I. Introduction
Almost everyone believes that each individual possesses a right of self-defense such that, absent complicating factors, any prospective victim V may kill any guilty aggressor GA, if the V must kill the GA in order to prevent the GA from killing the V. A guilty aggressor is one who knowingly and voluntarily seeks to inflict some rights-violating injury upon another individual. Footnote 1 However, there are many other cases in which it is also true that individual V can prevent herself from being killed only by killing another individual, and yet it is — to say the least — not as intuitively obvious that the V may kill the individual whom she must kill in order to prevent herself from being killed. There is a familiar range of such cases; and as one moves along this range — away from the case in which the V prevents herself from being killed by killing the GA who will otherwise kill her — the proposition that individual V may kill the individual whom she must kill to prevent herself from being killed becomes intuitively and theoretically more problematic.
Here is a brief reminder of the salient cases along the range beyond the GA case.
Innocent Aggressor: The V is being attacked by an innocent aggressor, IA, who will kill the V unless the V kills the IA. This aggressor may be innocent because he acts under duress or he reasonably mistakes the V to be someone he may kill or his actions are guided by delusions for which he is not to blame.
Innocent Threat: The V is not (in the usual sense) being attacked by innocent threat, IT. But the IT’s movements (which involve little or no agency on the IT’s part) will kill the V unless the V kills the IT. Someone who has been pushed out of a helicopter and will fatally land upon the V unless the V vaporizes him with her ray gun is an innocent threat. (The falling man will survive if he lands unvaporized on the V.)
Unintentionally Killed Bystander: The V can prevent herself from being killed by the actions or movements of the GA or the IA or the IT only if she directs some counteraction against those otherwise fatal actions or movements. Unfortunately, the V’s counteraction will have the secondary, unintended, but foreseen effect of killing a bystander, the UKB. For instance, the GA is driving his Hummer at the V and will kill her unless the V fires her bazooka at the Hummer in the next ten seconds. However, the V foresees that debris from the Hummer will strike and kill the UKB who happens to be standing near the path of the Hummer.
Intentionally Killed Bystander: The V can prevent herself from being killed by the actions or movements of the GA or the IA or the IT only by directing some lethal action toward bystander, IKB. For instance, the GA is driving his Hummer at the V and will kill her unless the V fires her (less powerful) bazooka at the IKB who happens to be standing near the path of the Hummer. The GA’s otherwise fatal attack on the V will be thwarted because some of the IKB’s body parts will so smear the windshield of the Hummer that the GA will lose sight of the V.
Throughout this essay I will as much as possible limit my attention to cases in which the V must kill one other person in order to prevent herself from being killed. I seek, thereby, to sidestep questions about proportionality constraints on permissible defensive killing. For example, I sidestep questions about whether, if the V may kill the GA or the IA or the UKB or the IKB to prevent to herself from being killed, the V may also kill the GA or the IA or the UKB or the IKB to prevent herself from being blinded or from having a finger chopped off, and questions about whether the V may kill more of those individuals who would singly be liable to being killed by her if she must kill more of them to prevent herself from being killed.
Should the permissibility of the V’s self-defensive killings extend beyond her killing of the GA? If it should extend beyond the killing of the GA, how far across this range of cases should it extend? I attempt to answer these questions by delineating and contrasting two types of approach to questions about the extent of rightful self-defensive action and by displaying the greater plausibility of one of those approaches. In Section II, I contrast the “forfeiture” approach and the “elbow room” for self-defense approach. In Section III, I provide examples of forfeiture doctrines and indicate the problems that such doctrines encounter. In Section IV, I say a bit more about the general elbow room approach and provide a sketch of the particular elbow room for self-defense doctrine that I favor.
II. Elbow Room Versus Forfeiture Approaches
Theorists of self-defense are guided in where they draw the line between permissible and impermissible killing by their own intuitions about which killings by V would be permissible and which would be impermissible. Clearly, a theorist’s intuitions and consequent line drawing will in turn dispose that theorist to views about how to think about the permissibility of self-defensive killing that will lend support to and systematize those intuitions and underwrite his line drawing. At the same time, different background views about how to theorize the permissibility of self-defensive killing will support different intuitions and different conclusions about where the line should be drawn between permissible and impermissible killings. The plausibility of these background perspectives or approaches will not be entirely a matter of the plausibility of the intuitions they accommodate and systematize. Thus, these distinct approaches will be open to comparative assessment that is at least to some degree not merely a matter of dueling intuitions about particular cases. One’s intuitions about particular cases may be rationally reinforced or weakened by an appreciation of the relative plausibility or implausibility of the background perspective concerning self-defense that provides a rationale for those intuitions.
The basic difference between the elbow room and forfeiture approaches is a matter of a difference in the direction of explanation. The elbow room approach begins with the rationale for ascribing to individuals a basic right of self-defense and proceeds with an investigation of the range of defensive conduct that such a right needs to sanction. Since the sanctioned defensive conduct must be permissible, the rights that persons have against being subject to various forms of impinging treatment cannot include rights against that sanctioned defensive conduct. Persons must be morally liable to that conduct. In short, there must be moral elbow room for the right of self-defense. Footnote 2 On the elbow room approach the direction of explanation is from the right of self-defense to the permissibility of subjecting persons to morally sanctioned self-defense.
Suppose that, in circumstance C, individual V killing individual Y constitutes an exercise of the V’s duly articulated right of self-defense. On the elbow room approach, the V starts out with a right to kill Y in C and, therefore, Y starts out without a right against being killed in C. Y does not start out with a blanket right against being killed and then forfeit or lose that right when C comes to obtain. Of course, it is when C obtains that the V’s killing of Y is sanctioned and is not an impermissible treatment of Y. But, from the elbow room perspective, C’s obtaining does not trigger in Y the forfeiture of some previously held right. Rather, when C obtains, Y is in a situation in which he never had a right not to be killed.
In contrast, the forfeiture approach starts with a focus on Y who is a prospective subject of the V’s death-dealing efforts to thwart some lethal threat to the V. The approach seeks to identify a feature of Y that, if present, explains Y’s forfeiture of his background right against being killed by the V. Such a forfeiture by Y is taken to explain Y’s liability to the V’s countermeasures and, hence, the permissibility of the V’s action. Finally, it is in virtue of the permissibility of the V’s counteraction — so explained — that this counteraction counts as an exercise of the V’s right of self-defense. The right of self-defense does no independent work in determining whether thwarting actions are permissible. For this reason, this forfeiture approach does not take seriously the right of self-defense.
In this essay I argue against the forfeiture approach by highlighting problems across a range of particular forfeiture doctrines — including the common problem of seeking to account for justifiable self-defense by focusing entirely on the prospective subject of defensive counteraction. I argue on behalf of the elbow room for self-defense approach by offering a sketch of a promising instance of this approach.
III. Forfeiture Approaches
A. Forfeiture by guilt
Sometimes a doctrine of permissible self-defense will seem to exemplify the forfeiture approach by seeming to rest on claims about the endangering party’s forfeiture of rights. However, those apparently basic claims about forfeiture will themselves surreptitiously rest on claims about the endangered party’s right of self-defense. The work that seems to be done by judgments about forfeiture is really done by judgments about permissible self-defense. Consider a guilty aggressor case in which the GA is driving his tank across an unobstructed field with the known intention of squashing to death the V who cannot escape except by firing her bazooka at the tank and killing the GA. It is commonly said that it is permissible for the V to fire her bazooka at the tank and kill the GA because the GA has forfeited his right not to be killed. But I think the explanation really runs from the judgment that the V’s firing her bazooka is an act of permissible self-defense to the judgment that subjecting Y to the firing is not impermissible and, from there, to the judgment that, under these circumstances, Y has no right against being subject to that firing. Y’s not having a right against the V firing the bazooka in these circumstances is expressed by saying that in these circumstances she “forfeits” this right. But, surreptitiously, what is doing the work is the judgment that in these circumstances the V’s firing of the bazooka would be an exercise of her right of self-defense. The evidence for this claim is that what one will say about Y’s forfeiture of rights under varying circumstances will be driven by what one will say about whether counteractions by the V in those circumstances are exercises of the V’s right of self-defense.
In “Self-Defense and Rights,” Footnote 3 Judith Thomson wonders what the friend of forfeiture will say if the tank that the GA is driving toward the V breaks down before the V fires her bazooka at it. Presumably at that time it becomes impermissible for the V to fire her bazooka and kill the GA. Does one say then that, when the tank breaks down, the GA un-forfeits or comes back into possession of his right not to be killed? Won’t the change in what one says about the GA’s forfeiture or un-forfeiture be driven by one’s judgment that in these circumstances the V’s killing the GA would not be an exercise of the V’s right of self-defense? Since that right only encompasses a right to kill if killing is necessary to thwart a rights-threatening attack, if the GA’s tank has broken down, V’s killing the GA is not an exercise of her right of self-defense.
Suppose that instead of having to fire her bazooka directly at the main body of the advancing tank (which will cause it and the GA to blow up), the V could also thwart the advance by firing the bazooka at a tread of the tank. That will stop the tank short and knock out the GA when his head slams into the tank dashboard. In these circumstances, the V’s right of self-defense only allows her to fire at that tread and, so, in these circumstances, it remains impermissible for the V to kill the GA. Here one could say that it is permissible for the V to fire at the tread but not directly at the tank because, in the given circumstances, the GA forfeits his right not to be knocked out but not his right not to be killed. Yet surely what accounts for the difference between what the V may permissibly do when she needs to kill the GA and what the V may permissibly do when she only needs to knock out the GA is the difference between what she needs to do (what she needs to do). The difference between what it is permissible for the V to do in the two cases is not at all made by any independently identifiable difference between the GA’s forfeiture in one case and his forfeiture in the other case.
Note also that, if we think that forfeiture turns on guilt, our doctrine of forfeiture will not accommodate the fact that the extent of the injury that may permissibly be inflicted on the attacker tracks the extent of injury that needs be inflicted to thwart the attack rather than the degree of guilt of the attacker. After all, the guilty attacker who can be stopped by being knocked out is arguably as guilty as the attacker who can only be stopped by being killed. Indeed, if it were guilt-based forfeiture that explains the permissibility of the V’s infliction of injury on the attacker, it would be permissible for the V to inflict the same degree of injury on the determined tank driver who, unbeknownst to himself, is driving a tank made of cotton candy. Again, one may say that the determined driver of the cotton candy tank does not forfeit his right not to be killed. But this is just a way of expressing the impermissibility of killing that driver that obtains because this killing would not be an exercise of the V’s right of self-defense.
Of course, if forfeiture is supposed to do the job of rendering the V’s counteractions permissible and forfeiture is linked to guilt, then no counteractions by the V against the IAs (or the ITs) will be permissible. If one holds, as seems plausible, Footnote 4 that the prospective victim of an otherwise lethal act by an innocent aggressor may permissibly injure (or even kill) that attacker and one seeks to explain that permissibility by invoking the aggressor’s forfeiture, one must abandon that linkage between forfeiture and guilt.
B. Thomson on forfeiture on the part of (prospective) rights violators
Perhaps a forfeiture doctrine can be developed that does not depend on a strong linkage between forfeiture and guilt. Perhaps there is some defective or unsatisfactory feature that characterizes the actions of both the GA and the IA (and even the movements of the IT), that can be detected independently of any appeal to the V’s right of self-defense and that can, therefore, be invoked to explain the permissibility of the V’s defensive action against the GA and the IA (and even the IT). Judith Thomson offers a good example of such a broader forfeiture doctrine in her 1991 essay, “Self-Defense.” Footnote 5 In that essay, Thomson maintains that guilt cannot be the feature that renders endangering parties liable to injury- or death-dealing counteraction by the V; for the IA and, indeed, the IT are clearly liable to such counteraction and yet neither are guilty. Rather than guilt, Thomson selects being on course to be a violator of the V’s rights as the basis for rights forfeiture; and she takes this feature to be realized in the case of the GA, the IA, and the IT. Footnote 6
I want to emphasize here that the direction of explanation runs from the selected basis of forfeiture to the forfeiture, and from the forfeiture to the endangering party no longer having a blanket right not to be killed, and from the endangering party no longer having that right to the permissibility of the endangered party killing the endangering party. The only right of the endangered party that plays any role in Thomson’s story is that party’s right not to be killed. Because of that right the killing of the endangered party would constitute a violation of the endangered party’s right. Hence, that killing would instantiate the forfeiture basis. Footnote 7 No independent understanding of the endangered party’s right of self-defense plays any role in the vindication of the V’s killing of the GA or the IA or the IT. The right of self-defense is the tail wagged by the forfeiture dog.
In “Self-Defense,” Thomson rejects the permissibility of saving one’s life from the otherwise fatal attack of a GA or an IA or an IT if one’s targeting of that endangering party would also kill an innocent bystander. For, after all, neither a bystander who is killed intentionally (an IKB) nor a bystander who is killed unintentionally (an UKB) has the feature of being on course to violate the endangered party’s rights. Thomson’s conclusion here is quite striking. For, in her earlier “Self-Defense and Rights,” she had quite casually affirmed the permissibility of counteractions against attackers that unintentionally kill bystanders. In that early paper, she said that, if an innocent baby has been strapped to the front of the tank that the aggressor is on course to drive over you, “you can presumably go ahead and use the [tank destroying] gun, even though this involves killing the baby as well as the [tank driving] Aggressor.” Footnote 8 However, in “Self-Defense” we get the characteristic focus of forfeiture doctrines on the party who will be killed by the counteraction of the prospective victim. Only if some forfeiture basis can be ascribed to that party will the killing of that individual be permissible. Since no mere bystander will instantiate the feature of being on course to violate the V’s rights, no bystander forfeits his right not to be killed and, hence, no killing of a bystander by the V can be permissible.
In “Self-Defense,” Thomson offers some further support for her contention that, if a V’s intentional killing of a bystander to save herself from being killed by a GA, an IA, or an IT is impermissible, so too is a V’s unintentional killing of a bystander in the course of her thwarting such a lethal threat. The further support turns on two claims. (i) One needs to appeal to an across-the-board moral difference between unintentionally producing an (untoward) outcome and intentionally producing that outcome in order to affirm the permissibility of the V’s unintentional killing of a bystander in the course of her attack on the endangering party, while denying the permissibility of the V’s intentional attack upon of a bystander to thwart the endangering party. And (ii) there is no such across-the-board moral difference between the unintentional production of an (untoward) outcome and the intentional production of that outcome. Footnote 9
At this point, I merely note that Thomson may be mistaken about (i). For the moral difference between the V’s unintentional and the V’s intentional killing of a bystander may depend much more specifically upon a reasonable articulation of the V’s right of self-defense according to which the V may attack a lethal threat even if that attack also unintentionally kills a bystander and yet may not attack a bystander to disrupt a lethal threat against her. Ignoring the possibility that this moral difference might be underwritten by a proper articulation of the right of self-defense is another way of not taking seriously the right of self-defense.
C. Otsuka on forfeiture through moral responsibility
In “Killing the Innocent in Self-Defense,” Michael Otsuka offers a powerful critique of Thomson’s doctrine from “Self-Defense” and an alternative forfeiture doctrine. Otsuka agrees with Thomson that endangered party V may not save herself from death at the hands of an attacker if she can only do so by killing an innocent bystander. Even the V’s unintended killing of a bystander is impermissible — because, after all, no bystander does anything that forfeits or (in Otsuka’s language) “strips” away his right not to be killed. Otsuka seeks to leverage this point of agreement to yield the conclusion that ITs and even many IAs also do not conduct themselves in ways that strip away their rights not to be killed. Hence, the V’s killing of lethal ITs and many lethal IAs remains impermissible.
To begin with, Otsuka contends that Thomson is mistaken to ascribe to the IT and the IA the feature of being on course to be a violator of the V’s right not to be killed. An IT no more has this feature than would a rock that is falling lethally toward an imperiled individual. Footnote 10 And, an IA no more has the feature of being on course to be a rights violator than would an angry grizzly bear that poses a lethal threat to an imperiled individual. Footnote 11 According to Otsuka, the feature on which forfeiture is based, that is, the feature that strips a person of his right not to be killed, is being morally responsible for being on course to cause the V’s death. Whereas bystanders are simply not on course to cause the V’s death, ITs and many IAs are not morally responsible for being on their death-dealing courses. Due to this lack of responsibility ITs and many IAs are not stripped of their rights not to be killed. In all these cases, the V remains bound not to kill even though she must kill to thwart being killed.
To be clear, according to Otsuka, an aggressor can be entirely “innocent of blame” and, in this sense, be an IA and yet still be morally responsible for the lethal threat he poses. Such an IA is stripped of his right not to be killed. Otsuka’s example of such a blameless and yet morally responsible threat is a diplomat who knows she is the likely target of assassins. At some diplomatic shindig, you approach this dignitary to shake her hand. But “a third party projects a stunningly realistic holographic image of a pistol onto your hand” Footnote 12 which is seen by the diplomat. In reasonable response to this perceived danger (apparently emanating from a GA), the diplomat draws her own gun and is about to kill you. According to Otsuka, the diplomat is innocent of blame in taking this course of action. However, she is still morally responsible for taking this course of action because in so acting she will be exercising her agency. She will be “ . . . an agent who identifies with the intention from which she acts, is of sound mind when she acts, and could have avoided endangering your life.” Footnote 13 She is not a mere bystander to her own agency as is a deluded individual who attacks you because a chemical injected into her brain makes her think that you are the Devil and induces in her an unnatural hatred of Devils.
For Otsuka, to be a truly innocent aggressor and, hence, not to be stripped of one’s right not to be killed, one must be a bystander to one’s endangering action. Since the diplomat who reacts reasonably to the holographic gun is not a bystander to her endangering action and, hence, is responsible for that threatening reaction, she is stripped of her right not to be killed. Thus, at the diplomatic shindig you may quickly pull out your real gun and kill her first. Endangered parties may kill both GAs and not-quite-innocent — blameless, but morally responsible — aggressors, whom they need to kill in order to thwart the otherwise lethal advances of these individuals. It is only because, as Otsuka sees it, moral responsibility for life-endangering attacks extends somewhat beyond the blameworthy that permissible defense against life-endangering attacks extends somewhat beyond the killing of GAs. Footnote 14
In support of the stripping away of the diplomat’s right not to be killed, Otsuka says,
When one is in possession of rational control over such a dangerous activity as the shooting of a gun at somebody, it is not unfair that if the person one endangers happens to be innocent, one is by virtue of engaging in such dangerous activity stripped of one’s moral immunity from being killed. A responsible agent takes a gamble by placing this moral immunity on the line when she engages in such avoidable risky activity. It is, on the other hand, unfair to strip an [innocent] Aggressor or [innocent] Threat of moral immunity from being killed if she endangers the life of another, since she exerts no rational control over her endangering activity. Footnote 15
Yet isn’t it unfair that the individual who reasonably and blamelessly exercises her agency to put herself on course to kill an apparent aggressor becomes liable to being killed while the individual who has been thrust into endangering action or movement by a chemical being injected into her brain retains her right not to be killed? Shouldn’t the exercise of reasonable and blameless agency count on behalf of this diplomat — so that there is a stronger case for the responsible diplomat retaining her right not to be killed? And what about fairness for the individual who approaches the diplomat for a handshake but who encounters a diplomat who has been injected with that enraging chemical? Isn’t it unfair to that individual to require that he allow himself to be killed even though he would not be so required had the troublesome projector projected the holographic image rather than the enraging chemical? I am, of course, not pressing for specific answers to these questions. Rather, I am suggesting that appealing to fairness here is a game of deuces wild.
In addition, we should note that it is difficult to locate the line between blameless but morally responsible aggressors and those who, according to Otsuka, are blameless and not responsible. Recall that Otsuka tells us that the blameless but responsible aggressor “ . . . identifies with the intention from which she acts, is of sound mind when she acts, and could have avoided endangering [her target’s] life.” Footnote 16 Consider the diplomat who has a chemical injected into her brain that makes her perceive the approaching individual as the Devil and makes her really hate the Devil. I think that Otsuka would take this diplomat to retain her immunity against being killed. Yet it seems that, when she proceeds to kill the approaching individual, she will be identifying with the intention from which she acts and will be of sound mind when she acts. Also, it seems as true of her as it is of the diplomat who responds to the holograph that she could avoid attacking the approaching individual. Footnote 17 So it seems that, on Otsuka’s criteria, this diplomat will be responsible and will be stripped of her right not to be killed. The implication that I wish to point to here is that it is hard enough for V to determine whether she may attack an attacker if the permissibility of her counteraction depends on the blameworthiness of that attacker. And it will be harder yet for the V if the permissibility of her counteraction depends on the more elusive feature of her attacker being morally responsible for — albeit, perhaps not blameworthy for — the attack. If the permissibility of the V’s counteractions against lethal threats depends upon her identifying just when the target of her counteraction has forfeited his right, the more delicately subtle one’s doctrine of forfeiture, the more frequently the V will not be able to determine whether she may act in defense of her life.
Despite questions about how common moral responsibility is among blameless attackers, it is clear that Otsuka favors a strikingly narrow scope for permissible self-defense. This comes across vividly in Otsuka’s bullet-biting response to an IT case. Suppose you are erecting a flagpole. You have not yet secured it in its perfectly vertical position. If you do not continue to hold it in that position, it will tip over. Unfortunately a villainous individual has just pushed a well-mannered fat man out of a plane directly above you and your flagpole. If you continue to hold the flagpole in place, the threatening non-responsible fat man will be fatally impaled on the pole whereas, if you do not continue to hold it in place, he will miss the flagpole, crush you, but walk away uninjured.
Otsuka accepts the implication of his doctrine that it would be impermissible for you to hold the flagpole in place; you must let it slip away from its vertical position to avoid being an impermissible killer of the fat man. Indeed, on Otsuka’s doctrine, if the fat man jumps from the plane because (having been injected with that chemical) he thinks you are the Devil who must be vanquished and is thereby a non-responsible aggressor, it will also be impermissible for you to continue to hold the flagpole in place. You must allow that fat IA to crush you to death. All this again highlights the characteristic feature of forfeiture views, namely, total focus on the normative standing of the person whom the endangered party must kill if she is to escape being killed, and total inattention to any independently identified right of self-defense in the endangered party.
D. McMahan on forfeiture through moral responsibility
Certainly the most systematic and sophisticated recent treatment of permissible self-defensive killing is Jeff McMahan’s Killing in War. Footnote 18 I can only attend to a few themes within McMahan’s highly complex discussion that intersect with the issues I have been addressing. McMahan rejects the view that “posing a threat [of an objectively unjustified harm] is the basis of liability to attack in war.” Thus, he rejects the vindication for killing IAs and ITs that Thomson offers in “Self-Defense.” However, McMahan also rejects the view that “noninnocence in the sense of moral guilt or culpability is necessary for liability to attack in war.” Footnote 19 Instead, he holds a position on liability to injury- or death-dealing counteraction that is very similar to Otsuka’s.
Something less [than moral guilt or culpability] is sufficient, namely moral responsibility for a wrong, particularly an objectively unjustified threat of harm. . . . In short, the criterion of liability to attack in war is moral responsibility for an objectively unjustified threat of harm. Footnote 20
Like Otsuka, McMahan relies upon a crucial distinction between culpability and blameworthiness, on the one hand, and moral responsibility, on the other hand.
It is important to note that the claim that a person is fully excused for an act of objective wrongdoing implies only that the person is not culpable, that he or she is entirely blameless. It does not necessarily imply that the person is absolved of all responsibility. A person may be responsible for his objectively wrong action even if he is not blameworthy. Footnote 21
So, McMahan’s view displays the core characteristic of the forfeiture approach. Y’s being liable to counterattack turns entirely on some fact about Y, namely, his being morally responsible for X’s endangerment. Y’s liability does not at all derive from any independent force that attaches to X’s right of self-defense. X’s right of self-defense with respect to Y is entirely a matter of Y’s forfeiting his right not to be killed and X’s not forfeiting of her right not to be killed. Those who engage in “morally justified self- and other-defense, . . . do nothing to forfeit their right not to be attacked or killed.” Even though they “are ‘doing harm’ and ‘pose a danger to other people’ when they oppose the military actions of unjust combatants, they do not thereby become legitimate targets of attack but retain their innocence in the generic sense.” Footnote 22
To illustrate the role of moral responsibility as the basis for forfeiture, McMahan offers an example that parallels Otsuka’s entirely nonculpable yet morally responsible gun-toting diplomat: this is Resident who hears on the news that a homicidal maniac has just escaped from prison and is in Resident’s neighborhood. Resident also hears from very reliable sources that, once this maniac approaches anyone’s home, he breaks in and kills everyone in sight. Fortunately, the far-sighted Resident has also seen accurate pictures of the maniac. Unfortunately, the unknown and entirely benign identical twin of the maniac (who knows nothing of his brother’s escape) has just run out of gas in Resident’s neighborhood. The twin approaches Resident’s home to ask to use his phone and Resident draws a bead on him with his trusty Remington. According to McMahan, if the Resident proceeds to fire and kill the twin, he will be entirely nonculpable; but he also will be responsible for the killing due to its being an expression of his agency. For this reason, Resident will be liable to being killed by the twin who, to his surprise, finds a bazooka in his pocket and can only thwart Resident’s attack by lethally firing that bazooka at Resident. I agree with McMahan’s judgment that the twin may fire that bazooka; but I don’t find it plausible that this permissibility turns on the agent responsibility of the Resident.
I do not see that there is enough distance between the fully nonculpable but responsible attacker and the fully nonculpable and non-responsible attacker to support the judgment that the former is liable to death-dealing counteraction but the latter is not. I do not see how an individual being an agent of an attack in a way that still leaves him entirely nonculpable for that attack renders him more liable to death-dealing counteraction than an individual who is fully nonculpable and not an agent of the attack that issues from him. And, again, there is the issue of how difficult it will be for the endangered party to determine whether she is faced with a fully nonculpable yet responsible attacker or a fully nonculpable and non-responsible attacker.
Consider McMahan’s case of the Cell Phone Operator:
A man’s cell phone has, without his knowledge, been reprogrammed so that when he next presses the “send” button, the phone will send a signal that will detonate a bomb that will kill an innocent person. Footnote 23
McMahan holds that the Cell Phone Operator, like the nonculpable Resident — who is liable to counteraction by the benign twin — “acts as a morally responsible agent.” Nevertheless, according to McMahan, this Cell Phone Operator is “a Non-Responsible Threat because he has no way of knowing that he poses a threat to anyone.” Footnote 24 There is a “complete absence of moral responsibility for posing a threat . . . because he cannot know that he poses a threat.” Footnote 25 Hence, according to McMahan, this Operator is not liable to counteraction. So, if I am the benign twin, I may fire my bazooka at and kill Resident but, if I am the person about to be blown up by Cell Phone Operator and I need to kill Operator to avoid being killed, I must allow Operator to proceed — even though, unlike some totally programmed individual, Operator “acts as a morally responsible agent.”
McMahan’s central claim is that the prospective victim of a lethal “Nonresponsible Threat” may not kill that attacker even if killing the attacker is the only way for the victim to save her life. Footnote 26 If Manchurian Candidate has been programmed to kill you and you can only save yourself by killing Candidate, you must allow him to kill you. Footnote 27 McMahan recognizes that “the claim that a Nonresponsible Threat is not liable to defensive force is highly counterintuitive” and that, therefore, he needs to advance arguments “in favor of the claim that a Nonresponsible Threat cannot be liable.” Footnote 28
McMahan’s defense begins with the proposition that “in general the presumption against killing a person is stronger than the presumption against letting a person die.” More specifically, “the presumption against intentional killing is stronger than that against unintentional killing and much stronger than that against letting a person die.” Footnote 29 McMahan then applies this maxim to the question at hand by construing the endangered party’s choice to be a choice between letting a person, namely, herself, die and intentionally killing a person, namely, the endangering party. “If one’s life is threatened by a Nonresponsible Threat so that one must choose between intentionally killing the Threat and allowing oneself to be killed by him, the presumptions oppose killing in self-defense. . . .”. Footnote 30 McMahan continues, “To overcome or defeat the presumption against killing [a threat to save one’s own life], there must be some important moral difference between the Threat and oneself.” Footnote 31
Not surprisingly, for McMahan, only the threatening party’s responsibility for the endangerment can overcome the presumption against the endangered party killing the threatening party. So, the presumption is overcome only in the case of culpable aggressors and nonculpable but still responsible aggressors. Notice, however, that the endangered party will also be morally responsible for her lethal counteraction. Since there is no difference in responsibility between the endangered and the responsible endangering party, it is not clear what the moral difference is between those two parties that overcomes the presumption against the endangered party’s counteraction. Moreover, even if there were some difference in responsibility between these two parties, why not think that this difference exactly counterbalances the presumption — leaving the two parties equally at liberty to kill?
Rather than pursue these specific problems, I want to focus on McMahan’s problematic invocation of the killing versus not saving maxim and on the general character of McMahan’s delineation of permissible self-defense. McMahan’s invocation is problematic because when one nods in agreement with the maxim that it is worse to kill than not to save, one has in mind cases in which one can save person X (who may be yourself) only by killing another person Y who is a bystander, that is, who is not on course to cause X’s death. We have such a case if one can save X only by fatally removing Y’s heart and transplanting it into X. In such cases, one reasons that it is too bad if one ends up not saving X. But it would be morally monstrous to fatally remove bystander Y’s heart; so one must not do that.
Suppose, however, that X and Y stand in a quite different relationship — namely, X is you and Y is Manchurian Candidate who has been programmed to kill you and will kill you unless you kill him. One might assert that the maxim that it is worse to kill than not to save extends to cover this sort of case as well as the transplant sort of case. However, one can only reasonably assert this if one has some independent reason to hold that X’s killing of Y in such an endangerment case is worse than X’s not saving himself. Absent a persuasive independent reason, one cannot plausibly take the maxim to extend to such cases; and, so, one cannot invoke the maxim in such cases to establish a prima facie indictment of X’s killing the endangering Y.
Note also that the issue in an endangerment case is not merely a matter of not saving versus killing. In a transplant case, all that emerges from Y being killed is X being saved. All that is avoided by the killing of Y is the non-saving of X. But in an endangerment case, what emerges from Y being killed is X not being killed. When X kills Manchurian Candidate, she does not merely avoid not saving herself; she avoids being killed. Thus, it is mistaken to think that cases of this sort simply embody the choice between not saving and killing.
My point about the general character of McMahan’s forfeiture approach to permissible self-defense is simple and predictable. At no point does any independently identifiable and forceful right of self-defense play any role in this approach. The fact that one party is endangered by another party does nothing to support the permissibility of the former’s counteraction against the latter. Indeed, via McMahan’s flawed application of the killing versus not saving maxim, a presumption is said to exist against the endangered party’s counteraction — even if the endangering party is a GA. And the only thing that can overturn that alleged presumption is the presence in the endangering party of responsibility for the endangerment. Footnote 32 Even the overcoming of the presumption against resistance by X involves no hint of any independently forceful right of self-defense in X, no hint of X’s standing as a being with a right of self-defense.
IV. Taking Seriously the Right of Self-Defense
A. The elbow room approach
In a recent paper entitled “Elbow Room for Rights,” I respond to a challenge posed to advocates of robust self-ownership and property rights (in extra-personal objects) by David Sobel. Footnote 33 Essentially, the challenge is that, for each individual A, A’s robust self-ownership and property rights place severe moral restrictions on what others may do. No other person may perform any action that at all physically impinges upon A’s person or property without A’s consent. However, the challenge continues, if this is true, just about any action that the advocate of robust rights wants to say is an exercise of the performer’s rights — for example, driving your car up your driveway — will turn out to be a violation of some A’s self-ownership or property rights. So, if everyone’s rights are to be respected, almost no one will ever get to exercise the rights that are to be respected. Sobel thinks this shows that moral rights must be construed less robustly.
In “Elbow Room for Rights,” I argue that the proper insight to derive from Sobel’s challenge is not that moral rights must be construed less robustly, but rather that they must be delineated more carefully. Since ascribed rights must be rights the exercise of which is generally permissible, the proper delineation of those rights cannot be such that their exercise is generally impermissible. Since people will possess generally exercisable rights over themselves and their (just) holdings only if the minor physical intrusions involved in their doing as they see fit with themselves and their (just) holdings are morally permissible, people’s rights over themselves and their (just) holdings must leave them liable to minor physical intrusions. In short, there must be moral elbow room for rights that includes people’s liability to minor intrusions. One does not start with a blanket right against any intrusion such that any intrusion requires (actual or hypothetical) consent. Rather, one reasons from people’s exercisable rights to do as they see fit with their persons and (just) holdings to persons’ liability to such minor intrusions. Footnote 34
A central theme of my critique of forfeiture approaches is, similarly, that one should not start with a blanket right against being killed that renders all self-defensive killings impermissible unless the party subjected to the counteraction forfeits his right not to be killed. Rather, one should start from the other side, that is, from a sensible articulation of the right of self-defense, and then identify which defensive killings persons must be liable to for there to be the necessary elbow room for the right of self-defense.
B. Partiality and rights
I have merely indicated the general nature of an elbow room for defense approach. Now I need to sketch my preferred version of that approach. Part of the reason for the sketchiness is that my preferred elbow room approach turns on certain contentious claims about the role of fundamental rights within a reasonable moral code; and these are contentions that I cannot attempt to justify here. Part of what excuses the sketchiness is that the primary purpose of the sketch is to indicate the availability of a general alternative to the forfeiture approach. One does not need to work out in detail a particular version of the elbow room approach to indicate the availability of this alternative approach.
The inclusion of moral rights at a fundamental level within morality takes seriously the existence of individuals as beings with systems of ends or hierarchies of interests of their own to which each is reasonably partial. If it were not for the reasonableness of persons’ partiality — if it were not, to speak somewhat loosely, because partiality merits protection — morality would not call for rights that protect each individual’s pursuit of her own conception of the good even when greater good for others could be achieved by compelling that individual to serve that greater good. Since each person may rationally devote herself to her own system of ends or hierarchy of interests — even when she could instead sacrifice those ends or interests to the more extensive ends or interests of others — and this partiality merits protection, there is a rationale for strong protective individual rights at a very basic level of morality. Footnote 35 Such rights protect individuals in their chosen pursuit of their own ends or interests by designating for each individual a distinct domain of what we might call “action resources” with which that individual may do as she sees fit without asking the leave of others. Moral rights provide each individual with moral protection against all other persons for her discretionary control over the action resources within her moral domain — within what is, therefore, her sphere of freedom. It is through respect for such rights that one recognizes in one’s conduct toward others the standing of other individuals as beings with rational ends of their own, as beings with lives of their own to live.
Let us focus on the fundamental moral right of each individual to dispose of her own person — of her body, capacities, and energies — as she sees fit in the service of her own conception of the good. To assert this right is simply to assert that each person herself is within the domain of action resources over which she rightfully possesses discretionary control. This is the fundamental right to focus upon, given this essay’s concern with killing to defend oneself against being killed. For being killed by another (without one’s consent) is paradigmatic of being denied one’s right to dispose of one’s own person as one sees fit.
In virtue of this fundamental right over one’s own person, should we ascribe to each individual a blanket right not to be killed? The answer is “no.” For such a blanket right would leave no elbow room for a right to act defensively against transgressions of one’s right over one’s own person. Rather than asserting such a blanket right, we need to consider what right of self-defense is plausibly ascribed to individuals and what elbow room must exist for that right. To have a right not to be killed is to be in a position to demand of others — those with the correlative obligation — that they not kill one. It is not merely to be in a position to offer good moral advice. Rather, it is to be in a position not to have to tolerate — not to have to submit to — noncompliance with that right. It is not enough to be in such a position that it be permissible to flee from the transgression. For the other’s action being contrary to one’s good moral advice to him is enough to vindicate such flight; and one is here, by hypothesis, in a position not merely to advise but, more strongly, to demand. Furthermore, flight from the transgression will often not be available. In fact, we are especially interested in those cases in which the only way not to submit is to fight. If, in virtue of one’s self-ownership rights, one may morally demand that another not kill one, one must have a right not to submit to that killing. And, if killing one’s prospective killer is necessary to escape submission to being killed, one must have a self-defense right to kill that prospective killer. One must have a right to thwart death-dealing attacks even if one can only thwart them by lethally counterattacking the attacker. Therefore, no one has a blanket right not to be killed, that is, a right that has to be forfeited by an individual — through, say, culpability or moral responsibility — before that individual may be killed in self-defense. The non-impermissibility of killing the individual who, if not killed by you, will deprive you of your rightful discretionary control over your own life is the moral elbow room required by the right of self-defense.
I have said that we find the killing versus not saving maxim agreeable insofar as we take it to apply specifically to transplant-type cases. In those cases, our judgment is that one may not save X by way of killing Y (even if one is X). Everything I have said about self-ownership and the right of self-defense and their implications is consistent with that transplant case judgment. For the demand that X properly addresses to Y under her right of self-ownership is: You Y may not deprive me of discretionary control over myself. The demand is not: You Y must sustain or extend my discretionary control. X cannot claim that her right of self-defense renders it permissible for her to extract Y’s heart for transplant into X. X may be required not to save herself if saving herself would amount to killing another who is not otherwise about to kill her; but X may not be required not to save herself if saving herself would amount to killing another who is otherwise about to kill her.
C. The scope of the right of self-defense
The argument that under her right of self-defense X may kill Y to prevent herself from being killed by Y does not at all depend on Y being a GA. Nor does it depend upon Y being morally responsible (in the sense that Otsuka and McMahan have in mind) for the threat that he poses to X. It is enough that Y’s action will deprive X of her rightfully held life unless X kills Y and that X need not submit to such a deprivation. So, an appreciation of the right of self-defense moves us at least to the permissibility of the defensive killing of IAs.
What about ITs? Imagine two situations involving yourself and a very large lump of (unowned) coal. In the first case, you could use smaller pieces of coal to keep yourself from freezing during a cold night. So, you permissibly break up the coal and burn it. In the second case, the large lump is falling rapidly toward you. So, you permissibly whip out your ray gun and vaporize it. Now envision two cases involving yourself and another self-owning individual. In the first case, you could use smaller pieces of the other individual to keep yourself from starving. However, you may not permissibly cut him up and eat him. In the second case, that individual has been thrown from the plane and is falling rapidly toward you with the same lethal prospects for you as in the second coal case. (But, if he lands on you, the crushing of your body will save his life.) Does your right of self-defense render it permissible to use your ray gun on the falling individual?
Construing your right of self-defense as rendering this use of your ray gun permissible does not much rollback the falling individual’s right not to be killed. For, he retains his right not to be killed in all those (vastly more common) situations in which neither his actions nor his movements are on course to deprive another of her rightfully held life. Quite strikingly, he retains his right not to be killed even if you must kill him to save yourself from a deadly danger that is not of his making. Moreover, even though the falling individual is an IT, it seems that the permissibility of your vaporizing the falling individual is reinforced — compared to the permissibility of your vaporizing the large lump of coal — because you have a right against the falling individual that he not kill you, whereas you have no such right against the falling lump of coal. Footnote 36
Here is a further consideration on behalf of the permissibility of your using your ray gun on the falling IT. Imagine a variant on this case in which you are able to flee from the spot toward which the IT is headed. Surely it would be morally permissible for you to flee even though your declining to stay in place to cushion his fall will be the death of the IT. It is permissible for you to flee even though your fleeing is pretty close to the boundary between not saving him and killing him. Suppose now that as you move toward the escape tunnel the door to that tunnel shuts tight. Now you can escape only by using your ray gun on the falling IT. But can an observer plausibly say to you that, although you were free a minute ago to save yourself even though that would have been the death of the IT, now you must just let him crush you to death?
Here is one last and somewhat different sort of argument for why it should be permissible to kill IAs and ITs. The argument is that the impermissibility of killing IAs and ITs morally empowers GAs to deprive Vs of their rights of self-defense. If IAs and ITs may not be killed in self-defense, GAs can morally disarm their victims by constructing and operating through IAs and ITs. Although you will not have to let the GA kill you when he comes at you with his guilty cohort, you will have to let yourself be killed whenever the GA sends his Manchurian Candidates at you or when he drops (nonconsenting) fat men from helicopters directly above you. The more guilty — the more evil — the aggressor is vis-à-vis his willingness to construct IAs or ITs and send them against you, the more likely it will be that you will have to submit to the GA’s death-dealing plans. If GAs are not to have such powers, the right of self-defense must include the permissibility for the target of such plans to kill unfortunate Candidates and fat men that the GA sends at that target.
My conclusion here is that elbow room for the right of self-defense requires that individuals be liable to be killed when their actions or movements are on course to being objectively unjust destructions of others’ lives. Of course, to assert the permissibility of these defensive killings is not to assert that all or even most of them yield more-fair or more-impersonally valuable outcomes than would be yielded by submission to the attack. To think that the permissibility of these killings hinges on their upshots being all or mostly more fair or more impersonally valuable is another way not to take seriously the right of self-defense.
Finally, what about the foreseen, but unintended, killing of innocent bystanders? What about attacks upon attackers that also result in the killing of bystanders? To begin with, let us lay out some of the powerful cases against one’s engaging in the direct (intentional) killing of a bystander, even if such killing is necessary to thwart an otherwise fatal attack on oneself. First of all, such killing involves an attack on a person who is not at all involved in posing a threat to you. A non-attacker is selected as the target of your attack. And, in this way, what you are doing must be as impermissible as the act of the party who attacks you. If you deny the permissibility of that attacker’s conduct or movement, you must also deny the permissibility of your attack on the bystander. Moreover, the attack on the bystander is a case of saving yourself from an impending loss by killing someone who is not at all involved in making that loss impending. It is relevantly like fatally extracting Y’s heart in order to transplant it into yourself. If, absent consent, that extraction is impermissible, then so too must be the direct attack on the bystander that saves you from the lethal threat.
Is there a significant enough difference between such a direct fatal attack on a bystander and an attack on an attacker that also has the (side) effect of killing a bystander — a difference that makes it reasonable to think that the bystander’s right not to be killed forbids the direct attack but not the indirect killing? Perhaps. But, note that in what follows I am not going to argue that there is a general, across-the-board difference between intended and nonintended killings such that, quite generally, nonintended killings are permissible while intended killings are not.
I have argued that IAs and ITs must be liable to being killed because, if they are not, GAs will be able to morally disarm their intended victims. That argument extends to unintentionally killed bystanders. If a bystander has a right not to be killed as the side effect of the V’s counterattack on the GA, the GA will be able to morally disarm the V by surrounding himself — by shielding himself —with such bystanders. Footnote 37 I think that the V cannot be deprived of her right of self-defense by the GA strapping an innocent bystander on the front of his advancing tank. Of course, it is a horrible thing that the V’s counterattack cannot now be performed without the death of that bystander.
Is it only when the attacker has arranged for there to be a bystander who will be killed by that counteraction that the V’s right of self-defense sanctions counteraction against an aggressor even though that action will also kill a bystander? It is easy to think of different circumstances in which it is at least as plausible that the V may proceed with her counteraction even though this will also kill a bystander. In these cases, the bystander has arranged his own vulnerability. The bystander may mistake the aggressor for a prospective victim and seek to act as a shield for this actual aggressor. It is hard to see how the bystander’ s mistaken attempt to shield an actual aggressor could deprive the actual victim of her right to thwart by way of counterattack the otherwise fatal attack on her. Yet there remain all those cases in which no one really arranges for the bystander to be so situated that a counterattack on an attacker will also kill the bystander.
We have two very plausible principles. First, one cannot be required to submit to the destruction of one’s person by an attacker; so, one cannot be required not to attack an attacker who will otherwise destroy one’s person. One has a right of self-defense. Second, one may not attack a non-attacker even if one must engage in that attack to save oneself from impending destruction. One has a right of self-defense. The first principle leaves room for one not being required to submit, even when one’s counteraction against the attacker will also kill a bystander. The second principle also leaves room for such counteraction because that counteraction will not be an attack on the bystander.
Extending the moral elbow room associated with the right of self-defense so that it is not impermissible to kill a bystander as the side effect of one’s defensive counterattack does involve a further and discomforting rollback of the right not to be killed. The world has been full and will no doubt continue to be full of bystanders who will be liable to collateral killing if prospective victims may engage in counteraction against their attackers even if bystanders are foreseeably killed. On the other hand, this shows how pervasively prospective victims will be morally disarmed if it is not permissible for them to defend themselves against attackers when doing so will also and horribly kill bystanders. At least in the case of large-scale attacks, if it is impermissible for the prospective victims to engage in counteraction that involves the foreseeable death of bystanders, the prospective victims will be left with no effective right of self-defense. And aggressors will have free reign. If one believes in the right of self-defense, this is a strong reason to hold that, at least in the case of large-scale attacks, it is permissible for the victims to fight back even if that will mean the unintended death of bystanders.
There are lots of other factors and subtleties to consider. One of these is whether some people who may permissibly be killed in self-defense may, nevertheless, themselves permissibly kill, if they must, to thwart their being killed. Perhaps it is permissible for the innocent fat man who will crush you unless you vaporize him to shoot you to thwart that vaporization. Footnote 38 It is clearer that the bystander need not allow you to carry out the counteraction that will unintentionally kill him. If that bystander is not one of those voluntary shields, he has as much right as anyone to insist that he cannot be required to submit to death at the hands of another. Here we may have an awfully nice case of two parties each permissibly acting in ways that will kill the other. Footnote 39 This is perfectly possible as long as the permissibility of killing in self-defense is not a matter of the killing of one party being fairer or more impersonally valuable than the killing of the other.
