According to the Doctrine of Double Effect (DDE), there is a
morally significant difference between (i) the consequences of an action
that were intended by the agent and (ii)
the consequences that were not intended, but are merely foreseen side-effects of the action.
More precisely, according to the DDE, if the consequences of
an action include a state of affairs S,
and S is in the relevant way a bad state of affairs, then (other things
equal) the action is worse if it is
the successful execution of an intention
to bring about this state of affairs S
than if S is merely foreseen (and not intended) by the
agent. Being the successful execution of a bad
intention of this sort is a bad
feature of an action – the sort of feature that can make the action
impermissible, or can at least make an impermissible action more seriously wrong
than it would otherwise be.
In my view, the DDE is entirely true. But the DDE has been
attacked on many fronts. One of these attacks certainly raises a fundamental
problem that any full defence of the DDE must solve – namely, the “closeness
problem”, which was originally raised by H. L. A. Hart in “Intention and Punishment”, and has since then been discussed by many philosophers, such as Philippa
Foot, Warren Quinn, and Jonathan Bennett, among others.
More recently, however, a different sort of
attack has been launched by Judith Thomson (“Self-Defense” in Philosophy & Public Affairs 1991,
and “Physician-Assisted Suicide” in Ethics
1999). This attack on the DDE has now received a whole-hearted endorsement
in T. M. Scanlon’s book Moral
Dimensions (Harvard UP, 2008). However, Thomson’s attacks on the DDE seem to me to
be completely misguided. In this post, I shall examine one of her objections,
and then I shall explain what is wrong with it.
Out of all of Thomson’s objections to the DDE, the one that I
shall consider here is the following (“Self-Defense”, pp. 293-95):
Suppose that Alfred intends to
kill his wife, but by mistake he gives her a drug that in fact saves her life.
Surely Alfred’s bad intention does not make it wrong for Alfred to give his
wife the drug that saves her life! This supports the thesis that quite
generally “intentions are irrelevant to permissibility”; and as Thomson claims,
“if that thesis is true … the DDE collapses”.
1. There are at least two serious problems with this argument.
First, as I formulated it, the DDE implies that it is a bad feature of an action
if the action is the successful execution
of a bad intention. But in the case that Thomson asks us to consider, Alfred’s
action is quite obviously not the successful
execution of a bad intention – on the contrary, his action produces the
very reverse of what he intended, since his action saves his wife, when he
intended it to kill her!
So prima facie, it
is hard to see how this objection can get a grip on the DDE at all. Moreover, since
the difference between intentions that are successfully executed and those that
are not successfully executed seems likely to be a morally significant difference, it is clear
that our intuitions about cases in which someone fails to execute a bad intention cannot go very far to support to such
a sweeping thesis as Thomson’s claim that “intentions are irrelevant to
permissibility”.
2. Secondly, it is not clear that Thomson’s thesis that “intentions
are irrelevant to permissibility” is incompatible with the DDE anyway. Thomson formulates the thesis as follows:
It is irrelevant to the question
of whether the agent may do A what
intention the agent would do A with
if he or she did it.
But does the DDE really have to deny this thesis? Suppose that it
is true of you that if you did A, you
would do it with a bad intention. Must the DDE then say that you may not do A? Surely the answer is No. What the DDE
must say is merely that you may not do-A-with-this-bad-intention;
it need not say that there would be anything wrong at all with your doing A with a different intention.
Now, some people may complain that this response to Thomson conflicts
with a certain widely-accepted rule of inference – namely, the rule that the conjunction of “If you were
to do A, you would in fact be doing B” and “You may not do B” entails “You may not do A”.
But in my view, this rule of inference is completely invalid.
Indeed, this rule of inference licenses an objectionable sort of bootstrapping – since according to
this rule, a certain purely counterfactual truth about the agent is enough to
fix what the agent ought to do, regardless of whether this counterfactual truth
ought to be true of the agent or not.
In my view, the most that follows from “You ought not to do B” together with “If you were to do A, you would in fact be doing B” is the disjunction “Either you ought
not to do A, or you ought not to be such that if you did A, you would be doing B”.
What is curious is that Thomson herself seems to
agree with me in rejecting this mistaken rule of inference (p. 294). However, she
fails to see that without this mistaken rule of inference, there is absolutely no
reason to think that the thesis that “intentions are irrelevant to
permissibility” (as she formulates it) is in any way incompatible with the DDE.
As I noted above, I believe that Thomson's other objections to the DDE are equally misguided (as is Scanlon's endorsement of these objections). I shall try to find the time to explain what is wrong with Thomson's other objections at some later point.
Two really small points.
First, I’ve always found assessing the DDE really hard as I’ve never been able to tell the difference between intended consequences and merely foreseen (and not intended) side-effects. I would appreciate if someone could help me with this distinction. What is the sort of plausible account of intentions that would make sense of the difference? I tend to think of intentions as some sort of attitudes of being settled on a plan.
Second, even if I’m sympathetic to the second objection, it does seem to require that acts and intentions are separable. In many cases, this seems to be possible – some actions you can do by intending different things. Yet, some people think that actions are identified by their intentions and for some actions there seems to be a logical connection between the action and some core of the intentions for which these actions are done (as von Wright argued I think). I worry that someone defending DDE might think that the actions which usually come under this principle are such actions.
Ralph,
Interesting post. A question: So what do you want to say about the trolley case? Is it permissible to pull the lever and divert the trolley onto the side track with only one person on it? I suppose that you would have to say that it depends on what the agent’s intentions are. Is that right? Suppose that the person on the side track is Bush. In case 1, the agent is Rush, someone who loves Bush. He pulls the lever with the intention of saving the five on the main track, lamenting the unintended but foreseen consequence that this will have: viz., the death of Bush. In case 2, the agent is Al, someone who hates Bush. He pulls the lever with the intention of killing Bush. The fact that five others will be saved in the process is incidental to his decision to pull the lever. Is it your view that it is permissible for Rush to pull the lever but impermissible for Al to pull the lever. I’m not sure that’s the right result.
I like this post! But here are two quibbles:
(i) In the interests of finding language that is as accommodating of all proponents of DDE as possible, I don’t think you should formulate DDE as a claim about what makes actions better or worse. Rather, I think you should simply formulate DDE as a claim about what makes actions right or wrong, or more or less seriously wrong.
(ii) You write: ‘… according to this rule, a certain purely counterfactual truth about the agent is enough to fix what the agent ought to do’. Shouldn’t this instead be ‘…subjunctive truth about the agent…’? This is a counterfactual truth about the agent only if the agent does not in fact do A.
Ralph,
I think these are excellent points, but I’m curious about one smallish thing. It seems you’re arguing that successfully A-ing from bad intentions can be worse than A-ing from other intentions whereas Thomson and Scanlion seem to be denying that acting from certain intentions ensures that the agent acted impermissibly. Do they have to deny the point about intentions as things that make things worse? Can’t they just say that you’re right but that the fact taht the actions are made worse than they could have been doesn’t mean that the overall verdict is that the agent ought to have refrained from acting or acted for different reasons?
I can imagine a view that accepts everything you’ve said and everything that Scanlon accepts from Thomson. It says that the intentions and motives can make it such that the agent acted impermissibly by providing a pro tanto reason not to perform the action. If it gives only a pro tanto reason, there could be a decisive case in favor of doing the action from bad intentions. (I think Steve Sverdlik has defended a version of this view. So, he agrees with Thomson that if the only reason the President decides to send the army to fight Hitler is that he just really likes war he ought to send the army. But, if you refuse to sell your house because you intend to keep minorities from moving into the neighborhood you grew up in the badness of the intention can give you a reason to refrain from declining the offer and there’s no overriding reason that justifies acting against that reason.)
[For some reason Mike O’s post didn’t display. Sorry for the redundant comment.]
Great comments! Here are some quick responses.
1. Jussi asks about what the distinction between intended consequences and merely foreseen (and not intended) side-effects actually comes to. (This is surely not a “really small point”, by the way!) I basically completely agree with Jussi’s point that intending is the “attitude of being settled on a plan”. It is a challenging question in the philosophy of mind and action to spell out this account of intention in more detail, but Jussi’s phrase seems correct and helpful to me.
2. If Jussi is right about what intentions are, then we should object strenuously to the tendency of many discussions of the DDE to conflate intentions with motives. (This conflation is evident throughout Chapter 1 of Scanlon’s new book, and Clayton seems to be conflating these two notions in his comment as well.) The motives of an action include many more mental states than just the intentions that the agent is executing in performing the action. For example, an agent’s motives typically include such mental states as her desires, emotions, evaluative beliefs, and so on; and none of these mental states are intentions. (I hope to explain in a later post why many of Thomson’s other objections to the DDE are vitiated because they rest on this conflation of intentions with motives.)
3. Another confusion that seems to run through the literature on the DDE is the confusion between act types and act tokens. So, for example, I would say that in the case that Doug imagines, the particular act token that Al performs is wrong (because it is the successful execution of the bad intention to kill Bush); but of course the act type of diverting the trolley is one that it is quite permissible for Al to perform in this case (although it is not permissible for Al to divert the trolley from the bad intention to kill Bush).
Jussi also raises the question of whether acts and intentions are “separable”. Once we have distinguished between act types and act tokens, this ceases to be a difficult problem. Act tokens are not separable from intentions: indeed, an act token just is the execution of a certain intention. But act types are almost always separable from intentions: it is obviously possible to do acts of the same type from different intentions. We can take the DDE as making a point about both act types and act tokens. According to the DDE, some act tokens are wrong precisely because they are the executions of bad intentions; and it is typically worse to do the act type of bringing about S (which is a bad state of affairs of the relevant kind) from the intention of bringing about S than from some different and less objectionable intention.
4. Now, Mike and Clayton are certainly right that my formulating the DDE in terms of factors that make actions “better” or “worse” was controversial. I was presupposing something like my own view here — since in my view, whenever an action is wrong, that is because the action is bad in some way.
At all events, I do think that there can be cases where the agent’s intention is what makes the difference between an act-type’s being permissible and its being impermissible. (I.e., there are some act-types, like dropping a bomb that destroys both a munitions factory and some innocent civilians, that it is permissible to perform with one intention, but impermissible to perform with another intention.) So I do accept the more familiar version of the DDE as well. However, I also find my more controversial version of the DDE more explanatory and to that extent more intuitively persuasive.
5. Finally, of course, Mike is quite right that I should have said “subjunctive” and not “counterfactual”…
Ralph,
You write:
So suppose that Al is deciding whether or not to flip the switch, and he asks me what he ought to do. We both know that if he flips the switch, he will, in doing so, intend for Bush to die. How should I advise him, then? Should I say: “No. You ought not to flip the switch, because, if you do, you will thereby successfully execute your bad intention to kill Bush.” That seems wrong to me. It seems that he should flip the switch even though he will thereby be executing his bad intention. Of course, I admit that he’s a bad person and that he should not intend for Bush to die. But whether he should intend for Bush to die seems to me to be a separate question from whether he should flip the switch given that he disposed to form this intention.
Ralph–
Insofar as there is a disagreement between us, I believe it is a different disagreement than you have with Thomson. (This follows from what you say about making an action “worse,” and Mike and Clayton are right to pick up on this.) In my chapter, I am trying to explain why DDE should have seemed very plausible (to me) when understood as a thesis about moral permissibility, if (as it now seems to me) it is not correct (when understood in this way.) I emphasize that all of this depends greatly on how “permissibility” is understood. My explanation is that the plausibility of DDE depends on a failure to distinguish between the question of permissibility, as I understand it, and an assessment of the way the agent went about deciding what to do. The latter is important–in particular, important for blame–but distinct from permissibility as I understand it. As I say on pp. 24-25) if one takes the important moral category to be that that of a good action, then things may look different, since the latter kind of assessment may be relevant to this question. So, I say, it is not clear how the disagreement over DDE should be understood: is it a disagreement about permissibility, understood as a distinct notion, or is it really a disagreement about what the relevant moral category is? What you say at the end suggests that as for as you are concerned the latter is true: your defense of DDE is premised on taking “good action” as the fundamental category and understanding “permissibility” as merely a corollary of this. I of coruse do not share this premise, but I think that clarifying the nature of the dispute in this way is a step forward. This was one of the things I hoped to promote by what I wrote.
Tim
My understanding of the DDE, based on what Anscombe says about it, is that it is essentially negative. That is, rather than telling us what to do, it simply tells us that certain acts are not necessarily wrong just because they have bad consequences. (See p. 225 of Human Life, Action and Ethics: “It will be apparent that this principle tells you rather what you can’t do than what you can. Also, it is particularly devised for the causing of death; causing of other harm is not covered by it.”) So Thomson’s example about an act with a good consequence (Alfred’s wife is saved) is, as you suggest, completely irrelevant (at least to the form of DDE that Anscombe is concerned with).
The idea that “It is irrelevant to the question of whether the agent may do A what intention the agent would do A with if he or she did it,” far from being incompatible with the DDE, could actually be part of it. Murdering people with a good intention, for instance, is clearly incompatible with the DDE. It is part of the doctrine that certain acts are forbidden no matter what.
In a sense, then, intentions are irrelevant to permissibility according to the DDE. They become relevant when a permissible action that would have been reasonably expected to produce more good than bad causes, or can reasonably be expected to cause, the death of an innocent person. Then, according to the DDE, we need to ask whether this death was intended.
The relevance of the trolley case, it seems to me, is not that it shows the DDE leading to bad results but that it provides a case in which the DDE does not help. Qua act of killing one person, pulling the lever would be forbidden. Qua act of saving five people, pulling it would be permissible. If the act in question is forbidden, the DDE says we must not do it. But it does not tell us how to describe or understand the act of pulling the lever. It does not tell us to describe the action as whatever the agent has in mind. Let’s say she wants to save the five and so diverts the trolley to the other track. If we rightly call this an act of saving lives, the lives saved outnumber the life lost, and the taking of that life was not intended, then the DDE says this is OK. But do we rightly call this an act of saving lives, since it looks an awful lot like an act of murder too? The DDE says nothing about this, as far as I can see.
Doug —
Your point here is very similar to another one of Thomson’s objections (see especially “Self-Defense”, p. 293). Briefly, I would say that the way in which you should respond to Al is by saying: “You should divert the trolley. But you mustn’t do it in order to kill Bush; you may do it only in order to save the people on the main track.”
You may ask, What if Al can’t help what intentions he would be acting from if he diverted the trolley? This would be very similar to one of the objections that Scanlon raises in Ch. 2 of his new book. I hope to explain why that objection is mistaken in a subsequent post.
Tim —
Thank you so much for your comment! I enthusiastically agree with you that it is important to distinguish between (i) the critical (typically retrospective) use of ‘permissibility’ or ‘wrongness’ in judging an act-token, and (ii) the deliberative (typically prospective) use of these terms in judging an act-type (in relation to the situation of a particular agent at a particular time). However, I do not accept that this distinction has the significance for the DDE that you believe it to have.
I also do not believe that the specific disagreement that I have with you hinges on my view that whenever an action is wrong, that is because the action is in some way bad. I am pretty sure that I could have formulated my response to Thomson’s objections without relying on this view of mine.
So, I believe that we do genuinely disagree about the deliberative (typically prospective) use of the term ‘permissible’. I say that even when the term is used in this deliberative way, you could be in a situation in which it is permissible for you to-do-A-in-order-that-p, but not permissible for you to to-do-A-in-order-that-q (where the “in order that…” clause indicates the intention or purpose). I take it that you would deny this.
Moreover, I suspect that I disagree with you about the retrospective judgments that we can express by means of the critical use of the terms ‘permissibility’ and ‘wrongness’. In my view, these judgments are not the same as judgments on the agent’s practical reasoning: they are genuinely judgments on the act-token itself. Surely every adequate theory should accept that one critical retrospective judgment that one can make on an act token is to say that the agent “did the right thing for the wrong reason”. (This is why it is so important to distinguish sharply between one’s intention in acting and one’s reasons for having that intention.)
In addition, these critical (retrospective) judgments on an act-token are also not to be identified with judgments of blameworthiness. Even if your act was wrong, you might have an excuse — such as ignorance or diminished responsibility — in which case you would not have been blameworthy at all.
Ralph–
First, about blame. I certainly agree that negative critical assessment of (as I would say) the way an agent decided or (as you would say) this aspect of the act, do not entail blame. All I said was that such assessments (depending on what the agent saw as reasons for so acting) can be relevant to blame. I wished after that I had said instead, “are relevant to the meaning of the action, and hence possible to blame.” I do not think that the difference between the way we put this (whether it is said to be a difference in the action itself) is very important. I do not hang anything on it and I suspect that everything I want to say could be put in the way you favor. I certainly agree that a person can do the right thing for the wrong reason (I defend this strongly in Ch 2.)
Second, as I make clear in Chapter 2 of my book. I agree that the reasons for which the agent would do something, if he were to do it, can affect the permissibility of his doing it. This can be so in several ways. To mention two: because of what I call the “predictive” significance of intention, and because the agents reasons affect the meaning of what he does and in some cases permissibility depends on meaning. I am not committed to denying that there may be other kinds of cases in which this is so. What I do deny is that permissibility in the sense I am concerned with depends on the reasons for which the agent would do A were he to do A in the particular way that DDE maintains. I think that the cases in which this appears to be so are better explained in other ways.
T.
Ralph, great post! I actually have a paper on exactly this topic arguing for the same conclusion as you, though not via the same route, which I am revising at the moment. I’ll send it to you when I’m done with it. Just three comments on your post.
First, as Tim Scanlon has already pointed out, he, Thomson, and you’ll be probably also want to add Kamm in the mix are concerned with the issue of permissibility/impermissibility. They can agree with you that having bad intentions is in some sense worse, but still maintain that this in no way affects permissibility/impermissibility. Indeed, they think that having bad intentions can affect a person’s character (Thomson, Kamm) or more narrowly, an assessment of the way the agent went about deciding what to do (Scanlon). So, at least for the purpose of this debate, the challenge for those who disagree with them is to show how intentions can affect the permissibility/impermissbility of an act and not just how intentions can be make something better or worse.
Secondly, for precisely the reason you have given about mistaken intentions, the particular example of Thomson’s that you’ve used is probably not the best example for her claim that intentions are irrelevant for permissibility. Consider though another example from her physical assisted suicide piece, what I call in my paper the
Thomson asks, does the permissibility of administering the drug depend on the doctor’s intention? In particular, does it depend on whether the doctor intends to relieve the pain by intending to cause the patient to die, or intends to relieve the pain by giving the drug which will, inevitably, also cause the patient to die? In this case, the problem of mistaken intentions does not seem to arise. And, according to Thomson, it seems counterintuitive to suppose that intentions determine the permissibility in such a case.
Thirdly, anticipating that someone might respond the way you have done, namely, that when you intend something bad, you are intending to perform a different act, Thomson offers what I call the
According to Thomson, while a doctor who acts on revengeful intentions in such a case is a bad person, as long as the act relieves the patient’s pain, the act remains morally permissible. If she is right, Thomson can say that it does not matter whether intending something bad has resulted in a different act. It remains the case that this case shows that having bad intention is irrelevant to permissibility/impermissibility. And assuming that cases like this one are ones in which intentions should do some moral work, if they are to do any work at all, it seems, so the argument goes, that intentions is irrelevant to permissibility/impermissibility generally, and the challenge for those of us who disagree with her (like myself) is to show why this is not so.
To fill out the last remark in my previous post, which may have seemed too elliptical:
In particular, once we accept the relevance of the distinction between combatants and non-combatants that is central to the usual understanding of laws of war, we can explain the difference between strategic bombing and terror bombing without reference to intention. Combatants in war are permitted to use deadly force of a kind not normally allowed provided that it is reasonable to believe that this is necessary to achieve some military objective, and likely civilian casualties are minimized and proportionate to the objective. (These are points about what justifies such action not about what the agent sees as justifying it.) In “terror bombing” there is no military target in the relevant sense ti provide adequate justification (and/or civilian casualties are not minimized.) Whether civilian casualties will be minimized in the event may depend on the attitudes of the combatants (how much care they are disposed to take.) This is what I referred to as the “predictive significance” of an agent’s reasons for acting. I do not deny its relevance. But this is not the kind of significance that DDE claims for agents’ intentions.
Duncan —
I’m sorry, you must have posted your comment while I was replying to Doug and Tim. Anyway, let me just make two brief points in response to your comment.
1. I do not agree with Anscombe’s interpretation of the DDE. I believe that the DDE should be interpreted as applying more widely, to acts that bring about other bad states of affairs as well as deaths.
2. However, I emphatically agree with you that it is a mistake to try to get the DDE to explain everything. In particular, I agree that we need more than just the DDE to explain why it is permissible to divert the trolley in the original trolley case. The fact that you do not intend the death of the person on the side track is not sufficient to explain this (as Frances Kamm has convincingly shown with the Grenade case).
I hope that I’ll be able to take up your other points later on….
1. What’s the difference between intention and motive?
2. What is a “bad intention”?
3. Yes, it sounds “bad” to have “bad intentions.” How does the opponent deal with this issue? I suppose permissibility might mean something quite technical for the opponent. (Objectively permissible.)
4. Can we really have act tokens? What exactly constitutes an act token? Nietzsche in particular didn’t think ethics could be based on actions because they are too messy and complected. They tend not to be generalizable. Is that absurd?
I’ll respond to Matthew and James as soon as I can. But for now, I’ll just briefly respond to Tim’s last two comments.
I fear that I muddied the waters in my earlier reply to Tim, by introducing too many extraneous issues. So let’s not worry here about use of terms like ‘wrong’ and ‘impermissible’ in retrospective judgments on act-tokens. (This is what I focused on in the second half of my reply to Tim, and what Tim focused on in the first half of his first reply to me.) Let’s just focus on the use of these terms in judgments on the impermissibility of act-types, with respect to the situation of a given agent and time. This is what Tim calls the “deliberative” use of such terms: this use is typically prospective — and at all events, when used in this way, ‘It is impermissible for you to F’ does not entail either that you do actually F, or that you do not F.
Now I insist that even when the terms are used in this way, it can be true that it is permissible for you to do-A-in-order-that-p but not permissible for you to do-A-in-order-that-q — even if you know that doing A will bring it about that q whether or not you intend that q. This will typically be true when the state of affairs of its being the case that q is a bad state of affairs of the relevant kind, while the state of affairs of its being the case that p is not a bad state of affairs in the relevant way.
Tim’s position seems to be that permissibility cannot behave like this; so there is clearly a disgreement between him and me here. (He has not explained why he thinks this in his comments on this blog; he explains this at length in his book.)
What he has pointed out in his last two comments on this thread is that he accepts that intentions can be relevant to permissibility in a different way — for example, he thinks that intentions can play a “predictive” role (roughly, intentions can be treated just like other features of the agent’s situation that help to predict what the results of the available alternative actions would be).
As it happens, I am rather sceptical about whether the agent’s own current intentions really can play this role. Here I think that Tim’s refusal to distinguish between intentions and motives (or motivating reasons) becomes important. I would certainly agree that some of my current motives can play a “predictive” role — my emotions and desires can certainly play such a predictive role, for example. But intentions are very importantly different from emotions and desires, as it seems to me….
However, even though I am sceptical whether intentions can play this “predictive” role, the claim that they can play this role does not directly threaten the DDE; and so I don’t really need to argue against this point here.
Matthew —
1. Thomson thinks that intentions are only relevant to judgments on the moral character of agents, and not to the permissibility of actions. I disagree with her about this. The agent’s intentions might be completely “out of character”, as people say: this will not stop the agent’s intentions from being relevant to permissibility in the way that I have described.
I haven’t tried to explain why the DDE is true here; I have only aimed to respond to certain objections to it. I am confident that I have a good explanation of why it is true, but I haven’t tried to give that explanation here!
2. Thomson’s physician-assisted suicide cases are very confusing because it is not at all clear that it is a bad state of affairs for the patient who is suffering from a terminal illness to die. Many of us suspect that it is good for the patient to die in these circumstances. If that is true, then the DDE, as I understand it, simply does not apply. Thomson is confusing the DDE which applies generally to bad intentions (cases in which “one’s will is guided by evil”, as Thomas Nagel put it) with the specific Catholic doctrines about the sanctity of life that some philosophers like Anscombe also accept.
In general, however, what I say about those who act in a way in which it is permissible for them to act, but do so with a bad intention, is the following:
James, here are some quick replies to your questions:
1. An intention is a plan, or part of a plan. When you intend to do something, you’ve settled the question of what to do — you’ve set yourself to do the thing in question. Even if motives include intentions, they also include many other mental states as well: for example, your motives will very often include your desires and emotions as well as your intentions. In general, your motives for doing A include all the mental states that motivated you to do A — i.e. all the mental states that in that distinctive “motivational” way explain why you did A.
2. A bad intention, as I am using the term, is an intention to bring about a certain state of affairs S, where S is in some relevant way a bad state of affairs. Unfortunately, it is a difficult matter to explain exactly what sort of “badness” of states of affairs is relevant here. (Indeed, my view is that this is what the notoriously difficult “closeness” problem ultimately turns on.) Still, I hope that this is a somewhat helpful answer to your question.
3. Of course my opponent will agree that of course it’s bad to have bad intentions. However my opponent will go on to say that it doesn’t follow that it could ever be wrong for you to bring about a bad state of affairs by executing a bad intention to bring about that state of affairs, if it wouldn’t also be equally wrong to bring about this bad state of affairs irrespective of what intention you had.
4. An “act token” is just a particular act, as opposed to a general type of act. I don’t see why Nietzsche’s point about the messiness of actions should cast doubt on the idea that there are indeed act tokens!
Ralph,
That’s helpful as it is becoming clear to me how you want to proceed. I have two comments.
1. You said that you don’t like Thomson’s PAS cases. How about the following from Kamm, which has a similar structure?
Here it can’t be good for the innocent children to get killed. Yet Kamm thinks that it is permissible for the bombardier to bomb the munitions factory, even with the bad intention. If she is right, this case seems to be incompatible with your
2. Your view is starting to look similar to William Fitzpatrick’s. (I don’t yet think it’s the same). He says the following in his Analysis paper:
There are two worries about his view that you will want to avoid/address if you want to go that route.
First, as Kamm has also pointed, this view suggests that what determines permissibility are the characteristics of the act itself and its effect, as these are what make it possible for one to decide whether there exists a justification in terms of a sufficiently worthy end that can be pursued through so acting without intending anything illicit as a means. In other words, in deciding whether an act is permissible by considering whether someone without a bad intention could perform the act, one is in effect primarily considering the properties of the act, e.g., whether it leads to a greater good. If so, this view seems to reinforce the idea that one ought to use the properties of the act to determining whether an intention with regard to this act is permissible, and not vice versa. As such, this view does not seem to help to establish that intentions can independently determine the permissibility of an act.
Secondly, consider the following case by Kamm:
In other words, Clever Terror-Intending Tactical Bombing is the same as the Terror-Intending Tactical Bombing Case, except that the bombardier is not committed to trying to kill innocent children independently if tactical bombing missions are not available.
Given that in this case, there is no action that the agent’s intention might direct him to perform that can be independently determined to be impermissible, this case would seem to come out as a permissible case for Fitzpatrick.
However, as Kamm has argued, intuitively, it seems that this ‘clever variant’ should be just like Terror-Intending Tactical Bombing. In particular, why should the fact that the bombardier is clever affect the permissibility of his acts, even though he has the same intention to kill the innocent children as in Terror-Intending Tactical Bombing? It seems that either both of these cases should be permissible or they should not be permissible. If this is right, the fact that Fitzpatrick’s view appear to give asymmetric explanations regarding the clever variant and Terror-Intending Tactical Bombing seems therefore to count against such a view.
This is a really interesting thread. I’m especially interested in this recurring question of ‘what is an intention?’
I wonder if some of the disagreement can be helped by focusing on the narrower question of ‘when is an end intended rather than merely foreseen?’ I think Scanlon’s answer is that an end is intended if the agent takes it to be a reason for acting. This gives us a much more specific definition than ‘planning’ or something of the like. Perhaps we can formalize it:
Agent A intends to bring about consequence x by phi-ing just in case A takes x to be a reason for phi-ing.
Does this help clarify anything? Maybe not. But it makes Scanlon’s position look very attractive to me, as now we can easily distinguish between the permissibility and meaning of an action by distinguishing between the reasons there are for acting and the considerations one takes to be reasons for acting. Do Scanlon’s opponents just say that there are reasons against acting for particular reasons? This sounds a bit weird to me, but I suppose that’s not a real cogent objection!
Ralph,
What does “an act of this type” mean? (I am thinking particularly of your 5:15 AM today comment.) Every act token belongs to many types, surely (and to many act types).
Travis,
Is that really what Scanlon says? I thought the interesting question for DDE was whether a means (which is intended but not taken to be a reason for acting) is significantly different from a side-effect (which is not intended). And I thought the distinction was supposed to be something like this: an intended means is incorporated into your plan, and is causally upstream from the end at which the act is aimed; a foreseen side-effect is also incorporated into your plan, but is not a cause of the intended end.
Have I got the wrong distinction here? I thought this was what the DDE was all about.
Travis,
I think there is something to your idea even if I don’t think it could stand as it is. Your formalisation would entail that I would intend to bring about the consequence that I am at home by walking there just in case I take the fact that I will be at home to be a reason for walking there. This sounds odd – the reasons I have for walking home do not seem to include the fact that I will be there.
But, in a way I think you have the right idea. I think this makes me want to formulate the DDE in terms of motives instead of in terms of what the agent intends. Thus, we should distinguish between the consequences that motivated the agent and the consequences that were merely foreseeable unmotivating consequences. A similar action would then be worse (or even impermissible) if some of the considerations that motivated the agent were harmful things. Here it would make sense to think of what motivated the agent in terms of what she judged to be a reason. And, thus, DDE would say that an action would be made worse by the agent taking its harmful consequences to be reasons for the action (they would be what motivated her). I actually quite like this way of putting the DDE and I think it fits much in Scanlon.
Hi Jamie,
One could construe DDE in causal terms (Kamm tries to do this in Chapter 5 of her Intricate Ethics), but not everyone does. Thomson, for example, argues against construing DDE this way. In her Self-Defence paper, she says
“One possibility is to construe the doctrine [DDE] as concerned, not with intendings, but with sheer causal order; I ignored this possibility in the text above, since I think it pretty obvious that the doctrine so construed has no future at all” (p. 295, n. 9).
Jamie —
Huh?! By an “act of this type” I mean an “act that belongs to this type” — what else might I mean?!
Of course every act belongs to infinitely many types. That was why I phrased the point in the way that I did. As I understand the logic of these matters, a coarsely individuated type T can be permissible (w.r.t. the situation of the agent at the relevant time) even if not every more finely individuated sub-type of T is permissible. I.e., it can be permissible to do an act of type T (e.g. to give your mother a drink) even if there are certain more specific ways of doing an act of type T that are not permissible (e.g. to give your mother a poisoned drink).
Travis —
I agree with you that Scanlon’s conception of an intention seems to be something like the one that you ascribe to him. However I also agree with Jamie that this is a very non-standard conception of intentions, and one that leads to a very unusual interpretation of the DDE.
Matthew —
1. I disagree with Kamm about this case. It is indeed permissible for the bomber to bomb the factory, but not with this bad intention. If he bombs the factory with the bad intention of killing the children, he has acted wrongly.
2. Yes, my view is very similar (but not quite identical) to William FitzPatrick’s.
I’m not sure that I understand your first problem for views of this kind. But it sounds as if you’re falling into a similar mistake as the other critics of the DDE — viz. to suppose that forming an intention about what ultimate end or goal to pursue is not part of the same sort of deliberative activity as choosing which act to perform. In choosing an act, one is typically also deciding on a whole plan of action, which includes both the act that one is planning immediately to perform and the end or goal that this act is designed to achieve.
In my view, we should interpret Kamm’s case of the Clever Terror-Intending Tactical Bomber as a case involving what I call “super-fragile intentions”. In these very unusual cases, the intention is subtly different from the more normal intentions. For various reasons, I actually think that the intention in these cases is marginally less bad; so I view them as intermediate cases — i.e. as a bit less bad than the prototypical Terror Bomber and a bit worse than the prototypical Tactical Bomber. So I disagree with Kamm’s claim that these cases should be classified with the Terror-Intending Tactical Bomber. But it would take a very long discussion to explain why I think this…..
Ralph,
You replied as follows to Travis’s suggestion that, for Scanlon, an agent’s intention is what he takes to be his reason for action:
Rather than conceiving of an agent’s intention as his reason for action, this, according to you, is how we should conceive it:
But in responding to Doug, you wrote:
It strikes me, however, that, Al’s plan, what he’s set himself to do, is to divert the trolley. Moreover, what you identify as the intention from which Al would be acting is, in fact, his reason for diverting the trolley – where this reason is, e.g., ‘in order to kill Bush’ or ‘in order to save the people on the main track’.
So I think you slip into speaking of intentions as reasons for action.
Moreover, you’re right to describe the intention from which Al would be acting as his reason for action.
It’s also, of course, true that intentions are plans.
If Al were to say ‘My intention is to divert the trolley forthwith, and I shall do so with the intention of saving the five rather than killing Bush’, the first instance of ‘intention’ would refer to his plans and the second to his reason for action.
So I don’t think you should insist that intentions are plans, not reasons for action. Depending on the context, they might be either.
Scanlon on intentions as reasons and as plans
Near the beginning of Chapter 1 of Moral Dimensions, Scanlon conceives of a person’s intentions as both her plans and her reasons for action when he writes:
It turns out, however, that it makes a big difference, to Scanlon’s argument in Chapter 2 against the relevance of intention to permissibility, whether an agent’s intention is understood as his reason for action or his plan.
At the beginning of Chapter 2 of Moral Dimensions, Scanlon writes:
Throughout much of the rest of the chapter, Scanlon argues against the relevance of intentions to permissibility when intentions are understood as reasons for action. He conveniently summarizes this argument on the last page of the chapter:
As I have noted above, Scanlon also conceives of intentions as plans. And when he conceives them as such, his above objection to the relevance of intentions to permissibility does not apply. For he maintains that sometimes one ought to abandon ones intentions, where these are understood as one’s plans. He says this of a man who buys rat poison in order to kill his wife:
Since this passage appears only four pages after Scanlon’s identification of ‘intent’ with ‘an agent’s reasons for acting’, a reader might be forgiven if he initially interprets Scanlon to be saying that the rat man should abandon his reason for action. So interpreted, this passage contradicts the p. 88 claims that I have quoted above.
I puzzled over this apparent contradiction for a bit, but it then dawned on me that Scanlon must be speaking of the rat man’s plan when he speaks of his intention in this passage. This reading is confirmed a few pages on, when Scanlon says the following of the rat man: ‘He could instead abandon his murderous plan, and as I have said [i.e., in the passage above], this is something he should do.’ [p. 44] At this point, everything fell into place.
Ralph,
Is there something wrong with this idea: We make a decision to do what we think is probably going to have bad consequences, but it actually has good consequences. The action itself was good, but the decision was bad. Wouldn’t your opponent then agree that it is bad to execute a bad state of affairs in the sense that making the decision to do so was itself a bad action? They might want to just say it’s a different action.
I guess Nietzsche rejected act types rather than tokens. Not sure how I mixed the two up (right after reading what you had to say about how we mix them up). You require that we accept act types. Can we be sure that there is no messiness of act types?
Mike–
Thanks very much for your patient exposition and clarification of my view. You are right about what I had in mind.
Just to add a point or two. In the passage from p 88 I am appealing to the fact that we can’t choose which ultimate reasons to act from in order to explain why we should not say, of a person who sees only bad reasons for saving someone, that the only permissible course of action is for him to save the person for the right reason. This is not a general argument against the possibility of actions being impermissible because of the reasons for which they are done. In particular, because I think that some actions are impermissible for this reason (e.g. in some expression and expectation cases.)
In the later passage you cite I make clear that we can choose plans and thereby choose to act for particular (derivative) reasons. I think that the fact that an action is part of a larger plan that is impermissible can explain (e.g. via the wrongness of facilitating wrongs) the impermissibility of that smaller action. But I don’t believe that DDE reasoning fits this pattern. It does not start from a plan that is held to be impermissible on grounds of its foreseeable effects, and count an action as wrong because it facilitates it.
Tim
Mike —
I never said that no intentions were reasons: indeed, when you intend an action as a means to an end, it seems that your intention to achieve the end is at the same time both a plan and one of your motivating reasons for the action (the “in order to” construction is particularly appropriate in these cases). What I said was just that not all of our motivating reasons are intentions.
What this shows, I think, is that Scanlon’s argument faces a dilemma. Either (i) he identifies the sort of intentions that he is talking about with plans, or (ii) he identifies them with some kind of motivating reasons other than plans.
If (i) he takes the first horn of the dilemma, then it just isn’t plausible to say that, in the cases that matter for the DDE, we can never choose the reasons that motivate us. After all, it seems that in deciding on a plan of action, we choose which ends to pursue, and by making such a choice we form an intention to pursue these ends — an intention that is in a sense (as we have seen) among our motivating reasons.
Or alternatively (ii) he could take the second horn of the dilemma, and identify the intentions that he is talking about with motivating reasons other than plans. But then it becomes wildly implausible to claim that these are the sorts of intentions that are being discussed in the traditional literature on the DDE.
Ralph,
In your latest post, you write:
You also reject Travis’s proposed analysis of intention. Here’s his analysis:
You reject this as ‘a very non-standard conception of intentions’. Now what an agent takes to be his reasons for action doesn’t necessarily exhaust his motivating reasons. (An agent might be motivated by jealousy to disparage the work of a colleague without taking jealousy to be a reason to do so.) So I think you want to say that not even everything an agent takes to be his reason for action constitutes an intention of that agent. Perhaps it would be helpful to explain why you don’t regard even every member of this narrower class of motivating reasons as an intention.
“What was his intention in doing A?” can ask for, and be answered by giving, his reason for doing A. This may also be explained by saying how doing A fits into a larger plan. What that plan is will affect his (proximate) reason. (I say this on p. 42) But this discussion of intention does not seem to be to be central to what is at issue regarding DDE.
I reject DDE because (1) I see no plausible theoretical defense for it; (2) the cases that seem at first to support it (bombing cases, transplant cases, man on the bridge cases and so on) are better explained in other ways; and (3) one can explain why it would have seemed plausible at first that DDE offered the best explanation of these cases, even if it does not do so. None of this requires coming down firmly on one side or the other of a supposed controversy about intention.
Tim
I omitted to say: DDE claims that it matters to the permissibility of an action whether it is part of the agent’s plan that a certain result will be produced or whether there is just good reason to think that this will be so. If the former, the fact that the act will bring about this effect will also be the agent’s reason for doing it. If the latter not. So in order to decide what DDE permits we do not need to choose between these.
Second addendum: I meant “in order to decide what DDE permits in the cases I am relying on.” Perhaps there are other cases in which DDE would have implausible consequences if “intent” is taken in one of these ways, which can be avoided by taking it in the other. But in the cases I rely on it does not seem to me to matter.
Enough!
PS to my request to Ralph for clarification in my last post above:
There’s Frances Kamm’s distinction between doing something ‘in order to’ bring about x and doing something ‘merely because’ x will occur, where she maintains that the former is an intention and the latter is something that one takes as a reason for action but not an intention. (Scanlon takes note of, while neither endorsing nor rejecting, Kamm’s distinction in n. 4 of Ch. 1.)
I wonder whether it was its failure to attend to Kamm’s distinction that you had in mind when you described Travis’s as ‘a very non-standard conception of intentions’? Or did you have something else (or more) in mind?
Tim, I found your example of strategic v. terror bombing helpful. I agree that we can evaluate the moral permissibility of actions independently of intention. This does not show us that intention is morally insignificant, but it does demonstrate that we can divorce the permissibility of actions from the question of intention. (This last is Kant’s project in the Rechstlehre). I think that intentions are morally significant, not for evaluation of the action, but for the evaluation of the agent.
A quick addendum: I was thinking about the relationship between private and criminal law, since the latter includes the idea of mens rea but the former does not. It seems to me that criminal law is parasitic on private law: private law defines the forms wronging others can take, and criminal law discourages intentionally wronging others.
Again, I’m not sure I would want to say the action is worse when I wrong someone intentionally, but I would surely say that I was worse for doing so than if I had wronged another accidentally.
Mike —
The statement that an agent x “takes y as his reason for F-ing” seems to me to be alarmingly ambiguous. What does ‘taking’ mean here? (i) It could mean that x “responds to y as one of his reasons for F-ing”, i.e. (presumably) “is motivated by y (as by a motivating reason)”. Or (ii) it could mean that x “believes y to be one of his reasons for F-ing”. Or finally the statement might mean (iii) that x believes that y is a normative reason for F-ing.
In short, I regard talk of what an agent “takes as his reason” to be hopelessly unclear and ambiguous, sliding around between at least three different uses, where it refers to (i) the agent’s motivating reasons, (ii) what the agent believes to be his motivating reasons, and (iii) what the agent believes to be his normative or justifying reasons.
Given that I had to resolve this ambiguity somehow, I decided to resolve it in such a way that Tim and Travis are claiming that the intentions that one is acting on can simply be identified one’s motivating reasons for action. I agree that other resolutions of this ambiguity would be possible. But neither of these other resolutions would lead to a more plausible theory of intention. Each one would lead inexorably to a dead end of some kind. The second interpretation (ii) cannot make sense of intentions in non-self-conscious agents who have no beliefs about their own motivating reasons at all. The third interpretation (iii) cannot make sense of akratic intentions, to pursue ends that one believes oneself to have no normative or justifying reason to pursue. In short, every reasonable interpretation of this theory of intention seems to me quite hopeless.
At all events, in criticizing the conception of intention that is advanced here by Tim and Travis, I was certainly not endorsing Kamm’s distinction between what one does “only because” some effect will result, and what one does “in order that” that effect will result. (In fact, I believe that there are some very serious problems with the way in which Kamm draws this distinction. But it would take us too far afield to explain this point here.)
Tim —
I concede that I haven’t yet presented my argument for this claim yet, but I believe that this dispute about how to interpret the term ‘intention’ takes us to the heart of the disagreement between you and me about the DDE. Once we are clear about what intentions are, then your arguments against the DDE become much easier to answer. But fundamentally, in this thread I have mainly been interested in criticizing Judith Thomson’s arguments against the DDE. I shall try to present my criticisms of your arguments in more detail elsewhere.
Thanks, Ralph, for your latest, which clears things up. Three further and related questions, if you don’t mind:
(1) Would you have any quarrel with the following?: A person’s intention is – among other things (such as a plan of his) that an intention of his can be – a reason for action referred to in a truthful answer that begins with ‘In order …’ to the question ‘Why did he phi?’
(2) Assuming that you wouldn’t have any quarrel, would you also agree that such an intention (i.e., the reason for action) is not necessarily (also) ‘a plan, or part of a plan’? Imagine, for example, that you and I are idly passing the time one afternoon while sitting on a bench in St James Park. You ask me: ‘Why did you suddenly turn your head?’ And I truthfully answer: ‘In order to avoid the sight of that police horse relieving itself’. Here my intention in turning my head is the reason for action that I supplied in my answer to your question. My intention here is not a plan. It’s not something I set myself to do. Perhaps you’d agree that it’s not a plan but insist that it’s nevertheless part of my larger plan to spend a pleasant afternoon with you. I think, however, that it would be a stretch to describe that as part of that larger plan, and, if you stretch that far, what’s to stop mere motivating reasons that aren’t intentions from being part of that larger plan as well?
(3) Why not just say (as I think Tim says above, and as I believe Anscombe said, though my command of Anscombe is hazy) that there are (at least the following) two distinct (but perhaps related) things that a intention can be?: (i) a plan (as in ‘My intention is to finish this paper this afternoon’), and (i) a reason for action along the lines described in (1) above.