1.

This paper assumes without argument a certain view about the nature of action. This will limit its interest for those who do not accept that view. But the second half of the paper discusses a difficulty which arises for them just as much as for anyone else. So there should be something here for everyone.

The view that I assume is a view about the nature of action. It is an agent-causalist view, but not the agent-causalist view. Traditionally, agent-causalism has been the view that actions are events caused by agents. But this is not the view I assume (and which I in fact accept). The view I assume is one which I first learnt from an article by Maria Alvarez and John Hyman, ‘Agents and Their Actions’.Footnote ¹ It holds that to act is to cause a change, and an action is an agent's causing a change. The action is not the change caused; that change, which is an event, is (in Alvarez and Hyman's terms) the ‘result’ of the action. We need to respect here the distinction between the causing of a change and the change that is caused by, and so the result of, that causing. The cause of the change is the agent. So here we have a form of causation where the cause is not an event, but an agent. The result, the thing caused, is an event (probably). Is the causing of that result an event? Alvarez and Hyman say not, and I agree with them, but not for the reason they offer. They effectively try to prove that this must be so; but I only want to suggest that nobody should think that a causal relation is an event. Anybody who holds the more standard view that causal relations are relations between events should avoid saying that the relation between those events when one causes another is another event, to be added to the cause-event and the effect-event. Otherwise they will face the question what is the relation between the cause-event and the causing-event, and I don't see a plausible answer to that question – any more than there is a plausible answer to the question what the relation is between the causing-event and the effect-event. So I suggest that there is a considerable incentive for all of us to avoid thinking of causings as events.

To come at the matter another way, causal relations are too thin to count as events. The sort of thinness I am getting at here might be what is in the mind of those who think of causal relations as essentially explanatory. An explanatory relation is not the same sort of thing as a ‘making-happen’ relation. The latter might be thought of as a sort of forcefulness in nature, and such a relation looks as if it should be thick, rather than thin. But even if causal relations are not themselves merely relations of explanation, they will still be thin. On the form of agent-causalism that I accept, the relation between an agent and an effect caused by that agent is not itself another event somehow interposed between the agent and the effect. The exercise of a causal power should not itself be understood as a happening. For if so, we would want to ask how the agent manages to get that happening to happen; and if the answer is that she does it by causing it, we are back where we started, or, rather, off on a regress. And on a traditional Humean approach, relations of causation are relations of regular succession, and those relations are not themselves events in addition to the events in that succession. My conclusion is that we should all avoid thinking of causings as events.

2.

I said that the view that to act is to cause a change is one that I learnt from an article published in 1998. But it appears that this view was around 100 years earlier, since Harold Prichard ascribes it to John Cook Wilson. In his ‘Acting, Willing, Desiring’, Prichard writes: ‘We should at first say that to do something is to originate or to bring into existence, i.e. really, to cause, some not yet existing state either of ourselves or of someone else, or, again, of some body.’ And he ascribes this view explicitly to Cook Wilson. But the relevant paper seems (like so much of Cook Wilson's oeuvre) to be lost. So we cannot be sure whether Cook Wilson actually tied himself to the thought that to do something is to cause some not yet existing state to come to exist. After all, there is the alternative possibility that some actions, at least, are the causings of a process rather than of a state; I mention this alternative now because it will become relevant later. So I think we should not take Prichard's talk of a state as committing him to any divisive metaphysics. Still, he mentions this view of Cook Wilson's in order to criticise it, and it is that criticism that originally intrigued me. Prichard continues thus:

Now in a way I should be sympathetic to this complaint, because it looks as if Prichard is here supporting the view I have already propounded, namely that causing is too thin a relation to be an activity. Still, the present paper is concerned with the rights and wrongs of this issue, and with those of another remark of Prichard's in the same paper, together with something that Alvarez and Hyman say on a related topic. Prichard, having announced that, though the causing a change may require an activity, it is not itself an activity, still wants to know what activity is required for causing a change, and he decides that this activity is that of willing. And he suggests without further ado that:

So it is by willing that we cause the changes we cause, and although we may thereby manage to cause other changes such as movements of the body, the changes we cause directly are all changes in our brain – neural changes, as I will say. This second suggestion is the topic of the second half of my paper. My interest in it derives from a remark on the same topic by Alvarez and Hyman:

I found this remark very puzzling. It seems to allow that the bodily movement can have two causes, one a neural event – as one event causing another – and one that is not an event at all but an agent. And the question is whether this suggestion is both compulsory and coherent. If it is indeed coherent, then it doesn't matter whether it is compulsory; but if it is both compulsory and incoherent, we agent-causalists are in trouble.

The general topic of this paper, then, is how to capture the role of neural events in one's account of an action as the causing of a change by an agent. And there is a question along the way, namely: is there something that drives us to the view that all our basic activities are acts of the will? Prichard supposes that when one causes a change, there must be some activity by which one does so, and that this activity can only be willing. So when I raise my hand, my raising of my hand is my causing my hand to rise, and such a causing is not itself an activity, but requires another activity – which he announces can only be an act of the will. Evidently we are not supposed to press the question how I do that act. We would have thought that I cannot ‘just cause’ the relevant volitional change in me, because there is ‘no such activity as originating or causing a change’. Prichard's own suggestion, that what we ‘directly cause’ is a change in our brain, seems to be an account of what sort of a change we can ‘just cause’, since ‘directly causing’ seems hard to separate from ‘just causing’.

There is one terminological issue that needs to be brought out here. I return to Prichard's remark first quoted above:

To understand this remark, we need to disambiguate the potentially ambiguous phrase ‘some movement of our body’. In the now standard way that we owe to Jennifer Hornsby, we need to distinguish between transitive and intransitive senses of this phrase. When I move my body, I cause my body to move. The first occurrence of ‘move’ in this sentence is transitive, and the second is intransitive. When I move my body, I cause a change in the location of my body; this is transitive moving. When my body moves, its location changes, and nothing is said about what causes that change; this is intransitive moving, like the moving of the moon – mere motion. We say that when I move my body, I cause my body to move, and this does not mean that when I move my body, I cause my body to move itself. It only means that I cause a motion of my body. Though my body then moves, it does not cause that motion – I do. To render this distinction perspicuous as we go along, we mark it by subscripts: when I move my body, there is me, the mover, and the body, the thing moved, one thing moving another and this is transitive moving – bodily movements_T – where the subscript T stands for ‘transitive’. When the body moves – perhaps because I move it – we are thinking of bodily movements_I – with the subscript I standing for ‘intransitive’. I will also use the phrase ‘bodily motion’ for this intransitive sense. So when I move my body, I cause a bodily motion.

Now in which sense should we understand Prichard's remark about willing some movement? Am I willing a bodily motion to occur, or willing to move my body? The point of this question is to raise doubt about the idea that when I move my body, I do it either by willing some bodily motion or by willing to move my body. Neither of these things seems to be what I do. I do not will (one act) to move my body (another act), nor do I will a mere event (a change in my body's location). We seem to need a third possibility; and if none can be found this casts doubt on the whole idea of willing a bodily movement.

3.

Bruce Aune writes: ‘the principle behind Prichard's claim is …‘Statements of the form ‘Thing or person S causes event E’ are to be understood as elliptical for statements of the form ∃A(A is an activity by S and A causes E)’.Footnote ⁵ But Prichard does not challenge the idea that agents cause changes in favour of the view that causing is always done by activities. He only wants to identify the activity by which agents cause the changes they cause, because he thinks that, though there is such a thing as causing a change, still causing a change, though it requires an activity, is not itself an activity. The activity we are looking for is willing.

But we might wonder why Prichard should suppose that causing a change requires the same activity each time. For he could be right in his claim that causing a change is not an activity, but requires an activity, but wrong to suppose that it must be the same one each time. Perhaps I cause my arm to rise by raising it, and cause my hand to turn (when opening the door) by turning it. So I am suspicious of Prichard's apparent assumption that whenever we cause a change, there is one and the same activity involved each time. If it were, then perhaps the only candidate activity would be that of willing. But we need reason to believe this assumption.

What is more, Prichard himself is wary of such assumptions elsewhere. He famously compared the question ‘Why should we obey the law?’ with the question ‘Why should we read books?’. It is clear that there are lots of different reasons to read books, depending on the book (and on us); and similarly there are different reasons to obey the law, depending on the law and the circumstances (and on us, again). So why should there not be different activities involved in the causings of different changes? I see no reason why.

So a Prichardian response to Prichard's challenge is to allow that there is indeed no activity of causing a change, but to insist that every causing of a change is an activity of some (other) sort, and every activity is the causing of a change. So: walking is causing a change by walking; it is an activity by doing which the agent causes a change in her location. Writing is causing a change by writing: and so on. So the general form of action is: causing a change in one way or another. This reminds me of Jonathan Lowe's suggestionFootnote ⁶ for the basic form of an agent-causal statement: Agent A caused event E by V-ing.

Could this be what Prichard meant by saying that though causing requires an activity, it is not an activity? No, because for him it must be the same activity every time. Lowe, by contrast, is happy – so far as the form of the causal statement goes – to allow that it might be a different V-ing for each causing. He argues, however, that all our basic actions are acts of will. A basic action can be understood for these purposes as a doing which is not done by doing something else. So I draw your attention by causing my arm to rise; and I cause my arm to rise by raising it (rather than in some other way). I raise it by an act of will. (I do not recommend this train of thought, and mention it only because of its relevance to what is to come later.)

Lowe's account of the form of an agent-causal statement allows that nothing can just cause anything. If an action is the causing of a change by an agent, there must be some way in which the agent causes that change. I cause my arm to rise by raising it – that is one way – but I could also do it by lifting it with the other arm, or by getting you to lift it. It is the way in which one does it that gives the causing its substance, its thickness (to revert to the idea that causing is itself a thin relation).

This returns us to the idea that causings are not events. If I write, I cause a change by writing, and the change caused is an event, but the causing of it is not. Causal relations do not have enough substance, enough thickness, for their instantiations to be events. So as an agent I cannot just exert a causal power and get something to happen. I have to act in a certain way if I am to cause a change. But whether this is compatible with the claim that I can just will, and that by willing I can do other things, such as raise my arm, is dubious.

If, when I cause a change, I do so by engaging in some activity, are we to say that it is the activity that is the real cause of that change? This is not the way the story is supposed to go. The ground idea is that the cause of the change is the agent, and the activity involved is the agent's causing that change in a certain way. Is there some danger that this response to Prichard generates a regress? I doubt it. If it does, the regress will derive from two thoughts: that one causes B by doing something else A, and that all doings are themselves causings. But the view above does not maintain that one causes B by doing something else A. It just holds that every change caused is caused in a certain way.

Furthermore, if there were a genuine regress here, how is it that willing would be able to stop it when nothing else can? Willing is an activity (if it happens at all), and there is no activity by doing which one wills. But why should one not say the same about causing one's fingers to move? If I can just will, why cannot I just move my fingers? To move one's fingers is to cause one's fingers to move by acting in a certain way, namely, by moving them. If we point out that there are various ways of causing one's fingers to move, the answer is that in this case one did not move one's finger by doing something else of which the motion of the finger is an effect. So understood, one's moving one's finger is a basic action.

4.

The notion of a basic action which I used in the previous sentence is taken from Alvarez and Hyman. More fully expressed, a basic action is the causing of a change whose result is not the effect of some other action by the same agent. (Remember that the ‘result’ of the action is the change of which the action is the causing.) An ordinary bodily movement_I, or motion, can be the result of a bodily movement_T, and such movements_T are basic actions. An ordinary bodily movement_T consists in the moving_T of the body by the agent.

As far as this goes, we don't need to introduce considerations of neurophysiology. But Alvarez and Hyman complicate things enormously by maintaining (or perhaps one should simply say pointing out) that on their story there are two causal relations, event-causation and agent-causation, and one and the same event can be caused both by an agent and by a prior event. If we reserve the term ‘causer’ for an agent-cause, and restrict the noun ‘cause’ to event-causes, a bodily motion can have both a cause (perhaps neural) and a causer (agent). If that bodily motion is the result of the agent's causing (that is, the result of the exercise of a causal power by that agent) and not the effect of some other action by the same agent, it is a basic action of that agent's.

Now when I move my body, the relevant bodily motion is an effect of neural events. If I cause those events, my causing those events would be my basic actions. And if I don't cause those events, how is it that I can be the causer of their effects? I seem to be unnecessary, and we would be forced to conclude that no bodily movement_T is a basic action.

We might optimistically suppose that the answer to this is that the two causal relations, event-causation and agent-causation, are so distinct that they do not compete. If they do not compete, the fact that our bodily motions are the effects of neural changes in our brain is not in tension with the fact that they are the results of our causings. The latter causings are ones done by causers rather than by causes. So we can have both causal relations in play simultaneously without either interfering with the other. We do not need to deny that those bodily motions that are the results of our agency are also the effects of neural changes in our brains. Those motions can be our doings (they are not our actions, of course, but the results of those actions) even though they have causes other than us.

I fear, however, that this suggestion has only to be understood to be rejected. It renders agent-causation epiphenomenal. As it represents things, there is the causal system (the event-causal system, that is) going on remorselessly in its own sweet way, and we somehow plaster agent-causation on top of that, to get ourselves into the story as agents. But we might reasonably notice that the bodily motion is going to happen anyway, once the relevant neural changes are under way, and there remains nothing really for us to do. Our role as causers is mere window-dressing.

Matters would be different if we could claim to be the causers of the neural changes whose effects are the relevant bodily motions. But if we were, those causings of neural changes would be our basic acts, for the bodily motions are the effects of those neural changes, and though we can be said to be the causers of events that are the effects of the results of other causings of ours, our basic acts will be the causings of neural changes.

I do not want to raise it as an objection to this picture, that our neural changes have other neural events as causes. This is true, but if we are to retain any sense of ourselves as agents, rather than as the conduits of extraneous causal forces, we have to find some point at which we do something – and, I would say, at which we do something for a reason.

Consider, in this connection, what Alvarez and Hyman have to say. They allow that the bodily motions that are the results of my basic actions are also the effects of certain neural events in my brain. And they allow that I do not cause the neural events to occur. What I do, according to them, is to ensure that those neural events have occurred by performing an action whose result is sure to be the effect of such a neural event. But what does cause them then? Three answers are possible: nothing, me, and something else. If something else, we seem to be locked into the causal nexus. The only way out of this impasse is to allow a plurality of causal relations, so that though an event is already perfectly well explained causally, there is room for a second causal explanation of that same event, but of a different sort. One might hopefully suppose that since the two causal explanations at issue are of very different sorts, there can be no competition between them. The technical term for this sort of approach is insouciance.

5.

But even the insouciant might prefer to tie the two causal relations together in some way, if that were only possible. So the suggestion I am going to consider is that we think of the agent as causing a causal process, or a series of causally linked events: the neural at one end causing the overt physical motion at the other, with the muscular in between. On this account, the basic action is not well understood as the causing of a bodily motion such as the rising of an arm; it is the causing of a process of which that motion is the terminal point.

What we cause when we act, on this account, is not a sort of causing to cause. We do not cause neural events to cause muscular ones, and we do not cause those muscular ones to cause arms to go up. These things are not up to us. What we cause is a causal complex, and we cause every part of that complex, but we do not cause the causal relations that bind it together. It is one thing to cause an event to occur, one that will have certain effects, and another to endow that event with causal powers. Similarly, it is one thing to cause a process to occur, and another to endow some element of that process with causal powers.

An immediate worry about this proposal is that it thinks of the agent as causing a process, when there are parts of that process of which the agent is more or less entirely ignorant. In fact, the only thing the agent really knows about is the final stage. But this epistemological reason for focusing on that final stage and thinking of the basic action in such terms applies just as well to the muscular as to the neural. We already knew that we are largely ignorant of the way in which we get our arm to rise. That the same applies to the neural events which are such an important part of the process should not disconcert us. The focus on the final stage had already to be abandoned, despite the fact that the sort of agentual knowledge that Anscombe tried so hard to understand is never of neural changes and only rarely of muscular ones.Footnote ⁷ So what I cause is not just, or really, the first stage of the neural/muscular/physical process, nor the last stage, particularly, but the whole thing, of which I may know only the later stage, and aim only at that.

One's basic action, on this approach, is the causing of the whole process. And this may seem to lead to immediate difficulties. Do I cause the neural events which are the first stage in the process? If so, how do I do that? The answer has to be that at the most I did that by causing the process of which those neural events were the first and the bodily motion the last stage. I suppose that one could try to suggest that I cause whatever neural changes are required for my hand to rise. But I prefer to stick to the letter of the ‘process view’ and maintain that I cause the whole process, and that my relation to that process is not identical to my relation to any of its stages. It is outside in rather than inside out.

Still, there is enough awkwardness about this for us to scout for possible alternatives. A different account might have it that the occurrence of the neural events is part of the story of how I cause the bodily motions. Not everything that is part of that story is itself a doing, or a causing, of mine, or the result of such a causing. This supposes that my action is really the causing of the bodily motion, and that is what I am trying to deny; so this new account is definitely different. But it still faces the familiar problem, which is that if I don't cause the neural changes, something else does, and this locks us into the causal nexus.

A better suggestion might be the the neural events are not part of the story of how I cause the bodily motions; they are part of the story of how those motions get to happen. It is not about how I do it. It is about how it works that the body moves_I. After all, there is no story of how I raise my hand – I just raise it, that's all. But if the neural events are part of how it works, must I not also cause them, in causing the process of which they are a part? Well, perhaps not. I do not necessarily cause to happen everything that needs to happen if some event of which I am the causer is to happen. I do not cause the absence of traffic jams that is required if I am to arrive in London by lunchtime. And I do not necessarily cause to happen everything that needs to happen if I am to cause that event – e.g. that I stay alive until the end of the process.

But this suggestion too runs up against the question what does cause the neural events if I don't. Again we face the fact that if the neural changes are not caused by me, they must be caused by something else, and the causal nexus looms.

So it seems that the best resort is to stick to the view that I cause the process of which the neural events are stages – a process which consists in a causal relation between neural changes at one end and bodily motion at the other. I don't cause the neural events in any other sense than this – and by the same token I don't cause the bodily motions in any other sense than this either.

There seems to me that the way in which children learn to control their bodily movements_I might chime rather well with the account I have been trying to promote. Very young children move around haphazardly. A process occurs at the end of which they can coordinate hand and eye, and move intentionally. What is it that they are learning during this process? They are not learning to make happen those neural events that are necessary for their hand to move towards the toy in front of them. Perhaps what they are learning is how to make happen a process which consists in neural changes leading to bodily motions. As I have suggested several times now, they don't need to be aware of every stage of that process in order to acquire the ability to execute it. And they don't need to be aware of it as a process either. It is a routine, but not one whose several stages will ever be present to consciousness. Nobody can say how it is done, but it might be possible to say what it is that we do.