Wednesday, December 18, 2013

Substance causation, agent causation and time

Aristotelians about causation think all causation is substance causation. Events are causes only derivatively. What does the real causing are substances. This should make Aristotelians very sympathetic to the use of agent causation in the theory of free will. And insofar as the theory of agent causation is just that the agent is the cause of free actions, the Aristotelian who believes that we are substances[note 1] is surely going to agree that we are the causes of our free actions, and we are both agents and substances, so the agent is the cause of her free actions.

So far so good. But there is more to agent causation in regard to free will. Typically, agent causalists invoke agent causation to solve problems such as the randomness problem for libertarians. Agent causation is what makes an action be genuinely one's own action rather than a random blip. But the Aristotelian's embrace of substance causation is too broad. For not only does the Aristotelian think that her free actions are caused by her, she also thinks her non-free actions are caused by her, and even things like the circulation of the blood, which isn't an action at all, are caused by her. Moreover, since she is an agent, she thus thinks all of these things are caused by an agent. But if agent causation metaphysically lumps free actions with non-free ones, and doesn't distinguish them metaphysically from the circulation of the blood, then agent causation can't do the job it's designed for. The Aristotelian believes in agent causation, of course, and may do so with good metaphysical reason, but this agent causation cannot be used to solve the problems that the free will theorists want it to solve.

This line of thought might lead some Aristotelians about causation to accept a version of Cartesian dualism on which we are souls. For then one might hold, contrary to Aristotle, that only our free actions are caused by us and that the circulation of the blood and so on is not caused by us, because we are immaterial beings whose only direct effects are in whatever the equivalent of the pineal gland on this theory will be. This is not in the Aristotelian spirit, though, and it leads to unhappy ethical conclusions (bodies as akin to property).

I think there is something else one should say here. One shouldn't say that agent causation just is causation by an agent. Rather, agent causation is causation by an agent qua agent. You cause your free actions qua agent and you circulate your blood qua mammal, though of course you are both agent and mammal. It is a bit odd to say that you don't perform your non-free actions qua agent, though. After all, how can there be an action without an agent? Aren't all actions, free or not, the actions of an agent qua agent? Maybe. But maybe the distinction is still of some help, for maybe the kinds of mere randomness we want to rule out with the distinction isn't an action at all when looked at more closely.

There is another issue around here. There needs to be more to substance causation than the simple structure substance x causes event E. For paradigmatic substances persist over a long time, but many of their effects happen only at particular times in their existence. And there is an explanation of why the substance causes an effect at one time or another. For instance, I caused oatmeal to be assimilated to me earlier to day because I was hungry. The explanation includes not just the substance, but a state of a substance (and maybe other substances—but that's for another day). I caused the assimilation of my breakfast not only qua agent, but qua hungry agent. Likewise, I circulate the blood not just qua mammal, but qua mammal with a brain stem that sends such-and-such electrical signals to the heart.

Apart from considerations of free will, then, Aristotelians should say that the structure of substance causation is something like: substance x qua in state S causes event E. If we have an Aristotelian constituent ontology, then the state will be a mode (an essence, a necessary accident or a contingent accident) of the substance, and the causation relation will be a ternary relation between the substance, the mode and the event.

But now that we have all of this detail in place, we can go back and ask whether we still get benefits of an agent causal theory in regard to free will. That's not so clear. For instance, when I qua hungry agent caused my breakfast to be assimilated to me, the work distinguishing this from non-actions like circulation is being done by my state of hungry agency. It is because this state is involved in the causation—and not just involved, but involved in the right way (my being a hungry agent could cause me to grow a tail if I was rigged the wrong way)—that I am acting. But event causalists can say something exactly parallel. What distinguishes my eating the breakfast from my circulating the blood is that the former is caused by the event of my being a hungry agent qua my being a hungry agent.

So I do not know that Aristotelian agent causalists can claim to do better than event causalists. In fact, for certain ends they might well want to join cause with the event causalists.

3 comments:

Peter said...

Part I:
Nice post. Sorry for the length of this reply. Some thoughts (which involve lots stage setting; this way, if I presuppose something you object to, it will be more obvious):

Maybe it is right to think of the job of agent causation as making sure an act is “genuinely one's own action rather than a random blip,” but this might be too strong. I suppose it depends on just how much work ‘genuinely’ is doing. When I think of agent causation as an aid to incompatibilists, it is to ensure more control over an action than determinists can provide. Presumably the determined actions of a dog (if they are determined) are still genuinely its own. But the dog can presumably not meet the control condition for free will.

If determinism precludes meeting the control condition, then it seems that the event-causal incompatibilist cannot account for our meeting it either, or so the argument goes (although maybe it can be desire-worthy in ways unrelated to control, as Mele notes).

Arguably event-causation does not add to the control had by a determined agent. A determined agent has control over an event in virtue of the event being caused by an earlier agent-involving event. (Something like, agent A having property P at T causes event E.) The event-causal libertarian will give the same story, just that now, even given agent A having property P at T, E might not occur. Even granting that E’s possibility of not occurring is also traceable to agent A having property P at T, this indeterminism does not seem to enhance control. If anything, it might decrease the level of control.

Presumably, what the agent-causal libertarian can add is that the agent does have more control than the event-causalist. Perhaps this is not so (van Inwagen), or perhaps even if it is so, the concept faces lots of other problems (Clarke) or perhaps it is helpful and coherent, but unlikely (Pereboom). In any case, it seems that it can enhance control in virtue of some sort of direct control, that is, control that is not mediated by events (or, if mediated, mediated in the right way on some kind of integrated approach).

Peter said...

Part II:

Okay, so can the Aristotelian help himself to this move?

It seems right to say that we shouldn’t just say that all human actions are caused by agents, but that certain ones are caused in certain ways, others in other ways. I am not sure this is captured by saying qua agent, but it might be. For example we could talk about actions caused by agents in virtue of different kinds of properties (reasons vs chemicals in stomach) and so we might be able to ground the difference between voluntary and non-voluntary activities.

You mention that the structure of causation might be something like this: substance x qua in state S causes event E. I am not sure what to make of the ‘qua.’ Is this something like “in virtue of?” (I suppose this might just be exchanging one slippery notion with another, but I feel as if I have a better handle on the latter.) Suppose it is. If so, then I cause some other event in virtue of possessing some feature.

This version of substance-causation still, it seems, could have three versions:

1) Determinist – given that you had that feature at that time in that environment, it was not possible for you to avoid causing that further event.
2) Indeterminist1 – given that you had that feature at that time in that environment, it was possible for you to avoid causing that further event, but you did not control which one occurred in any robust sense.
3) Indeterminist2 - given that you had that feature at that time in that environment, it was possible for you to avoid causing that further event and you did control which one occurred in a robust sense.
The question that remains is whether the Aristotelian can help himself to the third option in the way an event-causalist cannot. Perhaps so. If features give substances causal powers, and one of those causal powers allows for more control (will?), then perhaps. What if the causal structure of an action were something like this:

substance x in virtue of power y (bestowed by feature f, if features and powers are distinct) causes event E for reasons r

If this is right, then we could say:

Alex, in virtue of his power y (will?) chose to eat breakfast for reasons r (he desires to stop being hungry, and believe that eating would satisfy this desire [or whatever this would be on an external construal of reasons]).

If this is right, then it is quite different from the structure of an involuntary action.

If we are able to distinguish the ways that features/properties/powers relate to later actions from the ways that reasons do, and we are able to give an account of a feature that bestows a control-conferring power, then I don’t see why the Aristotelian cannot help himself to some of the supposed benefits of agent-causation in a way that seems impossible for the event-causalist.

One might worry that this control-conferring power is ill-defined. So it is. Even so, it seems that there is a possibility of the Aristotelian adopting such a view. The event-causalist does not seem to be in a similar position.



Sam said...

I'm having a hard time wrapping my mind around the idea that "the agent is the cause of the free action." Since it's the action that we say is free, that must be where the will is located since that is the volition.

But I thought the whole notion of free will was that the will is not caused. How can an action be free if it is caused?

The strange thing about saying "the agent is the cause of the free action," is that it seems to imply that there is a distinction between the agent and the action such that one causes the other.

If the volition or act of will is the same thing as the free action, then what does it mean to say that the agent causes it? Is the agent causation itself an act of the will, or is it a "random blip"?

If it's an act of the will, then to say "the agent causes the free action" seems equivalent to saying, "The choice causes the choice," which doesn't seem to solve any problems with random blips vs. control.

But if it's a random blip, then we're saying, "A random blip causes the choice," which doesn't seem to solve any problems about control.

So I don't know what is meant when people say, "the agent is the cause of the free action."

Shouldn't "free" modify "the agent causing" instead of "action"? In that case, "The agent freely causes the action." Wouldn't that be more accurate? Because that way you don't have anything that's free being causally determined.

But then you're still stuck with the random blip problem.

I really think compatibilism is the only coherent way out of this quagmire. An action is ones own to the degree that a person's own desires and motives play a hand in bringing about that action. The less hand one's own desires and motives have in bringing about the action, the less those actions are one's own. The more hand our desires and motives have in bringing about our actions, the more those actions are our own. It follows that our actions are completely our own when our desires and motives have everything to do with our actions, i.e. when they determine our actions.