The traditional problem of freewill has been so adequately covered in recent philosophical literature that some excuse must be offered for re-opening it, and I do so because, although I believe that the traditional problem has been solved, I believe also that the solution leaves open certain further problems that are both interesting and important. It is to these problems that I propose to devote most of this paper ; but, even at the risk of flogging dead horses, I feel bound to say something about the traditional problem itself. The problem arises out of a prima facie incompatibility between the freedom of human action and the universality of causal law. It was raised in an acute form when universal determinism was believed to be a necessary presupposition of science; but it was not then new, because the incompatibility, if it exist at all, exists equally between human freedom and the foreknowledge of God. As it appears to the 'plain man ' the problem may be formulated as follows: "Very often I seem to myself to be acting freely, and this freedom, if it exists, implies that I could have acted otherwise than I did. If this freedom is illusory, I shall need a very convincing argument to prove that it is so, since it appears to be something of which I am immediately aware. Moreover, if there is no freedom, there is no moral responsibility; for it would not be right to praise or blame a man for something that he could not help doing. But, if a man could have acted otherwise than he did, his action must have been uncaused, and universal determinism is therefore untrue."

Broadly, there are two methods of resolving this, as any other, antinomy. We can either assume that the incompatibility is a genuine one at a certain level of thought and try to resolve it at a higher plane in which either or 'both the terms 'freedom' and necessity ' lose their ordinary meaning or we can try to show by an analysis of these terms that no such incompatibility exists. If the latter method is successful, it will show that what is essential in our concept of freedom does not conflict with what is essential in our concept of causal necessity and that the incompatibility arises only because, at some stage in our development of one or both of these concepts, we have been tempted into making a false step. This method seems to me the better, (provided always that it is successful), on the ground that it does not resort to any metaphysical conception imported ad hoc to solve this problem, which might be objectionable on other grounds. In the first two sections of this paper 1 shall give a brief outline of the analysis of the problem that I believe to be correct; and for this analysis I claim no originality. The method of presentation will, however, throw into relief the partial nature of the solution and help to indicate the further problems to be discussed in the last three sections.

Freedom, so far from being incompatible with causality, implies it.(R.E.Hobart, Mind, 1934, p.1) When I am conscious of being free, I am not directly conscious that my actions are uncaused, because absence of causation is not something of which one could be directly aware. That the plain man and the Libertarian philosopher are right in claiming to know directly the difference between voluntary and involuntary actions, at least in some cases, I have no doubt; but we can never have this direct knowledge that something is uncaused, since this is a general proposition and, like other general propositions, could only be established by reflection on empirical evidence. Fortunately it is not necessary here either to attempt an analysis of causality or to answer the question whether or not it is a necessary presupposition of science. It is now widely recognised that the considerations which lead scientists to suppose that strict determinism is not true are irrelevant to the problem of freewill, since these considerations lend no support to the view that the phenomena with which we are concerned are not predictable; and it is to predictability, not to any special theory of the grounds of predictability, that the Libertarian objects. He claims that if our actions are predictable we are 'pawns in the hands of fate' and cannot choose what we shall do. If, it is argued, someone can know what I shall do, then I have no choice but to do it.

The fallacy of this argument has often been exposed and the clearest proof that it is mistaken or at least muddled lies in showing that I could not be free to choose what I do unless determinism is correct. There are, indeed, grounds for supposing that strict determinism in psychology is not correct; but this, if true, constitutes not an increase but a limitation of our freedom of action. For the simplest actions could not be performed in an indeterministic universe. If I decide, say, to eat a piece of fish, I cannot do so if the fish is liable to turn into a stone or to disintegrate in mid-air or to behave in any other utterly unpredictable manner. It is precisely because physical objects have determinate causal characteristics that we are able to do what we decide to do. To this it is no answer to say that perhaps the behaviour of physical objects is determined while that of volitions is not. For we sometimes cause people to make decisions as well as to act on them. If someone shouts: " Look out! There is a bull", I shall probably run away. My action is caused by my decision to run; but my decision is itself caused by my fear, and that too is caused by what I have heard. Or, again, someone may try to influence my vote by offering me a bribe. If I accept the bribe and vote accordingly, the action is caused by the bribe, my avarice and my sense of obligation to the donor; yet this would certainly be held to be a blameworthy action, and therefore a voluntary one. A genuinely uncaused action could hardly be said to be an action of the agent at all; for in referring the action to an agent we are referring it to a cause.

In calling a man 'honest' or 'brave' we imply that he can be relied on to act honestly or bravely, and this means that we predict such actions from him. This does not mean that we can predict human actions with the same degree of assurance as that with which we predict eclipses. Psychology and the social sciences have not yet succeeded in establishing laws as reliable as those that we have established in some of the natural sciences, and maybe they never will. But any element of unreliability in our predictions of human actions decreases rather than increases the reliability of our moral judgements about them and of our consequent attributions of praise and blame. An expert chess-player has less difficulty in defeating a moderate player than in defeating a novice, because the moves of the moderate player are more predictable ; but they could hardly be said to be less voluntary. In calling an action 'voluntary' we do not, therefore, mean that it is unpredictable; we mean that no-one compelled the agent to act as he did. To say that, on the determinist view, we are 'mere pawns in the hands of fate' is to confuse causality with compulsion, to confuse natural laws (descriptions) with social laws (prescriptions) and to think of fate as a malignant deity that continually thwarts our aims. What the protagonist of freedom requires, in short, is not uncaused actions, but actions that are the effects of a peculiar kind of causes. I shall be as brief as possible in saying what these causes are, since this has often been said before. But it is one thing to state the criteria by which we decide whether or not an action is voluntary and another to say why this distinction is important for ethics. The problem which the analysts have not, in my view, sufficiently considered is that of analysing the peculiar relation of 'merit' or 'fittingness' that is held to exist between voluntary actions and moral responsibility.

If someone overpowers me and compels me to fire a gun which causes a death, I should not be held guilty of murder. It would be said that my action was not voluntary; for I could not, had I so wished, have acted otherwise. On the other hand, if I kill someone because I hope to benefit under his will, my action is still caused, namely by my greed; but my action would be held to be voluntary and I should be blamed for it. The criterion here is that, while in the first case the cause is external to me, in the second it is my decision. A similar criterion would be used to distinguish a kleptomaniac from a thief. A kleptomaniac is held to be one who steals without having decided to do so, perhaps even in spite of a decision not to do so. He is not held morally responsible for his action because his action is not held to be voluntary. But in this case it is not true, as it was in the last, that his action is called involuntary because it is caused by some outside force. The cause of kleptomania is obscure; but it is not external compulsion. And, if the cause is not external, how can we say that the kleptomaniac is `compelled'? As used by psychologists, the term 'compulsion' is evidently a metaphor, similar to that by which we speak of a man's doing something when 'he is not himself '. Evidently our moral judgements imply not merely a distinction between voluntary and compelled actions but a further distinction among actions that are not compelled.

A third example will make it clear that some such distinction must be made. Suppose that a schoolmaster has two pupils, A and B, who fail to do a simple sum correctly. A has often done sums of similar difficulty before and done them correctly, while B has always failed. The schoolmaster will, perhaps, threaten A with punishment, but he will give B extra private tuition. On the traditional view his action might be explained as follows: "A has done these sums correctly before; therefore he could have done them correctly on this occasion. His failure is due to carelessness or laziness. On the other hand, B is stupid. He has never done these sums correctly; so I suppose that he cannot do them. A's failure is due to a moral delinquency, B's to an intellectual defect. A therefore deserves punishment, but B does not." This is, I think, a fair summary of what the 'plain man' thinks about a typical case, and the points to which I wish to draw attention are these :-

(a) Neither failure is said to be 'uncaused'.
(b) The causes assigned are divided into two classes, moral and. intellectual. (Cases of physical deficiency, e.g. not being strong or tall enough would go along with the intellectual class, the point being that such deficiencies are non-moral.)
(c) Praise and blame are thought appropriate to moral but not to non-moral defects.
(d) The criterion for deciding whether a defect is moral is 'Could the agent have acted otherwise ?'

I do not wish to suggest that the reasoning attributed to the plain man in this case is in any way incorrect, only that, particularly in regard to points (c) and (d) it needs explaining.

It is evident that one of the necessary conditions of moral action is that the agent 'could have acted otherwise' and it is to this fact that the Libertarian is drawing attention. His case may be stated as follows: " It is a well-known maxim that 'I ought' implies 'I can'. If I cannot do a certain action, then that action cannot be my duty. On the other hand, 'I ought' as clearly implies 'I need not'; for if I cannot possibly refrain from a certain action, there can be no merit or demerit in doing it. Therefore, in every case of moral choice it is possible for the agent to do the action and also possible for him not to do it; were it not so, there would be no choice; for choice is between possibilities. But this implies that the action is uncaused, because a caused action cannot but occur." The fallacy in this argument lies in supposing that, when we say 'A could have acted otherwise', we mean that A, being what he was and being placed in the circumstances in which he was placed, could have done something other than what he did. But in fact we never do mean this; and if we believe that voluntary action is uncaused action that is only because we believe erroneously that uncaused action is a necessary condition of moral responsibility. The Libertarian believes that an action cannot be a moral one if the agent could not have acted otherwise, and he takes no account of possible differences in the causes that might have prevented him from acting otherwise. The Determinist, on the other hand, holds that the objective possibility of alternative actions is an illusion and that, if A in fact did X, then he could not have done any action incompatible with X. But he holds also that differences in the various causes that might have led to X may be of great importance and that it is in fact from the consideration of such differences that we discover the criterion by which we judge an action to be voluntary, and so moral.

We all blame Nero for murdering Agrippina, and the Libertarian holds that this implies that Nero could have abstained from his action. But this last phrase is ambiguous. Even if we admit that it would have been impossible for anyone to predict Nero's action with the degree of assurance with which we predict eclipses, yet an acute observer at Nero's court might have laid longish odds that Nero, being what he was and being placed in the circumstances in which he was placed, would sooner or later murder his mother. To say that Nero might have acted otherwise is to say that he could have decided to act otherwise and that he would have so decided if he had been of a different character. If Nero had been Seneca, for example, he would have preferred suicide to matricide. But what could 'If Nero had been Seneca . . . possibly mean ? Unfulfilled conditionals in which both terms are names of individuals constitute, admittedly, a thorny philosophical problem; but it is clear, I think, that if 'If Nero had been Seneca' means anything at all, it is a quasi-general proposition which can be analysed either as 'If Nero had had the character of Seneca' or 'If Seneca had been emperor' or in some similar fashion. None of these analyses are incompatible with the Determinist's contention that, as things stood, Nero could not have abstained. But, adds the Determinist, the cause of his inability to abstain was not external compulsion nor some inexplicable and uncharacteristic quirk. His action was predictable because it was characteristic, and it is for the same reason that he is held to blame.

But the Libertarian's case is not yet fully answered. He might reply: "But, on this analysis, I still cannot blame Nero which in fact I do, and feel that I do justly. If the murder was caused by his character, he may not have been to blame. For his character may have been caused by hereditary and environmental factors over which be had no control. Can we justly blame a man if his vicious actions are due to hereditary epilepsy or to the influence of a corrupt and vicious court?" To this the answer is that we can and do. So long as we persist in supposing that, to be moral, an action must be uncaused, we can only push the moral responsibility back in time; and this, so far from solving the problem, merely shows the impossibility of any solution on these lines.

This is made abundantly clear by Aristotle's discussion of the subject, which I shall paraphrase somewhat freely in order to show that it must raise a difficulty which Aristotle does not squarely face. Aristotle says that, if a man plead that he could not help doing X because be was 'the sort of man to do X', then he should be blamed for being this sort of man. His character was caused by his earlier actions, Y and Z, that made him the sort of person who would, in the given situation, inevitably do X. But suppose the criminal pleads that at the time of doing Y and Z he did not know that these were vicious actions and did not know that doing vicious actions causes a vicious character? Then, says Aristotle, all we can say in such a case is that not to know that actions create character is the mark of a singularly senseless person. But this is clearly inadequate. For the criminal might proceed: "Very well then, I was a singularly senseless person; I neither knew that Y and Z were vicious actions nor that, if I did them, I would become the sort of person to do X. And, anyhow, at the time of doing Y and Z, I was the sort of person to do Y and Z. These actions were just as much caused as was X. You say that blaming me for doing X is really blaming me for having done Y and Z. Now apply the same argument to Y and Z and see where it leads you. Furthermore my ignorance at the time of doing Y and Z which, according to you, is the real source of the trouble, was not my fault either. My father did not have me properly educated. Blame him, if you must blame somebody; but he will offer the same reply as I have done, and so ad infinitum." This argument carries no conviction; but it admits of no reply, and it is here that the temptation to invoke a metaphysical deus ex machina becomes inviting. If we proceed on the assumption that, to be moral, an action must be uncaused, either we shall find a genuinely uncaused action at the beginning of the chain or we shall not. If we do not, then, according to the Libertarian, there can be no moral praise and blame at all (and it was to account for these that Libertarianism was invented); and, if we do, then we must suppose that, while almost all our actions are caused, and therefore amoral, there was in the distant past some one action that was not caused and for which we can justly be praised or blamed. This bizarre theory has in fact been held; but the objections to it are clear. We praise and blame people for what they do now, not for what they might have done as babies, and any theory of moral responsibility must account for this. Secondly the same man is subjected to judgements both of praise and of blame; therefore the subject of these judgements cannot be one solitary act, and, thirdly, even if we were able to discover this one hypothetical infantile act, would it in fact be a fit subject for either praise or blame? If it were genuinely uncaused, it could hardly be either, since it would not be an action of the agent.

So far we have discovered nothing more startling than the fact that moral actions are the effects of a peculiar kind of causes, namely the voluntary actions of the agent. To sum up this part of the argument, I cannot do better than quote the words of Prof. Ayer: " To say that I could have acted otherwise is to say, first, that I should have acted otherwise, if I had so chosen ; secondly, that my action was voluntary in the sense in which the actions say of the kleptomaniac are not; and, thirdly, that nobody compelled me to choose as I did: and these three conditions may very well be fulfilled. When they are fulfilled, I may be said to have acted freely. But this is not to say that it was a matter of chance that I acted as I did, or, in other words, that my action could not be explained. And that my actions should be capable of being explained is all-that is required by the postulate of determinism." With this I agree; but it leaves unsolved what is perhaps the most important part of the problem. Granted that we sometimes act 'freely' in the sense defined by Ayer, in what sense is it rational or just or moral to praise or blame voluntary actions but not involuntary ones? It is surely not enough to say: 'Actions of such-and-such a kind are given the name 'voluntary' and 'are praised and blamed ; others are not.' We need to explain the relation of 'fittingness' that is held to obtain between voluntary actions and moral judgement. Suppose that A and B each kill some one. We apply Ayer's tests and decide that A's was a voluntary action and B's not. We hang A, and B is immured in an asylum or regains his liberty...