At a recent symposium on Victor Tadros's book The Ends of Harm, Victor and I were debating whether the Means Principle (MP) is best thought of on a subjective interpretation (for A to use B as a means, A has to intend that B play some role in bringing about a good) or an objective, causal-role-based interpretation (for A to use B as a means, B has to serve as a causal means in bringing about some good that might be offered to justify A doing what she does). Victor argues for the subjective interpretation; I argue not exactly for the objective, but for the relevance of causal roles in a principle that has more or less the same range of application as the means principle.
This led to our discussing whether the critics of the subjective view–principally JJ Thomson, Frances Kamm, and Tim Scanlon—have ever offered any good reasons for their views (no, says Victor) or whether they have, or at least whether their arguments provide a good starting point for building an argument that the subjective interpretation faces an uphill battle (yes, say I). Victor then suggested that I post this as a topic for debate on Pea Soup. So here we are.
I want to begin, then, by rehearsing the classic arguments. I then address two cases that Victor offers to argue that we need to appeal to intentions to understand our judgments about permissibility (this is taken from my reply to Victor, which will be published in a forthcoming issue of Law and Philosophy). Finally, I offer a liberal, anti-perfectionist argument—which I draw from a paper I published in Philosophy and Public Affairs, “The Doctrine of Illicit Intentions”—that we should not think of intentions as fundamentally relevant to permissibility.
I. The Classic Arguments
As I read them, the classic arguments have two threads. The first makes room for the idea that we can do without intentions, at least on a basic deontic level. The second presents a reason to accept that we should do without them on that level.
A. Thread one: Making room
I like the way Tim Scanlon framed the first thread. He argued that intentions have only derivative or secondary significance. In the context of discussing and rejecting the Doctrine of Double Effect (DDE), and in particular in the context of discussing its use in the law of war and the prohibition on terror bombing, he says that dropping bombs where there is no munitions plant or other legitimate military target, but where noncombatants would die, is impermissible “because the circumstances do not provide a justification” for killing people in that way. (Moral Dimensions, p. 29.) In other words, there is some appropriate principle of justification that needs to be appealed to, and it would refer to harms causes in a particular context; it would not refer to the intentions of the agent causing the harm.
What is important is how Scanlon articulates what he means by saying that intentions are not fundamentally relevant, not whether he has actually found a principle that can do without intentions. He puts the point this way:
[A] person who intends to kill noncombatants in order to shorten the war … acts wrongly—she has intentions that she should abandon. But this truth should not be taken to suggest that intention has a fundamental role in determining the impermissibility of this action, in the way claimed by the [DDE]. The intention is wrongful because the act intended is wrongful, and the act is wrongful because of its likely consequences, not (fundamentally) because of the intention. (id)
Again, this presupposes that some principles can be articulated that would plausibly explain why the acts that agents might intend would or would not be wrongful independently of the intention with which they are performed. Such principles would have to do more than pay attention to likely consequences. They would have to capture much of what the DDE and other intention-focused principles capture. But if we assume for the sake of argument that such principles can be found—part of what I presented at the symposium was an abbreviated version of what I call the Restricting Claims Principle, which claims to be an objective, causation-focused alternative to a subjective interpretation of the MP—then the point is that we can understand the misleading appeal of intentions by seeing it as based on a confusion. Their appeal is based on confusing a sort of secondary significance that they do have—that it is wrong to intend to do that which is wrong—with the thought that they do primary work making acts wrong.
In support of this, Scanlon argues that we can understand a number of ways in which intentions are relevant to the permissibility of actions that are clearly secondary to other primary wrong-making features of a situation. In that vein, he describes what he calls expression and expectation cases. These are cases in which one presents oneself as motivated in a way that others will find acceptable, when one is really motivated in another way. What is really doing the moral work in such cases is the right of others, in certain limited circumstances, to demand that those with whom they interact have certain motives—for example, if you want to socialize with me, it must be because you like my company, not because you want me to give you money—and the right they have not to be deceived. These rights do not specially concern intentions. They could just as well concern clothing: if you want to socialize with me, you have to dress informally, as I’m an informal person and I choose not to socialize with people who dress formally.
In other cases, the “predictive significance” of an intention—that it makes it too likely that an agent will do something impermissible—can explain why it is impermissible to act on some intentions. My favorite example of this—not, I believe, one of Scanlon’s examples—is the intention of a doctor to help her patient die, the intention in play in physician assisted suicide. It is at least plausible that such an intention can be the basis for a legal ban on such actions, on the ground that if we allow physicians to act this way, too many of them will help people to die rather than offer them help which would allow them to regain the desire to live. In other words, from the state’s point of view, it might be predictable (a) that more patients will die, who should be helped to live, if doctors are allowed to act intending to aid their patients in dying; (b) that this loss will outweigh the gain of helping patients to die who should be supported in their decision to end their lives; and (c) that there is no way to effectively carve out a small range of cases where this balance would be reversed. I don’t actually believe this, but I grant that it is plausible.
A third kind of case—one that Scanlon does discuss—involves collective bad effects of acting on a certain kind of intention. The most obvious example is racism as a basis for action. It makes sense both for law and morality to say that it is impermissible to make decisions over a wide range of matters that are normally open to the public—going into stores, buying houses, applying for jobs—on the basis of race (leaving open the Q of whether affirmative action might sometimes be appropriate). But to be clear, the problem is not that discriminating on the basis of immutable features is always impermissible. If someone wanted to run a shop and let in only people over 6 feet tall, that would be seen simply as odd. The problem with racism is that it has been so pervasive that it needs to be prohibited quite thoroughly to provide people of races that have suffered from racism a fair chance at a good and dignified life.
None of this is not to say that all of what Scanlon says on this topic is plausible. What he says about attempts, in particular, is rather implausible: that it is wrong to do something attempting to commit a wrongful act because one is making it more likely that one will commit the act (the “predictive significance of intent.” (id., p. 43)). This seems a poor account of why attempts are impermissible. It would be better simply to say that it is impermissible to form and act on the intention to do something impermissible. Nonetheless, I think Scanlon was right to say that whenever intention seems to be relevant to the permissibility of acts, one can always explain its relevance in terms of some prior judgment of wrongness, or some independent right that a others have that a person not act on a particular intention in a particular situation.
It is important to be clear about the logical structure of the argument before moving on to the second thread in the classical argument. And it may help to establish just what the point is to say more about the contrasting view found in the DDE, which holds that there are two possible wrong-making features of actions. One is the balance of harms and goods; the other is the intention of the agent. The DDE holds that even if an act will cause more good than harm, it is impermissible if the agent intends the harm. And it holds this to be true not because an agent who intends the harm is disposing or committing herself to do impermissible things, but because it is a basic principle of morality that such intentions are impermissible. Theorists who support the DDE offer reasons to accept it: it disrespects persons or basic human goods or something like that to intend harm. Nonetheless, according to the DDE, there is no way to get a full account of the range of impermissible acts without appealing to the intentions of the agents who perform them, and thus intentions cannot be secondarily relevant to permissibility. They make acts impermissible, just as causing harm can make acts impermissible. The view Scanlon and I embrace, in contrast, holds that intentions may make acts impermissible, but only because they dispose or direct agents to perform acts that can independently be determined to be impermissible, or they have a tendency, when widely shared, to cause harms or injustices, or others have a right, in limited contexts—a point that will be explored in part III—that agents not act on certain intentions. In all of these ways, intentions are of secondary importance to wrongful action.
B. Thread two: Proper focus of concern
The second thread in the classic arguments is that the view that intentions are a fundamental wrong-making feature of actions directs an agent to think too often, in the wrong cases, about the intention with which she is acting. Agents should, according to this view, not focus inwardly on their intentions, but outwardly on how to act in ways consistent with the respect that others deserve. J.J. Thomson deserves credit for pushing this line (See her “Self-Defense,” Philosophy and Public Affairs 20 (1991): 283-310).
One way to bring the point into sharp relief is to think about an expression and expectation case, when it does make sense for an agent to focus inwardly on the intention with which she acts, and to contrast that with another case when it doesn’t. So let us start by thinking about an agent who wants to be social with me, and whose reason for doing so is that she hopes that I’ll give her money. Especially if I have made it plain how much I would resent such mercenary socialization, she should respect my right to restrict the circle of people with which I socialize in that way. She should therefore look into her heart, as it were, and take not of her reasons for actin. If she sees what I just said was there, then she should choose not to socialize with me. Alternatively, if she can find that she values my friendship after all, for the joy it brings her or for the affection she has for me, then she can permissibly seek my company.
Contrast that case with someone dropping bombs on strangers. Does it really make sense for him to be looking into his heart and saying that he may not drop the bombs if he would do it for bad reasons, such as hatred or the desire to use the lives lost to promote victory for his side? I say no. The cases are simply different. On the one hand, it is true that such a bomber should say to himself that there is something wrong with him if he values the deaths of civilians from an enemy nation, either intrinsically or instrumentally. He should try to work on his values, to bring them more in line with those that are morally sound. But, on the other hand, when it comes time to fly his mission, if he happens still to have those bad values, he need not say: “I must stand down unless I can either change my values or choose not to act on them.” He can permissibly say: “It doesn’t matter for the permissibility of what I do what my reasons are; what matters in this kind of context is that I am dropping bombs where it is permissible to drop them. No one can complain that I performed an act that was not justified, and no one has any rights over the intentions with which I act—this is not a consensual type of interaction.”
There is room to object that I am simply wrong here about the ill-motivated bomber. I will return to consider this in the third part of this post. For now, I want simply to conclude that this is, as I see it, the core of the classic set of arguments that intentions are not fundamentally relevant to permissibility.
II. Victor’s Cases for Intentions
Victor now comes back with a set of examples that he thinks show that intentions are, at least sometimes, fundamentally significant for judgments of permissibility. I will start by describing two cases that he develops, then give a reply in terms of my Doctrine of Illicit Intentions, then consider an objection from Victor, and offer a reply.
A. Victor’s cases
The first example involves a duress case. Victor imagines a gang of criminals who have each pledged to kill the family members of anyone in the group who tries to back out of group activities. He then imagines that the group has decided to rob a post office. We are to suppose that all but one member of the gang think that this is a good idea, but one member has decided that she does not want to engage in criminal activities. Still, given the threat to her relatives, she engages in the robbery. She should be able to make use of the defense of duress; the others, despite facing the same threat, should not. In other words, she acts permissibly (with justification, not just excuse—I’ll assume, for the sake of argument, that armed robbery can ever be justified by the necessity defense) because she acts for the reason that she is trying to save her family; they act impermissibly because they don’t act for that reason, despite the fact that they have the same sort of reason to act. What determines the permissibility of the act—what distinguishes the two sets—is the reason that is motivationally active in them.
Victor’s second case involves two people, A and B, who independently put poison into the water pipe leading to a victim, V’s, house—the Poisoned Pipe case. They each put in sufficient poison to kill V, and they are each aware of the other’s actions. Moreover, “A’s poison alone or B’s poison alone would lead to a very slow and painful death for V. Their poison together kills V swiftly.” (The Ends of Harm, p. 159) If they were both acting trying to kill V then they would both be acting wrongly. But we can imagine that B acts with benevolent motives. She can’t stop A or in any other way save V, but she can add her poison to the water to provide V a quicker, more painless death. It is at least plausible that the beneficent B is acting permissibly. (Again, I am willing to accept, for the sake of argument, that this would be permissible. But there are good reasons to doubt it. It would not be permissible to poison someone, without her consent, who is dying a slow, painful death of cancer. It is hard to see why this situation should be different.) If so, then in this case, as in the duress case, what is permissible depends on the intention of the agent.
B. Reply with the DII
The Doctrine of Illicit Intentions (DII) provides an alternative explanation that allows us not to have to say what Victor says about the significance of these cases. This is not the place to spell out the DII in detail. What I can do is spell out the basic idea and show how it applies to these cases.
A core distinction the DII makes is between an illicit reason and an illicit intention. An illicit reason is a reason for action that flouts the limits of moral permissibility. The bomber who drops bombs on an acceptable target area because he wants to kill the noncombatants who live there acts on an illicit reason. An illicit intention, in contrast, disposes an agent, in ways that depend on the circumstances she finds herself in, to perform impermissible actions. This same bomber would not act on an illicit intention as long as he is committed to dropping bombs only where they can justifiably be dropped. He would have an illicit intention if and only if his intention would direct him to drop bombs where they may not permissibly be dropped, at least if circumstances unfold a particular, foreseeable way.
It is important to note that the DII presupposes that intentions often are not simple linear affairs with the shape: Do act A to achieve goal G. They are often complex affairs with a shape more like this: To achieve goal G, do A1 in condition C1; do A2 in C2; do A3 in C3, etc.; and if you have done A1, then do B1 in C1a, do B2 in C1b, etc.; and look out for conditions K1-Kn, in which case do nothing to pursue G. I call the range of actions that an intention might direct its holder to perform an intention’s scope. The thought behind the DII is that an agent normally has no good reason to form an intention with a scope that includes impermissible actions. Because it is easy for an agent to restrict the scope of her intentions so that they exclude the performance of impermissible actions, at least those that she can easily see she might be directed to perform, and because she can have no good reason to include impermissible acts within the scope of her intentions, it is reasonable for morality to require her not to form intentions with impermissible acts in their scope.
It is also important to note that the DII does not direct agents to focus inwardly on their intentions; it directs agents to focus only on what they will do. It involves their intentions only insofar as it directs them to commit themselves to framing intentions that exclude impermissible acts from their scope. It does not matter how they do that; it matters only that they do that. Yet the DII still concerns intentions because it gives us a ground for condemning someone who performs an otherwise permissible act on an illicit intention. It says not that the final act itself was impermissible, but that acting on an illicit intention, one that could well have directed her to perform an impermissible act, was itself an impermissible choice.
Now we can apply this to Victor’s two cases. Start with the case of the gang members, one of whom is robbing a post office under duress. The one who is acting under duress is acting on an intention whose scope seems intuitively to contain no impermissible acts. We can assume that the only reason she is robbing the post office is to protect her family, a condition under which that action is, we can assume, permissible. If freed from such coercive threats, she would presumably not engage in post office robbery or any other impermissible actions. By contrast, her fellow gang members are ready and willing to engage in post office robbery. Even if the mutually enforcing threat were to drop away, they would still rob the post office because they think the money they get is justification enough for them. Thus they act on illicit intentions; she does not.
The same thing is true of beneficent B in Poisoned Pipe. Assuming that her act is permissible, it is permissible only in this odd circumstance in which by adding poison to the pipe, she helps V suffer a less bad death. Beneficent B would add poison to the pipe only in such a highly unusual circumstance, so her intention is licit. By contrast, her maleficent counterpart would perform the act of poisoning whether it would help V or not. She then clearly acts on an illicit intention, and that is why she and not her beneficent counterpart act impermissibly.
C. Victor’s objections and my replies
In the symposium on his book, Victor objected to the DII on two grounds. First, he said that I can’t distinguish someone who would order coffee willing to shoot the barista if she doesn’t sell him the coffee, but ready to pay normally if she does, from the bad actor in Poisoned Pipe who kills for the money: the acts are objectively justified in both, and both are ready to perform acts that aren’t. Second, he says that I’m just getting the wrong in Poisoned Pipe wrong: it’s killing, not acting on an illicit intention.
Here I must pause to say how grateful I am to Victor, as I hadn’t thought about the DII paper in some years, and I didn’t deal with this problem or the problem of comparative culpability in it. Nonetheless, I now have the following view. Taking the second problem first, let’s assume the poisoner is not aware of the justification for the act—assuming again that the justification is successful. He is adding the poison to kill and get the money. I say that if he kills and was unaware of the justification for doing so, he is guilty of attempted murder. Why attempted murder and not murder? My answer is that no death was caused that could not justifiably be caused. Thus murder does not fit. But from the agent’s point of view, he was trying to cause an unjustified death. Thus he should be held culpable for attempted murder.
Now what if he is aware of the justification but it was just a fortuity of the case; he would have killed anyway? Then we’ve got the coffee case in view. But I think a big difference in culpability can be traced to the reason acted on. If the reason is a murderous one, and the person got lucky, finding that in this case it turns out that the act is permissible, then the act is culpable on the level of attempted manslaughter. Why manslaughter? Because the intention was reckless, and but for good luck, he would have killed impermissibly. (And such a reduction in culpability might not be available in cases like that of the gang, where people set it up that they have the reason to do the crime—much as one cannot escape mens rea requirements by getting drunk so that one doesn’t know what one is doing.)
If, however, the reason is itself innocent—to get coffee—then acting on the illicit intention is still morally impermissible, but the culpability is low enough that the law should be reluctant to criminalize it at all. I’d say that the law should criminalize acting on an illicit intention, grounded on a morally neutral or good reason, only if the person was essentially taking active steps to be ready to act impermissibly.
Victor can still object that this view is quite peculiar. I would have those who kill with nothing but bad intent in cases like Poisoned Pipe guilty only of attempted murder. This just shows, he can say, that I’m getting the cases wrong. Again, I treat the wrong as one of acting on an illicit intention, but the wrong is killing.
Here I must simply acknowledge that my view is in this way counter-intuitive. We are not used to thinking of such cases as involving anything less than murder. But he and I both agree that it would be permissible to encourage the person to add the poison in such a case (again, assuming the justification works). And in that regard my view is less counter-intuitive. For all I have to say is that I’m encouraging someone to attempt to perform an impermissible act, while actually performing a permissible and good act. He has to say that it is permissible to encourage someone to murder. This is not to say that my view is now intuitive. It is only to say that there are peculiarities however one slices the terrain up, which probably indicates that our intuitions don’t all make sense. Therefore I am not too worried if I have to admit that my view is somewhat counter-intuitive.
III. The Liberal Anti-Perfectionist Case Against Intuitions
Let us go back to the terror bomber who drops bombs where they may permissibly be dropped. And let us suppose that he was committed to constraining his actions to conform to those that would be performed by a tactical bomber who only hit targets where the civilian damage would not be disproportionately high. One might say: why not require him not to act on the illicit reasons that he accepts?
There are a number of bad answers worth rejecting before spelling out my answer. One such bad answer is to say that if he accepts illicit reasons for action, then he cannot know that they are illicit, and thus he cannot be acting impermissibly. But there is no gap between intentions and actions on this score. One might think it is permissible to hit noncombatants where there are no legitimate targets in the area. That mistake does not make one’s action permissible. So why should a mistake about what reasons are morally licit?
One might also say that if he thinks that he has sufficient reason to target noncombatants, then he cannot choose not to do so, at least not while still performing the bombing run. Scanlon makes this argument: that an agent cannot have a reason for an action and choose not to act on it without also choosing not to perform the act that the reason would direct her to perform (Moral Dimensions, pp. 56-61). But Victor and I both reject this claim. We both think that an agent can take herself to have a reason R1 to do X, take herself to have another reason R2 to do X, and choose whether or not to do X acting on R1. That is, we both think that, at least in many cases, she can frame an intention to do X by appeal to R1 or not, as she chooses.
But now if the bomber could choose to fly the mission—which, we should suppose, there is good reason for him to fly—and if he could choose to do so without acting on his desire to kill enemy noncombatants or his belief that doing so would be a good thing, then why not say that he must fly the mission for better reasons, such as that it will take out legitimate military targets? Why not say that he acts as a murderer if he flies it for bad reasons?
Indeed, one might think that as a supporter of the DII, I should be quite open to this, as I think that one can be guilty of something close to murder, attempted murder, if one acts on illicit intentions that would lead, under some circumstances, to actual murder.
But I reject that view because I think there is a meaningful difference between illicit reasons and illicit intentions. The latter concern disposing oneself to perform acts that are impermissible. That seems to me to inherit the wrongfulness of simply choosing to do something impermissible. It is a close cousin of that other choice, and the presence of permissible acts in the scope of the intention does not seem enough to neutralize the wrongness of having impermissible ones there—especially given that it is not hard to frame an intention that excludes foreseen or easily foreseeable impermissible acts from its scope.
Illicit reasons, on the other hand, are not about acts, they are about how one thinks, how one reasons. I don’t deny that morality has something to say about this process. But I do deny that it can say it in the voice of a judgment of impermissibility. I think our basic freedom of conscience and of thought precludes that sort of deontic judgment.
What I am saying here is that there is an important sense in which we have a right to do wrong. And contrary to the dominant view of that right, the right is not just about what others can do to us. It is not primarily a right to be free from outside interference. It is fundamentally a right to choose to be immoral in certain ways. And this right is grounded in our right to lead our own lives, and to decide how and when to work on perfecting ourselves. There are limits to our permissible self-indulgence. We may not indulge vice to such an extent that we lose our ability to respect the rights of others and otherwise do what we are morally required to do. But within those limits, we are basically permitted to be vicious. The obligation not to be is an imperfect one, an obligation to take as one of our ends our own perfection. But that imperfect obligation does not give rise to perfect duties—to duties that it is impermissible to violate—unless, again, one needs to act to enable oneself to conform to the perfect duties that otherwise exist.
Now it can be objected, drawing on the inspiration of Warren Quinn, that people do have rights to the intentions of others. On Quinn’s view of the DDE, the terror bomber who kills by dropping bombs where a tactical bomber might permissibly drop them wrongs his victims because they have not only a claim not to be killed, but an especially strong claim not to be killed by a person acting on a vicious or illicit reason.
My response is that this objection seems deeply illiberal. Generally speaking, people have rights over the intentions of others only when the interaction should be consensual and they can withhold their consent to have the interaction on a variety of grounds, including that the other is acting on an intention they object to. But in cases that involve inherently nonconsensual interactions, like dropping bombs that kill people near military facilities, or turning trolleys away from some and onto others, it’s hard to see why those who are affected have any extra claim over the intentions of the actors. I think society has a claim that the actor at least know that the act is permissible; otherwise the actor is essentially recklessly or intentionally killing, and that should be a crime. But if the actor knows the act, abstracted from the intention with which it is performed, is permissible, and there is no other problem with her intention (like racism or having other impermissible acts in its scope), then it is hard to see why the victim gets to dictate what reasons the actor may act on.
The victim may claim that it adds insult to injury to be acted upon by one who thinks so hatefully or instrumentally about him. But we just aren’t entitled to demand that others not have insulting thoughts about us. Take a variant on a racism case, where no collective action problems will arise: suppose someone doesn’t like me because I remind her of a past lover who she now hates because of the shabby way he treated her. She chooses not to sell her house to me because of her irrational dislike for me based on my physical resemblance to someone else. If I want the house, I could reasonably be insulted. But I have no right to demand that she sell it to me. She is permitted by law to refuse to sell on that idiosyncratic reason. And in truth I think she is morally permitted to refuse to sell to me for that morally idiosyncratic reason. She shouldn’t take out on me her feelings for someone else. But if she has those feelings, she is morally entitled to indulge them in this way. It is her house, her feelings, and her choice to make.
At bottom, I cannot see why the fact that one is a victim of a harm worse than not being able to buy the house one wants—being killed—gives one more right to control the reasons for action of another. There are reasons to be concerned with intentions, to prohibit people from acting on them. But from what I can see, the right not to be thought of badly as another does what she otherwise has a right to do does not give rise to a new basis for saying that intentions matter quite generally to the permissibility of actions.
CODA: some are also tempted by a sort of Aristotelian position that acts are to be distinguished from reflexes by the fact that they are performed with an intention, and often it is the intention that defines the act as the type of act that it is. For example, burglary is distinguished by the specific intent with which one enters a building: the intention to commit a felony therein. Banning burglary is banning an action that can only be identified by the intention of the actor.
To this I say yes, but we can still describe physical act types, like entering a building, and we can still describe relevant circumstances, like doing it without permission, and we can still identify a range of conditions that would justify or fail to justify such an act, all without knowing the intention with which the person entered the building. We can then add to the considerations the fact that the person performed an act with a particular intention to see if it makes a difference. Sometimes it might, sometimes it might be relevant only to level of culpability if the act is already on the impermissible side of the line. None of that undermines in any way what was said above.