think their opposites; we cannot see, for example, how anything could at once be x and not be x. But our inability to think this, we are told, is not the same as x's inability to be this; it might, in spite of us, both have a quality and lack it. But what precisely does this mean? It means presumably that though we cannot think x's being at once black and not black, it might in fact be both. But a moment's reflection will show this to be meaningless. Remember that according to the theory we cannot think x's being both black and not black. If so, then in asserting that x may be both we are saying what has just been admitted to be unthinkable and therefore without meaning. On the other hand, if the statement is accepted as meaningful, if we really can think x's being black and not black at once, then we have contradicted ourselves and are again saying nothing meaningful. For we have said both that the law of contradiction binds our thought and that it does not, since we have successfully thought its opposite. Thus the theory that the law is a law of thought only, and not of things, cannot even be intelligibly stated.

The law of contradiction, then, is more than a statement about usage, more than a convention, and more than a law of thought. It is a statement about the nature of things. It says something about the real world. What it says is that truth is coherent in the sense that if two assertions are inconsistent with each other they cannot both be true. And if they cannot both be true, that is because the world about which they are reporting cannot itself have incompatible attributes. Truth must be consistent if the reality it describes is consistent. The laws of logic give us knowledge about the way things are put together.

(4) There is a current theory that denies this and urges a logic without ontology. It is a pragmatic theory which holds that the laws of logic are means to ends; they are not logically necessary, but only practically necessary, indispensable instruments in ‘regulating the pursuit of human ideals’. What are these ‘human ideals’? They are such things as the order, coherence, and systematisation of our knowledge. This is not, so far, very enlightening. For consistency is an essential part of ordered, coherent, and systematised knowledge, and so far as this is true, what we are told is that we should obey the law of contradiction—that is, should be consistent—because if we do we shall achieve consistency. The argument is thus circular. But the fact surely is that none of these things is the ultimate end of thinking. It is perfectly meaningful to ask why we should wish our conclusions to show order, coherence, or systematisation. And to this question the one natural and plausible answer is, because only thought that possesses these characteristics can be true. Thought as such and always aims at truth, that is, the grasp of things as they are. It is admittedly the pursuit of an end, and the laws of thought regulate that pursuit; so far we agree with the pragmatists. But why this beating about the bush, this vagueness and circularity in their account of the end? Presumably because if they gave the natural answer, that the end of thought is to see things as they are, they would have to admit that logic is concerned with ontology after all. But then we must admit this, or else deny that the end of thought is to know. The ultimate reason why we should try to think consistently is that unless we do we shall not think truly, which means that we shall not see the world as it is.

The first step in our rational construction is thus that, whatever else it may be, the world is a consistent whole, in the sense that it harbours no real contradiction. But this is a very abstract and meagre bit of information. What else can be said about this whole?


5 I think we may advance a step and say also that it is a causal whole. By this I mean not only (I) that every event is causally connected with some other, but (II) that every event is thus connected, directly or indirectly, with every other. Since both these propositions play important parts in the scheme of things I am outlining, I must develop their meaning more particularly.

6(1) Every event is causally linked with an event that precedes it and another that follows it. This needs some explanation and defence. What causal law connects is special characters of those facts or events. What causation connects is particular facts or events; Jones takes a country walk in August and on his return has a fit of sneezing; he suspects that something about the walk led to the sneezing, and he tries to single this out. His work would be vastly facilitated if he could see that the sneezing fit is the sort of effect that must follow from a certain kind of antecedent; but, as Hume pointed out, we do not possess such insight. We must proceed not directly by seeing what must be the cause, but indirectly by seeing what cannot be the cause. We do that by taking for granted that the cause A and the effect B are connected through certain of their characters a and b, and that if one of two suspected characters happens without the other, they are not causally related. It could not have been taking a walk as such that caused the fit of sneezing, for Jones has often taken walks without sneezing afterwards; nor could it have been walking through this particular field, nor the smell of the pine woods, nor going bare-headed, for these things too he has been exposed to without any such result. Jones may soon reach the end of his resources and surrender the matter to a medical researcher, but the method of search would be the same. The researcher would go on eliminating suspected factors until he came to one—say the exposure to a certain saturation of ragweed pollen—that was regularly present when such attacks occurred and absent when they did not occur.


7 Anyone who uses this method must admit on reflection that he is making two assumptions, and it will be to our advantage to state them explicitly. (1) The first is the uniformity of nature. This does not mean, of course, that nature is uniform in the sense that as a whole it repeats itself, or that it is not continually producing what is novel, unexpected, and in the light of existing knowledge unpredictable. The state of the world at this instant presumably never occurred before and will never occur again. What uniformity means is that if a thing acts in a certain way today it will under like conditions act in the same way at any other time. To say that nature is uniform is thus only another way of saying that things and events behave lawfully, that if an event occurs in accordance with a certain law today that law will not be suspended tomorrow. ‘Same cause, same effect.’ If nature were not thus uniform, if an attack of sneezing were produced today by exposure to ragweed pollen and tomorrow by an ice-jam in Behring Strait or the bribery of an official in Indonesia, it would be idle to search for the cause of anything, not merely because no method could find it, but because it was not there to be found. The cause is the uniform cause, the cause without which the effect never happens, and given which it always happens. For practical purposes it may be expedient at times to assume that the same effect can be produced by a plurality of causes or that the same cause can produce different effects, as when it is said that the sneezing may result from either ragweed or a cold, or that a given amount of alcohol produces different effects in different persons. But if nature is really uniform, there cannot be exceptions of this kind. One can only believe that if the sneezing in the first case is really the same in kind, a common element is present to account for it in the cold and the exposure to pollen. And if alcohol acts differently on different persons, that does not prove that the same cause has different effects, for the cause is not the same; the cause in one case is alcohol acting under organic conditions abc, and in the next case under conditions xyz. Uniformity demands that once the relevant conditions are fully and exactly stated the resulting law is invariant.


8(2) Besides the uniformity of nature, there is another assumption that is commonly made by those who would explain events through causes. It is called the law of causality, and means simply that all events do have causes. If it were suggested to Jones that he might fail to find the cause of his sneezing fits because it was something recondite and technical, he would no doubt agree; but he would certainly flout the suggestion that he might fail because there was no cause to find. That he would regard as absurd. And if asked to say whether he thought this applied to all other events as well as to the particular event he was investigating, he would no doubt say yes again. Indeed he can see on reflection that his confidence that this event has a cause is based on the assumption that events generally have causes. What he is thus assuming is the law of causality. That law is the most general of all natural laws—a law of laws, to the effect that there are laws governing all particular events.

It was remarked above that an important part of the world view here being sketched is that the world is a causal whole, and that the meaning of this statement consisted in part of the law of causality, that all events have causes. We have now seen something of the role this law plays in the attempts of the plain man and the scientist to explain the world about them. Both use it daily and confidently as the cornerstone of their attempts at understanding. In this I think they are right; and in view of the importance of the law to my own scheme of things, I must reluctantly pause over it a little longer. Why should we believe it to be true? The mere fact that it is constantly and successfully appealed to does not prove its truth. Indeed when we reflect upon it further, we soon come upon difficulties that in the minds of many have seemed sufficient reasons for rejecting it. For it is not only (a) incapable of proof; it seems to be also (b) inconsistent with new developments in science, (c) inconsistent with miracles, and (d) inconsistent with free will. These points deserve consideration. But in accordance with the plan of this chapter, I can only set down my own views briefly and rather dogmatically.


9 (a) It is true that the law of causality cannot be established in the manner of other causal laws. Many attempts have been made to prove it in this way, but there is really no parallel between the types of arguments used. The scientific argument for A's causing B is: assuming that there is a law covering the occurrence of B, the dissociations I have found between B and all other possibilities than A leave me no alternative to fixing upon A as the cause. This is perfectly valid reasoning. The corresponding argument for the law of causality would be: assuming that all events are governed by law, I may infer from the lawfulness of the events I have studied to the conclusion that all events are governed by law. And this is obviously circular; the conclusion is begged in the assumption with which the argument starts. So far as I know, all attempts to justify induction inductively have failed in similar fashion. The law of causality is the general principle used to justify generalisations from one case to others. Grant it, and the inference proceeds smoothly; call it in question, and reasoning cannot get underway. This suggests that if the law is to be established at all it must be by a priori rather than empirical argument. To this suggestion we shall return.

10 (b) There is an apparent conflict between the law of causality and the results of quantum physics. The causal laws with which we are familiar connect changes of observable things, but we now know that these things are composed of enormous aggregates of elements, so minute as to be far below the threshold of observation. Our familiar causal laws are thus statistical statements about the behaviour of vast aggregates of particles. But statistical statements throw little light on individuals. One could not argue from the fact (if it were a fact) that Americans who reached forty died on the average at sixty-three to the conclusion that Jones, aged forty, would die at sixty-three; he might have a stroke tonight or be still sprightly at ninety. Similarly, we cannot argue from the fact that an army of ten billion protons and electrons shows a majestic stability in its behaviour to the conclusion that its individual components will show a similar absence of skittishness. Indeed these components have so far resisted all attempts by physicists to catch their behaviour in formulae. Heisenberg showed that the more exactly one determines the velocity of a particle, the less exactly can one know its position; and it is notorious that when atoms were conceived on the analogy of the solar system, electrons had a baffling way of leaping from orbit to orbit without any assignable cause. Such discoveries have led many physicists to question whether, since causation connects only determinate events happening at determinate times, one can rightly speak of causation at all in the submicroscopic realm. A layman will be well advised not to rush into this vast and dimly lit region with a small and flickering candle. But there are certain things that he may justifiably say.

(i) Even those who hold that causality is suspended in the atomic realm still admit its sway in that world with which alone we are directly concerned.

(ii) There is a great disparity between the two realms, and since our experience is confined to events of the macroscopic order, we can only represent those of the atomic order through similes and analogies that cannot be accurately checked. Protons, electrons, and photons are not things in the sense in which tennis balls are things, yet it is hard not to think of them in this way. Thus we face an awkward alternative in construing the tenants of the atomic world. If we think of its members in terms of the things and changes we know, we are likely to talk nonsense, while if we regard them as wholly dissimilar, we can say nothing about them at all. Let us try to be clear about this.

Regarding the causeless leaping of electrons from orbit to orbit, C. D. Broad remarks: ‘I see no objection to physicists using such language when about their own business, if they find them convenient ways of briefly expressing certain complicated facts or hypotheses. But, if the terms “electron”, “nucleus”, “orbit”, “motion”, and “jump” are taken literally, I have no hesitation in saying that such statements are, and can be seen by everyone to be, absurd.’2 That a tennis ball or the planet Mercury should leap from one point to another without any cause whatever would certainly seem to the physicist incredible, and if he uses the same language of an electron, conceived as a thing whittled down in size, one can only say that it remains incredible.

On the other hand, if we hold that the entities of the atomic world, and their ways of behaving, are utterly unlike those of our familiar world, they become inconceivable. And again physicists would not accept that. They hold that at least the more abstract categories, those of logic and mathematics, apply to the atomic world, for these are their chief reliance in exploring it. Now causality is also a category, and though less abstract than these others, is, like them, a major means of understanding the world. It should not be abandoned except from absolute necessity. And it is not clear that this necessity has yet arisen.

(iii) The reply may come that whether it has in fact arisen is a matter to be decided by specialists, those with an intimate knowledge of physics and the philosophy of science, not by knight-errant philosophers. That is no doubt true. But if the issue turns into a battle of authorities, the names are not all on one side. Many names could be thrown into the scale against causality in the atomic world, but there are names on the other side too. Einstein, Planck, and Rutherford among physicists, and Broad and Nagel among philosophers of science, would, if I am not mistaken, agree with the statement of Russell that ‘There is nothing whatever in the Principle of Indeterminacy to show that any physical event is uncaused’.3

11 (c) We now come to an objection to the law of causality of a very different kind. Most of the theologians whose views we have considered in this volume would decline to accept it because it leaves no place for miracles, miracles being conceived as suspensions of natural law by a supernatural power. To be sure, a miracle is not thought of as an uncaused event, but rather as one caused by God. Nevertheless the act of God is conceived as an intervention in the natural order, the overriding of a natural law to produce a result which, except for that intervention, would not have occurred.

Now it is easy to abolish miracles by definition; one need only define a natural law as universal to make any suspension of it self-contradictory. But we know too little about the causal connection to justify such dogmatism. It is not logically impossible, so far as we can see, for a man to walk on water or even for a decomposing body to come to life again. Hume was therefore right in saying that the decision whether an alleged miracle had actually occurred rested on the weighing of probabilities. And he thought that the evidence in favour of the ‘law’ would always outweigh the testimony in favour of its violation. It would always be more likely that there had been some deception, some error in observation, or some mis-reporting of the facts than that a generalisation firm enough to present itself as a law of nature had been violated. If the testimony for the miracle did seem to outweigh that for the law, it would always be more probable that the eccentric occurrence was governed by laws yet undiscovered, which supplemented and corrected laws already known, than that there had been a sheer break-through from a non-natural order. We can accept this, I think, though with some demurrer about the ‘always’. A criterion of the occurrence of miracles should not be so framed as to exclude the acceptance of them even if they occurred. Still it is difficult to conceive an event, even the sudden disappearance of an advanced cancer or the reanimation of a dead body, that could not more credibly be assigned to unknown natural causes than to no natural causes at all. Regarding many of the miracles of healing recorded in the New Testament, for example, it is clearly possible to accept the facts while assigning them to laws now imperfectly known about the relation of mind and body. Regarding such miracles as Jonah and the ‘great fish’, Joshua and the halting of the sun, and the parting of the Red Sea, the probability that they are myths rather than actual suspensions of natural law is overwhelming. There is, indeed, no empirical way of disproving a miraculous factor in these or other events. That could only be done by showing that the laws alleged to be violated are necessary laws, and this is not now possible. Nevertheless the threat posed by miracles to the law of causality does not seem very alarming. The probability is that if they cannot be dismissed by the law they can, in the manner just suggested, be absorbed by it.

12 (d) For many persons the main objection to the law of causality is its apparent exclusion of free will. ‘If all events are governed by causal law,’ they would say, ‘and my volitions are events, as they clearly are, then my volitions must be so governed, and everything I do or say must be already determined’. Now it is plain that the law of causality does imply determinism, for determinism is only the statement of that law as it applies to a special class of events called volitions. Many persons who would otherwise accept the law unhesitatingly draw back from it when they realise that it commits them to this conclusion. I do not think that on reflection they should. For indeterminism is a less desirable and determinism a less undesirable position than these persons probably suppose.

If the law of causality is true, the state of one's mind and body and their surroundings at time t1 completely determines the state of one's mind and body at time t2. This the indeterminist denies. He holds that from the moment t1 there may emerge some state of one's body or mind which is connected by no law with what went before and would therefore be unpredictable even by a complete knowledge of the circumstances at t1. To anyone who believes that all events are caused, that is, occur in accordance with causal law, this is an arbitrary position. What has led the indeterminist to accept it? I think that two kinds of consideration have been uppermost, one drawn from fact and the other from assumptions in ethics.

The consideration drawn from fact is that when one makes a decision to act one commonly ‘feels free’ to act in any of a variety of ways. Even the determinist admits that he feels so. But this feeling does not stand up very well when put in the witness box. It is essentially the sense of the absence of any circumstances constraining one to act in this way rather than another. But this sense of the absence of constraining factors is clearly based on a lack of awareness of them; and this lack of awareness is inevitable and is quite compatible with their presence.4 When we make practical decisions, we are always forward-looking; we are engrossed with what our actions may lead to; and there would be no point at such a time in grubbing about in our subconsciousness for causal factors that may be acting on us. With our minds wholly engaged in one direction, we naturally have no eyes for what is going on in another. But if at a later moment we do turn our eyes in the other direction, we have no difficulty in bringing factors to light that were working on us at the time, and we can only suppose there were many more which retrospection can less readily find. This of course does not prove determinism, for one can never by empirical study exhaust the causes of an event, and there is always a loophole by which the indeterminist can escape. But it does show that the feeling of freedom, if offered as evidence of causelessness, has very little force.

For many reflective persons, however, the chief objection to determinism is the indignity that it seems to inflict on minds that like to think of themselves as moral and rational. To them it would make man a robot, as fully determined by mechanical law as the puppets of a Punch-and-Judy show. And of some forms of determinism it must be admitted that the charge is justified. Professor Skinner, for example, in his Beyond Freedom and Dignity sees so clearly that his behaviourism is incompatible with dignity as well as freedom in their traditional senses that he tries elaborately to deflate both. But this kind of determinism, which is built on the violent paradox of identifying consciousness with bodily behaviour, has little plausibility.5 Determinism as accepted by the ablest moralists of this century—such writers as Rashdall and Moore, Ross and Prichard, Bradley and Bosanquet, Dewey and Broad—fully recognises that human motives affect the course of conduct. What makes determinism objectionable to most people is the sense that it dehumanises them, that it assimilates their behaviour to that of some mindless machine like a computer; and their antipathy to such crude reductionism is well grounded. Determinism in the hands of responsible thinkers rejects that position as emphatically as they do. And when they see that the line of causality runs, not around or beneath the human spirit but through it, so that every degree of elevation in their motives can still register itself in their conduct, the base of their suspicion largely disappears.

But not wholly, it will be replied. For those motives themselves will still be determined, and though it is less degrading perhaps to be a mental or spiritual robot then a mechanical one, it is still a deplorable state to be in. Now I think that when people take this view they are not considering the alternative. They object to their actions being determined; very well, consider the situation of the man whose actions are not determined. If not determined, they are of course uncaused. If they were uncaused, that means that he did not cause them, and that in any proper sense they are not his actions at all. Hence he is not responsible for them, and he can never be justly praised or blamed or punished for them. His attempts to form or improve his character may not only fail to improve his conduct but must fail, for if his actions are uncaused, whether at the stage of motives or of behaviour issuing from motives, they must remain unaffected by any measure taken to improve them. And if his own actions are thus undetermined, so are those of others; and his habitual predictions of what others will do from their behaviour in the past will be without foundation. If determinism has its undesirable aspects, none of them seems as undesirable as this detachment of one's conduct from one's past self and character, and its delivery over to chance. I say chance advisedly. For so far as an event issues from t1 that is unconnected by law with what preceded, it is chance of the deepest kind—not merely the chance that is based on ignorance but the chance that means disconnectedness in the very nature of things.

The argument that the law of causality is to be rejected because it conflicts with free will thus seems to me to be based on factual and ethical misapprehensions, and to lose its force when these are removed. And with their removal, the last of the objections to the law of causality that we are considering has also been removed.


13(2) The causal monism proposed some pages back consists of two propositions. The first of these, the law of causality, we have considered. The second is the proposition that every event is causally connected, directly or indirectly, with every other.

This suggestion at first glance seems most improbable. We often think of a present event as if it were the last bead on a string that stretches straight and far into the past, or the final movement in a row of billiard balls in which each transmits its motion by impact to the next. From the last ball that has moved we can look back along the series and trace the line of causation as far as we care to go. The image is useful in suggesting that every event is only the most recent in a line of causes that stretches into an illimitable past and will continue into an illimitable future. But in an important respect it is misleading. The causation of an event is never merely a linear succession. We say that the cause of the ball's motion is the impact on it of another. But the cause of a motion includes everything without which it would not have happened. And when we ask what is essential in this sense to the motion of the ball, we clearly cannot confine ourselves to the impact of the ball before it, for the motion would not have been what it was without the retarding pressures of the air and the felt surface on which the ball rolls, without the minute tilt of the table, or the sudden breeze that came in at the door, or the given temperature of the atmosphere, or the gravitational attractions on the balls exerted by the earth, the sun, and indeed all the fixed stars. To pursue this last a little further, if the law of gravitation holds at all, we must suppose that, allowing for the time it takes for gravitational influence to travel, nothing in the universe is wholly irrelevant to the motion of the ball. The sun is exerting a pull upon it that varies directly as the sun's mass and inversely as the square of its distance; but so is the chair by the window, Mt Everest, Orion, and the Pleiades. Thus at one stroke we are taken out to numberless other things and events that would have to enter into the full causal story of this apparently trivial event.

These seemingly irrelevant factors are linked with the motion of the ball directly, for the gravitational influence from each is conceived as following a single causal line, like that of the light from a star. But every event that is gross enough for us to detect is the result not of a single causal line but of a convergence of such lines. Instead of following the gravitational track, for example, we could start, as we usually do, from the impact of the impinging ball, or again from the support exerted by the table, or the resistance of the air, or the friction of the felt surface, or the influence of the breeze through the door. Now just as the movement of the ball is the product of a pencil of forces, most or all of these causal factors would also be found to involve a pencil of forces each of whose radiating lines runs back indefinitely into the past and into remote spaces. The tracing of the full causation of an event is thus not like following the smooth path of a meteor, but rather like tracing the diverging roots of a tree.

14 This enables us to see what is meant by indirect causal connection. Events of importance are happening at every moment on the sun; it is believed, for example, to be losing mass at the rate of four million tons a second. Suppose that among these events there is happening at this moment an explosion of gases on the sun's surface. So far as direct causation is concerned, this event can have no influence on the present motion of the billiard ball, for there is an interval of some eight minutes between the occurrence of the explosion and the arrival of its effects on the earth. Are events on the sun and events on the earth therefore unrelated causally at any given time? Directly, yes. But indirectly, no; for the two sets of events have causal ancestors in common. The explosion of gases has a chemical and thermal history that runs through the five trillion years of the sun's biography. Take one state of the sun, say the one existing a million years ago. That state is a causal ancestor of the present explosion on the sun. But it is also a causal ancestor of the movement of the billiard ball, since without the light and heat made possible by that state of the sun, there would be no human life on the earth, and hence neither billiard players to wield cues nor billiard balls to be moved by them. The explosion on the sun and the movement of the ball, though neither is the direct cause or effect of the other, are thus clearly interrelated through their common causal ancestry.

Is this indirect relation strong enough to enable us to say that if either of the two events had not happened the other would not? I think we can say this, if not with certainty, at least with a presumption in favour of its truth. For consider: the motion of the ball is produced by a plurality of causal lines, which themselves tend to ramify out like the roots of an oak to a great depth and in many directions. Suppose that for a short time we follow one of these roots. The ball moved because it was struck by another; that in turn moved because it was struck by the cue; and the stroke of the cue followed a volition in the mind of the player. That volition sprang from the state of his mind and body at the moment before, which might be described generally as an interest in the game. What was it that conditioned this interest? The answer would have to include the means of the player, his residence, his education, his skill, his immediate opportunity, and much else. And these depend on what? His parentage, for example, which depends on his parents’ parents, and their parents, in a line that could not be stopped short of the primeval woods and the still more primeval slime. And the woods and the slime were what they were, as we have seen, because of the light and warmth of the primeval sun.

Now let us make a contrary-to-fact supposition. Suppose that the player had not wielded his cue in the place, time, and manner that he did. If we believe in the law of causality, what would that commit us to? The denial of the consequent entitles us, of course, to deny the antecedent. We can say that if that effect had not happened none of the necessary conditions (the without-which-not conditions), either of the event or of those conditions themselves, could have happened as they did. And if one follows that through, it means not only that the man, his parentage, his community, and his nation would have had to be different in their history; it means that the story of life on earth would have had to be different. And since the warmth and light of the sun were indispensable conditions of the story's being what it was, they too would have had to be different. But if they were different, it could only be because the sun itself had a different constitution from that which it in fact had. Now start down the different line of causation governing the history of the sun. Every state of this history from that remote time to the present would likewise have had to be other than it was. The extent of the difference from what actually occurred, either at the start of the downward movement or at any later moment, is of course beyond calculation. One can only say that, in view of the ramifying lines of causation as one moves forward, the present total consequence in the state of the sun would have had to be massive. Indeed we could not say that any event now happening on the sun would be what it is. In short, deny all the necessary causes of the movement of the billiard ball, work out all the changes involved in the replacement of one remote cause by another, and you cannot say with confidence that anything in the sun or on the earth would remain the same. Indeed by an easy extension of the argument we can say that if any event anywhere were cancelled every event everywhere would in some way and degree be affected by it. In the network of causal relations, nothing is wholly irrelevant to anything else.

It may be objected that we are overstating the implications of a change in the causal series. ‘The player is playing billiards because he met Jones, who invited him to play; if he had met Smith and been invited by him instead, surely the game would have gone on just as it did; why this preposterous invocation of the sun and the fixed stars?’ The answer is that on the large chessboard of the world you cannot in this nonchalant way replace one piece and assume that the other pieces will remain the same. If the billiards player had met Smith instead of Jones, the lines of causality both in Jones's life and his own would have had to be different not for a step or two but for millions of steps, and to suppose these steps different while their successors remained the same is to play fast and loose with causality. If we take the causal web seriously, the burden of proof would be on anyone who sets a limit anywhere to the radiating implications of any causal change.


15 We are now in a position to move forward. If the world is an interconnected causal system, can we say that it is a system in any further sense? It seems to me that we can. For I cannot accept the view that causal relations are merely contingent relations. If we know that A causes B, we know that the two are in some sense necessarily connected, and therefore that if the facts and events of the world are interconnected causally they are interconnected necessarily. I realise of course that this step from causality to necessity is controversial, and that it is to be taken only with care.6

Anyone who tries to take this step finds himself blocked at the outset by the burly figure of Hume. For Hume maintained, to the satisfaction of the now dominant schools of philosophy, that in no case of causation do we have any insight whatever into why the effect follows the cause. We see just as little necessity linking the blow of the hammer and the sinking of the nail, or the succession of one idea to another in our thought, as we do in the side-by-side appearance of a red leaf and a green one on an autumn tree. This doctrine has become part of the standard equipment of young philosophers. And of course if one starts from the Humean premise that whatever is in our ideas must come from sensation, this conclusion does follow. The relation of necessity cannot be sensed. But to design a theory of knowledge that excludes necessary relations, even if they exist, and then say that inquiries made in accordance with that theory find no trace of them, is not inspired philosophising. The questions we have to ask are two: (1) Keeping our notion of experience broad enough to include necessary relations if they occur, does our experience of particular cases reveal the complete opaqueness to understanding that Hume claims to have found? and (2) When we analyse the notion of causality as such, do we find a priori necessity in it or not?

16(1) The plausibility of Hume's view varies greatly as one passes from one type of case to another. In the field of purely physical relations, it is perhaps right. We do not know why the hammer sinks the nail or, to put it more technically, why the electrons moving around the nucleus of an atom repel the electrons of another atom. Electrons do act in that way, but why they do so is beyond our present understanding, and we can only accept the fact. This seems indeed to be true of all other instances of purely physical causation. Even laws of high generality are similarly opaque. Newton's law of the inverse squares, and the modification of it introduced by the theory of relativity, are statements of mathematical precision, but they are not statements of mathematical truth, and the question Why? can still be raised about them with no prospect of an early answer. The same is true of the action of body on mind. We know that if we stub our toe violently, we suffer pain, and that if a certain region in the brain is stimulated electrically, sensations of sound occur. No one has the least idea why.

Does this opaqueness remain as one passes to higher levels of causal experience? I do not think it does. It is true that one does not reach the simple and clear-cut necessities that belong to the abstractions of logic and mathematics, but neither does one remain in the darkness of mere conjunction. Take the influence, not of body on mind, but of deliberate volition on behaviour. A man decides to lift his arm and hold it out straight before him, and proceeds to do so. Can it reasonably be said that he is wholly in the dark as to why the arm behaved as it did? That hardly accords with his actual experience. He would be ready enough to admit that he does not know how a mental state affects the course of neural change. But I think he will also say that there is something about the causation of the movement that he knows, and does not merely infer with high probability from past conjunctions. He would say that, exceptional circumstances apart, if his arm carried out the volition he could see that without that volition it would not have behaved as it did. So far, the insight claimed is merely that the volition is a necessary condition of the movement, in the sense employed in Mill's method of difference. But the claim actually goes much farther. It is a claim to some understanding of why, given that volition, a lifting of the arm followed, rather than a clenching of the fist or a leap in the air. I do not think this insight a mere illusion. The light it gives is admittedly limited. But it does go beyond the bare conjunction to which, on Hume's view, the knowledge of causation is confined.7

When we pass from psychophysical to psychical events, the conjunction theory is still less plausible. Mrs Jones receives a telegram from the Secretary of Defence reporting that Private John Jones, her son, has just been killed in action, and she is grief-striken as a result. This, according to Hume, is a mere de facto sequence, in which it remains as mysterious why the news should have produced the result as it is why electrons repel each other. But it is surely not a total mystery why that sort of loss should have produced that sort of grief. Romeo receives a note from Juliet saying that she will meet him at their trysting place tonight. He is excited and happy. Is that wholly unintelligible? Othello, when he realises that he, ‘like the base Indian, threw a pearl away, richer than all his tribe,’ resolves to treat himself with the contempt that such fatuousness deserves. Critics freely allow themselves to speak of the ‘inevitability’ of the tragic denouement, and such language does not seem inappropriate. Much of the conviction carried by a great dramatist's handling of human nature lies precisely in this inevitability; one sees that a human being with such-and-such motives, passions, and standards, confronted by such-and-such crises, must act in this way, and would be acting incoherently if he acted in any other. The literary critics show more perceptiveness here than the Humean philosophers. If human conduct is not a highly rational achievement, neither is it a mere refuse heap in which anything may be found along with anything else. It is that peculiar kind of chaos that is shot through with gleams of rationality, and the higher in the scale it rises, the more intelligible the conduct becomes.


17 Sometimes—and here Hume is most obviously wrong—the nerve of necessity in a causal sequence shows itself nakedly. I have argued this matter before, and may here be brief.8 A course of reasoning is a series of events, and therefore a process governed as truly by causality as any physical process. In a logical deduction the grasp of the premises precedes the presentation in one's mind of the conclusion. If all members of the zoological class of insects have six legs, and a spider does not have six legs, then (strangely enough and against our prepossessions) a spider is not an insect. We cannot doubt that the apprehension of the premises in their union was a cause as well as an antecedent of the conclusion's emerging in our minds. Is the emergence of that conclusion something as beyond our understanding as the occurrence of pain when we step on a tack? I do not think that anyone who says so can be looking attentively at the facts. If we are asked, ‘Why did that special conclusion emerge in your mind rather than the suggestion, say, that Popocatapetl might now be erupting?’ we should reply unhesitatingly, ‘Because the premises I just had in mind required it,’ and that answer, though not a complete one, is, so far as it goes, correct. And if it is correct, the causality that produced the conclusion has necessity within it. Our thought moved along one line rather than another because it was under constraint by the logical necessity linking the contents.

It will not do, of course, to say that the conclusion emerged because we saw that it was entailed by the premises, for we can only see this if the conclusion is already in mind as one term of the nexus; and then one is explaining the appearance of the conclusion by presupposing this appearance. There is no escape, I think, from saying that the timeless relation of implication between the content of the premises and that of the conclusion is itself a condition accounting for the course of the inference. It is not a sufficient condition, granted; for there are other factors, both conscious and unconscious, that enter in; and these others may be such that the premises are followed by the wrong conclusion or none at all. But it remains true that when we have arrived at a certain conclusion we can often see clearly and confidently that among the conditions of our doing so was the relation of entailment between the premises we had in mind at one moment and the conclusion that entered our mind at the next.

It is not true, then, that an empirical study of causal sequences reveals nothing in them beyond conjunction. Hume's analysis comes nearest to the truth in purely physical cases, but in mental sequences, and particularly as we approach the level of thought and action that is recognised as rational, we see with increasing clearness that necessity enters in as part of the causal strand.


18(2) The sense of necessity is stronger still when one turns from an empirical study of cases to an analysis of the idea of causation. Though most men would feel uneasy if told that in the action of one billiard ball on another there is no link stronger than succession, they find it hard to say precisely what this stronger link is. And when some neo-Humean presses his case by asking them whether there is anything contradictory in supposing that the second ball should refuse to budge or should fly straight up in the air, they are likely to admit that there is not, and to suppose that they were merely confused in supposing necessity to be present at all. But plain men are often right in their intuitions, even when faulty or helpless in formulating them. Fortunately, on this matter C. D. Broad has come to their aid. He was convinced that we have a priori knowledge even of physical causation. Assuming that causation is concerned with change, and that a change normally means the starting, stopping, or varying of a process, he asks whether we have any self-evident insight into the laws governing such occurrences, and concludes that we have such insight at four distinguishable points, (i) ‘Every change has a cause.’ (ii) ‘The cause of any change contains a change as an essential factor.’ (iii) ‘If a change issues from a moment t, then all changes which are factors in its cause are changes which enter into t.’ (iv) ‘A given change in a given process issuing from a given moment cannot have more than one total cause.’9 I agree with Broad that these are a priori insights, not empirical generalisations. If anyone were to offer empirical evidence that any of them was violated in a particular case, we should insistently disbelieve it, holding that the evidence must be mistaken or incomplete. Hume would deny that we have any such insights. And it is true that self-evident insights are to be accepted very gingerly. But they do exist, and when they present themselves as clearly as they do here, it is hardly possible to put them down as merely probable generalisations that experience may at any time rescind.

Nor do they exhaust our a priori knowledge of causality. I think we also know that if A causes B it is in virtue of its character as A. It is no accident that men do not gather grapes of thorns or figs of thistles. That is why causal laws, though holding of events, are always statements of connection between characters. Even though we do not know in the end why atropine dilates the pupil of the eye, we are so sure that the effect depends on something in the nature of atropine that if a liquid offered us as such did not dilate the pupil we should decline to admit that it was atropine. As H. W. B. Joseph said:

‘If a thing a under conditions c produces a change x in a subject s… the way in which it acts must be regarded as a partial expression of what it is. It could only act differently, if it were different. As long therefore as it is a, and stands related under conditions c to a subject that is s, no other effect than x can be produced; and to say that the same thing acting on the same thing under the same conditions may yet produce a different effect, is to say that a thing need not be what it is. But this is in flat conflict with the Law of Identity.…’10


19 I think Joseph is substantially right. Though we do not know in advance of experience what effect a will produce, we do know that it will produce whatever it does produce in virtue of its being a; and this, so far as it goes, is necessary knowledge. Professor Ayer takes exception to this view, and his criticism should be noted, though it requires some lingering over technicalities; and if the reader is impatient, he should turn at once to the next section. If Joseph's view is accepted, Ayer says,

‘I can no longer have any doubt that if the event I am observing really is an instance of A it will be succeeded under these conditions by an instance of X; but the trouble is that I must at the same time become correspondingly more doubtful whether this really is an instance of A. For the evidence which I formerly took to be sufficient to establish the truth of this existential proposition will ex hypothesi not be sufficient to establish it, now that I have widened the connotation of A by making it logically necessary that every instance of A should, in the relevant conditions, be conjoined with an instance of X.’11

This last is no doubt true. Even if we accept necessity in causation, our grasp of it will not enable us to predict with certainty that A will produce X, and hence we cannot be sure that the true cause of X is present until we find X present in fact. But this is not an argument against Joseph's view, and perhaps is not meant to be. Such knowledge as he claimed here was of the extremely general kind that is confined to the statement, ‘if A causes X, it does so in virtue of its character’; and it is no objection to this view that we can never demonstrate what will follow from a particular given event and hence must wait for the effect before we can be certain. This is no evidence against either the presence of necessity or our knowledge of that presence.

Ayer goes further, however. He dismisses Joseph's theory on the ground that it is essentially a proposal to use words in a new and inconvenient way; its truth ‘depends entirely upon how one chooses to construct one's language’. According to the theory, ‘all causal expressions are treated as partial definitions of some general term’. But if we were to treat them so, ‘we should have a language in which all synthetic causal propositions would take the form of singular existentials. We should not then be able to express at all what we now express by making causal propositions apply synthetically to an infinite number of possible cases.…’ We should find, further, ‘that it would not be possible for us, as it is at present, to abandon a “causal law” without making a change in our usage of words. There would, indeed, be nothing describable as “the abandonment of a causal law”, except the redefinition of some term.’12

This criticism does not, I think, achieve its end. (a) To say that there is an element of necessity in A's producing B is not an attempt to redefine A. If a geometer discovers some new property which follows from a figure's being a circle, he is not proposing a redefinition of the circle; indeed if all that follows from a definition had to be included in the definition, we could never define anything. It may be replied that Joseph must have been attempting a definition, for he made A's very identity depend on its causal behaviour; if one says that A causes B in virtue of being A, B's absence commits one to saying that the apparent A could not have been really A. True; but this does not commit one to including all the consequences in the definition unless one holds that all necessary propositions are analytic of their subjects. Joseph would not have accepted that view, nor do I. Ayer seems to be contending that on a rationalist view of causality, if Julius Caesar produced a stream of effects lasting from his time to our own, these must all be parts of the very conception of Caesar, and therefore that none of his friends had a right to recognise him as Caesar until the scroll of history had unrolled. In one sense, no doubt, this is true; we do not fully understand the nature of anything until we understand what it entails, and hence our conceptions generally will be kept tentative and modifiable. But unless we could at least provisionally define things before knowing all that the definitions entailed, we could never develop these entailments. No sensible rationalist proposes that discursive reasoning be abandoned because everything it might discover is not provided in one packet at the outset.

(b) Nor does the truth of the rationalist view of causality ‘depend entirely upon how one chooses to construct one's language’. That view is a theory of how nature is put together, and how nature is put together does not depend on how we put our words together. One can indeed order one's words appropriately if one knows the order of nature first, but it does not make sense to say that whether that order is of one kind or another ‘depends entirely upon how one chooses to construct one's language’. The statement that causation in nature involves logical entailment is either true or not true, and it remains what it is, whatever language one uses about it. How otherwise could one choose between ways of constructing one's language? One can make language answer to fact; one cannot rearrange fact to suit one's linguistic preferences. These are truisms. But they have to be set down when questions about the nature of things are translated into questions about the use of words.

(c) Professor Ayer says that if we should adopt the necessity view, ‘we should have a language in which all synthetic causal propositions would take the form of singular existentials. We should not then be able to express at all what we now express by making causal propositions apply synthetically to an infinite number of possible cases,’ and ‘it would not be possible for us, as it is at present, to abandon a “causal law” without making a change in our usage of words’. Ayer's admirable succinctness makes him somewhat cryptic here, and for fairness’ sake it may be well to state in my own words what I take him to mean. (i) Why should he say that on the necessity view such a causal proposition as ‘atropine dilates the pupil’ is a ‘singular existential’? Because if the atropine dilates the pupil in virtue of its nature as atropine, then so acting is part of its nature, and we cannot be sure it is atropine at all until we find it so acting. And we find it so acting only as we observe particular cases. Hence ‘atropine dilates the pupil’ must, if responsibly said, be only a report of one or more particular observations. (ii) Hence, again, it can no longer be what causal statements have commonly been, universal or general statements with application ‘to an infinite number of possible cases’. And (iii) if we found a case in which atropine did not dilate the pupil, we should know that what we had could not be atropine, and, to be clear, we should have to find a new name for it. Hence we could not ‘abandon a “causal law” without making a change in our usage of words’.

In all this there seems to me more subtlety than cogency. (i) Though it is true that on the necessity view you cannot be sure that you have atropine until you find it dilating the pupil, neither can you be sure beforehand on the empiricist view that what you conceived as the cause is really before you. Whether A produces x necessarily or only uniformly, you cannot be sure the A you meant is in operation without awaiting the result, and its being so is subject in both cases to the same veto through the non-appearance of the expected effect. (ii) It is hard to see that the statement ‘atropine is conjoined with dilation of the pupil’ should be a law capable of infinite applications and ‘atropine as such dilates the pupil’ should not. The latter is more obviously universal than the former, and if made with a little interpretation escapes the requirement of particular verification that Ayer would impose on it. What it means for the rationalist is something like this: in the substance called atropine, which has been found to dilate the pupil, there is something in virtue of which this effect will, under like conditions, be produced always and necessarily. This seems to me an a priori statement and, like all such statements, genuinely universal. (iii) Nor is it easy to see why on a rationalist theory we could not abandon a causal law without a change in our usage of words. That would follow only if the causal behaviour not only were necessary but were included in the very definition of the cause, which, as we have seen, it need not be. The word ‘circle’ does not have to be redefined and a new word found whenever a new property of the figure is discovered or an old one found to have been mistakenly deduced. If atropine failed in certain cases to dilate the pupil, the verbal way of dealing with this would be the same on either theory; we should say that atropine dilated the pupil only under certain conditions—e.g. when it was in liquid form, or when in a solution of three per cent or more, or if it had not been exposed to sunlight.

20 If what we have said is accepted, it is not unreasonable to hold that causal laws are also necessary laws. This does not mean, of course, that they are self-evident. We do not know that fire burns or water wets in the way we know that two and two make four. But in addition to our knowledge that necessity does in certain privileged cases enter into causation and our a priori grasp of some aspects of causality, we have what can only be called an invincible surmise that if events of certain kinds go uniformly together it is not by chance. Hume would have us think that no number of repetitions of fire's burning or water's wetting would affect the certainty or even the probability of their behaving the same way in the next instance. And on a theory of chance which rules out the possibility that the dice are loaded, there seems to be no way of avoiding this conclusion. If you throw a die ten thousand times and get ten thousand successive fives, it is no more likely, if loading is excluded, that you should get a five on the next throw than a two or a six. But the plain man would have no doubt in such a case that the die was loaded, and would regard it as absurd to rule that hypothesis out. He would take a roughly similar view about causation. One can point out to him that in an infinite run of cases no finite run can establish any probability, and hence that he has no right to expect anything rather than anything else; but he will go right on expecting as before, and—no one really doubts—with results that will confirm his expectation. The hypothesis of chance cannot be disproved, but once the hypothesis that more is at work is allowed, it is overwhelmingly supported by the course of events. We must admit that we do not in the end know why fire burns or water wets. But it will take more than Hume to convince men that it is merely a matter of chance; if fire burns and water wets, that is because of the distinctive nature of each.

Our conclusion so far is that the world is a unity in three senses. In the first place, it is a coherent world between whose facts there are no inconsistencies. Secondly, it is a causal unity, in the sense that all events are causally interrelated, directly or indirectly. Thirdly, its elements are also necessarily related, since causal relations, we have argued, would be seen as necessary if fully understood.


21 In discussing consistency, causality, and necessity, we have been dealing with abstract and general relations which, though they tell us little about individual things, tell us much of importance about the structure of the whole. Let us turn now from these general features to a more familiar scene, the perceptual world of common sense. Each of us at any moment is, or has, a field of consciousness in which the most conspicuous objects are such things as tables and chairs, rocks and roads, trees and animals and other people. This common-sense order is usually accepted without question as being what it appears to be. It is the base camp from which all the explorations of science or philosophy must start, and the haven to which all speculations must return when in need of empirical verification. It is the home base for everyone—the region where we feel most secure, where things and persons seem most stable and predictable, and doubt and mystery are at a minimum. It is a region where the speculations of the philosopher seem a luxury or an excrescence.

Some philosophers have come close to agreeing with this judgement of their work. A school of thought has appeared in recent decades which would make accordance with common sense the test of philosophy rather than make philosophy the inquisitor and judge of common sense. In his notable essay, ‘A Defence of Common Sense’, G. E. Moore offered a long list of beliefs that had been held by philosophers—the beliefs, for example, that matter did not exist, that space and time were unreal, that memory might be an illusion, that the existence of other minds was dubious; and he went on to assert that he knew these beliefs to be false on the ground that they outraged common sense. He was more certain, he said, that common sense was right about them than that any of the subtle arguments offered by philosophers against them were valid. Other philosophers, following what they took to be Moore's lead, developed a cult of ‘ordinary usage’, maintaining that so long as philosophers stuck to the standard meanings of words instead of warping and twisting them into new meanings of their own, they were on sure ground. The quagmires of philosophy consisted of verbal confusions.

These views found pleased acceptance among persons—and they were many—for whom the subtleties and ardours of philosophy held no attraction. We are all at times members of that camp. Nevertheless it had no tenable case. To suppose that common sense has the answer to sophisticated questions that it has never thought of raising is a strange position, to which Moore himself was unable to adhere. The common man takes it for granted, for instance, that matter exists, and that when he looks at a match-box, he is perceiving a material thing. Moore examined this perception and concluded that it may be interpreted in any one of three ways, between which, after a vast expenditure of pains and patience, he found himself unable to choose. Now common sense can hardly be a supreme court for philosophy if its pronouncements are so vague that it must thus rely on philosophers to tell it what they mean. Nor is the appeal to common usage in any better case. One can talk nonsense or falsehood in the most impeccable standard English. It is perfectly good usage, for example, to say ‘I see a star’. But as understood by the plain man, that statement would abolish both modern physics and modern astronomy.13 Common sense and common usage make a good beginning for philosophy, but if it must remain accountable to them to the end, they prove a mere ball and chain. It is true that anyone who departs from the common-sense view of things should be prepared to carry the burden of proof, but this burden is sometimes light.


22 That holds, I think, of the common-sense conviction that we are now about to examine, a conviction whose truth or falsity is of the highest importance to anyone trying to understand the world he lives in, even in broad outline. This is the conviction that we see things as they are, or, to put it a little more precisely, that what we are immediately aware of in perception is a physical world that exists apart from perception just as it is revealed to us. There is, I believe, no good ground for this conviction, and once that becomes clear, the world proves to be a very different place from the one common sense accepts.

The issue has been a controversial one for centuries. Even among philosophers of the present century there is a wide spectrum of opinions, reaching all the way from the extreme realism of E. B. Holt, who thought that even the objects of illusion had a subsistence independent of consciousness, to the solipsism of Christine Ladd-Franklin, who accepted the existence of nothing outside her own field of consciousness. It is impracticable to weigh all these theories, and I shall content myself with outlining my own position.

The crucial point is the fact of error. All men are realists, in the sense of believing that they perceive things directly and as they are, until perception misleads them, and sooner or later it always does. Straight sticks look bent in water; railway tracks seem to converge; santonin makes things look yellow; for alcoholics on lost weekends the wallpaper may squirm with bats’ heads; water is warm to a cold hand and cold to a warm one; one man sees the signal in the road as green while his colour-blind partner sees it as grey. One or other of the conflicting impressions, if not both, must be in error. Here common sense and science agree, for they proceed on the same assumption, namely that no physical thing can have incompatible qualities. A surface cannot be green and grey at the same time, nor the same water hot and cold, nor the same rails parallel and converging, nor the same wallpaper quiescent and squirming, nor the same stick both straight and bent. Attempts have been made, indeed, to deny this. The most thoroughgoing of them was probably the theory offered by Russell in 1914 that a physical thing could be defined as the class of its actual and possible appearances, all of which were to be accepted as parts of it and independent of being perceived. This theory was given its coup de grace by Lovejoy in The Revolt against Dualism, and does not call for present discussion. We shall assume, with common sense and science, that physical things, if there are such things, cannot be characterised by qualities that are self-evidently incompatible.

This means that some at least of the qualities we seem to perceive in physical things cannot really belong to them. How do the errors arise? Take the case of the colour-blind man. Some twenty-four out of twenty-five men would agree in calling the signal green, and would accept this as its real colour. How does the twenty-fifth man come to call it grey? The answer in principle is clear. It is because the constitution of his eye is different, so that he does not respond in the normal way to the light waves that impinge on his retina. In his case, the colour he sees is not on the surface of the physical thing; it is a private impression produced by an abnormality in his nervous system. This is the standard account, which would be unanimously accepted, I think, by men of science.

23 Nevertheless it has surprising implications. If the immediate condition of one man's seeing grey is having a nervous system of a certain constitution, then the immediate condition of the normal man's seeing green must be his having a nervous system constituted slightly otherwise. Thus the immediate condition of seeing either grey or green is certain events in the nervous system, not the fact that a physical surface at some distance from us is coloured one way rather than another. If colour does exist in nature, we never see it. The experience of colour comes at the end of a long causal chain which contains as successive segments the movements of light waves, the occurrence of chemical changes in the retina, and the passage of electro-chemical pulses along the optical nerve. Have we any good reason for thinking that the colour appearing in the last event of this series is even like any quality in the object whose reflection of light formed the first event? None whatever. Of course the fact that a causal chain is interposed between sensation and physical thing does not prove their non-resemblance either. But it does, in sum, show two things. First, we never see the colour of anything in nature. Our experience of a colour is an event separated by time, space, and an elaborate causal mechanism from anything in the physical world that we suppose ourselves to see. Secondly, in view of this causal mechanism, there is no ground for supposing even a qualitative similarity between the object in nature and the appearance in our minds.

It is obvious that these considerations do not apply to colour alone. They apply in much the same way to pains, sounds, odours, and tastes, and also to felt hots and colds, wets and dries, smooths and roughs, hards and softs. There is no reason to believe that our experience of any of these qualities is a revelation of what exists in nature apart from us. Their appearances in our experience are always conditioned, as are those of colour, by the states of our brain, and without these states cannot be presumed to exist at all. Nor is there any more reason in their cases than in that of colour to suppose that the machinery of perception transmits a qualitative similarity along the causal chain.

24 We have said that sensations of certain qualities are conditioned by states of eye or brain. But eye and brain are physical things, and is it to be simply assumed that we can know them through perception? Science and common sense generally assume this without question. They assume that when a surgeon looks at the eye or the brain he is seeing an object in physical space, whose shape and size are directly presented to him. It is easy to see, however, that this cannot be the case. Our perceptions of shape, size, and motion are just as truly perceptions, and just as truly separated by a causal series from the physical objects they are supposed to reveal, as colours, sounds, or odours. No surgeon directly sees the shape or size of the eye or the brain he may be working on, and to suppose that he does would throw his theory of perception into hopeless confusion. He cannot consistently accept a causal theory for the ‘secondary’ qualities and a wholly different theory, instrumental or photographic, for the ‘primary’ qualities, for these two types of quality are plainly produced in his experience in the same way. It would not make sense to say that when he looks at an eye or a brain and sees a coloured shape, the colour diffused over that shape exists in his own experience only, while the shape over which it is diffused belongs to a different and material order. Both characters equally are contents of his perception, and if the appearance of one of them is conditioned by changes in his own brain, so is the appearance of the other.

This step also, natural as it is, has implications that carry us far. For if the shapes and sizes that we see belong to our private experience, then the whole spatial world of our immediate apprehension belongs likewise to our private experience. The extension occupied by these shapes and sizes is the space we perceptually live in. But this is not physical space, the space of stars and atoms. Perceptual space is a region where rails converge, spoons in water are bent, and one senses a variety of shapes as one walks around one's desk. There are no such things in the world of physics. The space we live in, the space of our perceptual world, is peculiar to each man; it is born with him and dies with him.

The same must be said of time. The time that enters into perception and memory is not the objective time of physics and astronomy. Perceptual time runs swiftly when we are amused and slowly when we are bored; time remembered seems long if it was crowded with events, and short if nothing happened; and memory may reverse the order of events. These accelerations, retardations, and reversals have no part in the impersonal time of science.

The conclusion from this line of thought is that the whole landscape of immediate experience, all sensed qualities and the time and space that they occur in, are of the stuff of consciousness. Many ingenious attempts have been made to avoid this conclusion, and if an adequate defence were to be offered, it would have to consider much else—the alleged distinction between act and object in sensation, the current objections to ‘sense-data,’ and the many attempts by naive realists, critical realists, behaviourists, neutral monists, and others to preserve the plausibility of direct contact with the physical world. None of these attempts seems to me to have been successful, and some of the recantations made by their proposers are instructive. A single example will suffice. We have noted that Bertrand Russell in early life developed one of the most thoroughgoing forms of realism on record, holding that every sensed quality existed independently of experience. From this position he was compelled by criticism and reflection to beat a retreat, and after considering the matter for some forty further years, he wrote as follows:

‘I hold… that whatever we know without inference is mental.’ ‘What I know without inference when I have the experience called “seeing the sun” is not the sun but a mental event in me. I am not immediately aware of tables and chairs, but only of certain effects that they have on me.’ ‘Among the things that I see at a given moment there are spatial relations which are a part of my percepts; if the percepts are “mental”, as I should contend, then spatial relations which are ingredients of percepts are also “mental”.’14

Such testimony is the more valuable as coming from a pioneer of realism, who would have clung to it if he could have found any plausible way of doing so.

25 ‘No man is an island,’ said John Donne. ‘Every Englishman is an island,’ said Novalis. In our sense, both were short of the truth, for all men are islands. We never actually see the same chairs or tables, the same hills or sky, for we cannot share the contents of another's field of consciousness. In a way, we can cross the gulfs between us by throwing out precarious bridges in the way of words and signals, and using such assumptions as that if the thing in your field, called a chair, is related to the things around it in about the same way as something so named in my field, it is the same chair. Strictly speaking, it is not. One's knowledge of another's mind is always reached by inference, implicit or explicit, never by immediate insight. A possible exception is telepathy, which does seem to occur in certain cases and which, if it does occur, is philosophically important. But of its conditions almost nothing is known.

The world we have now reached is somewhat similar to Berkeley's, in which each person is confined to a field of consciousness of his own. Berkeley escaped a complete confinement of this kind by a leap of causal inference, and since this leap had in our judgement an a priori sanction, we should agree as to its validity. It is clear that when we open our eyes and see the sun, something other than ourselves is acting upon us. Berkeley could plausibly take this external cause as the will of God because he had ruled out the only important alternative, the action of matter. He ruled this out by the esse est percipi argument, which takes the very being of anything to lie in its being perceived. Indeed he thought this self-evident and unperceived matter to be a contradiction in terms. The argument is far from self-evident to me. I have no difficulty in conceiving even the blue of the sky and the greenness of grass as existing unperceived; and though I think they exist in fact only in consciousness, it is not on this ground that I think so. And just as Berkeley is unconvincing in his way of dismissing matter, so he is also in his alternative to matter. What we want in a causal explanation of perception is something that will explain why, when we open our eyes, we sometimes see a mountain and sometimes a postage stamp. Berkley would explain the appearance of these and of all other objects in nature in the same wholesale way; they are all impressed upon us by the divine will. But an explanation that applies equally to everything does not differentially explain anything.


26 Now science does attempt this differential explanation. Its accounts are detailed and elaborate, and though they fall short of demonstrable proof, they have been far too massively confirmed for philosophers to ignore. Philosophy and science form, or should form, one continuous, co-operative attempt to understand the world; and philosophy cannot afford to neglect the extraordinary success of science in correlating perceptions with inferred changes in the physical order. In 1906 a philosopher of distinction, McTaggart, could write that the existence of matter was ‘a bare possibility to which it would be foolish to attach the slightest importance’. At present the foolishness would lie in writing that way at all. Protons and electrons, neutrons and positrons have never been perceived and presumably never will be, but at a time when scientists, by assuming that they exist and behave in certain ways, can harness them to such purposes as the driving of ships and the destruction of cities, they are hardly to be dispelled by a deprecatory waving of hands.

It must be admitted, however, that knowledge of them is as yet extremely tenuous. Nothing at all is known about their qualitative character. They are scarcely more than x's, arranged in dimly discoverable patterns and capable of motion. Though matter must be admitted to exist, it is neither the sturdy, solid matter of common sense nor the tiny atomic fragments of such matter envisaged by most materialists from Democritus, on, but a gossamer net of insensibles, or, as one theorist has put it, ‘waves of probability undulating in nothingness’. There is no reason to believe that the chair I see, with its clean-cut lines and its hard and smooth brown surface, exists in nature. But we can form hypotheses as to the causes of our perception, and if these hypotheses are granted some degree of antecedent probability, they can be progressively verified. If we assume that there are light waves moving in independent lines from the various parts of the object to our eyes, that these parts are arranged in some sort of pattern that can be correlated with the parts of our percepts, that sensations of sound, odour, coldness, hardness, smoothness are produced by the motion of certain particles or waves, we find that these hypotheses can be confirmed, not only through continuing sensations of one's own, but through those reported by others. They may be further confirmed by batteries of tests made with cameras, scales, thermometers, reflectors, and other sensitive instruments. The confirmation is never complete, nor is it possible to assign the hypothesis a definite degree of probability, but when it enables us to predict how men and instruments will react to its presence, and the results all hang together as if the hypothesis were true, we are justified after a time in believing with an approach to certainty that it is true.

27 Epistemological idealism, if that means the view that all objects of sense experience are mind-dependent, seems to me unavoidable. What of ontological idealism, the view that everything is of the stuff of mind? That seems to me more dubious. There is of course something absurd in saying that the earth which produced us and our forefathers is a set of impressions in the minds of some of the creatures it has spawned. With the aid of telescopes we can see objects in the sky from which the light now reaching us began its journey before human minds appeared on the earth at all. Unless geology and astronomy are gross deceivers, matter long preceded such minds in the scheme of things.

Is it possible to say whether the constituents of matter—protons and electrons, neutrons and photons—independent as they are of our consciousness, are independent of consciousness as such? I do not know the answer to that question. The physicists who have considered it have held differing views. At a time when metaphysicians, who had furnished the chief support for idealism, were beginning to falter about it, physicists, who had not questioned the material character of atoms, began defecting in an idealist direction. Sir Arthur Eddington argued from the fact that the ultimate entities of physics, though independent of our own minds, were caught up in a web or system of mathematically formulable law to the conclusion that such a system and its terms must exist within a mind.15 Sir James Jeans wrote: ‘the universe can be best pictured, although still very imperfectly and inadequately, as consisting of pure thought, the thought of what, for want of a wider word, we must describe as a mathematical thinker.’16 Professor E. A. Milne believed that the laws of the physical world formed so fully integrated a whole that they were capable of being stated as a deductive system, and he so far succeeded in constructing such a system as to convince himself of a rationalistic view of the world.

A philosophical rationalist can only be grateful for expert scientific aid, particularly in fields where his own light fails him. Eddington, Jeans, and Milne have been attacked for their thesis in terms suggesting that there is something perverse in trying to show that the physical world is intelligible. Needless to say, I have much sympathy with their endeavour. What gives me pause, however, is the assumption that if the universe could be shown to be an order governed by mathematical law it would thereby be shown to be not only intelligible but also mental, that is, to exist only in consciousness. This assumption may be true, but it is not self-evident to me, and I do not see how, except by self-evidence, it could be shown to be true. All the distinguishable things in the universe fall under the scope of arithmetic. Would the relations reported by arithmetical statements, for example 2 + 2 = 4, remain in force if there were no human beings to apprehend them? Astronomers certainly believe that the planets pursued their courses round the sun before human life appeared, that these planets were different from each other, and that two of them and two others made four. Would this relation have held among them, even though no mind were aware of it? I do not know how to deny that it would. Man did not invent the multiplication table. He invented the words for it, to be sure, but not the relations it expresses; and if those relations, abstract and necessary as they admittedly are, can exist apart from his own consciousness, it is not clear to me that they must exist in another consciousness. It seems possible that the world should be a rational whole, in the sense that its parts are linked together by necessity, without being itself a mind. In short rationalism does not seem to entail idealism. In saying this, I do not mean to reject ontological idealism as demonstrably untrue. I mean that, unlike Eddington and Jeans, I see no straight path from the logical articulation of things to the conclusion that everything exists in consciousness. It may or it may not.

Our present position, then, is this. The universe is an order in which all events are interconnected by links of causation and necessity. In this universe there have emerged conscious minds, each limited in immediate awareness to the qualities and structures that appear in its own experience. From these qualities and structures it is possible to infer certain causes and certain abstract corresponding structures in the external world, though this world is cut off from direct access, and seems bound to remain only dimly and inferentially known.

From the book: