Four-Dimensional Agent Causation
I. Introduction
Agent-causal libertarianism is not very popular. (That’s putting the point mildly!) And, of course, the trouble is not really with libertarianism in general. Though many find libertarianism unacceptable for a variety of reasons, it at least seems a defensible theory. But the particular variety of libertarianism that requires that agents themselves enter into causal relations with events has seemed utterly mysterious to most philosophers. Peter van Inwagen, for instance, says that he finds the notion of agent causation “more puzzling than the problem it is supposed to be a solution to”.[1] Though I don’t find the idea so obscure, I do think that agent-causal libertarians have their work cut out for them. In what follows, I’d like to offer some help to the agent-causal libertarian (whether the help I offer will be welcome is, of course, a different question). Specifically, I’d like to consider and formulate a response to C. D. Broad’s famous argument for the impossibility of agent causation.[2] What I will suggest is that if agent-causal libertarians are willing to give up their commitment to an endurantist picture of persistence through time, Broad’s argument can be easily resisted. My hope is that this will help dispel some (though certainly not all) of the mystery surrounding agent-causation.
II. Broad’s Objection(s)
Let’s begin by considering Broad’s argument. The most relevant part of Broad’s essay for our purposes runs as follows:
Now it is surely quite evident that, if the beginning of a certain process at a certain time is determined at all, its total cause must contain as an essential factor another event or process which enters into the moment from which the determined event or process issues…How could an event possibly be determined to happen at a certain date if its total cause contained no factor to which the notion of date has any application? And how can the notion of date have any application to anything that is not an event?[3]
It seems clear that what Broad is here taking issue with is the fundamental difference in temporal nature that seems to exist between events and agents. Whereas events happen at specific times, the “notion of date” has no straightforward application to agents. Though Broad isn’t explicit about this, I’m inclined to agree with others who have discussed Broad’s objection that there are actually two distinct objections here: one about causation and one about explanation.[4] The causal objection might be reconstructed as follows.
1)All events occur at a particular time.
2)So, if an event is caused, then it is caused to occur at a particular time.
3)If an event is caused to occur at a particular time, then some part of its cause must also occur at a particular time.
4)But only events occur at particular times – agents don’t.
5)Therefore, if an event is caused, then its cause must be an event.[5]
This causal argument is the one that seems closest to Broad’s original remarks, but the general form of this argument can also be used to construct a similar (and, I think, more powerful) argument in terms of explanation as follows:
1)All events occur at a particular time.
2)So, if an event is caused, then it is caused to occur at a particular time.
3)If an event is caused to occur at a particular time, then some part of its cause must explain why it occurred when it did.
4)But only events (which are directly in time) can provide such an explanation – agents can’t.
5)Therefore, if an event is caused, then it is caused by an event.
One or the other of these two arguments has seemed convincing to many philosophers. And I do think that they pose a serious challenge for the agent-causal libertarian. However, I think the challenge can be met.
I will proceed as follows. First, I will consider Broad’s objection in a bit more detail. I will then argue that if the agent-causal libertarian is willing to give up endurantism, Broad’s objection can be resisted. In order to make such a suggestion plausible, I will need to consider why agent-causal libertarians typically take themselves to be committed to endurantism and why I think they ought not to consider themselves so committed. I then conclude that rejecting endurance is a promising way for the agent-causal libertarian to avoid Broad’s objection, and thus render agent-causation slightly less mysterious.
III. The Objection in More Detail
Since I’m dubious that the causal version of Broad’s objection can be formulated in a non-question-begging way, and for the sake of time, I will restrict my discussion to the explanatory version. According to this version, the problem with agent-causal libertarianism is that it can’t explain why the agent causes an action at one particular time rather than another. Suppose that I agent-cause an action – turning a doorknob, say – at t3. According to the agent-causal libertarian, it is perfectly coherent to suppose both that I existed at t1and that all of the relevant considerations regarding whether to turn the knob were salient to me at t1. Why, then, didn’t I act at t1 instead of t3? The agent-causal libertarian seems to have no response to this question. There is nothing that explains why I acted precisely when I did. The event-causal libertarian, on the other hand, is not supposed to encounter a similar problem. The event that is the cause directly precedes (or is simultaneous with) the event that is the effect and thereby explains the timing of the effect. To be sure, the fact that the agent had certain relevant characteristics at the time of action does seem enough to explain why the action occurred. But the objection here is that this explanation is not enough explanation. We also need an explanation for why the action occurred at the precise moment that it did, given the agent-causal libertarian’s view that an agent can have the relevant characteristics and powers required to act without actually acting.
With a bit deeper understanding of how the explanatory version of the objection is supposed to work, let me move to my suggested solution.
IV. Temporal Parts to the Rescue
I will argue that if agent-causal libertarians are willing to give up their commitment to an endurantist theory of persistence through time, then both versions of Broad’s argument can be easily resisted. In the next section, I will consider why they typically take themselves to be committed to this particular account of persistence through time, but for now let me spell out the rewards they will reap as a consequence of rejecting it.
First, let me make a few general remarks about persistence through time so we will be in a better position to see what bearing it has on the issues we’ve discussed so far. Theories of persistence attempt to explain how a person persists from one time to a later time.[6] They all agree that persons do persist, in one way or another, but they diverge as to how that persistence occurs. Endurantists, on the one hand, hold the view that a person persists through time by being “wholly present” at each time. All of me, so to speak, exists right now, and all of me will exist at some point in the future. I am a three-dimensional object that moves from one point in time to the next, without leaving anything behind. Perdurantists, on the other hand, hold the view that a person persists through time in virtue of having temporal parts at each time. Not all of me exists right now, nor will all of me exist at some point in the future. I am a four-dimensional object that is extended in time as well as space, possessing both temporal as well as spatial parts. What exists at any given time is just a temporal part of me, [My Name]-at-t.[7]
Most, perhaps all, agent-causal libertarians take themselves to be committed to endurantism. We will see why they take themselves to be so committed in the next section, but for now, what I want to point out is that it is precisely this commitment that lands the agent-causal libertarian in the trouble that Broad’s objection poses. Broad’s objection stems from the fact that on the endurantist picture, persons are related to time in a completely different way than events are related to time. Persons don’t have temporal parts; events do. A person is three-dimensional; an event is four-dimensional. Persons are wholly present both at t1 and at t2; events are only partly present at t1 and partly present at t2. But if the agent was wholly present some time before he caused the action, how can we explain his causing the action at the precise time that he did? Why didn’t he cause it before? As we’ve seen above, these are important challenges to the agent-causal libertarian. But they are challenges that arise only if one accepts an endurantist view of persistence through time. Admit temporal parts into your ontology, and a response to Broad’s objection immediately presents itself.
The response I have in mind is probably already easy to see, but let me spell it out a bit. For the sake of argument, let’s suppose that perdurance is true. On the perdurantist picture, persons are temporally extended objects with a temporal part at each time in their life-span. One immediate benefit is that we can reject that contention that persons are related to time in a much different way than events are. On the perdurantist picture, both persons and events are four-dimensional, have temporal parts, and are only partly present at t1 and partly present at t2. Because persons are now analogous to events (in some relevant ways), it’s no longer so mysterious how a person can stand in a casual relation to an event. (Or, at least, the reasons we used to have for thinking it was mysterious are no longer good reasons.) How can it be that a temporally extended person can cause an action at a very specific time? The answer is quite straightforward – by having a temporal part at t1 that causes the action at t2. Why didn’t he cause it before? Because a different temporal part was present at prior times. Once we put persons more directly in time, the way the perdurantist urges us to, agent causation becomes more intelligible. Let’s look again at the explanatory version of Broad’s objection.
Again, according to this version, an agent cannot be a cause because if an event is caused to occur at a particular time, then some part of its cause must explain why it occurred when it did. Since an agent can’t provide the requisite explanation, agents can’t cause events. But if we accept perdurance, then agents become much more explanatorily powerful. We can explain the precise timing of an action by appealing to the precise timing of the temporal part that caused the action. What explains the agent’s causing the action at t3? The agent had a temporal part at t2 (or perhaps t3 in the case of simultaneous causation) that caused the action. Why did the agent cause the event at t3rather than at t1? Again, it was the agent’s temporal part at t2 that caused the event, and that temporal part didn’t exist prior to t2, and so couldn’t have caused anything at t1. It seems that perdurance allows the agent-causal libertarian to explain everything that Broad’s objection accuses him of not being able to explain. In this way, temporal parts can help block the explanatory version of Broad’s objection.
More work would need to be done in order to spell out the temporal-parts-as-agent-causes picture here, but it looks to be a promising project, given the resources it has to respond to Broad’s objection. I hereby recommend to the agent-causal libertarian that he adopt an ontology of temporal parts. Although I think this is a good recommendation, I doubt that it will be very welcome simply because most agent-causal libertarians appear to think that they are committed to endurance for independent reasons. But I am puzzled by this, and in the next section, I’d like to examine this supposed commitment to endurance. I will argue that the agent-causal libertarian need not take himself to be committed to endurance, and thus is welcome to adopt my suggested strategy of responding to Broad’s objection.
V. The Ontological Commitments of Agent-Causal Libertarianism
Most players in the free will debate do not argue for one view of persistence through time over another, but they do take sides. So, for instance, we have this quotation from Randolph Clarke:
For one thing, agent causes are taken to be enduring substances – entities whose persistence through time consists in their being wholly present at a plurality of times…As the problem of persistence is a large one, I shall here only note that the commitment to enduring objects is a vulnerability of agent-causal accounts. If it is impossible for objects to endure, then such an account cannot be adequate.[8]
And a similar sentiment is expressed by Timothy O’Connor:
Let us take stock of the ontological commitments of my account of agent causation…Concerning particulars (specifically, agents in our technical sense), we require that they include things that endure through time, wholly existing at each moment of an extended temporal interval, as opposed to things that ‘perdure’ by having temporal parts that exist at each moment of the thing’s existence.[9]
And O’Connor again, in a more recent article:
All [agent-causal] theorists require that we think of agents as things which endure through time, such that they are wholly present at each moment of their existence.[10]
But why do agent-causal libertarians think that endurance is required to make sense of agent causation? I can find few arguments for this view. In fact, as far as I can tell, O’Connor is the only person who explicitly addresses this question. In a recent article, he elaborates a bit on why he thinks any adequate agent-causal theory must count endurance among its ontological commitments as follows:
Clearly, such a temporally extended object [the four-dimensional worm of temporal parts theory] is not suited to play the role of an agent cause of ever so many particular episodes in its own life. But nor are any of the momentary stages suitable, as these are not distinct from total states of the object at a particular time, and agent causation is supposed to be different from causation by states or events within the agent. Hence, there is nowhere to 'put' agent causation in the temporal parts theorist's ontology.[11]
Let me attempt to make the argument here a bit more precise so that we can better evaluate the claim that agent-causal libertarians must also be endurantists.
As I see it, O’Connor is here presenting a dilemma – the first horn takes on the notion of a temporally extended person causing an event and the second horn takes on the notion of a temporal part of a person causing an event. But since those are the only two possible ways in which a perduring person might be an agent cause, O’Connor concludes that endurance is required to make sense of agent causation. Let me take each horn of the dilemma in turn.
First, regarding the claim that a temporally extended person cannot cause an event. As we’ve seen, perdurantists think that people (and so agents, presumably) are four-dimensional objects with both spatial and temporal parts. But O’Connor maintains that such a view doesn’t have the resources to accommodate agent causation. Why not? Well, it’s not easy to see what the objection exactly is here since no argument is explicitly stated, but let me attempt a charitable reconstruction.
When an agent causes an action, the action occurs at a specific time in the agent’s life. But how can a four-dimensional object possibly be the cause of actions that occur at specific times? Certainly four-dimensional objects might be able to stand in tenseless relations to other objects that occupy four-dimensional regions of spacetime, but causation is much more involved. It involves production, or bringing about, which is a notion that only makes sense if the thing doing the bringing about is somehow more intimately related to the time of causation than a four-dimensional object would be. Perhaps it will help to draw an analogy with event causation. Events, all parties agree, are four-dimensional. They last for a certain period of time, and have temporal parts. But it doesn’t seem to make sense to talk as if the whole (four-dimensional) event caused something that happened at, say, some time during the middle of the event. It isn’t that the lecture I attended last week (the whole four-dimensional event) angered me – rather, it’s that a particular temporal part of the lecture angered me, namely, the part in which the lecturer made an offensive remark. Though we might colloquially say things like, “The lecture angered me”, we don’t literally mean that. Rather, we mean that the lecturer’s making an offensive remark angered me. Similarly, if people are four-dimensional objects, it makes no sense to talk as if they cause events that happen at, say, some time in the middle of their lifetime.
So that’s the first horn of the dilemma. Or, at any rate, that’s my attempt at fleshing out O’Connor’s remark that a four-dimensional object “is not suited to play the role of an agent cause of ever so many particular episodes in its own life”. But is the argument successful, or is there some way a perdurantist can respond? I am inclined to think that the perdurantist can give a perfectly good response as follows. It does make sense to talk as if a four-dimensional object causes actions that occur at particular times, but just like many other claims associated with four-dimensional objects, the claim is made true in virtue of what the object’s temporal parts do. So, for example, it makes perfect sense to say of a four-dimensional object that it is red today, because its today temporal part is red simpliciter. Likewise, it makes perfect sense to say that a four-dimensional agent causes an action that occurs at (say) t2 because its t1 temporal part caused that action. So long as it makes sense to ascribe these properties to the temporal parts of a temporally extended object, it also makes sense to ascribe them (though in a temporally qualified way, of course) to the four-dimensional object itself.