Sunday, November 25, 2007

Causal Theories of Action (I): Difficulties With Acting for a Reason

In defending Davidson’s causal account of reasons, Mele suggests that the strongest defense for this account is provided by a challenge Davidson raises to non-causalists: “If you hold that when we act intentionally we act for reasons, provide an account of the reasons for which we act that does not treat (our having) those reasons as figuring in the causation of the relevant behavior.” (Mele 2003, 69) Davidson does make quite a lot of this challenge, arguing that the causal account—i.e., an account on which reasons are causes of actions—gives us a way of explaining why an intentional action occurred. Lacking any other explanation, we must take up the causal account as the best one. One might add, also, another common point to strengthen this one: a non-causal account in principle cannot explain, or fully explain, why something occurred (Honderich, Chapters 2 and 4).

Mele goes on to claim that “the challenge is particularly acute when an agent has two or more reasons for A-ing but A-s for only one of them” (Mele 2003, 70). Mele illustrates this with an example:

Al has a pair of reasons for mowing his lawn this morning. First, he wants to mow it this week and he believes that this morning is the most convenient time. Second, Al has an urge to repay his neighbor for the rude awakening he suffered recently when she turned on her mower at the crack of dawn and he believes that his mowing his lawn this morning would constitute suitable repayment. As it happens, Al mows his lawn this morning only for one of these reasons. In virtue of what is it true that he mowed his lawn for this reason, and not the other, if not that this reason (or his having it), and not the other, played a suitable causal role in his mowing his lawn? (Mele 1997)

Mele’s point is that if I have more than one reason for acting but act for only one of these reasons, it is unclear how we could explain this occurrence without seeing the reason for which I acted, but not the other reason(s) I had, as playing a causal role in the production of my action. Let’s call examples of this sort Multiple Reasons Scenarios (MRSs).

Interestingly, very much the same MRS can be used to defend a non-causal account of reasons. Witness the following, from John Searle:

[One] way to see the existence of the gap is to notice that in a decision making situation you often have several different reasons for performing an action, yet you act on one and not the others and you know without observation which one you acted on… Suppose for example that you had a whole bunch of reasons both for and against voting for Clinton in the presidential election. You thought he would be a better president for the economy but worse for foreign policy. You liked the fact that he went to your old college but didn’t like his personal style. In the end you voted for him because he went to your old college. The reasons did not operate on you. Rather you chose one reason and acted on that one. You made that reason effective by acting on it… The remarkable thing about this phenomenon is: in the normal case you know without observation which reason was effective, because you made it effective. (Searle, 16)

Unlike Mele, who uses an MRS to demonstrate the need for a causal account (or, rather, to point to an explanatory problem that—in his view—is best resolved by a causal account), Searle uses an MRS to show that, as far as our experience is concerned, there is a gap between our reasons and our actions. There are, of course, ways of bringing these accounts together—some libertarians (e.g., van Inwagen, Kane), for example, accept that reasons cause actions, but insist that they do so indeterministically. Both the causal theory and the gap are thus preserved.

When the same kind of scenario, an MRS, seems to support conflicting positions, however, it may seem worthwhile to look at what sort of bedrock the scenario provides; is it really solid enough to support any conclusions whatsoever? I do not think so. There is a hidden assumption within any MRS that makes it, to my mind, impossible to use it as a premise in any argument.

The assumption is this:

Given multiple reasons R[1-n], agent S can act for/on reason Rp and not for/on any other reason Rq, where p,qЄ{1-n}

You might be thinking: what kind of hidden assumption is this? Of course we are capable of acting for some particular reasons out of a set of reasons we might have to do something! That’s obvious! But now, suppose you have to justify this claim. However you go about doing it, you certainly cannot appeal to the thing that makes it seem so obvious, i.e., introspection. I will deal with this in the next post. First, I want to point out that it is possible that, when we have multiple reasons for doing something, we never act on only one of them. If so, then it would be false that we play any active role in selecting one reason out of a group to act on, or that one reason out of a group causes us to act. That this is possible does not make it true, of course. But without a way of dismissing this possibility, one has no right to use it as evidence for anything else.

In the next post, I will look at various ways of defending the MRS in question and suggest that their failure may lead us to the thought that any causal account of action must be backed by a theory of the will.

Honderich, Ted. (2002) How Free are You?

Mele, Alfred R. (2003) “Philosophy of Action” in Kirk Ludwig, ed. Donald Davidson, 64-84.

Mele, Alfred R. (1997) Agency and Mental Action, Philosophical Perspectives, 11, 231-49.

Searle, John. (2001) Rationality in Action.


  1. Interesting. But consider the following example:

    Joe is someone who is convinced by Singer's arguments that eating meat is wrong. These arguments give him a reason to abstain from eating meat. But those succulent steaks are so good, that Joe is unable to act on that reason.

    This state of hypocrisy and akrasia goes on for several years, in the course of which Joe becomes a philosophy instructor who takes his teaching duties seriously. He is teaching applied ethics this semester, Singer's arguments against eating meat included. Now because Joe takes his teaching duties seriously, he wants to impress on his students that philosophical arguments can be used to revise one's erroneous beliefs and improve one's life. He is also aware that his hypocrisy (w.r.t. Singer's arguments) will get in the way of impressing this important lesson on the students. So he stops eating meat for this reason.

    In the above argument, can't you say that Joe has two reasons against eating meat, one mentioned in the first paragraph, the other mentioned in the second? And isn't it clear that it is the second reason which actually motivates Joe to become a vegetarian? At least, it seems conceivable, and psychologically plausble, no? (Also, I believe I know at least one person to whom this example applies.)

  2. Thanks for the great comment, Boram. Here are a few points off the top of my head:

    First off, the scenario you present is already different from the typical MRSs. There are not two unconnected reasons, since without the first reason there would not even be a reason to stop eating meat at all. Thus, whatever reason Joe acts on of the ones you give, he will--at least indirectly--be acting on the first reason. So this isn't really a counterexample to my point, I think.

    Second, it seems that Joe is bothered by his perceived hypocrisy; perhaps he is ashamed of it. This may well give him an extra reason to stop eating meat.

    Third, there is a question of why Joe takes his teaching duties seriously. Also, of why he thinks that his hypocrisy will interfere with those teaching duties. The reason this gives him to stop eating meat is itself fairly complex.

    Now, I'm perfectly fine with the idea that we might have extremely complex reasons for acting. Or, we might just as well say that we have multiple reasons here, arranged in some sort of hierarchical structure, where some particular reason is clearly strongest--but it might also be true that this particular reason just adds that extra bit of weight needed to get the whole ball rolling, so to speak. E.g., Maybe Joe is convinced by Singer's argument, and he's worried about his hypocrisy, and then--let's just say--the fear that his hypocrisy (or his concern about his hypocrisy) will interfere with his teaching goals finally gets him to act on the first two reasons.

    What I'm trying to do here is muddy up your example. Mainly, because real life motivation tends to be pretty muddy, I think.

    Incidentally, though: in both Searle's and Mele's examples, the idea of the person acting for only one of the reasons is both conceivable and psychologically plausible. That doesn't show that it ever happens this way. It's the idea that our reasons for action might be purified, separated out from the other reasons, that I find somewhat dubious.

    (Also, if you know a person to whom this happened, you can mess with them by pointing out that giving up eating meat in order to improve one's teaching, and not because of the ethical arguments for not eating meat, is kind of hypocritical under the circumstances.)

  3. A fine response! I am looking forward to your next post.

  4. I wonder how we might compare MRS's in action theory to MCS (multiple cause scenarios) in nature. Aren't all phenomenon the effect, in some way of, multiple reasons or causes? The pipes burst because the water in them freezes, and because they are made of copper, and because they were too close to the surface, and because.....

    I mention this because the MRS's seem only to be more problematic than natural causal sequences if we assume that all events in nature naturally have only one cause--thus making it seem like, in order to remain consistent with this idea about nature, we think that we have to make sense of there likewise being only one cause to actions.

    Put differently, isn't the MRS scenario in action--which reason, really, was the reason behind my action--a bit like asking of MRC in nature, which cause, really, was the cause behind the bursting of the pipes?

    If I understand you correctly, this is sort of the point you were making in the post, right?

  5. Michael--yeah, that is pretty much my point.

    Honderich reasonably wants to talk about causal circumstances rather than causes, since a bunch of states of affairs must normally be in place in order for something to happen. What we normally call the cause, then, is something like the most relevant event involved in producing another event. Oxygen and dryness are both necessary for the striking of the match to cause fire, but the striking is the most relevant since, what we are looking for, is a way ceteris paribus to repeat the effect.

    We do the same with actions: that I wanted to see you suffer is more relevant to explaining why I stabbed you with a thumb tack than, say, that I happened to have a thumb tack on hand, or that I enjoy seeing blood... But relevance is context dependent: if I normally stab people with thumb tacks, then my lack of a thumb tack on hand may be the best explanation for why I did not stab you with a thumb tack.

    What's weird about MRSs is that they seem to treat reasons as single, individuated causal events rather than just the features of a mental state that provide the best explanation of my action, given all sorts of background assumptions and goals of explanation.