But now we must ask how it is that I come to make my choice. Either it is an accident that I choose to act as I do or it is not. If it is an accident, then it is merely a matter of chance that I did not choose otherwise; and if it is merely a matter of chance that I did not choose otherwise, it is surely irrational to hold me morally responsible for choosing as I did. But if it is not an accident that I choose to do one thing rather than another, then presumably there is some causal explanation of my choice: and in that case we are led back to determinism.
(Philosophical Essays, 1954, p.275)

In any case, it is not always true that a philosophical inquiry will, for all practical purposes, leave things just as they were. A conspicuous counter-example is one which I mentioned earlier in passing, the problem of the freedom of the will. As often happens in philosophy, this problem takes the form of a dilemma. On the one hand, we are inclined to believe that all spatio-temporal processes, and therefore also human actions, are governed by natural laws; and from this we are inclined to infer that given the initial circumstances whatever actually happens could not have happened otherwise. On the other hand, all our moral assessments of our own and other people's conduct and all our legal practice depend on the assumption that people are responsible for their acts; but this seems to imply that, even given the existing circumstances, they could have acted otherwise. But these conclusions are mutually contradictory. It has to be shown, then, either that there is some flaw in this reasoning, or that at least one of the premisses of one or other of these arguments is false.

Again, I do not propose to offer you a solution of this problem. To do so would mean going very carefully into a set of notions of which the correct analysis is by no means obvious; the notion of causality and of natural law on the one hand, and on the other the whole group of concepts; which figure in the teleological explanation of human action, as well as our idea of moral responsibility. The concept of a person would itself come under review. One would have to consider exactly what was involved in the shift from causal to statistical laws in micro-physics, and whether this was relevant to the present issue; whether there were any logical reasons for denying that every human action could be predicted; whether there was anything at all about a human being that could not be explained in physiological terms. My own view, which I give you without argument, is that we have as yet no very strong reason to believe in the emergence of a physiological theory which will account for every facet of our behaviour, but that the possibility of it cannot be excluded a priori. [I try to deal with some of these questions in my essay 'Man as a Subject for Science'. ]

The point which I now wish to make is that it may very well turn out, in the course of such an inquiry, that the picture of ourselves which underlies much of our moral thinking, the idea of our actions as proceeding from the unfettered choices of self-propelling agents, did not stand up to critical scrutiny. If this did prove to be the case, it would not follow that the idea of freedom would have to go by the board. We could still draw a distinction between the actions which a person chose to do, never mind how the choice came to be made, and those which were forced upon him, in the sense that he was subject to unusual pressures, or even deprived of any power to choose; and there might be utilitarian grounds for our responding to actions of these different sorts in different judicial ways. But certainly a great strain would be placed on the conventional ideas of merit and of guilt; and if these ideas were given up or greatly modified, it is hardly to be expected that our moral and our legal outlook would remain unchanged.
(Metaphysics and Common Sense, chapter I, "On Making Philosophy Intelligible," p.14)

The idea that man somehow stands outside the order of nature is one that many people find attractive on emotional grounds; so that it has to be received with some caution. It is, however, fairly widely accepted nowadays, even by philosophers who are supposed to be able to discount their emotional prejudices, and this for various reasons. One of them is of course the belief in the freedom of the will. It is argued that since men are free to behave as they choose, they are always capable of nullifying any generalization about their conduct to which they are alleged to be subject. If any such generalization is produced, it is only to be expected that someone will proudly or perversely exercise his option of rendering it false.

The trouble with this argument is that it simply assumes the falsehood of the position which it is intended to demolish. If the attribution of free will is construed in such a way that a man can be said to have acted freely only if his action is not susceptible of any causal explanation, then there will indeed be no question but that if men ever act freely, their behaviour is not totally subject to causal laws. This still allows for its being subject to statistical laws, but on the assumption, which the proponents of this view tacitly make, that a man is free on any given occasion to try to do anything whatsoever that he believes to be feasible, the possibility of there being even statistical laws about human behaviour which would be of any scientific value is effectively excluded. But now it is surely fair to ask for some justification of these very strong assumptions. What reason have we for believing that men ever do act freely, in this sense? There may be a prima facie case for holding that men are capable of acting freely in some sense or other; but it is by no means clear that an action which passes this test of freedom, whatever it may be, cannot also be governed by some causal law.

Compatibilist and incompatibilist views among philosophers and the general public may be a source of confusion.

Many philosophers have in fact held that what we ordinarily mean by speaking of an action as freely done is not incompatible with its being causally determined; some have gone even further to the point of holding that when we say that an action is free we actually imply, or presuppose, that it is determined; others who take the view that determinism excludes free will, as this is ordinarily understood, have concluded just for this reason that our ordinary notion of free will has no application. I do not myself think that we stand to gain very much by making a conscientious effort to discover what people ordinarily mean when they talk about free will: it might very well turn out that some people mean one thing by it, and some another, and that many people's idea of it is very confused. The important question, so far as we are concerned, is whether human behaviour is or is not entirely subject to law. If we conclude that it is, or even just that there is no good reason to suppose that it is not, then we may find it expedient to introduce a sense of acting freely which squares with these conclusions. We shall presumably want it to apply, so far as possible, to the same actions as those that most people would now regard as being free, though not necessarily with the same implications, but we shall rather be correcting ordinary usage than merely following it. If, on the other hand, our conclusion is that human behaviour is not entirely subject to law, then again we shall have to decide what provision, if any, this enables us to make for freedom of action; for example, if we make the absence of causal determination a necessary condition of an action's being freely done, we shall have to consider whether we can still preserve the connection between freedom and responsibility. But the point is that before we can usefully embark upon such matters, we must first decide the issue of determinism. How far and in what sense is man's behaviour subject to law?

It may appear, indeed, that this is not an issue which one could hope to settle a priori. Surely, it may be said, we can never be in a position to show that any piece of human conduct is inexplicable: the most that we can claim is that we have not been able to find any explanation for it, but this does not imply that there is no explanation, or even that the explanation is one which it will always be beyond our power to discover. But while this remark is perfectly sound, it may also be thought to miss the point. For what is most commonly maintained by those who wish to set limits to the extent to which men's actions are governed by law is not that these actions are inexplicable, but rather that the kinds of explanation which they call for do not conform to the scientific model. That is to say, they do not account for an action as resulting from the operation of a natural law. It is allowed that explanations of the scientific type are sometimes appropriate, as when we account for some piece of deviant behaviour by relating it to a disorder in the agent's constitution, but cases of this kind are said to be the exception rather than the rule. In the normal way we explain a man's action in terms of his intentions, or his motives, or his beliefs, or the social context in which the action takes place. Consider, for example, the simple action of drinking a glass of wine. As performed by different people in different circumstances, this may be an act of self-indulgence, an expression of politeness, a proof of alcoholism, a manifestation of loyalty, a gesture of despair, an attempt at suicide, the performance of a social rite, a religious communication, an attempt to summon up one's courage, an attempt to seduce or corrupt another person, the sealing of a bargain, a display of professional expertise, a piece of inadvertence, an act of expiation, the response to a challenge and many other things besides. All these are accepted as good explanations: if the circumstances are right, they render the performance of the action intelligible; but only in the case of the alcoholic is it clear that the explanation is of a scientific character. In the other cases we find the action intelligible because we are given a reason for its performance; it is explained in terms of the agent's purpose, or by reference to a social norm, or through some combination of these two.
(ibid. p.221-223)

A philosophical question which I have not here discussed, partly because I do not think that I have anything new to say about it, is whether the denial of determinism is implied in our usual ascriptions of moral and legal responsibility.

Ayer calls here for some experimental philosophy to determine the popular opinions and common sense

In common with many other philosophers I used to hold that it was not, that in this respect the antithesis between the claims of free will and determinism was illusory, but in so far as this is a question of what people actually believe, I now think it more likely that I was wrong. This is indeed a matter for a social survey which, as I said before, would probably not yield a very clear result. I should however expect it to indicate that if it were shown to them that a man's action could be explained in causal terms, most people would take the view that he was not responsible for it. Since it is not at all clear why one's responsibility for an action should depend on its being causally inexplicable, this may only prove that most people are irrational, but there it is. I am, indeed, strongly inclined to think that our ordinary ideas of freedom and responsibility are very muddleheaded: but for what they are worth, they are also very firmly held. It would not be at all easy to estimate the social consequences of discarding them.
(ibid. p.238-239)

Ayer on Chance, Design, and Cause