Dr Daniel von Wachter,
Institute for Formal Ontology and
Medical Information Science (ifomis.de)
University of Leipzig,
Härtelstr. 16, D-04107 Leipzig
dvw@epost.de;
http://users.ox.ac.uk/~orie0460/
Abstract: The claim put forward by Chisholm that some events are caused, not by other events but, by substances, namely agents, is discussed. It is argued that, contrary to what one should think, it is in fact not a claim about agency and causation but one about certain statements. An alternative theory of agent causation is proposed.
Imagine Ludwig has a cup of tea for breakfast. He pours it; he eats his egg until it seems to him that the tea should have the right temperature; he moves his hand to the cup, puts his fingers at the handle, and then, careful not to spill anything, he does something with his arm; namely, he raises it, and if all goes well he then drinks the tea without burning his lips.
The rising of Ludwig’s arm surely has a cause. But what is the cause? Defenders of agent causation, such as Thomas Reid (1788), Richard Taylor (1966), Roderick Chisholm (1976a), and many more recent authors (see Swinburne 1997, ch. 5; Thorp 1980; Meixner 1999; Clarke 1996; O'Connor 2000) have argued that the rising of Ludwig’s arm is caused by Ludwig himself. Some events are caused, not by other events, but by concrete things, by substances, more specifically by intentional agents.
Agent causation is usually advocated by believers in libertarian free will. One might think that it is postulated in order to allow for agents originating causal processes. The opponents of agents causation believe that it is metaphysically extravagant to postulate agent causation as another kind of causation besides event causation, and that it is impossible that agents originate causal processes (Honderich 1993, ch. 3). However, I shall argue that the usual theory of agent causation, as Chisholm has campaigned for it, does justice neither to free action nor to non-free action. I shall then sketch an alternative account of causation in free actions, which many will find even more extravagant than Chisholm’s.
The Chisholmian theory of agent causation
Chisholm holds that actions involve what he calls ‘undertakings’. An undertaking is a mental event as it occurs, for example, in Ludwig’s mind when he raises his arm, or as it occurs in somebody’s mind who tries to raise his arm even if the arm does not move because it is paralysed. (Others have used ‘trying’, ‘purposing’, or ‘volition’ instead.) An undertaking is an intentional event whose intentional object is the desired effect, for example the movement of one’s arm. When Ludwig successfully undertakes to raise his arm, then Ludwig is the cause of the arm’s rising.
A free action, for Chisholm (1976a, 202; 1976b, 62), is an action which involves an undertaking for which there is no preceding ‘sufficient causal condition’. That is, Ludwig raised his arm at t freely if, and only if, there was in a period which includes, but begins before, t ‘no sufficient causal condition’ either for Ludwig raising his arm or for Ludwig not raising his arm.
A quick, and I believe valid, objection against this view is that it fails to require that the agent has control over what he does. The mere fact that there was no ‘sufficient causal condition’ for a certain undertaking does not entail that it was up to the agent that he undertook this rather than something else. It could have been just a matter of indeterministic processes, that is, just a matter of chance, that one undertaking rather than another one occurred.
Chisholm (like Swinburne 1997, ch. 13) clearly allows for the possibility of an undertaking being caused by preceding events. For him, a non-free action is exactly that, an action which involves an undertaking for which there was before a ‘sufficient causal condition’. So in non-free actions too the agent is cause of his action and its results.
Now, at this point we should wonder what agent causation really is supposed to be. Surely, if Ludwig, when he raises his arm, is fully determined to do so, then there is no other causation involved than ordinary event causation. Some events, such as brain events or desires, cause Ludwig’s undertaking, the undertaking causes events in the nervous system, which cause events in the muscles, which cause the rising of the arm. There is no reason to postulate a kind of causation other than event causation.
The linguistic turn
So why does Chisholm claim that there is agent causation even in non-free action and that it is irreducible to event causation (Chisholm 1976a, 199)? The reason is that they do not mean by this claim what one should think they mean. They do not mean that when Ludwig raises his arm the arm’s rising being caused is dissimilar to a billiard ball’s rolling being caused. Instead, they mean that statements of the form ‘Ludwig did so-and-so’ cannot be transformed ‘without loss of meaning’ into statements of another form (Chisholm 1976a, 199; 1978, 622f). The Chisholmian claim that there is irreducible agent causation really is not a claim about action but about certain statements describing actions. It is a child of the linguistic turn.
I suggest we should not make a linguistic turn, or if we have made one already, we should make an ontological turn and look, not at statements describing actions but, at the things they describe and which make them true. The adherent of the linguistic turn might want to hold that investigating statements describing actions is one way, or the best way, or the only way, to investigate causation in actions. I reply that we can see that the linguistic approach here not only does not reveal anything about action, it even leads to false results. The linguistic turn leads us to say that there is agent causation also in non-free actions because statements like ‘Ludwig raised his arm’ can in no case be transformed into statements of another form. But in fact in a non-free action the undertaking as well as all events involved in the action are solely and deterministically caused by preceding events; there is only one kind of causation involved, namely event causation. So what the linguistic turn leads us to say is false if it is taken to imply anything about action or causation, rather than just about statements.
Turning away from the linguistic turn
So let me try to describe the causation involved in free actions. If the rising of Ludwig’s arm, including Ludwig’s reasoning and undertaking, was the result of ongoing causal processes; if it was—like the movements of the planets on Easter 2000—solely and deterministically caused by preceding events, then, I suggest, we should not say that he raised his arm freely, whatever else was the case. That is, I am using ‘free’ in the libertarian sense.
I use ‘event’ and ‘state of affairs’ here interchangeably in order to refer to a thing’s (or things’) having a certain property (or properties) at a certain time. One refers to an event by specifying which property is instantiated where, or by what, when, i.e. during which period of time. If an instant of time is specified, instead of a period of time, it is implied that there is a period which includes the instant and during which the property is instantiated. For example, in order to refer to a physical event taking place between t1 and t2 one has to specify, at least vaguely, which quality is instantiated where at each time between t1 and t2. One event may be constituted by different properties being instatiated at different places at different times. Further, I am assuming that there are causal processes. A causal process is a continuous series of states of affairs each of which can truly be said to have caused any later state of affairs in the process.
What, then, happened when Ludwig raised his arm freely? The rising of the arm was the result of a causal process, which I call the ‘action process’. It comprised events in Ludwig’s muscles, in his nervous system, in his brain, etc. That is, for each phase of the rising of the arm there were preceding events in the muscles, in the nervous system, in the brain, etc., which caused this phase of the arm’s rising.
Now, if the action was free, then some phase of this process is not just the result of preceding events. The process must somehow have a beginning, and it’s occurrence—that is, the origination of the action process—must somehow be due to the agent. I call this first phase of the action process the initial event, or ‘initial state of affairs’.
But not the whole initial event has to be due to the agent. Consider a match being lighted at t1 so that an explosion at t2 is caused. The lighting of the match is only a part of the cause at t1 of the explosion at t2. The complete cause includes much more, e.g. the presence of certain gas, etc. Similarly, in a free action only a part of the initial event has to be due to the agent. This event which is a part of the initial event has no preceding cause, that is, it is not caused by an event. So why does it occur? The answer must be: because of Ludwig. Ludwig just brought about this event; he made it pop up. If you investigate the action process going backwards from the action result you come to a stage which has a part for which you cannot find a preceding cause, but you find that it’s occurrence is due to Ludwig. The event is Ludwig’s choice. I call it therefore ‘choice event’. I say interchangeably about a choice event that ‘it is Ludwig’s choice’, that ‘it is the result of Ludwig’s choice’, that ‘Ludwig brought it about by choice’, or that ‘Ludwig chose it’. So, on the account I am proposing, free agents are able to bring about certain events just by choosing, and such an event has no event cause.
What are choice events in the case of free human agents? I shall only indicate here what answers to this question are available. The answer depends on one’s solution of the mind-body problem. A materialist will say that human choice events are physical, presumably brain events. A non-materialist may, but need not, say that they involve non-material stuff. If choice events are not mental then we are not aware of them when we act, as we are not aware of the brain events involved in our actions. One candidate for being choice events are undertakings, i.e. certain mental events. On a non-materialist view they may, but don’t have to, be taken to be distinct from any physical events; on a materialist view they are identical to certain brain events. In any case, a man’s freely moving his body in a certain way involves a choice event which in conjunction with other events, presumably physical events, causes subsequent physical events in the body. I implied this above by saying that the choice event is part of the initial event of the action process.
Further, one may hold that the occurrence of a choice event, sometimes or always, is an intervention into an ongoing causal process. That is, there may or may not have been a causal process (deterministic or indeterministic) that, had nothing intervened, would have led to an event instead of which the choice event occurred. Moreover, one may or may not hold that, preceding a choice event, there are causal processes leading to events involving the same things or stuff as the choice event, but different properties. That is, if a’s being C at t is a choice event, there may be causal processes (deterministic or indeterministic) leading to a’s being D, E, F, etc., at t.
Freedom and chance
Many authors have tried to do justice to the intuition that if an action was free the agent ‘could have done otherwise’, by postulating that certain causal processes which are involved in the action are ‘indeterministic’. By an ‘indeterministic’ process a process is meant which leads with a certain probability to one event or to certain other events instead. It is ‘chancy’, like (presumably) processes on the quantum level. Some authors hold that in a free action the process of deliberation, before the decision, is indeterministic (Dennett 1978; Fischer & Ravizza 1992; Mele 1999); others hold that the decision itself is the result of an indeterministic process (Clarke 2000). Similarly, Chisholm (1976a, 202; 1976b, 62) and Swinburne (1994, 25; 1997, ch. 13) hold that an action is free if the undertaking involved is not ‘fully’, i.e. deterministically, caused by earlier events.
The trouble with all these accounts is that they fail to take into account the control a free agent has over what he does. They postulate a possibility that the action could have failed to occur, but they do not entail what people usually imply when they say ‘he could have done otherwise’, namely that it was up to the agent what he did. If somebody’s action was preceded by an indeterministic process of deliberation, or if the decision or the undertaking was the result of an indeterministic process, that makes the occurrence of the action a matter of chance and not a free action. This becomes clear also through the fact that it would not make the agent any more responsible for the action than he would be if the action was determined. An agent is responsible for an action not through the mere possibility that the action could have failed to occur but through it having been up to him what he did. An action which occurs as a matter of randomness is no more free than an action which is determined.
Perhaps these accounts are based on the mistaken assumption that if event A at t2 was the result of a deterministic process between t1 and t2 then after t1 nothing could prevent the occurrence of A. But there are no processes like this. Even a maximally deterministic process, such as the movement of billiard balls as described by Newtonian mechanics, can be interrupted. It only requires a sufficiently strong intervention. A cat, for example, can interrupt the deterministic process of a rolling billiard ball.
If there is at t1 an indeterministic process running towards Ludwig raising his arm at t2, then Ludwig, if he is to raise his arm freely, has to interrupt this process (as he would have to if it were a deterministic process) and raise his arm via a choice event. This choice event is then the result of Ludwig’s choice and not a result of the process. It is hence false what some authors (Clarke 1996, 27) say that the presence of chance in the process leading to an action ‘leaves room’, and is required, for control and for the possibility of alternative actions.
I have not yet answered the question whether agents are causes of what they bring about freely. Non-free actions involve only event causation. It may still be convenient to say that Ludwig is the cause of the rising of his arm even it he raised his arm non-freely; but no special kind of causation is involved. For free actions the question arises whether the agent is cause of the choice event, which has no event cause. The relation between the agent and the choice event is certainly different from the relation between, say, the lighting of a match and the subsequent explosion. But as an agent, like an event cause, is that to which it is due that one thing rather than another occurs—namely one choice event rather than none or another one—I suggest that it is adequate to call the agent ‘cause’, or more specifically ‘agent cause’, of the choice event, as well as of the events later in the action process.
Chisholm, R. 1976a "The Agent as Cause", in M. Brand and D. Walton (eds.), Action Theory, Dordrecht: Reidel, 199-211.
Chisholm, R. 1976b Person and Object, La Salle (Illinois): Open Court.
Chisholm, R. 1978 "Replies", Philosophia, 8, 620-636.
Clarke, R. 1996 "Agent Causation and Event Causation in the Production of Free Action", Philosophical Topics, 24, 19-48.
Clarke, R. 2000 "Modest Libertarianism", in J. E. Tomberlin (ed.), Philosophical Perspectives, 14: Action and Freedom, Malden: Blackwell Publishers, 21-46.
Dennett, D.C. 1978 "On Giving Libertarians What They Say They Want", in, Brainstorms, Brighton: Harvester Press, 1981.
Fischer, J.M. and Ravizza, M. 1992 "When The Will is Free", in J. E. Tomberlin (ed.), Philosophical Perspectives, 6: Ethics, Atascadero: Ridgeview.
Honderich, T. 1993 How Free Are You?, Oxford UP.
Meixner, U. 1999 "Kausalität der Ereignisse oder Kausalität der Personen", Metaphysica, 0, 105-120.
Mele, A.R. 1999 "Ultimate Responsibility and Dumb Luck", Social Philosophy and Policy, 16 (2), 274-293.
O'Connor, T. 2000 Persons and Causes, Oxford UP.
Reid, T. 1788 "Essays on the Active Power of the Human Mind", in R. E. Beanblossom and K. Lehrer (eds.), Inquiry and Essays, Indianapolis: Hackett, 1983, 297-368.
Swinburne, R. 1994 The Christian God, Oxford: Clarendon.
Swinburne, R. 1997 The Evolution of the Soul (Revised Edition), Oxford: Clarendon Press.
Taylor, R. 1966 Action and Purpose, Englewood Cliffs, NJ.
Thorp, J. 1980 Free Will: A Defence Against Neurophysiological Determinism, London: Routledge & Kegan Paul.