¹ Some writers suggest that innocent people in situations of the sort being discussed here are somehow relieved of responsibility for committing homicide in self-defence because their choices are coerced. See, for example, Miller, Seumas, ‘Self-defence and Forcing the Choice Between Lives’, Journal of Applied Philosophy, ix (1992)Google Scholar. For a reply to Miller's article, see Montague, Phillip, ‘Forced Choices and Self-Defence’, Journal of Applied Philosophy, xii (1995)Google Scholar. Ryan, Cheyney argues along similar lines in ‘Self-Defense, Pacifism, and the Possibility of Killing, Ethics, xciii (1983)Google Scholar. The precise nature of Ryan's position on this issue is unclear, however, since he blurs certain distinctions in the area of responsibility, particularly that between causal and moral responsibility (seen. 2 below).

² I realize that being causally responsible for a situation is sometimes equated with causing that situation, and that being morally responsible is sometimes interpreted as merely being a candidate for moral appraisal. As I use ‘causally responsible’ and ‘morally responsible’ here, however, both imply agency; moreover, being morally responsible for a state of affairs implies being at fault or to blame for causing that state of affairs. It is important to distinguish (causal or moral) responsibility for the existence of forced-choice situations from responsibility for events that occur within such situations. Only the former plays a role in the account of justified self-defence being proposed here.

³ Unless otherwise indicated, the justifications, permissions, rights and requirements to which I refer throughout this paper are to be understood as presumptive, prima facie, or defeasible. I will assume, moreover, that if there exists a presumption in favour of x's doing y and an equally strong presumption against x's doing y (and other things are equal), then x is justified all things considered in doing y and x is justified all things considered in not doing y. The account of the morality of self-defence against culpable aggression that I have sketched is developed in Montague, Phillip, ‘Self-Defense and Choosing Between Lives’, Philosophical Studies, xl (1981)Google Scholar. The account is defended in Montague, Phillip, ‘The Morality of Self-Defense: A Reply to Wasserman’, Philosophy and Public Affairs, xviii (1989)Google Scholar; and Forced-Choices and Self-Defence, The Journal of Applied Philosophy, xii (1995)Google Scholar. In the discussion that follows, I will accept the proposed account of self-defence against CAs without further argument.

⁴ According to some writers, acting with the intention of harming others is immoral, which apparently implies that my innocent aggressors are not innocent at all. This position strikes me as mistaken, but I will not contest the point here. If my ‘innocent’ aggressors are in fact culpable, then killing them in self-defence is justifiable on the grounds described earlier.

⁵ At this point let me make explicit some presuppositions with which I am operating throughout this discussion, (a) Creating situations in which innocent people must choose between committing homicide or being killed is presumptively impermissible even if done innocently. Hence, even righteous IAs and AITs do something presumptively impermissible when they create forced choices for others; and this is why I characterize the distinction between righteous and wrongful IAs and AITs in terms of the presence of permissibility and impermissibility all things considered, (b) Actions (or action types) are justified (permissible) or unjustified (impermissible), whereas people are blameless (innocent) or blameworthy (culpable, at fault) for acting, (c) Actions are justified or unjustified in virtue of factors that are independent of the beliefs, attitudes, etc., of their agents, whereas people are blameless or blameworthy for acting at least partly in virtue of certain of the beliefs, attitudes, etc., with which they act. I realize that at least some of these presuppositions are controversial, but they are certainly no less reasonable than possible alternatives. And, of course, some things must go unargued in every discussion.

⁶ Anscombe, G. E. M., Ethics, Religion, and Politics, Minneapolis, Minnesota, 1981, p. 53Google Scholar. I am interpreting Anscombe's remarks as applying to IAs and to AITs because of her use of the term ‘non-innocent’. She need not have introduced this technical term if she meant to limit the remarks in question to CAs. According to Anscombe, in other words, although ‘innocent’ aggressors and threats are not blameworthy for their actions, neither are they innocent in the sense relevant to whether those whom they threaten are justified in defending themselves.

⁷ In n. 2, I distinguished responsibility for the existence of forced-choice situations from responsibility for events that occur within such situations. Now I am relying on the distinction between some x's being unjustified in forcing innocent person y to choose between her life and x's on the one hand, and x's being unjustified in killing y on the other. My comments on Anscombe's view question the relevance of x's (counterfactually) killing y unjustifiably. Later, however, I will affirm the relevance of x's (actually) forcing y to choose between her life and x's unjustifiably.

⁸ In contrast to many other writers on the topic of self-defence, George P. Fletcher heavily emphasizes the distinction to which I am referring in his discussion of legal justification. See, for example, his article The Right and the Reasonable, Harvard Law Review, xcviii (1985), esp. 977 fGoogle Scholar.

⁹ Some writers use the expression ‘liberty right’ as a label for claim rights of a certain sort, namely claim rights to act (as opposed to claim rights to be treated or to not be treated in certain ways). These writers typically use ‘liberty’ to refer to what I am calling liberty rights. I prefer my usage because it reflects how people commonly use the language of rights to refer to what is merely ‘all right’—that is, merely permissible.

¹⁰ The distinction I am emphasizing is overlooked by Jeff McMahan in the following remarks that he offers as a criticism of the account sketched above of self-defence against CAs (an account McMahan calls ‘JISTICE’):

JISTICE provides a compelling explanation of the permissibility of self-defense against a…[CA]. Yet it seems to suffer from a narrowness of scope. For most people believe that there are cases in which self-defensive killing is justified but in which the Attacker is not morally culpable for the threat he poses to the victim. In such a case, JISTICE provides no ground for assigning priority to the Victim. If self-defense is justified, the justification cannot come from JISTICE, which then fails to provide a complete account of the foundations of the right of self-defense. (McMahan, Jeff, ‘Self-Defense and the Problem of the Innocent Attacker’, Ethics, civ (1994), 263Google Scholar. The emphases are mine.)

As I will suggest presently, however, JISTICE can provide a complete account of the claim right of self-defence even if it is inapplicable to IAs and AITs, since self-defence against the latter is merely permissible. Or, alternatively, the account to be proposed of the permissibility of self-defence against IAs and AITs could be used to explain the mere permissibility of self-defence against CAs if such an explanation were all that mattered.

¹¹ Fletcher denies that justifications can properly be interpreted as permissions, but his argument for this denial applies only to legal justifications and permissions (Fletcher, 977 f.). I will not attempt to assess Fletcher's arguments concerning legal justifications and permissions since my concern here is with moral justification and Fletcher's argument (together with its underlying presuppositions) has no plausible moral analogue. For another discussion of some of these points, see Alexander, Laurence, Justification and Innocent Aggressors, Wayne Law Review, xxxiii (1987)Google Scholar.

¹² I can imagine someone objecting to the claim that Ed is justified in diverting the slide on the ground that the interests of justice are best served if he lets the slide destroy his house and collects compensation from Flora. While it might well be true, however, that Flora would owe him compensation if the slide destroyed his house, this would not demonstrate that Ed's diverting the slide would be unjustified. His being justified does not mean that he is required to divert the slide; and if he chooses not to do so, then the matter of compensation would arise. The interests of justice can therefore be served in either of two ways.

¹³ I offer no arguments here in support of the proposition that people have a claim right to life, and that this includes a right to prevent their lives from being adversely affected. This contention strikes me as needing no argument–although an explanation is required of why it is not question-begging in the present context. I provide such an explanation below. Let me also emphasize that the argument being advanced here could have been formulated in terms of the familiar and widely accepted idea that people are morally required to prevent their lives from being adversely affected. I have not formulated the argument in this way because I do not believe there is such a requirement.

¹⁴ Jeff McMahan criticizes an account (‘Unjust Threat’) that resembles the one being developed here in certain respects (although the two accounts are distinct) on the ground that it ‘fails to justify self-defense in certain cases in which it seems justified’ (McMahan, 274). In support of this criticism, he imagines a situation in which ‘Tactical Bomber’ is justified in carrying out a raid even though it will harm innocent civilians; and he claims that Unjust Threat ‘does not permit the civilians to engage in self-defense against him’ (McMahan, 275). This last claim is open to two very different interpretations, however: (i) Unjust Threat implies nothing about the permissibility of the citizens' defending themselves, and (ii) Unjust Threat implies that their defending themselves is impermissible. If (ii) were true, then the position might be defective as McMahan claims (although see the discussion of ‘Objection (B)’ that follows in the text). Proposition (ii) seems clearly to be false, however, since Tactical Bomber is not an unjust threat and the view is therefore inapplicable to his action. Proposition (i) might very well be true, but this would imply only that Unjust Threat is not a comprehensive account of self-defence against IAs or AITs.

¹⁵ McMahan, 266 f.

¹⁶ McMahan quotes the following remarks by George Fletcher in support of his point:

[I]t is true that the defendant…‘acts’ so as to cause harm, but if the act is excused, the harm is no more attributable to the defendant than it is to the victim.…Excused harms are circumstantial, for they derive not from the responsible choice of the defendant, but the defendant's and plaintiff's being thrown together by circumstance. (Fletcher, George P., The Search for Synthesis in Tort Theory, Law and Philosophy, ii (1983)Google Scholar.)

But people can harm others voluntarily–or even deliberately—and be blameless for doing so; and such harm is surely ‘attributable to’ those who inflict it. Moreover, a deliberately caused but excused harm does ‘derive’ from the agent's ‘responsible choice’–unless the responsibility to which Fletcher refers is moral rather than causal. Indeed, Fletcher seems to conflate moral responsibility with causal responsibility, as if people who are blameless for what they do somehow fail to exercise their agency. In fact, however, people are blameless for most of what they do–not because they are – excused but because their actions are fully justified and sometimes even praiseworthy. Hence if Fletcher's remarks (and McMahan's reference to them) are meant to imply that no facts about an agent other than her culpability are appropriately ‘internal’—can render her ‘liable to attack or harm’ (using McMahan's language)–then the implication seems clearly to be mistaken.

¹⁷ The relevance of these ‘side-effects’ is noted in the articles cited in n. 3.

¹⁸ Note that Dot's justification might consist in a mere permission (and, indeed, I will argue presently that this is the case), so claiming that Dot is justified in causing Cal's death to save herself is not necessarily to imply that Dot's position is morally superior to Cal's — that their positions are morally asymmetrical–and that Cal is somehow morally vulnerable relative to Dot. If Cal were a righteous AIT, then his position would be morally superior to Dot's in the sense that she would not be permitted to kill him to save herself. For Dot to be (merely) permitted to defend herself, she needs only to be boosted to a position of moral parity with Cal, and I am claiming that Cal's unjustifiably forcing her to choose between her life and his provides this boost.

¹⁹ Objection (C) might be formulated differently. It might be based on the claim that actions cannot be unjustified (that is, impermissible) in virtue of the nature of their consequences unless the consequences are foreseeable from the agents' standpoints. If this claim is correct, then the proposed explanation of justified self-defence against wrongful IAs or AITs must include a foreseeability condition since, if the forced-choice situations created by the IAs or AITs are not foreseeable by them, their creating them is not unjustified. I will not attempt to deal with this form of the objection, since doing so would involve addressing an issue in moral theory that is well outside the scope of this paper.

²⁰ I pointed out earlier that Dot's being permitted to kill Cal in self-defence does not imply that her position is somehow morally superior to his. I should now emphasize that Dot's being merely permitted to defend herself, so that third parties are permitted to interfere, does not imply that her position is morally inferior to his. In fact, their positions are morally on a par relative to what each of them or a third party is permitted to do to save his or her life. In contrast, the positions of targets of culpable aggression can be regarded as morally superior to those of the CAs in two respects: the CAs are not permitted to kill their targets while the targets are permitted to kill the CAs, and third parties are obligated not to interfere. And the positions of innocent bystanders can plausibly be regarded as morally superior to those of other innocent people who can save their own lives only by fatally using the bystanders as shields but whose doing so is impermissible.

²¹ Like Jeff McMahan (see n. 8 above), Robert F. Schopp points to incompleteness as a defect in the account of self-defence against culpable aggression proposed above. Schopp regards this criticism as applying to a certain class of theories of self-defence, namely those rooted in principles of distributive justice. Schopp, Robert F., Self-defense, In Harm's Way, ed. Coleman, Jules L. and Buchanan, Allen, New York, 1994, pp. 255–89Google Scholar. Schopp's criticism is at least blunted, however, if I am correct in claiming that selfdefence against IAs and AITs is also justifiable on grounds of distributive justice, even though these are not precisely the same grounds that justify self-defence against CAs.

²² I am very grateful to John Deign, Tom Downing, Frances Howard-Snyder, Hud Hudson, and Larry Alexander for their valuable comments on earlier drafts of this paper. I am especially indebted to Jeff McMahan whose thorough and penetrating criticism of an earlier draft led me to modify it significantly, not only in regard to the arguments and explanations advanced, but also in regard to the positions adopted.