Reconsidering Double Effect
by Michael Dorf
The latest issue of the Harvard Law Review includes an article by Harvard Law Professor Richard Fallon, in which he argues: (1) that Supreme Court constitutional case law with respect to the question of how to evaluate forbidden legislative intent is confused and contradictory; and (2) that in its place the Court should adopt a rule under which forbidden intent by itself is never sufficient to invalidate a statute but allowing that such forbidden intent can, under certain circumstances, be the trigger for the application of heightened scrutiny. The Harvard Law Review Forum--the online companion to the printed volume--includes a short paper by me in which I praise Fallon's descriptive account of the case law and raise a question about his normative proposal.
Before describing my critique, I should say that I have enormous regard for Fallon and his work, which has been instrumental to my own. Fallon and I are co-editors (along with others) of a Constitutional Law casebook. In my recent review of Judge Posner's book Divergent Paths in the Journal of Legal Education I was especially critical of Posner's critique of Fallon's work. (Posner's response is here. My rebuttal is here.) Fallon's work on constitutional theory played an important role in my own thinking about the subject, as reflected in this 1997 article in the Georgetown Law Journal and in this 1999 California Law Review paper that I wrote in response to Fallon's very helpful article (which I have frequently assigned to seminar students) in the same volume. Accordingly, no one should conclude that the question I raise about Fallon's latest Harvard Law Review article reflects anything but respect.
The question I raise about Fallon's legislative intent argument is this: If Fallon's key argument with respect to legislative intent were generally accepted, would that be potentially disruptive of many settled and sensible doctrines in and beyond constitutional law? Fallon relies on a book by philosopher T.M. Scanlon, in which Scanlon argues against the doctrine of double effect (DDE) and more generally against intent as a basis for evaluating the morality of conduct. Under DDE, an actor may sometimes be justified in knowingly bringing about an undesirable end so long as the actor aims at a permissible one. Here's how I summarize Scanlon's argument in my paper responding to Fallon. I say that, according to Scanlon, DDE
Here I want to put on my amateur philosopher hat and ask whether Scanlon's argument is sound. One way to do so is by asking about a paradigm case. I'll use the law of war rather than assisted suicide, because I think that the latter is controversial--many people think that there should be a right to assisted suicide even apart from double effect while others think that it is morally acceptable for a doctor to prescribe a lethal dose of narcotics to a patient only if the narcotic dose is necessary to treat pain, with death as a side effect, i.e., they rely on double effect. By contrast, few people argue that because it's morally permissible to kill civilians as proportionate "collateral damage" from an otherwise justified attack on a military target, it is therefore permissible to aim at killing civilians if doing so will directly bring about comparable military advantage (say, by demoralizing the enemy, as was standard military practice for millennia).
Sam Harris is something of an exception. In The End of Faith, he argues that torture is not only morally permissible but sometimes morally necessary, reasoning his way there from the acceptability of collateral civilian casualties. However, I don't regard Harris as expressing a view that is widely shared among people who have thought seriously about wartime killings. True, some prominent politicians, such as Ted Cruz and our president-elect, have been cavalier in expressing their willingness to harm civilians, but they were not stating thought-out moral or philosophical views in saying these things.
So, is it morally permissible to deliberately target civilians for killing in wartime if doing so would secure the same military advantage as attacking military targets with the civilian casualties occurring collaterally? No doubt, for some readers, my merely stating this test case suffices to dispatch Scanlon's argument. They will think that targeting civilians cannot be morally permissible and so Scanlon's argument necessarily fails. Indeed, I think I fall into this category. Nonetheless, let's suspend the impulse to reject Scanlon's argument out of hand in order to see where, if anywhere, it goes wrong.
To make things concrete, I'll pose a hypothetical case involving two generals. General A decides to bomb a munitions factory because doing so will undercut the enemy's ability to fight, even though he has reason to think that after taking efforts to reduce civilian casualties, there will still be some. The military lawyers advise him that, given the military advantage to be gained, the expected civilian casualties are permissible. General A orders the bombing, even as he regards the expected civilian casualties as a source of moral regret. General B faces the same choice and receives the same legal advice. However, General B is secretly a sadist and he is secretly delighted by the fact that civilian casualties will result. Indeed, General B joined the military and rose through its ranks precisely so that he would be in the position of ordering actions that result in civilian deaths. His personal motivation is to cause the civilian casualties. He wouldn't order the bombing if it didn't have military value, but not because he cares about the war effort, much less sparing civilian lives. Rather, he abides by the rules prudentially, as Holmes's "bad man" does. General B knows that if he orders civilians to be bombed without a legal military rationale, he risks prosecution for war crimes. At the least, he will be relieved of his command and thus lose future opportunities to cause civilians to suffer and die. Whereas General A orders the bombing despite the likely civilian casualties, General B orders it because of the likely civilian casualties.
If the generals' respective motivations were sufficiently well known (by the discovery of their diaries, let's say), the existing law of war would acquit General A of any wrongdoing and hold General B accountable for a war crime. Scanlon thinks that's a mistake. In each case the action and the result is the same and so if General A's actions were morally permissible so were General B's.
To be sure, Scanlon does not think that intent is completely irrelevant. He thinks that intent is irrelevant to the moral permissibility of an action but relevant to judging the character of an actor. Under Scanlon's approach, there is thus a difference in the blameworthiness of my hypothetical generals. General B's motive to kill civilians makes him blameworthy, whereas General A's motive to secure a military advantage while foreseeably but regrettably killing civilians makes General A blameless.
What shall we make of that conclusion? Some people reject DDE because they think that even actions that are intended to cause a permissible result are impermissible if they foreseeably also cause harm to innocents. As Norman Cantor and George Thomas pointed out in an important 1996 article in the Kennedy Institute of Ethics Journal, the criminal law in general does not require specific intent, i.e., one is guilty of a crime if one intends an act the consequences of which are reasonably foreseeable, where those consequences are the actus reus of the crime. (The abstract of the article is available here, but the full article is behind a paywall. Cornell has a site license, so I've got access to the article. Weirdly, the abstract I just linked lists two additional authors, but the article itself only lists Cantor and Thomas as authors.) Put differently, the law generally rejects DDE as a basis for allowing actions that have particular unintended but reasonably foreseeable consequences.
Here I don't want to defend DDE against the sort of "leveling up" argument put forward by Cantor and Thomas (although I think DDE is defensible against that kind of argument). Instead, I'm exploring whether DDE is vulnerable to the sort of "leveling down" argument made by Scanlon, which says that if General A's decision is morally acceptable then so is General B's.
To my mind, Scanlon's argument is least persuasive in its caveat. Why would an actor be blameworthy for committing an act that is morally justified? One answer might be consequentialist. The sort of person who is motivated in the way that General B is motivated might be acting in a morally justified way in my scenario, but his evil motive makes him likely to commit bad acts in other contexts, and so we should be wary of him. However, in my hypothetical example, General B is a prudential rule-follower, so there isn't this worry.
Moreover, it's not clear to me that one can separate blameworthiness from conduct. In Beating Hearts, Prof. Colb and I say that to make sense of virtue ethics--the ethical system that says that the rightness or wrongness of conduct depends on the character (or virtue) that an actor exhibits when engaging in that conduct--we need to make it parasitic on some external conception of value, such as utilitarianism or deontology. Otherwise, it's hard (I would say impossible) to know whether the act manifests virtue. I think something similar is true of Scanlon's argument. Saying that an actor is blameworthy for engaging in a morally justified act smuggles into the concept of blameworthiness some external conception of what makes the act itself morally permissible or impermissible--unless Scanlon is prepared to rely wholly on the consequentialist assessment of how someone like General B will act in other contexts, but I have stipulated away such concerns here.
Accordingly, I am unpersuaded that Scanlon's exception is justified. It seems to me that if an act really is permissible regardless of the actor's intentions, then those intentions cannot make the actor blameworthy.
Of course, Scanlon could be wrong about his exception but right about the general point. Maybe General B's actions really are no different from General A's. One reason we might reach this conclusion is an aversion to thought-crimes and the like. On this account, the law oughtn't to judge people based simply on their mental states. But this can't be completely right, because one could respect that judgment and still reject Scanlon's argument as a standard for individual moral decision making, quite apart from law.
I think the best argument for Scanlon's view is the one I used simply to describe it: If someone is trying to decide a course of conduct, it's odd for that someone to interrogate his own motivation rather than asking whether the conduct is morally justified.
In the end, I still resist Scanlon's conclusion, but I'm not sure that my resistance is principled rather than ultimately prudential. My caveat about General B--that on prudential grounds he will only engage his sadistic appetites when doing so can be morally justified if the same conduct were undertaken by someone with good motives--is quite unrealistic. In reality, a sadist will engage his sadistic appetite when he can get away with doing so, not just when doing so doesn't violate the law (or the moral standard). Likewise, given everything we know about cognitive biases, a person who wants to engage in an act for a bad reason will likely conclude that the act is objectively justifiable and thus permissible in many circumstances when it really isn't.
In practice, using double effect as an external standard of conduct will occasionally permit someone who acts with bad motives to get away with doing so, because those motives will be difficult to detect. But if I'm right that the sort of person who would engage in harm-causing conduct for the wrong reasons will tend to do so even when the harm-causing conduct is not cost-justified, then eventually such a person will be found out. The prisons are filled with Holmesian bad men.
The latest issue of the Harvard Law Review includes an article by Harvard Law Professor Richard Fallon, in which he argues: (1) that Supreme Court constitutional case law with respect to the question of how to evaluate forbidden legislative intent is confused and contradictory; and (2) that in its place the Court should adopt a rule under which forbidden intent by itself is never sufficient to invalidate a statute but allowing that such forbidden intent can, under certain circumstances, be the trigger for the application of heightened scrutiny. The Harvard Law Review Forum--the online companion to the printed volume--includes a short paper by me in which I praise Fallon's descriptive account of the case law and raise a question about his normative proposal.
Before describing my critique, I should say that I have enormous regard for Fallon and his work, which has been instrumental to my own. Fallon and I are co-editors (along with others) of a Constitutional Law casebook. In my recent review of Judge Posner's book Divergent Paths in the Journal of Legal Education I was especially critical of Posner's critique of Fallon's work. (Posner's response is here. My rebuttal is here.) Fallon's work on constitutional theory played an important role in my own thinking about the subject, as reflected in this 1997 article in the Georgetown Law Journal and in this 1999 California Law Review paper that I wrote in response to Fallon's very helpful article (which I have frequently assigned to seminar students) in the same volume. Accordingly, no one should conclude that the question I raise about Fallon's latest Harvard Law Review article reflects anything but respect.
The question I raise about Fallon's legislative intent argument is this: If Fallon's key argument with respect to legislative intent were generally accepted, would that be potentially disruptive of many settled and sensible doctrines in and beyond constitutional law? Fallon relies on a book by philosopher T.M. Scanlon, in which Scanlon argues against the doctrine of double effect (DDE) and more generally against intent as a basis for evaluating the morality of conduct. Under DDE, an actor may sometimes be justified in knowingly bringing about an undesirable end so long as the actor aims at a permissible one. Here's how I summarize Scanlon's argument in my paper responding to Fallon. I say that, according to Scanlon, DDE
rests on a confusion between the considerations relevant to whether an act is morally permissible and those relevant to how a moral agent should decide what to do. Take, for example, a doctor trying to decide whether it is morally permissible to administer a lethal dose of sedative to a patient. [DDE] asks whether the doctor intends to kill the patient or merely administer a pain-relieving dose. Scanlon would argue that instead of focusing on her own intentions, the doctor should simply ask whether the benefit (pain relief) justifies the cost (death).In my response to Fallon, I bracket the question whether Scanlon is right or wrong as a matter of first-order moral philosophy because I recognize that I am not, by training, a philosopher. Instead, I note that DDE plays a role in many areas of law, including assisted suicide, abortion, anti-discrimination, and the humanitarian law of war with respect to targeting. Intent as an evaluative standard, which is more generally an object of criticism by Scanlon, is ubiquitous in law. I say that if acceptance of Scanlon's argument would destabilize whole bodies of law--as it might--then one ought to proceed with great caution before accepting the argument in any domain, unless one has a sound basis for distinguishing the large numbers of cases in which one is reluctant to accept the argument.
Here I want to put on my amateur philosopher hat and ask whether Scanlon's argument is sound. One way to do so is by asking about a paradigm case. I'll use the law of war rather than assisted suicide, because I think that the latter is controversial--many people think that there should be a right to assisted suicide even apart from double effect while others think that it is morally acceptable for a doctor to prescribe a lethal dose of narcotics to a patient only if the narcotic dose is necessary to treat pain, with death as a side effect, i.e., they rely on double effect. By contrast, few people argue that because it's morally permissible to kill civilians as proportionate "collateral damage" from an otherwise justified attack on a military target, it is therefore permissible to aim at killing civilians if doing so will directly bring about comparable military advantage (say, by demoralizing the enemy, as was standard military practice for millennia).
Sam Harris is something of an exception. In The End of Faith, he argues that torture is not only morally permissible but sometimes morally necessary, reasoning his way there from the acceptability of collateral civilian casualties. However, I don't regard Harris as expressing a view that is widely shared among people who have thought seriously about wartime killings. True, some prominent politicians, such as Ted Cruz and our president-elect, have been cavalier in expressing their willingness to harm civilians, but they were not stating thought-out moral or philosophical views in saying these things.
So, is it morally permissible to deliberately target civilians for killing in wartime if doing so would secure the same military advantage as attacking military targets with the civilian casualties occurring collaterally? No doubt, for some readers, my merely stating this test case suffices to dispatch Scanlon's argument. They will think that targeting civilians cannot be morally permissible and so Scanlon's argument necessarily fails. Indeed, I think I fall into this category. Nonetheless, let's suspend the impulse to reject Scanlon's argument out of hand in order to see where, if anywhere, it goes wrong.
To make things concrete, I'll pose a hypothetical case involving two generals. General A decides to bomb a munitions factory because doing so will undercut the enemy's ability to fight, even though he has reason to think that after taking efforts to reduce civilian casualties, there will still be some. The military lawyers advise him that, given the military advantage to be gained, the expected civilian casualties are permissible. General A orders the bombing, even as he regards the expected civilian casualties as a source of moral regret. General B faces the same choice and receives the same legal advice. However, General B is secretly a sadist and he is secretly delighted by the fact that civilian casualties will result. Indeed, General B joined the military and rose through its ranks precisely so that he would be in the position of ordering actions that result in civilian deaths. His personal motivation is to cause the civilian casualties. He wouldn't order the bombing if it didn't have military value, but not because he cares about the war effort, much less sparing civilian lives. Rather, he abides by the rules prudentially, as Holmes's "bad man" does. General B knows that if he orders civilians to be bombed without a legal military rationale, he risks prosecution for war crimes. At the least, he will be relieved of his command and thus lose future opportunities to cause civilians to suffer and die. Whereas General A orders the bombing despite the likely civilian casualties, General B orders it because of the likely civilian casualties.
If the generals' respective motivations were sufficiently well known (by the discovery of their diaries, let's say), the existing law of war would acquit General A of any wrongdoing and hold General B accountable for a war crime. Scanlon thinks that's a mistake. In each case the action and the result is the same and so if General A's actions were morally permissible so were General B's.
To be sure, Scanlon does not think that intent is completely irrelevant. He thinks that intent is irrelevant to the moral permissibility of an action but relevant to judging the character of an actor. Under Scanlon's approach, there is thus a difference in the blameworthiness of my hypothetical generals. General B's motive to kill civilians makes him blameworthy, whereas General A's motive to secure a military advantage while foreseeably but regrettably killing civilians makes General A blameless.
What shall we make of that conclusion? Some people reject DDE because they think that even actions that are intended to cause a permissible result are impermissible if they foreseeably also cause harm to innocents. As Norman Cantor and George Thomas pointed out in an important 1996 article in the Kennedy Institute of Ethics Journal, the criminal law in general does not require specific intent, i.e., one is guilty of a crime if one intends an act the consequences of which are reasonably foreseeable, where those consequences are the actus reus of the crime. (The abstract of the article is available here, but the full article is behind a paywall. Cornell has a site license, so I've got access to the article. Weirdly, the abstract I just linked lists two additional authors, but the article itself only lists Cantor and Thomas as authors.) Put differently, the law generally rejects DDE as a basis for allowing actions that have particular unintended but reasonably foreseeable consequences.
Here I don't want to defend DDE against the sort of "leveling up" argument put forward by Cantor and Thomas (although I think DDE is defensible against that kind of argument). Instead, I'm exploring whether DDE is vulnerable to the sort of "leveling down" argument made by Scanlon, which says that if General A's decision is morally acceptable then so is General B's.
To my mind, Scanlon's argument is least persuasive in its caveat. Why would an actor be blameworthy for committing an act that is morally justified? One answer might be consequentialist. The sort of person who is motivated in the way that General B is motivated might be acting in a morally justified way in my scenario, but his evil motive makes him likely to commit bad acts in other contexts, and so we should be wary of him. However, in my hypothetical example, General B is a prudential rule-follower, so there isn't this worry.
Moreover, it's not clear to me that one can separate blameworthiness from conduct. In Beating Hearts, Prof. Colb and I say that to make sense of virtue ethics--the ethical system that says that the rightness or wrongness of conduct depends on the character (or virtue) that an actor exhibits when engaging in that conduct--we need to make it parasitic on some external conception of value, such as utilitarianism or deontology. Otherwise, it's hard (I would say impossible) to know whether the act manifests virtue. I think something similar is true of Scanlon's argument. Saying that an actor is blameworthy for engaging in a morally justified act smuggles into the concept of blameworthiness some external conception of what makes the act itself morally permissible or impermissible--unless Scanlon is prepared to rely wholly on the consequentialist assessment of how someone like General B will act in other contexts, but I have stipulated away such concerns here.
Accordingly, I am unpersuaded that Scanlon's exception is justified. It seems to me that if an act really is permissible regardless of the actor's intentions, then those intentions cannot make the actor blameworthy.
Of course, Scanlon could be wrong about his exception but right about the general point. Maybe General B's actions really are no different from General A's. One reason we might reach this conclusion is an aversion to thought-crimes and the like. On this account, the law oughtn't to judge people based simply on their mental states. But this can't be completely right, because one could respect that judgment and still reject Scanlon's argument as a standard for individual moral decision making, quite apart from law.
I think the best argument for Scanlon's view is the one I used simply to describe it: If someone is trying to decide a course of conduct, it's odd for that someone to interrogate his own motivation rather than asking whether the conduct is morally justified.
In the end, I still resist Scanlon's conclusion, but I'm not sure that my resistance is principled rather than ultimately prudential. My caveat about General B--that on prudential grounds he will only engage his sadistic appetites when doing so can be morally justified if the same conduct were undertaken by someone with good motives--is quite unrealistic. In reality, a sadist will engage his sadistic appetite when he can get away with doing so, not just when doing so doesn't violate the law (or the moral standard). Likewise, given everything we know about cognitive biases, a person who wants to engage in an act for a bad reason will likely conclude that the act is objectively justifiable and thus permissible in many circumstances when it really isn't.
In practice, using double effect as an external standard of conduct will occasionally permit someone who acts with bad motives to get away with doing so, because those motives will be difficult to detect. But if I'm right that the sort of person who would engage in harm-causing conduct for the wrong reasons will tend to do so even when the harm-causing conduct is not cost-justified, then eventually such a person will be found out. The prisons are filled with Holmesian bad men.