¹ Of course, not all bystanders are innocent – just as not all innocents are bystanders. This is so obvious that I shall permit myself to speak of killing bystanders and killing innocents and allow the context to reaffirm that what is always meant is killing innocent bystanders (or bystanders whose guilt does not render it permissible to kill them). For a vindication of grouping innocent aggressors, but not guilty bystanders, with guilty aggressors, see Jeffrie, Murphy, “The Killing of the Innocent,” The Monist, vol. 57 (Spring 1973).Google Scholar

² As such an account develops and illuminates the specific judgments which it takes as its preliminary data, we can be more secure in those initial judgments and more confident about extending those judgments to initially more puzzling cases (or pairs of cases). (It is because the development of such a justifying account is already in progress that, later in this paper, I permit myself less guarded judgments about particular cases.) And these more extensive judgments themselves can credibly guide refinements in the theoretical account. The hope, of course, is that the judgments and the justifying account will enjoy sufficient independent plausibility that they will reinforce one another, and not just be two ways of proclaiming the same prejudices.

³ In seeking to allow certain abortions traditionally disallowed under DE, Germain Grisez has argued that an effect is not intended as long as an alternative course of action which had the same acknowledged effect and lacked the putatively merely foreseen effect would have been pursued, if it had been available. For instance, if a woman having an abortion would have placed the fetus in an artificial womb were one available, then the death of the fetus from the actual abortion is not intended. (Grisez means to imply – probably correctly – that for most actual abortions the artificial womb would not be exploited. Thus, for Grisez, most fetal deaths remain intended.) But this criterion surely seems to imply that the death of Alyosha in Cases 3 and 4 is unintended. And these implications cannot be welcomed by Grisez or any advocate of DE. See his “Toward a Consistent Natural-Law Ethics of Killing,” American Journal of Jurisprudence, vol. 15 (1970).Google Scholar

⁴ See Philippa, Foot, “The Problem of Abortion and the Doctrine of Double Effect,” J., Rachels, ed., Moral Problems, 2nd ed., (New York: Harper and Row, 1975) pp. 61–62.Google Scholar

⁵ The language of doing something to a person in order to deal with a threat versus doing something to (or with) the threat follows Judith Thomson in her wonderfully provocative “Killing, Letting Die and The Trolley Problem,” The Monist, vol. 59 (Spring, 1976).Google Scholar One of the most suggestive points in favour of DE is how philosophers who do not want to be advocates of it, such as Thomson (and myself), find themselves returning to pronouncements which look to all the world like formulations of DE. (For Thomson's rejection of DE, see her “Rights and Deaths,” Philosophy and Public Affairs, vol. 2 (Winter 1973).Google Scholar

⁶ Any theory of responsibility for harm that does not have bizarre implications needs something like a distinction between standard and non-standard (or proximate and non-proximate) results. If Alice can distract Alfred only by chopping off Alyosha's finger and proceeds to do so for that reason, then Alyosha's loss of his (rightfully held) finger is intended. But if this loss leads to Alyosha's death only through a highly circuitous, non-standard, and lengthy causal route, then Alice cannot be said to have violated Alyosha's right to life.

⁷ These cases are discussed throughout “The Problem of Abortion and the Doctrine of Double Effect” and “Killing, Letting Die and the Trolley Problem.”

⁸ Cf., with the case in Foot, p.68. The specification of the releasing agent plays a crucial role in the further discussion of this case in Part IV.

⁹ For part of such a defense, see my “Bad Samaritanism and The Causation of Harm,” Philosophy and Public Affairs, vol. 9 (Summer 1980)Google Scholar, and “Deontologism, Negative Causation and the Duty to Rescue,” E., Regis, ed., Gewirth's Ethical Rationalism (Chicago: University of Chicago Press, 1984).Google Scholar

¹⁰ When Foot proposes that the killing/letting die distinction (and the associated greater stringency of negative duties over positive duties) do the work that some have assigned to DE, she does not clearly acknowledge that, celeris paribus, this substitution leads (again, on the assumption of fetal personhood) to an even stronger anti-abortion stance than that adopted by the advocate of DE. Foot seems to be distracted by her concern with the even more special class of cases in which both parties will die unless one of them (e.g., the fetus) is killed, and the special reason why, in these cases, killing the fetus is permissible. Cf., Foot, pp.69–70.

¹¹ This move to “proportionate reason” is attempted in Germain Grisez, “Against Consequentialism,” American Journal of Jurisprudence, vol. 23 (1978)Google Scholar, especially pp.49–56. My first encounter with the idea was in conversation with Joseph Boyle.

¹² I do not test subsequent doctrines against Case 6, the abortion case. For our sense of what is permissible in that case depends upon many further factors, e.g., our views about the personhood of fetuses and about the moral relevance of this purported personhood.

¹³ As stated, this right of self-preservation would be both a first and second order right.

¹⁴ If one were to adopt the widely held thesis that there is no morally significant difference between knowingly failing to prevent an injury and causing that injury, then it would seem to follow that all such failures constitute attacks against which self-defensive force is legitimate. The adoption of this thesis is another way to collapse the distinction between self-defense and self-preservation. For a survey of arguments for this thesis, and a critique of it, see my “Bad Samaritanism and the Causation of Harm.”

¹⁵ In an earlier essay, “The Moral Basis of National Defense,” I was more satisfied with something like the CD approach, just because I focused primarily on killings which are responsive to attacks by other agents. Nevertheless, the policy implications drawn in that essay are the same as those which should be drawn from the present essay. See R.Poole, Jr. Poole, Jr., ed., Defending the Free Society (Lexington, MA: Lexington Books, 1984).Google Scholar

¹⁶ On the claim that even a moderately permissible right of self-defense (i.e., one encompassing the killing of individuals posing innocent threats) would not vindicate killings of the sort needed by Alice in 1′, see Nancy, Davis, “Abortion and Self-Defense,” Philosophy and Public Affairs, vol. 13 (Spring 1984).Google Scholar “It is considerably less obvious how we are to justify killing to preserve our own life when the attacker has done nothing at all to threaten us but is, instead, a passive threat: someone whose mere movements qua physical object or mere presence constitutes a threat to our life” (p.190).

¹⁷ Deflection cases, including the various runaway trolley cases, are discussed in Thomson's “Killing, Letting Die and the Trolley Problem.”

¹⁸ I assume that the “proportionate reason” argument which was discussed at the close of section II does not work so as to allow the advocate of DE to escape acceptance of Act 5. But, if it does work out, then the advocate of DE must disallow Acts 2′, 2″ and 5′.

¹⁹ Hart, H.L.A. and Honoré, A.M., Causation in the Law (London: Oxford University Press, 1959)Google Scholar especially Chapter III, “Causation and Responsibility,” pp.58–78, and Chapter VI, “The Law of Tort: Causing Harm,” pp.126–170.

²⁰ Ibid., p. 72.

²¹ However, Hart and Honoré also write that “the free, deliberate and informed act or omission of a human being, intended to produce the consequence which is in fact produced, negatives causal connection” (p.129, my italics). Here, intention seems to be required for negativing causal connection. But, perhaps, they would say that the fact that an act has some intended consequence is sufficient to negative the causal connection between the background setting for the act and all of the act's foreseen consequences. So, if I intend the good consequence of escape from the concentration camp, my deliberate act of escape negatives the causal connection between the setting of the traps along my escape route and the foreseen injuries I suffer in travelling that route!

²² Contrast this with the case in which Jones has been promised $10,000.00 if he consumes poison and, to his great good fortune, I come along and dump some otherwise unavailable poison into his coffee. He then drinks the potion before him (partially at least) in order to collect on this promise. Here I will not have imposed the poisoning upon Jones.

²³ Even such intention may not shift responsibility to the intermediate agent. See Case 7 below in which Jim's intentional killing is thoroughly and purposively orchestrated by the commandant.

²⁴ The AC “vindication” of an agent looks more and more like a theory of excuse rather than a theory of any sort of justification. But the same point can be made about the DE and the CD “vindications.”

²⁵ Thomson, pp.215–216.

²⁶ Suppose that an employee of a mountain-climbing resort has told a guest to wait for her at point x and now, if she deflects the boulder from its path toward Alice and Allen, it will strike the guest at x. This employee, it seems, is less justified in deflecting the boulder than she would be if she had not given this instruction to the guest. In her “Killing, Letting Die, and the Trolley Problem,” Thomson also discusses the importance of the history behind the presence of the individual toward whom the trolley may be deflected. She contrasts a case in which this individual has been promised safety on the alternative track with a case in which this person is a trespasser on the alternative track. See pp.210–211.

²⁷ Compare this with the case and discussion offered by Bernard Williams in J.J.C., Smart and B., Williams, eds., Utilitarianism: For and Against (London: Cambridge University Press, 1973) pp.98–118.Google Scholar In Williams's case, the person Jim must kill in order to save many is among the many. That person will die no matter what. This special factor suggests a number of special ways to vindicate this killing. To avoid these routes, I have excluded this special factor from all the cases in this essay. Any doctrine which depended on this special factor would, of course, be of no use in vindicating the killing of innocent Soviet citizens by the U.S. government for the sake of saving inhabitants of the U.S. from death, injury, or loss of liberty.

²⁸ Thomson, p.208.

²⁹ And, again, as illustrated in the case of Jim, in some circumstances, an intermediate agent may aim at the death of a bystander and still not acquire responsibility and blame for that death.