You are here

X: Indeterminacy and Indeterminism

NO discussion of fundamental questions of ethics would be complete without some discussion of the problem of free will; and I must therefore say something about the subject, even if I do not feel that I have anything very new to contribute. One can at least attempt some survey of the present position of the problem. The problem has been once more brought to the front by the appearance among physicists of the doctrine known as that of indeterminacy, which might at first sight appear to offer an analogy in physics to what libertarians claim to exist in moral action. I have been helped by studying the symposium on the subject in an extra volume of the Aristotelian Society's Proceedings.1 Professor Broad opens the discussion, and starts, after a few preliminaries, by considering the a priori argument for determinism, which he rightly treats as far the most important. His main contention is that determinism is not self-evident, but follows only from a certain view of the universe which itself when closely examined turns out not to be self-evident. He urges that if a determinist were asked why he feels sure of the truth of determinism, his answer would be, ‘If everything else in the world up to a certain date had been exactly as it in fact was, and yet the subject S had then been in a different state from that in which it actually was, S would have had to have a different inner nature from that which it in fact had. But anything that had had a different inner nature from S would not have been S but a substance of a different kind and therefore a different substance.’ Here the second premiss raises difficult questions; it might be suggested, for instance, that S might have had an inner nature different in some respects from that which it had, without necessarily losing its self-identity. And to set this possibility aside we should have to examine what we mean by the inner nature of a substance, much in the manner in which Professor Broad proceeds to do so. But it is surely clear that the determinist does not need this second premiss. His first premiss is ‘if everything else had been the same and yet S had been in a different state from that in which it was, S would have had to have a different nature from that which it had’. Professor Broad makes the determinist go on to supply as his second premiss ‘but S could not have had a different nature’. But it is obviously enough for him to say ‘but S had not a different nature from that which it had’, without raising the question whether it could or could not have had it. From the new premiss+the first it will equally well follow that, everything else being as it was, S could not have been in a different state from that in which it was. The determinist is in fact saying that the question whether S is at a certain moment to be in one state or in another must depend on somethings and that if it does not depend on anything else in the universe (as ex hypothesi it does not, since everything else is supposed to be the same), it must depend on something in S. And this is surely as self-evident as anything could be.

I think, therefore, that Professor Broad's analysis of the inner nature of a substance is unnecessary for the discussion of determinism. At the same time, it will be only fair to follow his exposition in detail; and I do not question the value, in its own place, of his analysis. It may be summarized as follows:

‘To ascribe a property to a subject is to assert that its states are connected with its previous states and relations by a causal formula. A property of the first order is a causal formula in which none of the variables are themselves properties, but all are simple attributes like, say, greenness. Instances of first-order properties are the magnetic property, the properties of a solid, the properties of a liquid, the properties of a gas. Now any given subject may change in respect of any such first-order property; e.g. it may pass from the liquid to the gaseous state. But if so, there will be a second-order property which is the property of changing in respect of a certain first-order property under certain conditions. Thus it is a second-order property of water to lose the property of liquidity and acquire that of gaseousness at a certain temperature under any given pressure. There may be a third-order property which is the property of losing one second-order property and acquiring another under certain conditions, and there may be properties of still higher order. But there must be some ultimate property of the nth. order which determines changes in respect of properties of the (n−1)th order, which property in turn determines changes in respect of properties of the (n−2)th order, and so on, so that in the long run all changes the substance undergoes are determined by one of its ultimate properties. This ultimate property is, or these ultimate properties are, its inner nature, in accordance with which all its changes must take place.’

The determinist assumption is next formulated, as follows:

‘(i) Every substance has a set of ultimate properties, each of which is of finite order.

‘(ii) No substance can change in respect of any of its ultimate properties.

‘(iii) Any substance whose ultimate properties had differed in any respect from those which S in fact has would necessarily have been a different substance from S.

‘(iv) The value of any variable property of a substance at any moment is inferrible from one or more of its higher-order properties by substituting in the latter the determinate values of its states and relations immediately before that moment.

‘(v) The state of a substance at any moment with respect to any characteristic which is not a property is inferrible from its first-order properties at that moment by inserting in one or more of them the determinate values of its states and relations immediately before that moment.’2

From this complex assumption Professor Broad rightly says that determinism follows at once. But he thinks that certain parts of the assumption are not certainly true. On the first three propositions he comments that they are involved in the notion of substance, but that it is doubtful whether there are substances, and whether, if so, all events must be states of substances. Now to me, in spite of the very great difficulties involved in thinking out the nature of substance, it seems self-evident that there must be substances, and that all events must happen in or to substances. And I think that if empirical evidence were needed of the existence of substances, i.e. of beings which undergo change and yet retain their self-identity through it, we have the clearest evidence of it in the facts of self-consciousness, and particularly of memory, in which we certainly seem to have direct awareness that that which remembers is one and the same being that had the experience which it remembers. But the difficulties in thinking out the notion of substance are certainly sufficiently great to raise in many minds an honest doubt whether there are or ever could be substances, and whether the world is not to be interpreted solely in terms of qualities, relations, and events. I would therefore prefer not to rest the principle of determinism on the belief in substances. But I may remark in passing that if a case could be made out for libertarianism by criticizing the notion of substance, the gain for the theory of morals would be one that would not be worth having at the price it would cost, the price of forgoing any belief in a permanent self-identical self.

On the fifth proposition Professor Broad says that it does not seem to him self-evident that a substance must have so many first-order properties as to determine its states at any moment down to the last degree of determinateness. He thinks it possible that its first-order properties might suffice only to confine its states, with respect to various characteristics, within narrow limits, leaving free play within these limits. And he makes a similar remark about proposition (iv). Let us see what this suggestion implies. It means that, all the properties, states, and relations of a substance being what they are (and in saying ‘all its relations being what they are’, we imply ‘all things else in the universe, in so far as they can influence the object, being what they are’), any one of several different states can yet indifferently come into being in the subject. And this is to say that whatever state does emerge, its taking the precise form that it does is accounted for by nothing whatever in the whole preceding state of the universe. To say this is to imply the existence of causation as regards the broad features of any state of affairs, and to deny it about the details; and for this I can see no justification whatever. It would be more rational to go the whole way with Hume and say that ‘all events are entirely loose and separate’, one state of the universe, or rather one set of simultaneous states of the various things that are, succeeding another without being in any way determined by it. Our conviction that things cannot be as Professor Broad suggests they are, seems to me not to depend on the assumption of the existence of substances, but to be equally clear if we try to think of mere events coming into existence, determined as to their general nature but undetermined as to their precise form. Professor Broad's analysis surely only leads us back to the original question, which is the question whether we are or are not certain that there must be something to account for each event's happening precisely as it does.

If our conviction of the law of causation rested on a posteriori evidence, such a half-hearted belief in causation might be justified. For it cannot be claimed that our powers of observation and measurement are such that we could ever on the ground of experience assert complete determinism. We can measure only down to the degree of accuracy of our measuring instruments, including our eyes. We do not know the precise determinate value of the variables either in the earlier or in the later of two states of the universe, or of two states of a single substance, and as far as observation goes two antecedent states might be exactly alike and the two subsequent states slightly different, or two antecedent states slightly different and the subsequent states exactly alike. All that experience can enable us to do is to correlate limitation within certain limits in the antecedent state with limitation within certain limits in the subsequent. Now I venture to think that were it not that some recent physical observations have suggested to some observers that precise determination in the antecedent state is consistent with slight variations in the subsequent state, no philosopher would have suggested indeterminism, at any rate as regards physical events, as a possibly true view. But it is surely clear that empirical evidence can never establish indeterminism, any more than it can establish determinism. The most that observation can justify a man of science in saying is, ‘I find here a variation in respect of a certain characteristic, which I cannot correlate with any variation in the antecedent state of the universe which is at all likely to be relevant to it’. He can never be in a position to say ‘there is no corresponding variation in the antecedent state’; he can only say, ‘I cannot observe any’. Now while empirical evidence could never prove the truth of determinism, it has gone a very long way to confirm its truth, which I believe we know independently of such evidence. The whole progress of science has depended on the assumption that the precise form of every event has a precise cause to account for it, and the history of science has provided a series of triumphant vindications of this assumption. Over and over again, where common sense sees no variation in the antecedent to account for a variation in the consequent, science by more careful measurement and stricter reasoning has discovered such a variation to exist. At present physics has reached a point at which in regard to certain phenomena it cannot detect such a variation. But to sit down before this defeat and say ‘no variation existed; the event just happened so, without cause’, is to be untrue to the whole history of science, to give up because of a temporary defeat a principle which has everywhere else carried it to victory.

I have just described the belief in the law of causation as having been often vindicated by experience; but such an expression needs careful qualification. Strictly speaking, neither the law of causation nor its denial can be confirmed by experience. This follows from the nature of the law. The law may be formulated in various ways, which in principle come to the same thing. One way, which will serve as well as another, is to say, ‘for every variation between two events, there must be some variation between the antecedent circumstances, without which the variation between the events would not have existed’. This cannot be proved by experience, because no set of experiences could claim to have covered all past events, let alone those that are still in the future. Nor can it be disproved by experience, because all that can be said on the basis of experience is ‘no variation has been observed’, not ‘no variation existed’. Not only the law of causation, but any law of the form ‘for any A there is always a B’, when the number of A's is infinite, is incapable of being proved by experience. But furthermore, the law of causation cannot even, strictly speaking, be confirmed or even partly confirmed by experience. For the law says not merely ‘wherever there is a variation in the events, there is a variation in the previous conditions’, but ‘wherever there is a variation in the events, there must be a variation in the conditions without which the variation in the events could not have existed’; and experience, even where it verifies the existence of a variation in the previous conditions, can never assure us that without it the variation in the events could not have taken place. In other words, while experience can prove the existence of a constant conjunction within the limits of the experience, it can never discover the existence of a necessity; only reason or insight can do that.

If one says, then, that experience confirms or vindicates the law of causation, it can only be in a Pickwickian sense. Whenever experience verifies the existence of a variation in the conditions answering to a variation in the events, this confirms the belief in the law of causation only in the sense that it weakens the case of one who denies the law on the ground that experience presents us with numbers of cases in which the one variation exists without the other, by diminishing the number of instances on which he might otherwise rely.

What has given rise in the last few years to a doubt or denial, by some physicists, of the universality of the law of causation, is certain facts with regard to the behaviour of atoms. The original form of the so-called principle of indeterminacy was this: The atom was thought of as consisting of a nucleus, with electrons rotating round it in a certain restricted set of orbits. Radiation from the atom was supposed to take place when and only when an electron jumps from one to another of this set of orbits, and the principle of indeterminacy asserted that the jumping or not jumping of an electron from one orbit to another was undetermined. It is surely obvious that the only ground which could have led any one, in this situation, to think that the movement was undetermined was that there was no known difference between the constitution and circumstances of one electron and those of another, which could account for the one's behaving in one way and the other's behaving in another. And it is equally clear that the rational conclusion to draw was, not that there was no difference to account for this, but that there was a not yet known difference which accounted for it. But, since this way of stating the principle of indeterminacy seems now to have been abandoned in most quarters, we need not dwell on it. In the later form of the theory, the pictorial representation of the atom is given up. There is much diversity in the statement of the later form of the theory, and it may be of some assistance to the reader if I quote some typical statements. Mr. Birtwistle3 states the matter as follows:

‘At the instant when the position is determined the momentum of the electron is suddenly changed. This change of momentum is all the greater the shorter the wave-length of the light used, i.e. the more accurate the determination of position is. Thus at the moment when the position of the electron is known, its momentum can only be known to an order equal to the discontinuous change. Hence the more exactly a coordinate q is determined, the less exactly can its momentum p be found; and conversely.’

Mr. Dirac's4 statement of the matter is this:

‘We find that our wave packet represents a state for which a measurement of q is almost certain to lead to a result lying in a domain of width. Δq′ and a measurement of p? is almost certain to lead to a result lying in a domain of width Δp′. We may say that for this state q has a definite value with an error of order Δq′ and p has a definite value with an error of order Δp′. The product of these two errors is

Δq′ Δp′ = h.

Thus the more accurately one of the variables q, p has a definite value, the less accurately the other has a definite value.’

Mr. Frenkel's5 statement is as follows:

‘Since the position itself can be observed with only a limited accuracy of the order of magnitude λ and the corresponding uncertainty in the value of the momentum is of the order of magnitude h⁄λ, the product of the two inaccuracies—or rather of the inaccuracy of the observation of the coordinates of the particle and the resulting uncertainty in the evaluation of its momentum for the instant of observation—is approximately equal to h.… It is impossible to measure simultaneously with any degree of precision the position and the momentum of a material particle, because the measurement of the one implies, or produces, an uncertainty in the estimation of the other.’

If it is said that certain experiments with regard to atoms reveal that an electron has not at any moment both a determinate position and a determinate momentum (though, the more indeterminate the one is, the less indeterminate is the other), this saying seems to me susceptible of three different meanings. It may be taken to mean that science is not in a position to assign definite values to both the position and the momentum. This would indicate a regrettable gap in the state of our knowledge, and possibly in any future state of our knowledge, but can have no bearing on the metaphysical problem of indeterminism. Or it may be taken to mean that while an electron has a position and a momentum, these are not in fact definite; and this cannot be anything but nonsense. If, as is now thought, electrons are not, as they were earlier conceived to be, ‘particles concentrated almost into points’, but partake at the same time of the nature of ‘waves spreading throughout space’,6 then it may be improper to describe them as having position or momentum at a moment at all (they being entities too complicated to have a single position or momentum), but what is surely quite clear is that nothing in the world can have a position which is not a definite position or a momentum which is not a definite momentum. Or, thirdly, the phrase may be taken to mean that, while an electron has a definite position and a definite momentum, these are not completely determined by pre-existing conditions. It is only in this third form that the newer conception has any direct bearing on the problem of indeterminism; and in its third form it is in principle the same as the older ‘principle of indeterminacy’, in which electrons were thought of as small solid particles whose jumps from orbit to orbit are not determined; and precisely the same comment must be made on it as on the older theory.

Mr. Birtwistle and Mr. Frenkel state the matter with perfect propriety in the first of these three ways. They do not suggest that the electron has not a definite position and momentum, or that it has a position and a momentum which are not determined by previous conditions; they only say that we cannot find out precisely both its position and its momentum. Again, Mr. Dirac does not suggest that the position and the momentum are not determined by previous conditions; and when he says ‘the more accurately one of the variables q, p has a definite value, the less accurately the other has a definite value’, this must, I think, be interpreted in the light of the fuller preceding statement in which he says ‘q has a definite value with an error of order q′ and p has a definite value with an error of order p′’; i.e. q and p have definite values, but values which we cannot discover exactly.

The most authoritative statement, however, of Heisenberg's principle of indeterminacy is that by Heisenberg himself. Towards the end of his famous article7 he says:8 ‘An der scharfen Formulierung des Kausalgesetzes: “Wenn wir die Gegenwart genau kennen, können wir die Zukunft berechnen”, ist nicht der Nachsatz, sonder die Voraussetzung falsch. Wir können die Gegenwart in alien Bestimmungsstücken prinzipiell nicht kennenlernen.’ Now clearly to deny that the if-clause of the statement ‘if we know the present, we can calculate the future’ is ever fulfilled is not to deny the statement itself, and therefore not to cast any doubt on the law of causality. Heisenberg, however, goes on to ask whether behind the perceived, merely ‘statistical world’ a real world conceals itself, in which the causal law is valid, and proceeds to say: ‘Weil alle Experimente den Gesetzen der Quantenmechanik und damit der Gleichung (1)9 unterworfen sind, so wird durch die Quantenmechanik die Ungültigkeit des Kausalgesetzes definitiv festgestellt.’ The conclusion is clearly not warranted. If it is true, as we may well admit, that the state of a particle can never be known with complete precision in respect of all its properties, then we can never establish the law of causality inductively by showing that each later state is completely determined by an earlier state: but any one who regards the law of causality as self-evident and needing no proof, inductive or of any other kind, will not be shaken by being told that an inductive proof of it is impossible.

Such a matter is not to be settled by weight of authority, but by reason; but lest any one should suppose that the weight of scientific authority is in fact in favour of accepting indeterminism in the physical world on the strength of experiments on the atom, it may be worth while to quote the words of Einstein, ‘that nonsense is not merely nonsense; it is objectionable nonsense’,10 and those of Planck, the discoverer of the facts upon which the theory of indeterminacy has been founded. Planck pronounces with vigour equal to Einstein's against the theory and in favour of the universality of causation: ‘In point of fact, statistical laws are dependent upon the assumption of the strict law of causality functioning in each particular case. And the non-fulfilment of the statistical rule in particular cases is not therefore due to the fact that the law of causality is not fulfilled, but rather to the fact that our observations are not sufficiently delicate or accurate to put the law of causality to a direct test in each case.’11 The truth must surely be this, that the failure of science, so far, to discover causes for certain events is not a fact to be accepted with complacency or with enthusiasm and made the basis of a theory, but should be treated as a challenge to further inquiry in the hope of replacing our ignorance by knowledge. We must regard the principle of indeterminacy as one of the crude theories which are apt to be put forward in the years immediately after the discovery of some startling experimental fact, before the true meaning of the fact has been really digested.

Sir Arthur Eddington's contribution to the symposium is very different from Professor Broad's. The latter contents himself with saying that there may be some very slight element of indeterminacy in physical processes. Sir Arthur Eddington tries to persuade us that the whole doctrine of causal determination has in the current state of physics gone completely by the board and been replaced by that of probability. There is one type of discovery that lends some plausibility to this view. There have been, I suppose, several cases in which an older generation thought it had discovered that all particles of a type P behave under certain circumstances in a certain determinate manner Qd, while later discovery has shown that different instances of P behave in slightly varying ways Qa, Qb, Qc,…, and that the truth is only that a large number of P's will have characteristic Q in varying forms grouped closely round the form Qd. From this a hasty thinker may easily draw the conclusion that there are no laws prescribing the precise behaviour of single instances of P, but only laws stating the average behaviour of a large number of P's. And it is often the case that at a certain stage of our knowledge this is the only sort of law about their behaviour that we know. But even to discover such a statistical law as this is to obtain some confirmation of the view that being P determines any particular P to have the attribute Q in some form, and leaves us perfectly free to suppose that the particular mode and degree of Q which a particular P has depends on still unknown particularities of its nature or of its environment; which indeed is the only rational conclusion to draw.

There is, however, one line of thought which Sir Arthur works out with some care,12 which it is worth while to follow. He points out that the majority of the inferences which science draws are inferences about the past. And his view about such inferences amounts, I think, to dividing them into three classes. There are (1) inferences in which from a later event in the history of a substance we infer its previous possession of some attribute or set of attributes whose existence can be verified apart from the inference; e.g. the conclusion, from some test, that a certain salt was silver nitrate—which can be verified by other means than the test in question. There are (2) inferences in which from a later event in the history of a substance we infer its previous possession of some attribute which we cannot verify independently, but can know or reasonably think to belong to some already known class of characteristics. It is an inference of this kind if we infer that the Iliad and the Odyssey were written by two different though unidentified men; for we already know that difference of personal identity is likely to lead to difference of style. There are (3) inferences in which from observation of some character in a later event we infer that there must have been something to cause it, without knowing or even being able to conjecture what it was. To this last class belong the phenomena which have given rise to the new doctrine of indeterminacy. Here we can, if we please, say ‘S1 behaves in one particular way, and S2 in another; therefore there must have been some difference in the previous character of S1 and S2, though we have no idea what it is’. This, because we do not know what the two differing characters of S1 and S2 can be and can only describe them at the moment as ‘something that will lead S1 to behave in one way’ and ‘something that will lead S2 to behave in another way’, Sir Arthur describes as not being genuine inference. ‘I am well aware,’ he says, ‘that our ignorance is immense; but I see no point in attempting to enumerate the things which might exist without our knowing them.’13

What exactly is the cash value of the distinction? The position is simply this. (1) In the first case, where we can trace a difference between two results to a difference in the causes which can be independently verified to exist, we have a partial confirmation—if confirmation were needed—of the law of causality. (2) In the second case we think we are on the track of such a confirmation, since we have an inkling of what the difference between the causes will turn out to be. (3) In the third case we have no inkling of how or where the confirmation will occur. But the inference ‘since the effects are different the causes must be’ is just as sound as it is in the other two cases, though it remains, owing to our ignorance, for the present unverified by experience. The occurrence of such cases throws no doubt whatever on the law of causality; it simply fails to provide fresh confirmation of it. And, to speak for a moment pragmatically, one may surely point out that science would never have made the progress it has made if, whenever men failed to trace different causes for different events, they had been content to suppose that the events just happened to be different for no reason at all. The present failure of science to know why atoms sometimes behave in one way and sometimes in another is not different from many another case in the past history of science in which the momentary failure led people who had a firmer grasp of the causal principle than Sir Arthur has, to some of the greatest triumphs of science.14

As regards the physical universe, then, I think we may remain unrepentant determinists; must we be so as regards the universe of moral action? At the outset it must be said that we are much farther, in this region, from the possibility of empirical verification of the law of causation than in the region of physical action. When we conclude from the observation of two physical events that a greater force must have been at work to produce the one effect than to produce the other, we usually can verify by independent means the excess of the one force over the other. When we infer, because a man does the one and not the other of two acts, that he must have desired the one more than the other, we have no delicate instruments of measurement by which we can verify that one desire actually was stronger than the other. Sometimes we can do so by introspection, and then we are so far verifying the law of causation; but often it is only by observing what we proceed to do that we can know what we desired most. And here we are in the same position as we were in respect to the third of Sir Arthur Eddington's types of inference. We get no empirical verification of the law of causality, but nothing (in the mere fact mentioned) to lead us to doubt it. Those who are interested in the question of a posteriori reasons for determinism will find in Professor Broad's article a careful study of the facts which make such verification far less complete in the moral universe than it is in the physical.15 But it is clear that the essential question is whether we have or have not a priori knowledge of the law of causality and of its application to moral action; and the empirical arguments on either side cut very little ice.

The strength of the a priori argument for determinism in ethics rests on the consideration that the law of causality does not present itself to our minds as one peculiar to physical events but as one applying to all events as such. An event which escaped its sway would be an event of which no explanation could be given. Suppose some one says that one is here taking one's notion of an event from physical events and applying it without reflection to mental events; my answer would have to be that I at least make no use of the conception of anything physical when I say every event must have a cause; by an event I mean any change, anything that takes place. And it is surely clear that even for a libertarian the line is drawn not between physical events and mental, but between one special class of mental events and all other events whether mental or physical. No one wants to exempt from the law of causation the occurrence of a sensation on the occasion of a physical stimulus; or the occurrence of an inference on the occasion of the grasping of its premisses. In the one case the event is thought to be completely determined by the physical stimulus+the pre-existent nature and state of the body and of the mind; in the other by the occurrence of the grasping of the two premisses in their connexion. We certainly cannot choose what conclusion we shall draw from the premisses; if we see their truth and notice their relevance to one another we cannot but see the truth which by the laws of logic follows from them. It is only acts where choice is involved that can with any plausibility be said to escape the law of causation. And while there might conceivably be some plausibility in making the law of causation apply to all physical and to no mental events, it is very unplausible to make it apply to all physical and some mental but not to other mental events.

The strength of the case for libertarianism, therefore, cannot be said to rest on any ontological ground upon which we could expect acts of choice to be free from the law of causation. It rests on two things alone—on the supposed intuition of freedom, and on the thought that morality involves freedom. These are the things at which we must look more closely. I may just say in passing that, if we could believe in occasional or even in universal indeterminacy in the physical world, that view, which is equivalent to a belief in blind chance, would furnish no support either for our intuition of freedom or for our awareness of duty.

In considering the intuition of freedom, we shall do well to consider separately the two questions of freedom to choose, or decide, or resolve, and of freedom to do what we have resolved upon. As regards choice, what is it that we really are aware of? We are aware that a certain kind of activity called choice or resolution takes place, which is different from any other activity—different, in particular, from desiring, and from desiring one thing more than another; and we are aware that it is we and not anything else that performs this activity. We are aware that desires are not like physical forces beating on a physical body from without, to which it is merely passive. They are occurrences happening in the very same being which chooses among them; and they are occurrences such that the mere occurrence of one stronger than all others does not ipso facto lead to corresponding action. We are aware that we often reflect on them and come to see that a desire which as it originally presented itself was stronger than another, nevertheless harmonizes less well with the universe of our interests; and we are aware that when we come to see this, it becomes less strong, or some other becomes stronger, and that we then act on what was at first the weaker but is now the stronger desire, or follow the line of what was at first the greater but is now the lesser resistance. Can it be maintained that there is an intuition of freedom to which this account does not do justice? It seems to me that it cannot, and that reflection on the relations between opinion, desire, and choice shows that it cannot. Let us first take, as the simpler case, one in which the thought of duty does not occur, one in which we choose, for instance, between two courses promising different pleasant experiences to ourselves. The suggestion is that at one point in the nexus of events there comes a choice in which some element of the self operates quite freely, independent of the circumstances and also of the self's system of interests as they exist at the moment. Suppose (1) that the act of suspense, whereby we decline to act immediately on the strongest desire and instead subject the desires to reflection, has already taken place. It can hardly be maintained that then we can by an act of free choice (i.e. choice independent of the circumstances and of our system of interests) control the remoulding which the desires undergo as they are brought into relation with our universe of interests. The existing universe of our interests determines that. Not that we are passive in this process; it is by the activity of thinking about the alternative courses of action that we come to desire one action more than the other. We are not a field on which desires as independent entities wage their battle. But the I which thinks and desires is the I which has been moulded by its previous experiences and opinions and actions.

At the end of the process we are desiring one act more than the other. Can it be maintained, then, (2) that thereupon we are conscious of freedom nevertheless to do either of the two acts? Is it not clear that, in a case where the thought of duty does not come in, we inevitably do that which after reflection we most wish to do? And incidentally we may ask, ‘Would there be any moral value in a freedom to do, in such a case, what we do not most wish to do?’ It would be a freedom to act for no reason, and indeed against reason.

Can it be maintained (3) that, while the determinate self, with its universe of interests, determines what we do, if we perform the act of suspense, an indeterminate element in us performs the act of suspense? Is it not clear that we perform this act only if we think, for example, that the greater of two immediate pleasures may bring more pain in its train than the lesser of the two? And is it not clear again that we are not free either to think or not to think this, that whether we do or do not think it depends on the circumstances and on the determinate kind of being that we are when we compare the two pleasures? We cannot choose what opinion we shall form about the situation, and what we do think about it determines us to perform the act of suspense; or else fails to determine us to perform it, and then we do not perform it. Freedom to suspend or not to suspend action irrespective of any reason for doing either would, again, be a freedom not worth having. To sum up, then, we do not seem, on reflection, to be conscious either of a power of thinking what we please in the light of given evidence, or of a power of desiring what we please independently of our opinions, or of a power of doing what we do not most desire to do. Nor, if we had any of these powers, would it be of the slightest value. What would be the moral value of a power of forming opinions not based on the evidence, of desiring not in accordance with our opinion, or of acting to get what we do not desire?

‘But’, it might be said, ‘the situation is altered when we turn to cases in which the thought of duty occurs to us. Then, at least, we do not choose according to the strongest desire; we may choose to do our duty instead of what we desire.’ If one has done a conscientious act, and is then asked why one did it, the answer which first comes to one's lips is ‘because it was my duty’; and in this there is no mention of desire. Duty tends to be represented as something standing over against all objects of desire. But a little reflection shows that our answer was highly elliptical. It is clear that an act's being our duty is never the reason why we do it. For however much an act may be our duty we shall not be led on that account to do it, unless we know or think it to be our duty. And again, the thought that an act is our duty will lead us just as much to do it when it is not in fact our duty as when it is. Thus our answer ‘I did it because it was my duty’ must be changed into the form ‘I did it because I knew it, or thought it (as the case may be), to be my duty’. But even this is still elliptical. For it is a familiar fact that people often know or think an act to be their duty and yet do not do it. They will do it only if in addition to knowing or thinking it to be their duty they are impelled with a certain degree of intensity towards the doing of duty—with enough intensity to overcome the urge towards any of the alternative possible acts. Thus the fuller and truer answer would be, ‘I did the act because I knew, or thought, it to be my duty, and because I was more powerfully impelled, or attracted, towards it as being my duty than I was towards any alternative act.’ Now a question arises as to the nature of the impulsion. The simpler account would consist in saying that all impulsion is desire, and that therefore our answer may be put in the form, ‘I did the act because I knew, or thought, it to be my duty, and because I desired to do it, as being my duty, more than I desired to do any other act.’ And if this description is true, conscientious action would fall under the same description which we have seen to apply to actions in which the thought of duty does not occur; it would be action in accordance with the strongest desire present at the moment.

The alternative would be the view that what impels us to do our duty is not desire but a certain specific emotion which only the thought of duty arouses—the emotion which Kant calls Achtung, respect or reverence. Now I think it is clear that the thought of duty does arouse such a specific emotion. It arouses it in very varying degrees in different people, and in the same person at different times; and in some people the emotion is very weak. Not only do they habitually not do their duty, but they feel little or no shame or remorse at not having done it; while if the emotion in question had any considerable strength, this would at least lead to remorse after the act, even if it were not strong enough to overrule the desires that lead men to do the act. Still, in the ordinary man the clear recognition of a duty undoubtedly arouses to some extent the emotion of respect. Kant represents the choice between doing and not doing our duty as resultant on a struggle between desire and respect, and resists strongly the suggestion that it is resultant on a struggle between desire and desire. But this seems to be due to his tendency to accept too readily the Hobbesian view that all desire is for the agent's pleasure. Since the motive in dutiful action is clearly not that, he infers that it is not desire at all. But it is now generally admitted that Hobbes was wrong; and if we recognize that there is a variety of desires for objects other than one's own pleasure, there seems to be no objection in principle to recognizing a desire to do one's duty; and certainly introspection seems to reveal such a desire. It seems to me, further, that the emotion of respect aroused by the thought that a certain act is one's duty produces, in proportion to its own strength, a desire to do that act, and that duty is done when and only when that desire is stronger than all those with which it has to contend.

But, it might be said, if action is described as following upon the strongest desire, is not resolution or choosing or deciding made completely nugatory? Would not the same act follow if the step called choosing were entirely omitted? The answer to this is in principle contained in what I have said previously. It is the nature of each single desire to be concentrated on a single feature in an imagined future, and if desire were left to itself we should act to get the most desired of these isolated features. But in fact every such desired feature involves, in the getting of it and as a result of the getting of it, many other features which are not desired. To get the single thing which at the moment we desire most of all things, we might involve ourselves in getting many things that will give us pain, or in doing actions to which we have a strong moral repugnance. In the deliberation which precedes choice we set ourselves, more or less thoroughly according to our character, to choose not between isolated objects of desire but between acts each of which is thought to involve a whole set of consequences, and it is one act with all its expected consequences that is chosen in preference to all others with all their expected consequences. The choice is thus determined not by the strength of the isolated desires as they were before the process of deliberation, but by the strength of the appeal which one act, with all that it is expected to involve, makes on us, as compared with the appeal which the alternatives make. Thus what is resolved on may be, and often is, very different from what would have been done if deliberation and choice had not intervened. What choice depends upon, and reveals, is not the strength of isolated desires but the trend of the whole character, of the whole system of more or less permanent desires, including of course the desire to do one's duty.

The nature of choice, and its importance, may, it seems to me, be brought out by contrasting the way in which we actually behave with two accounts of behaviour which might suggest themselves as possible accounts. I assume that every desire must at any one moment be of a perfectly definite intensity, and that all desires must be comparable on a single scale of intensity, even though it must be admitted that we are quite incapable of distinguishing the intensities of two desires which are very much alike in intensity. Now it might be thought in the first place that action would be determined by the strongest single desire. That, however, is not what happens. I take a simple case. If I am attracted to act A by a desire, of intensity 3n, to get a certain pleasure for myself, and to act B by a desire, of intensity 2n, to give a certain pleasure to another person, and am attracted to act B also by a desire, of intensity 2n, to give a certain pleasure to a third person, I shall actually do act B, though the strongest single desire that is affecting me is attracting me towards the other act. That is the truth that is vaguely expressed by saying that it is the universe of my desires that determines my action, and not the strongest single desire. But, on the other hand, it must not be supposed that all the desires that are present in me co-operate to determine my action, as the movement of a body is determined, in accordance with the parallelogram of forces, by all the forces that are acting on it. If that were so in human action, our action would be an attempt to get to some extent each and all of the many things we happen to be desiring; and it is clear that such action would be futile in the extreme. What happens, in persons of strong character, is something quite different; they make up their minds that certain of the things they want are incompatible with certain other things that they want more; and they then suppress the desires for the former, so far as concerns their becoming influences affecting their behaviour. They cannot entirely suppress these desires; but they cause them to be, not motives modifying their action, but mere longings for certain objects, which remain even when these objects have been resolutely renounced. This, it seems to me, is the great difference between physical and mental causation, that in the latter there is no law of the composition of all the forces concerned, but some of the forces concerned are, by an act of choice, deprived of any effect on action.

If the line of thought I have tried to present is true, the libertarian belief in its complete form cannot be true. The libertarian belief is the belief that, the circumstances being what they are, and I being what I am, with that whole system of beliefs, desires, and dispositions which compose my nature, it is objectively possible for me here and now to do either of two or more acts. This is not possible, because whatever act I do, it must be because there is in me, as I am now, a stronger impulse to do that act than to do any other. Every one would probably agree that this is so when neither of the two acts is being thought of as a duty. It would be agreed that then we must do the act, to do which we have the strongest desire, or rather the act to which the strongest mass of desire leads us. But it is equally true when one of the two acts is thought of as a duty, even if we take the view that what impels us to do an act of duty as such is not a desire but a specific emotion of respect or reverence. For even so we shall do the duty if and only if this emotion constitutes an impulse to do the act of duty, stronger than the impulse which moves us towards doing any other act. And if act A can only be done if I have a stronger impulse to do act A than to do act B, and if act B can only be done if I have a stronger impulse to do act B than to do act A, it cannot be the case that I can here and now equally do either act. It must follow from my nature and my present condition that I must do act A, or else that I must do act B.

Not only is it metaphysically impossible that I should be capable of doing either act indifferently, but if I could, the doing of either could have no moral value. It would have no value, because it would not be the result of any thought about the nature of the act, and of any consequent impulsion to do it. It would be an unintelligent and unmotived leap in the dark.

Nevertheless we all, when we are not philosophizing, tend to hold this opinion, and cannot altogether prevent ourselves from continuing to hold it even when we have come to think it incapable of being true. What then is the truth, if any, which underlies my belief that I can here and now, being what I am, yet do either of two or more acts?

This belief, in the form in which we ordinarily hold it, is the belief that I can cause any one of two or more changes in the state of affairs. In some cases the change we are thinking of is a change in the condition of one's own mind; but far more often a change of something in the physical world is at least part of the change we think we can cause. And such changes can only be caused by first causing a change in the state of one's own body. If I think, for instance, that I can here and now tell either the truth or a lie, there is involved in that the thought that I can cause either of two sets of movements of my vocal chords. And undoubtedly one main source of my sense of freedom to act in either of two ways is the conviction, well based on experience, that my soul or mind has the kind and degree of control over my body which will enable me, if I set myself to tell the truth, to produce certain movements of my vocal organs, and if I set myself to tell a lie, to produce certain other movements of them. This conviction is not, strictly speaking, knowledge; for since I last made the attempt, paralysis may have set in and deprived me of this power. But normally the chances are much against this having happened, and for practical purposes the conviction is justified and is almost as good as knowledge.

The existence of this power has of course sometimes been doubted. It has been thought that mind cannot act on body, and various forms of parallelism of mental and bodily events have been put forward in opposition to the belief in the action of mind on body. But, as several recent discussions of the question have brought out,16 the theoretical arguments against action of mind on body are really very weak; and the argument for it from experience is very strong; and, to avoid a long digression into psychophysics, I am going to assume that common sense is justified in its belief that mind can act on body—provided of course that it is the body which belongs to the mind. Now it seems clear that one great source of our belief in freedom is our conviction—a well-founded conviction, as I think—that there are in any set of circumstances more than one bodily change, any one of which the mind can and will produce if it sets itself to do so.

In saying that mind can produce changes in body, I do not mean merely that a certain change in our mental state is one of the conditions upon which there supervenes, and without which there would not supervene, a certain bodily change. What we naturally think is that, while there are certain static conditions, both of our body and of our mind, without which the bodily change would not take place, a mental act of self-exertion stands out from the background of these static conditions and is that which actually causes the bodily change to take place. To reduce causation to mere necessary sequence is to eviscerate it of a good deal of its natural meaning, and no cogent reason has ever been given for this evisceration.17

When I say this, however, I ought to add that I have in the course of this discussion frequently used the word ‘causality’ in another and wider sense which is also justified by common usage. When I have referred to the law of causality I have meant by this the principle that no change takes place in the absence of conditions upon which it necessarily follows; and in this conception of causality there is not necessarily involved the thought of one thing acting on and producing a change in another. And indeed in the causations which we have been mainly considering there is no question of one thing, i.e. one substance, acting on another. When one says that an act of will is caused by the previous desires and thoughts of the wilier, there is no question of two substances being involved; for all that happens happens in or to one substance, the individual human being in question. One is then merely asserting a necessary connexion between earlier and later events in the history of this individual. When, on the other hand, one says that a mind by an act of choice produces a change in its own body, one is, I believe, asserting something more than necessary connexion; one is asserting the existence of activity and passivity. To avoid the ambiguity, one ought perhaps to use ‘determination’ as the more general word to include both what is commonly called transeunt causation and what is commonly called immanent causation, and should restrict the word ‘causation’ to what is commonly called ‘transeunt causation’, to the action of one substance on another, whether it be that of body on body, or that of body on mind, or that of mind on body, or (if this occurs) that of mind on mind.

The mind's power of controlling the body has great importance for the moral life. If it did not exist, our moral life, our range of moral activities, would be immensely impoverished. For if a man did not think that his mind can control his body, he would not set himself to make such movements as are involved in telling the truth, in paying his debts, in helping his neighbours, and the like; and if he discovered by experience that he cannot produce such results he would soon cease to set himself to produce them. His moral life would come to be restricted to setting himself to bring about changes in the state of his own character or of his own intellect. He would have a moral life of his own, but he would be cut off from the activities which make up by far the greater part of the moral life of most men.

Yet the question whether the mind can control the body is irrelevant to the question of the freedom of the will. For if we say that the mind can control the body, we are making a statement about the effect of an act of will, and are in fact saying that a certain causal nexus exists between acts of will and consequent bodily changes; while what Libertarianism maintains is not anything with regard to the effects of acts of will, but that acts of will are not themselves caused by the pre-existing conditions. Suppose that a man had, without knowing it, lost the power of speech. He could still set himself, or make the effort, to tell the truth, or alternatively set himself to tell a lie. The activity of mind he would be exercising would be the same in the case in which the control over his vocal chords has ceased to exist, as it would be if he still had the control; and the activity of choosing between these two activities would be the same as if the control over the vocal chords still existed. Morally, both the act and the decision to act are just the same as they would be if the control over the body still existed. Thus control over the body, though it is a great part of what we usually think of when we assert our freedom to do this or to do that, is not the morally essential thing that is being claimed. The essential thing that is being claimed is the power to set oneself to do this, or to set oneself to do that, i.e. to perform a mental activity which is just the same whether it is or is not followed by the appropriate bodily movement.

Is there any foundation for the belief that I am free to set myself to produce either of two or more bodily movements? If my previous argument is correct, the belief cannot itself be true. If my nature and condition and the circumstances are such and such, I shall set myself to produce one change; if they are different, I shall set myself to produce another. Yet, if it is not true that I can here and now set myself to produce either indifferently, it is true that in a certain sense I could set myself to produce either. For within the total range of my nature there are motives which in certain circumstances would lead me to make the one attempt; and other motives which in certain circumstances would lead me to make the other. Neither act seems to me impossible for me, because neither is clean outside the range of motives which I know to exist in myself. If I knew myself completely, I should know either that the one motive or set of motives is the stronger, in which case I shall set myself to do the one act, or that the other motive or set of motives is the stronger, in which case I shall set myself to do the other. But knowing that both motives are such as I am familiar with, and knowing further that both are present in my mind at the moment, and not knowing which is the stronger, or, if I happen to know which is the stronger now, not knowing which will have emerged as the stronger by the time when I act, the only reasonable belief for me to hold is that I am capable of doing either of the two acts.

Now if in the end I do the right act, then, if the defence of Determinism which I have put forward is sound, my doing the right act implies that, just before I did it, conditions which made it necessary for me to do it were present, and similarly my doing the wrong act, if I do that, implies that conditions which made it necessary to do that act were present. If the one act was objectively, in the precise circumstances, possible for me, the other was not; and if the other was possible, the first was not. Yet all the conditions which ever justify the assertion of possibility were present justifying the judgement that both acts were possible for me. For what are the conditions that justify the judgement that A may happen? A judgement of possibility may be of three kinds. It may be a judgement of the form ‘A may be the case here and now’, or of the form ‘A may have happened in the past’, or of the form ‘A may happen in the future’. We need not trouble ourselves with any consideration of the first two types, for it is clearly the third that is relevant to the consideration of free will. Now the one essential condition needed to justify the judgement ‘A may happen’ is that I do not know of any circumstance which makes the happening of A impossible. If the general argument I have offered for Determinism is true, there is no such thing as possibility in rerum natura. If certain conditions are present, certain things will happen, and no alternative to them will happen; if certain other conditions are present, certain other things will happen, and no alternative to them will happen. Possibility is always related to a judger, and to say ‘so-and-so may happen’ is just to say ‘I don't know that it won't’. Now it often happens that I don't know that A will happen, and don't know that it won't. Then I am entitled to say that A may happen and may not happen. Now this is actually the position we are in with regard to the acts of any other person. We may have watched him in a hundred similar situations in the past and seen him always behave in one way, and we may therefore think it much more likely that he will behave in way A than in way B; but though the circumstances seem just alike to us, they may seem different to him; or his character may have improved since the last similar situation occurred to him, or it may have got worse; and it is the literal truth that we can neither say that he will do act A nor that he will do act B, but can only say that he may do either. And we are, in principle, in the same situation with regard to our own future acts, and even with regard to our acts in the immediate future. We can see that the situation is never exactly like any in which we have been in the past. It may be like in that our relation to one person who forms part of the situation is just like the relation in which we have stood to some person in the past, but it practically never happens that the whole set of people who enter into the situation is related to us just like the whole set of people who entered into previous situations we have been in, and therefore we cannot infer that because we behaved in way A in the past we shall behave in way A now. Again, we may feel convinced that at the moment the motives inclining us to do act A are stronger than any inclining us to do any alternative act. But all motives or desires are subject to a constant alternating weakening and strengthening; we are perfectly familiar with the fact that a desire which at its inception is strong weakens as its novelty wears off. Then so long as any interval, however short, separates the present moment from that at which the act will be done, we do not know what the relative strength of the different motives will be when the latter moment arrives; and therefore do not know what we shall do, though we may of course think it highly probable that we shall do a certain act rather than any other.

Thus the logical precondition of my saying ‘I can either do my duty or fail to do it’ is always present. But merely not knowing that A will not happen is not a sufficient psychological precondition of my judging that A may happen. I should be logically justified in saying, ‘Mr. Chamberlain may make Sir Stafford Cripps his next Minister of Labour’. But I should not be likely to form this judgement unless I knew or thought, for instance, that Mr. Chamberlain had socialistic leanings; unless I knew or thought that there was something in Mr. Chamberlain's psychology which might lead to such an act. Such a psychological precondition of the judgement ‘I can either do my duty or not do it’ is in fact present. For I know that the desire to do my duty is a desire that has a certain strength in me, and I also know that other desires which if followed would lead me not to do my duty have a certain strength in me, so that the possibility of my doing my duty is not (as we sometimes put the matter) a mere logical or abstract possibility, but a very real one, and the possibility of my not doing it is also such a possibility. Thus both the logical and the psychological condition of my making the judgement are present. Or, as we may also put it, it is not merely possible, but has an appreciable probability, that I shall do my duty, and it is not merely possible, but has an appreciable probability, that I shall not do it. Therefore the judgement ‘I may either do my duty or not do it’ is fully justified.

The thought ‘I can do my duty’ has also a further justification. The practical utility of a belief is never, indeed, a sufficient reason for holding that belief. If I did not on other grounds think that I can do my duty, the thought that I should be more likely to do my duty if I believed I could do it would not lead me to believe that I could; and if it did, the belief would be unjustifiable, since there would be no logical connexion between the psychological cause of the belief and the belief itself. But if a belief is on other grounds justifiable, the fact that keeping the belief in our minds would have good results is a good reason for keeping the belief in our minds. Now I have tried to show that the belief that we can do our duty is on its own merits justified. I certainly do not know that I shall not do my duty, and I know that there is in me a motive, viz. sense of duty, which makes it appreciably probable that I shall do it. And the keeping of this belief before my mind will in fact make it more likely that I shall do my duty. Consider the effect of holding the opposite belief. Suppose I believed that I could not do my duty; then I should be thinking, in effect, this: ‘Though the act is the right one in the circumstances, and one which a better man ought to do, yet since I can't do it there is no use in my trying to do it.’ I should resign the struggle, and do the act which most appealed to me on other grounds. But if I keep before me the thought ‘that is the right thing to do, and I don't know of anything that makes it impossible for me to do it, and do know of something, viz. the sense of duty in me, which makes it appreciably probable that I shall do it’, the wish to do my duty is kept alive and allowed to have its full weight in determining my action.

It is true that the thought ‘it is possible for me not to do my duty’ is equally justified logically, and that this also is not a mere idle or abstract possibility, since I know there are motives in me inclining me to do something other than my duty. And it might at first sight seem that this thought would have as great a tendency to depress us and damp down our moral activity as the complementary thought ‘it is possible for me to do my duty’ has to intensify our moral activity, and would simply neutralize the effect of the latter. This, however, is far from being the case. The joint thought ‘it is possible that I shall, and possible that I shall not, do my duty’, which is the thought that the facts of the case justify, is precisely that which is most favourable to the doing of our duty when the time comes. If we felt certain that we should do it, that would simply encourage us to take things too much for granted and to neglect the concentration on the thought of duty which alone will secure our doing it when the time comes. If we felt certain that we should not do it, that again would discourage us from paying any attention to the thought of duty. The recognition that we may either do or not do our duty is just that which is needed to induce us to keep our moral armour in the best possible repair.

I may illustrate this by a homely analogy. Of the candidates in an examination, those who are most likely to make the needed effort are neither those who feel sure that they will pass nor those who feel sure that they will fail, but those who know that they do not know whether they will pass or fail. And similarly the thought that it is possible that I shall do what is right and also possible that I shall do what is wrong is precisely that which is most conducive to my exercising that sort of control over my thoughts and desires which is in turn most conducive to my doing what is right, provided of course that a wish to do what is right is also present. This thought, then, is not only true, but one which it is very desirable to keep constantly before one's mind.

What, then, is the difference between the perfectly proper statement which I may make ‘it is possible that I may presently do this, and possible that I may presently do that’, and the libertarian claim? Possibility is always compossibility. What I am saying when I make the above statement is that there are no existing conditions known to me with which my doing A is incompatible, and none with which my doing B is incompatible; and I probably should not make the statement unless I knew some important conditions to be actually present with which each action is compatible. What the libertarian says is that all the conditions, known and unknown, are compatible with my doing the one action, and also with my doing the other; and this cannot be true.

But the phrase ‘it is possible that I shall do what is right, and possible that I shall do what is wrong’ does not do full justice to our actual thought when it takes its usual form ‘I can do what is right’, though it expresses part of what that thought involves. I could be content with that phrase only if I thought of myself as a mere spectator of the play of forces within my mind, as a looker-on at a game between fairly equal sides may say ‘this side may win, or that side may win’. The fact that the spectator is also the agent makes the actual situation different. We say not merely ‘I may do right, and I may do wrong’, but ‘I can do right, and I can do wrong’. What exactly does this mean?

When I say ‘I can do this’, which I will now use as a brachylogy for ‘I can perform this mental activity of self-exertion’,18 I am not claiming that all the conditions necessary to my doing this are already present. If they were, I should be ‘doing this’, whereas the claim ‘I can do this’ is always made about something which one is at least not yet doing, and which perhaps one will never in fact do. If I say ‘I can do this’, either I am not doing it, or at least I suppose I am not doing it, and if so, I cannot be supposing that all the conditions necessary to my doing it are already present. But I am claiming that some of those conditions are present, and present in me. I am in fact saying ‘certain of the conditions of my doing this are already present in me, and if certain conditions not yet present are added I shall actually do this’. Now sometimes when I say ‘I can do this’, I could specify certain external conditions which must be fulfilled if the power is to be translated into act. If I say ‘I can walk a mile in thirteen minutes’, I should if I wanted to be more precise say ‘I can walk a mile in thirteen minutes if the road is not too uphill, if there is not too strong a wind against me, if I am not stopped on the way, if I am not too tired when I start’, &c. But when ‘doing this’ is not the effecting of a certain bodily movement or set of movements, as walking is, but is the purely mental activity of setting oneself to do a certain thing, such external conditions are irrelevant. Yet there must be some unfulfilled condition of my performing this activity of ‘setting myself’; else I should be doing it, and not merely saying ‘I can do it’. The general formula must hold good, that ‘I can do this’ is really a hypothetical proposition—‘if a certain condition is added to conditions already present in me, I shall do so-and-so’. And it is easy to see what the additional condition must be. It consists of my wishing to do this. ‘I can do this’ means ‘I have such a nature that if I want to perform the activity of setting myself to do this, I shall perform it’; and ‘I can refrain from doing this’ means ‘I have such a nature that if I want to refrain from doing this, I shall refrain from doing it’. Or, putting it briefly, ‘I can do this or that’ means ‘I shall do this if I want, and I shall do that if I want’—‘want’ being here a brachylogy for ‘want predominantly’. Now this claim is absolutely correct. For we have verified in experience over and over again two things, (1) that we have a faculty of setting ourselves to bring about changes, and (2) that this is exercised when we predominantly want to bring about these changes. Thus the claim which we instinctively make that we can perform either of two (or more) acts of setting ourselves to effect changes, seems to be absolutely correct, when we expand it to bring out what it really means. But what it points to is not the libertarian but the deterministic account; for we clearly imply that if the capacity of setting ourselves to effect changes has added to it a predominant wish to bring about a certain change, the setting oneself to bring it about will necessarily follow.

Suppose it is said that I am whittling away the claim of ordinary common sense, by adding the words ‘if I want’. Suppose it is said that the claim is a claim to power not subject to this condition. If this were so, the claim would, I think, have to take one or other of two forms: (1) ‘I can do this whether I want or not.’ But does any one make this claim? Does any one really think that he could, for instance, set himself to do his duty if he did not wish to do his duty? Suppose that some one suggests that the motive of a dutiful act is not a wish but the emotion of respect or reverence which the thought of duty inspires, he will only be substituting a new condition for the condition ‘if I want’ which I have suggested. He will be paraphrasing the phrase ‘I can do my duty’ by the phrase ‘I shall do my duty if I am sufficiently under the influence of the motive of reverence’. He cannot think that one who says ‘I can do my duty’ is claiming that he can or will do it in the absence of this emotion as well as of desire.

(2) The second possible interpretation is ‘I have in me already all the necessary conditions of my doing the thing, except that of wanting to do it, and I can produce this condition’. To this interpretation two fatal objections can be made. (a) We do not in the least think that we can produce wishes in ourselves by an act of choice, and (b) if this were the right analysis of the claim ‘I can do act A’, then the right analysis of the statement ‘I can produce the condition of want’ would be ‘I have in me already all the conditions of producing the condition of want, except the wish to produce this condition, and I can produce this wish’. Thus the claim ‘I can do this’ would involve the claim ‘I can produce the wish to do this’; that would imply the claim ‘I can produce the wish to wish to do this’; and so ad infinitum—a regress which quite clearly is not involved in our simple claim ‘I can do this’.

Thus both the attempts to remove the condition ‘if I want’ fail, and our analysis of the claim ‘I can do this’ remains good—that it means ‘I have in me a general capacity of setting myself to do things, and if I want to bring about change A, this capacity+this desire will lead to my setting myself to produce change A’. And if we thus interpret the claim ‘I can do act A’, it is strictly true that I can do act A and that I can refrain from doing it. Thus the instinctive feeling that we can either do an act or refrain from doing it (or do some alternative act) is thoroughly justified. But it does not in turn justify the libertarian account. For it involves the assertion of the determined sequence of action on desire. As regards the origin of the desire it says nothing, but it certainly does not claim that desire is originated by an act of free choice, and in fact we never think of our desires as so originated, though we do think of our acts as so originated.

Thus what is claimed in the natural statement ‘I can either do act A or leave it undone’ is that my doing or not doing act A depends entirely on myself. And this is in a sense true, because, while my effecting change B in my body, and indirectly change C in something beyond my body, depends on conditions in my body, and in that thing beyond my body, act A, being a purely mental activity of setting myself to effect changes B and C, cannot depend for its occurrence on the extent to which my body happens to be under the control of my mind.

Of course we are not claiming a complete non-dependence of our acts of self-exertion on any conditions outside our mind. Most, at least, of our wants would never arise if they were not suggested to us by the perception of bodies outside us, and this is true even of the desire to do certain things as being our duty. For it is by the use of our bodily senses (though not by that alone) that we become aware of the existence of selves other than our own, and thus of duties to other selves. The independence that is claimed is the non-dependence of our acts on any immediately preceding condition except the general capacity of setting ourselves to effect changes, and the wish to effect this or that change—both of which are conditions in our mind and nowhere else.

Thus a great deal of what is claimed in the claim to free will—not only the thought that either issue is still possible, but the thought that the issue will depend on us, our minds, our wishes, not our bodies, nor other minds or bodies, is absolutely true, as well as absolutely vital to the moral life. Again, the thought that the relative strength of our wishes is not already fixed once for all is both true and valuable. But it is a thought which a determinist can admit, as well as a libertarian. A determinist is not in the least bound to say that the effort after self-improvement is fruitless. He can admit quite freely that in a character that is bad on the whole there may yet be an element of desire for better things which, weak at first, may by the influence of example and of teaching, and of the effort after self-improvement which example and teaching may arouse, become the strongest element in the character.

The controversy between Libertarianism and Determinism is apt to present itself as one in which the metaphysical argument is in favour of Determinism and all the ethical arguments in favour of Libertarianism. This is far from being the case. It seems to me instructive to reflect, in this connexion, on two features of the moral situation. The one is our reliance on the characters of other people, the reliance implied in the use of such words as ‘trustworthy’. The other is our attitude when people surprise us by their behaviour. The reliance which we place on the decent behaviour of people whom we know and trust is so much evidence that we think that their actions, when they come to be done, are the result not of a will acting independently of their present character, but of the same continuing character which we have seen at work before. We are in no way detracting from the moral status of our friend if we say ‘I knew you could be trusted to do the right thing’. ‘Know’ is no doubt an exaggeration; for every heart has its secrets which no one knows, and the friends we know best may have their secret weakness which no former situation has revealed but which a new one will. But a confidence amounting almost to certainty is a tribute to our friend's moral worth, not a detraction from it; yet it is not compatible with the libertarian's view that any one may at any moment make a choice which is quite independent of his whole pre-existing character.

Again, if some one behaves, for better or for worse, in a way different from that which we had confidently expected, what is our reaction to this? We do not put down the unexpected act to the credit or discredit of an unmotived choice. We always assume that there must have been, before the act, some existing but hitherto unknown trait of character which has now, perhaps for the first time, manifested itself in act. In our reaction to people's unexpected behaviour, no less than in our expectation of their behaviour, we betray the conviction that action is the result of continuing, even if constantly modified, character.

Another feature of our ordinary moral thought which really harmonizes better with the deterministic than with the libertarian account is the importance for good or evil which we attach to the formation of habits. We all think that if we repeatedly behave in a certain way we shall make it more likely that we shall go on behaving in that way, and more difficult for ourselves to behave otherwise. An attempt may be made to harmonize this with Libertarianism by saying that habits ‘incline without necessitating’. But to describe them thus is to imply that a habit becomes one of the influences which operate on the will; and this is inconsistent with the thought of a transcendental will standing apart from the formed character and free to operate independently of it. The thought that habits incline without necessitating is in itself perfectly correct. It is simply one way of expressing the fact that a habitual tendency to behave in a certain way forms one element, and an important element, in the total character from which future action will spring, while yet there may coexist with it some other element which may prove stronger; some long unfulfilled but not extinguished longing after good which, brought to the surface by some feature of a new situation, may overcome a bad habit, or some lingering weakness which may lead us to yield to a new temptation though we have overcome many others. The thought of habits as inclining without necessitating, which is the true way of thinking of them, is also the most salutary, since it frees us from the despair which would overtake us if we thought of bad habits as completely necessitating, and from the carelessness that would come over us if we thought of them as leaving us as free to do well as we were before their formation.

I may add that the judgements which we make about the characters of other people (or about our own characters), in distinction from judgements about particular acts, imply the view that action is determined by character. For our judgements about character are mostly19 based on observation of actions, and we are justified in drawing inferences from people's actions to their characters only if their actions flow from their characters. When we call a man a bad man we do not mean that he is a man who has done more bad acts than good (for that might be equally true of a reformed sinner), but that he still has substantially the same character which was evidenced by bad acts in the past and may be expected to be evidenced by more in the future. In fact Libertarianism is inconsistent with belief in the continuity of human character; but our actual moral judgements are evidence that we do believe in its continuity, though we think it a continuity that admits of modification for better or for worse.

I am far from contending that the whole of our ordinary thought about moral action is reconcilable with the doctrine of Determinism. But it is worth while to point out that it is by no means true that all the arguments drawn from the moral consciousness tell in favour of Libertarianism, and only the metaphysical argument tells in favour of Determinism.

If my line of argument is right, we must find the uniqueness of moral behaviour not in freedom from the general law of causation, but in the unique character of the activities which constitute such behaviour—the activity of choosing or deciding, and the activity of setting oneself to do what one has decided to do, to which there is no analogy in the behaviour of any physical thing; and in the further fact, to which there is nothing analogous in the behaviour of a mere animal, that one of the thoughts under whose influence one can choose and set oneself to act is the thought of an action as right.

Besides the ‘intuition’ of freedom, the other main reason which leads people to believe in the freedom of the will is the thought of responsibility for our acts, which is involved in the facts of remorse, blame, and punishment. These, it might seem, are unjustifiable and indeed unintelligible, unless we are really free to do either of two or more different acts. We must therefore set ourselves to examine the implications of these things. Professor Nicolai Hartmann20 has argued that of these three phenomena, remorse is the one which is most clearly an evidence of free will. Blame and punishment, praise and reward, might, he argues, perhaps be explained as devices adopted for the encouragement of men to future good acts and their restraint from future bad acts, without involving a genuine imputation of freedom; but there can be no such utilitarian explanation of the free assignment by a man to himself of responsibility for his past acts, and of the accompanying remorse.

I think it possible that a society which had ceased to believe in the responsibility of individuals for their acts might retain praise and blame, reward and punishment, as utilitarian devices for the encouragement of virtue and the restraint of vice. But two comments may be made on this. In the first place, I think we should agree that the denial of responsibility is not the assumption on which we actually praise and blame, reward and punish. Our actual assumption is a belief in responsibility. And secondly, we should think it somewhat dishonest to continue to practise praise and blame, reward and punishment, if we had lost the belief in responsibility. We should be treating people as if they were responsible, when we had really ceased to believe that they were.

I think, therefore, that we need not isolate remorse from praise and blame, reward and punishment, but should treat all alike as involving a belief in individual responsibility. The question we must now face is whether this belief is compatible with the Determinism which we have been led on other grounds to believe in, or whether it involves freedom of indifference. I may begin by pointing out that responsibility is always divided. Obviously responsibility for results of action can never be assigned to one person alone; there are always circumstances (and these will usually include acts by other people) which co-operate to produce the result, of whose complete cause one person's act is merely the most striking part. But it is also true that responsibility for acts is divided. It is never right to assign to one person the sole credit or the sole discredit for any of his acts. Other people by teaching and example, the writers of the books he has read, and so on, have all helped to mould his character into that form of which his action is the expression. But it is equally certain that the sole responsibility for any act can never be assigned to any person or persons other than the doer of the act. For acts spring from opinion and desire, and there is no possibility of forcibly implanting either an opinion or a desire in the mind of another. No opinion or desire will find a lodgement in his mind unless his mind accepts it, or rather responds to suggestion with a reaction which is all his own.

To recognize, then, that people and things other than the agent have been part causes of his act may be ground for mitigating the severity of our blame or the enthusiasm of our praise, but it is never in itself sufficient ground for withdrawing praise or blame from him altogether. Other people, and outside circumstances, have never been more than part causes of his act; the act is the reaction of his character to them, and his character is partly responsible.

Mr. Wisdom has suggested21 that there is one and only one way in which a belief in responsibility can be reconciled with a belief in the law of causation. If the individual soul, he argues, is regarded as having been called into being, at the moment of conception, either by the act of its parents, or by the act of God, then all its subsequent acts are in the long run determined by events entirely beyond its own control, viz. the events that happened before it began to exist. But if it has existed from all eternity, then all its acts may be regarded as determined by previous conditions of which acts of its own formed some part, and its partial responsibility for all it does is thus preserved.

A belief in the eternal pre-existence of the soul no doubt has much to commend it on other grounds. Not only moral considerations, but our whole way of thinking of the soul, involve that the soul is something other than a succession of its states and acts, that it is something that has these states and does these acts; in other words, that it is a substance. Memory implies the thought, and if it is ever itself a form of knowledge it then implies the knowledge, that the very self which remembers also had the experience which it remembers, and thus establishes the existence of a soul-substance lasting as far back from the present as the earliest experiences we can remember. And if we grant the existence of a soul-substance going so far into the past as this, the belief in its infinite pre-existence is at least no harder to accept than any of its alternatives. The alternatives seem to be (1) the emergence of the soul from purely physical antecedents, or (2) its emergence by emanation or fission from some other soul or souls—either those of the parents or that of God, or (3) its creation by God out of nothing. Can we seriously accept any of these alternatives? Emergence by fission seems to be the only one of the three that is not incredible; and it plainly has difficulties of its own. Furthermore, it should be clear that pre-existence is the hypothesis that goes best with the belief in immortality.

There is thus a great deal to be said, on general grounds, for the belief in the eternal pre-existence of the soul. But it is evident that there are great difficulties in the attempt to understand how and why an eternally existing soul comes to inhabit this or that embryo organism. And, further, it appears to me that, while the doctrine of pre-existence gives a sort of juridical justification of responsibility, in the sense that it leaves the individual with a part-responsibility for all his acts, which he cannot shift on to any one but himself, it is only a juridical justification that it gives, and not one that answers to or accounts for our natural thought about responsibility.

What would be the natural reply of one who wants to assert the responsibility of the individual, to the suggestion that an eternally pre-existent soul can be responsible while a non-eternal soul cannot? I think he would say that at two points the suggestion fails to satisfy him. The eternally pre-existing soul had no opportunity of choosing its own character ab initio, any more than a non-eternal soul can have had one; there never was a time at which a characterless soul chose to become a soul of a certain character, for the eternally existing soul has always had a character, and its choice of how it should develop must have been limited by the character it already had. It is true that it cannot blame God or its parents for any badness there may be in its character, but will it not instead blame the nature of things, which has prevented it from ever having a completely free choice of character? But further, even if Mr. Wisdom could allow that the eternal soul had sometime in the past had a free choice of character, its character at each moment thereafter is inherited from its past, and its character at each moment determines its action at that moment, while what our natural thought craves for is some element in the self which can at any moment act contrary to the character it inherits from its past—can act, as it is sometimes expressed, in the line of greatest resistance. Thus the pre-existence of the soul really does nothing to remove the reluctance which our natural thought has to accepting a deterministic theory. It is not without significance that the Eastern religions which believe in eternal pre-existence are more fatalistic and less inclined to assert individual responsibility than the Christian religion, which, in its orthodox form at least, does not believe in eternal pre-existence.

This attempt to reconcile responsibility with Determinism can, then, hardly be deemed successful. And, holding fast to Determinism, I am inclined to think that the only account we can give of responsibility is this: that bad acts can never be forced on any one in spite of his character; that action is the joint product of character and circumstances and is always therefore to some extent evidence of character; that praise and blame are not (though they serve this purpose also) mere utilitarian devices for the promotion of virtue and the restraint of vice, but are the appropriate reactions to action which is good or is bad in its nature just as much if it is the necessary consequence of its antecedents as it would be if the libertarian account were true; that in blaming bad actions we are also blaming and justifiably blaming the character from which they spring; and that in remorse we are being acutely aware that, whatever our outward circumstances may have been, we have ourselves been to blame for giving way to them where a person of better character would not have done so. I cannot pretend that this satisfies the whole of our natural thought about responsibility, but I think that in claiming more, in claiming that a moral agent can act independently of his character, we should be claiming a metaphysical impossibility.

A philosophical genius may some day arise who will succeed in reconciling our natural thought about freedom and responsibility with acceptance of the law of causality; but I must admit that no existing discussion seems to be very successful in doing so. The most recent elaborate discussion known to me, that by Professor N. Hartmann, while it does justice to the strength of the case for Libertarianism, shows little appreciation of the strength of the case for Determinism.22

  • 1.

    Indeterminism, Formalism, and Value, 135–96.

  • 2.

    Ibid. 144.

  • 3.

    G. Birtwistle, The New Quantum Mechanics, 276–7.

  • 4.

    P. A. M. Dirac, The Principles of Quantum Mechanics 2, 103.

  • 5.

    J. Frenkel, Wave Mechanics, Elementary Theory2, 48–9.

  • 6.

    H. Dingle, Through Science to Philosophy, 289.

  • 7.

    Zeitschrift für Physik, 43 (1927), 172–98.

  • 8.

    Ibid. 197.

  • 9.

    q, p, ∼ h (ibid., p. 175); i.e. the product of the indeterminacy of the position of an electron, multiplied by the indeterminacy of its velocity, is constant.

  • 10.

    M. Planck, Where is Science Going?, 201.

  • 11.

    Ibid. 145; cf. ibid. 210. I may refer also to two contributions in the Travaux du IXe Congrès International de Philosophie—those by M. M. Barzin (vii. 15–20) and by Lord Samuel (vii. 21–7).

  • 12.

    Indeterminism, Formalism, and Value, 168–72.

  • 13.

    Ibid. 173.

  • 14.

    I have not discussed Mr. Braithwaite's contribution to the Symposium, because the other two seem to have raised sufficiently the points in dispute.

  • 15.

    Indeterminism, Formalism, and Value, 146–9.

  • 16.

    See, for instance, the discussion in Mr. Wisdom's Mind and Matter, 65–102.

  • 17.

    For a defence of the common-sense view of causality against the criticism of Hume see G. F. Stout, Mind and Matter, 15–36.

  • 18.

    This alone is in question, since, as I have pointed out (pp. 231–4), the mind's control over the body is not what is in question when we are discussing the freedom of the will.

  • 19.

    Only mostly, because some judgements about character are based on observation of the expression on people's faces, their gestures, &c.

  • 20.

    Ethik,2 673–8 (Eng. tr. iii. 172–8).

  • 21.

    In Mind and Matter, 122–30.

  • 22.

    There is in Mind, xliii (1934), 1–27, a very persuasive attempt by Mr. R. E. Hobart to show that, quite apart from metaphysical considerations, our natural thought about moral questions implies a belief in Determinism. He is, I think, rather sanguine in believing that he has shown the whole controversy about free will to be based on confusions; but I do not know of any better exposition of the extent to which our moral thought involves the belief that actions are the expression of, and flow necessarily from, a definitely characterized and continuing though modifiable self.