Baruch Brody and the principle of justifiable homicide

Abstract

In a series of papers in the early 1970s and in his important book Abortion and the Sanctity of Human Life (1975), Baruch Brody offered what remains to this day one of the most philosophically rigorous contributions to the debate concerning the morality of abortion and the ethics of homicide more generally. In this paper I would like to critically examine Brody’s argument that abortion is sometimes justifiable in some cases even when (1) one cannot claim self-defense, or (2) diminished responsibility, and (3) the abortion is a ‘killing’ rather than a ‘not saving.’ This justification, I argue, is limited to certain cases in which the life of the mother is at stake. The cautious principle which he finally formulates merits serious attention and consideration. While I find a great deal of value in Brody’s discussion, I will argue that there are several difficulties with the principle of justifiable homicide he constructs. Accordingly, I will further amend and supplement his final version by offering my own alternative principle.

Introduction

In a series of papers in the early 1970s and in his important book Abortion and the Sanctity of Human Life (hereafter ASHL) (1975), Baruch Brody has offered what remains to this day as one of the most philosophically rigorous contributions to the debate concerning the morality of abortion and the ethics of homicide more generally.¹ In this paper I would like to critically examine Brody’s argument that abortion is sometimes justifiable in some cases even when (1) one cannot claim self-defense,² or (2) diminished responsibility, and (3) the abortion is a “killing” rather than a “not saving.” This justification, I argue, is limited to certain cases in which the life of the mother is at stake.³ Brody begins by first laying out a standard anti-abortion argument that he finds promising. The argument is as follows: (1) all human beings have a right not to be killed except in the most extreme circumstances; (2) fetuses (and infants) are human beings⁴; therefore (3) abortion (and infanticide) is justifiable only in such circumstances.⁵ He then develops two cases and later variations of them in which it might seem morally permissible or even obligatory to kill someone. In neither case, however, does the individual killed violate anyone’s basic rights.⁶ Thus, no appeal can be made to the principle of self-defense to justify the killings. Following this, Brody next attempts to construct a principle that could justify both killings.⁷ The cautious principle which he finally formulates merits serious attention and consideration. While I find a great deal of value in Brody’s discussion, I will argue that there are several difficulties with the principle of justifiable homicide he constructs. In particular, I will find it necessary to further amend and supplement even the final version that he defends and offer my own alternative principle.⁸

Two test cases: the triggering mechanism and the surrounded village

I would like to begin by reviewing two cases in which Brody thinks homicide is permissible, apart from any appeal to self-defense. I will call this first case ‘the triggering mechanism case.’

By a series of accidents for which no one is to blame, it has come about that the five people in room R2 and the one person in room R1 will be blown up by a bomb in the next sixty seconds. The only way to prevent this is to defuse the bomb by blowing up its triggering mechanism which is in room R1. Unfortunately, this necessarily means that the person in R1 will also be blown up. In addition, the man in room R1 is having a massive coronary attack and will certainly die anyway within the same range of time. He is, however, the only one who will die; if you do nothing, all six people in question will die.⁹

In this scenario, the situation quickly turns tragic as there is no way to save everyone and even the most tolerable course of action necessarily involves killing someone. The second case, in which we once again confront a tragic set of circumstances, I will refer to as ‘the surrounded village case.’ Brody presents it as follows:

A small village is surrounded by a hostile group of brigands who demand that the villagers kill Pietro, an innocent villager whom they dislike. If the villagers do not do this, the brigands threaten to (and the villagers have every reason to believe that they will) destroy the village and everyone in it (including Pietro). The village is cut off from outside help and giving in to their demands is the only way to save the village.¹⁰

Here one could certainly understand why the villagers might kill Pietro. There is no lack of motive and they may feel that they have no other choice but to take his life in order to save themselves. But would their doing so be morally justifiable? Significantly, in his discussion of the case, Brody acknowledges the existence of intrinsic moral evils: “as for me, I am in agreement with those (like Anscombe and Geach) who maintain that there are some actions, like murder, that are never justified” [9, p. 20]. Despite this, Brody claims that the villagers’ killing Pietro is morally permissible and perhaps even obligatory. In particular, he disputes whether this particular killing is an instance of ‘murder’ insofar as ‘murder’ is unjustified killing and in this particular case no injustice is being done to Pietro since he is going to die anyway. Apparently, this claim strikes Brody as self-evident. The readers’ supposedly widely shared intuitions about Pietro’s case are, at any rate, assumed as the basis, in part, of subsequent argument. It seems, however, that the villagers are scapegoating and intentionally killing an innocent man (Pietro) and therefore cannot justifiably kill him.¹¹

Fortunately, though, I share Brody’s response to the triggering mechanism case. I would argue it is morally permissible to blow up the mechanism in room R1, even though doing so will kill the man trapped inside. So, while I object to his response to the second case, it still may be possible to make use of a version of a principle suggested by the first case. Thankfully, moral convictions or considered judgments are close enough to allow for this much. I hope to show, of course, that any ultimately acceptable principle of justifiable homicide distinguishes between the two cases. But I must first consider Brody’s own efforts to construct a principle that could justify killing in both cases and in certain difficult abortion cases as well.

Justifiable homicide: two case studies

Brody thinks the most plausible initial accounts of why it is permissible to kill in both cases are these. First (I),

In these cases, the person whose life you will be taking is going to die anyway. If you take his life, however, you can save those other lives. Taking his life is an optimal act since no lives are sacrificed and some are gained, and that is why it is okay to take his life. In general, it is permissible to take B’s life to save A’s life if B is going to die anyway and taking B’s life is the only way of saving A’s life [9, p. 111].

And second (II),

In these cases, it is a question of sacrificing one life to save many other lives. While the life to be sacrificed is tremendously valuable, so are all the lives that can be saved, and when we weigh them, we find that the many lives outweigh the one. This may, of course, seem like a harsh attitude, but it would be even harsher to forget about the many lives you can save by sentimentally refusing to take one life. In general, it is okay to take B’s life to save the lives of A and C (where A ≠ C) if taking B’s life is the only way of saving their lives [9, p. 111].

At first glance, both principles (I) and (II) have a certain initial appeal. Is it perhaps a matter of indifference which one to invoke? I would argue that there is actually a great deal at stake in choosing between the two principles. Significantly, (I) and (II) do not justify the same set of actions. As Brody acknowledges, (II) is basically a utilitarian principle which allows (and seemingly demands) taking lives to save lives. Whenever rights to life come into conflict, it directs one to maximize the number of lives saved. On this account, an act is right (and perhaps morally obligatory) which saves the greatest number of lives.¹² But (II), as a utilitarian principle, faces what Brody and I both regard as a fatal objection. As he notes, “it is unjust that the happiness of however many individuals should be brought about by causing suffering to a few, and all the more that it is unjust that life be preserved for many by being withdrawn from a few” [9, p. 16]¹³ In short, principle (II) allows and, even more strongly, requires one to violate basic human rights and to do what is unfair.¹⁴ As Brody notes later, “we must remember that our basic intuition is that one life cannot be unfairly exchanged for another” [9, p. 23].

And yet this concern is not obvious in the cases of the triggering mechanism and the surrounded village. In both scenarios, the consequentialist could argue that nothing unfair is done, since those who are killed would be killed anyway. In response, I think all that needs to be done to show how principle (II) is unfair is to alter the surrounded village case so that now Pietro will not be killed unless the villagers kill him. Even with this crucial modification, (II) still requires that the villagers kill Pietro. This seems plainly unfair and raises serious questions about the credibility of principle (II).

But once one alters Pietro’s case as I have proposed, the fundamental difference between principles (I) and (II) becomes clear. In particular, (I) does not allow the villagers to kill Pietro. After all, (I) requires that the one who is killed would die anyway. But this would not be the case in the revised scenario. It is only when the person to be killed would die anyway does Brody think that no injustice is done (Brody refers to this at various times as the “nothing is lost” principle) [9, p. 18]. Because of this he, rightly, I think, prefers principle (I) to (II). This is not to say, however, that (I) faces no objections.

Given the revised scenario, the consequentialist will doubtless pose the following questions. ‘Aren’t we being unfair to the rest of the villagers if we don’t allow them to kill Pietro and thereby save their own lives?’¹⁵ And ‘isn’t it far better to be unfair to one than to many?’¹⁶ Brody’s reply to this objection turns squarely on the distinction between ‘killing’ and ‘not saving’ and his lexical ordering of the obligation not to kill and the obligation to save lives. In response to Jonathan Bennett’s rejection of a meaningful moral distinction between killing a person and merely refraining from saving that person’s life,¹⁷ Brody argues that such a distinction is based on the more fundamental distinction between the negative prohibition not to take a person’s life and the positive obligation to save someone’s life. As he notes, “it is because the former is far stronger than the latter, and therefore nullifies the latter when the two conflict, that it is sometimes wrong to kill a person while permissible not to save another person” [9, p. 18]. In vital conflict cases, the first duty takes priority over the second, unless perhaps the person one kills has ‘nothing to lose.’ According to Brody, one must first avoid causing evil, e.g., death. Only then can one turn to preventing evil, often a less stringent and demanding obligation.

The argument Brody gives to support this ordering is not as developed as one might like. As he notes, the obligation not to kill is more commonly present than the obligation to save and is generally recognized as more important. It is implied by Brody that logical consistency would require one to recognize it here. The only way to check on this supposed pervasive character of the obligation not to kill, though, is to examine one’s own intuitions and moral and legal practices. And yet to some extent these intuitions and practices are just what are at issue. But to some extent, too, one’s convictions are shared and one’s practices are largely agreed upon. So, I do not think Brody can be accused of embracing a form of vicious circularity here.¹⁸

In addition, Brody strengthens the case for his lexical ordering by highlighting an important contrast between the obligation not to kill and the obligation to save. He rightly observes that, in general, B cannot kill A just to save B’s life, much less to avoid hardship, even though this would be preventing an evil. A must in some sense be violating B’s basic rights or, perhaps, going to die anyway. One does not, however, normally recognize any obligation for B to save A at the cost of B’s life or at the cost of considerable danger or hardship. Of course, if B has a special responsibility for A (such as marriage, a parent–child relationship, or certain professional roles such as a physician or police officer) this may be different. But the general point is this: what cannot justify killing can (sometimes) justify not saving. So at least in certain cases the obligation not to kill would seem stronger than the obligation to save. I cannot help but suspect, though, that there is more to say about the basis for the distinction between killing and not saving than Brody considers here.¹⁹ Later in the paper I will argue that the moral relevance of this distinction is related to whether one’s action can be construed as an intentional choice against an innocent human life. But a closer look at this distinction must wait, for I have only just begun to introduce the objections to principle (I).

Some initial objections and refinements

As it now stands, principle (I) needs further internal specification. As has been shown, it allows taking A’s life to save B’s life if doing so is the only way of saving B’s life, provided that A will die anyway. If something like (I) is to be acceptable, however, one must restrict the time period for A’s death. For eventually, one could object, we are all going to ‘die anyway.’ Unless (I) is to remain too open-ended then, it must be temporally limited. Brody proposes doing so by adding the rather vague phrase “in a relatively short time.” This yields the following principle (I'):

It is permissible to take A’s life to save B’s life if A is going to die anyway in a relatively short time and taking A’s life is the only way of saving B’s life [9, p. 116].²⁰

According to Brody, then, principle (I') still allows both for blowing up the triggering mechanism and, seemingly, for killing Pietro, thus satisfying the brigands’ evil demands. In particular, Brody argues that in neither case will one be treating unfairly the person they kill. Each would very shortly be killed anyway, thus, each has ‘nothing to lose.’ At the same’time, (I') makes it clear that one cannot just kill any A to save B, even when this is the only way to save B. For A has, in most cases, a great deal to lose.

Still, Brody’s phrase ‘in a relatively short time’ is by design rather open-ended. If there are difficulties in applying principle (I') because of it (as is likely), what should one do? Brody rightly suggests that one consider not just the actual length of time but rather the unrealized goods or opportunities that are at stake. One must ask how important to A is the time, otherwise open to him, that one would take from him. As the surrounded village case shows, the actual time may be very short and still enough for A to ‘put his life in order,’ whether literally or figuratively. Given this way of clarifying its temporal restriction, (I') seems a clear improvement over (I).

Having made this distinction, Brody still worries that even principle (I') is inadequate. In particular, it faces the following objection: suppose not just A but both A and B alike will die in a relatively short time barring a drastic intervention. Suppose, too, that killing A will allow for saving B and that killing B allows for saving A. Now if one takes principle (I') as it stands, one could choose either to kill A or to kill B. In this scenario, one might well be inclined to let one’s personal preferences, natural affections, or utility guide them. But Brody opposes allowing any of these considerations to be decisive in such cases. In particular, he is concerned that doing so might leave one vulnerable to the complaint that one is still unfair to the one they kill. On the other hand, to let both die because of a paralyzing concern for ‘fairness’ cannot be required. How then is one to avoid this moral quandary?

Brody’s suggestion is that principle (I') be amended to require in such cases a fair random choice of who is to be killed. This yields (I''):

It is permissible to take A’s life to save B’s life if A is going to die anyway in a relatively short time, taking A’s life is the only way of saving B’s life and either

• (i) taking B’s life (or doing anything else) will not save A’s life; or

• (ii) taking B’s life (or doing anything else) will save A’s life but one has, by a fair random method, determined that one should save B’s life rather than A’s life [9, p. 23].

Brody finally concludes, then, that (I") adequately states when, apart from self-defense, one can legitimately kill a human being.²¹ Given principle (I"), it is sometimes permissible to intentionally kill an innocent human being in a very limited range of circumstances.²²

A first potential alternative

This analysis of Brody’s discussion now raises the question of whether one can sometimes justify homicide, even when construed as an intentional killing, by an appeal to principle (I"). I would argue here that the answer is a qualified ‘yes.’ As will be shown, however, this is true only for a very restricted range of cases. After I first indicate when Brody’s principle can justify homicide, I will explore whether the range of cases in which homicide now seems justifiable is really as narrow as Brody concludes. Later, surprisingly perhaps, I will argue that (I") is in some ways not restrictive enough.

First, consider when principle (I") permits abortion. There are cases, though exceedingly rare, when both the unborn baby and the mother will die unless one or the other is killed. This killing may result from, say, crushing the infant’s head such as in a craniotomy or performing an operation on the mother which will kill her but free the fetus.²³ According to Brody, in such cases (I") is satisfied and an abortion is justifiable if it is chosen by a random procedure (i.e., something like flipping a coin or fair lottery which supposedly guarantees fairness, although Brody never offers a substantive argument for it).

Moreover, principle (I") is consistent with the standard argument restricting abortion outlined at the beginning of the paper. Unfortunately, I find that, with one important exception, any relaxation of (I") relies on frankly problematic arguments. Nonetheless, I think these arguments are worth exploring and are independently interesting. In developing them, especially given their problematic character, two considerations should be kept in mind. First, these arguments should reflect one’s recognition of the value of a human life and must not conflict with the strong reluctance to take human life that is the basis of the standard argument outlined at the beginning of the paper. Secondly, they should not result in principles which themselves prove too restrictive and hence only frustrate efforts to protect and promote the value of human life.

In arguing, first, that principle (I") is too restrictive, I will adopt Brody’s own ethical methodology. My plan is to introduce certain test cases to show the implausibility of his final principle. These cases will elicit, I hope, widely shared agreement. I will then propose a first alternative to (I") that can adequately deal with these test cases, if not with others.

Consider, then, the case of two courageous soldiers, A and B. Both soldiers see a triggered grenade that will kill them both within seconds unless one shields the other by hurling himself on the explosive. Is there a fair random method to determine who will be killed and who will be saved? It seems not, but surely that need not prevent either soldier from saving the other. So, one might conclude here that such a method is not always required. If nothing else, lack of time precludes it in this scenario. But more than this, I think, the very idea of a random method in some situations, at least, seems artificial and forced.

As a second example, consider a variation on the surrounded village case. The brigands demand that the villagers kill either Pietro or Maria (assume for now that it is permissible to kill one or the other.) Suppose, too, that there are these morally relevant differences between Pietro and Maria. Now, consider the following contextual points:

• (1) Maria feels terrified and betrayed at the thought of being killed by the villagers; Pietro does not feel this way.

• (2) Maria would be killed in a much more painful way than Pietro—for the sake of argument assume the brigands demand this.

• (3) Maria’s death would be far more emotionally painful for the villagers.

• (4) Maria, unlike Pietro, would leave dependents.

Is it really fair to require a random procedure for deciding whether to kill Pietro or Maria? In this regard, utilitarians sometimes appeal to considerations of fairness to resolve moral dilemmas or ‘ties.’ Thus, when two courses of action will each produce a like amount of pleasure over pain, that which more equally distributes the pleasure is preferable. Could one cautiously adopt a comparable tie-breaking strategy to resolve this particular moral dilemma? Even if considerations (1–4) are utilitarian in spirit, it may be possible to use them to resolve some cases in which fundamental human rights seem to conflict. One can say, given these four conditions that the fair thing to do ‘on the whole’ is to kill Pietro rather than Maria. In this scenario, utilitarian considerations may prove to be decisive. In addition, the demand for a random decision method may seem inappropriate or insensitive to many.

But, if all this is so, by a parity of reasoning one need not adopt a random choice method to justify abortion when both a mother and her unborn child will die if nothing is done. At the same time, I am not sure that one can never choose to save the fetus. After all, there are, generally at least, the following morally relevant differences between the mother and her fetus to consider:

• (1) Only the mother could feel terrified or betrayed at the thought of the physician’s killing her.

• (2) The mother will presumably die a more painful death than would the fetus.

• (3) The mother’s death may be more emotionally painful for others than is the unborn infant’s.

• (4) Only the mother could leave dependents.

Nonetheless, there seem to be cases where it is legitimate to require a fair random method to decide who is to be killed. One can imagine, for example, yet another variation of the surrounded village case. Now one must kill either Pietro or Jack, and both are, say, solitary and stoic characters who keep to themselves. An adequate alternative, then, to (I") must retain the random decision method but allow for cases in which morally relevant differences between A and B obviate such a method. I propose, accordingly, that we simply amend (ii) in (I") to read:

taking B’s life will save A’s life but one has, by a fair random procedure, determined that one should save B’s life rather than A’s life, unless there are morally relevant differences between A and B that make this unnecessary.

The amended principle (I"), hereafter (III), allows for abortion whenever both the mother and the unborn child will die if nothing is done and the abortion will save the mother. Thus, it is significantly less restrictive than (I"). At least this much of a relaxation of (I") seems defensible. Subsequent relaxations, as I have warned, will be more problematic.

A second alternative

Another abortion dilemma clearly raises the question of further relaxations of Brody’s (I") and some revision of my principle (III) as well. It is this sort of case that I would like to examine next. Sometimes it happens that unless one aborts the fetus, the mother will die; but if one does not abort the fetus though the mother will die, the fetus can be saved. Such cases arise, of course, only when the fetus is naturally or medically viable. The crucial feature of these cases is that while one can only save one of two lives, unlike the earlier cases, one need not take anyone’s life to do so. Whatever one’s intuitions, neither principles (I") nor (III) allow for abortion in such a case. This restriction is defended on the ground that killing the unborn child would be unfair, moreover, it could not be excused by saying that the fetus would die soon anyway. Nevertheless, some might claim that an abortion could be justifiable in such an extreme situation. Thus, it seems that even principle (III) is perhaps too restrictive. In arguing that (III) is too strict, I shall once again appeal to what intuitively seem to be counterexamples to it. Then I will explore what sort of alternative principle would be consistent with one’s intuitions about my proposed counterexamples. Finally, I will consider whether this new alternative principle would in some cases justify abortion where (III) would not.

First, consider a variation on the case of the two courageous soldiers, A and B. Suppose that B notices an already triggered explosive. Fortunately, he can escape its blast by leaping out of the way. Tragically, however, it will kill his comrade A, who is, say, pinned down under some wreckage, unless B hurls himself on the explosive. In this regard, many may think it legitimate, even praiseworthy and heroic, for B to shield the grenade with his body and save his fellow soldier. Moreover, one takes this view of the matter even if B foresees that doing so will very likely cost him his life.

But does B truly merit praise here? Could not a life have been saved, and perhaps this is the most one can hope for, without B’s killing himself? If so, is not this just what principle (III) requires? Indeed, one cannot say on B’s behalf that he would have been killed anyway. In fact, many may feel inclined to defend and perhaps even praise B’s action as heroic. Moreover, one might strongly object to the idea that B ‘killed himself’ in this context. While it is very likely B will die if he hurls himself on the explosive, it is implausible to claim that B’s action is a suicide.²⁴ B’s action seems misdescribed here. For suicide is at least morally ambiguous whereas B’s heroism is not.²⁵ One factor crucial to this case, and to our reaction to it, is that B does not directly kill himself. The killing here is indirect and unintentional. That is, B does not kill himself either as an end in itself or as a means to an end. Rather his end is to save his comrade A, and his means is to interpose himself between A and the explosive in order to save his comrade by shielding him from the blast. In addition, B’s death is not required to save A’s life. What does this case reveal? It seems to show, although principle (III) does not reflect this, that an unintentional and indirect killing which saves the life of another, even when the one killed would not soon die anyway, is sometimes justifiable. We must remember that in this scenario the victim volunteers himself. This is clearly of some significance; a fetus or a patient in a PVS state, for instance, could never do this.

Consider now a second example of an intuitively legitimate indirect killing inspired by Judith Jarvis Thomson’s influential essay “A Defense of Abortion” [81].²⁶ One could call this the case of the ‘unlucky bystander.’ In this case, A, an agent of the Society of Music Lovers, is attempting to kill B and harvest his vital organs in order to save the lives of others. Acting in self-defense, B is about to hurl a grenade at A as it is the only weapon available. But just before throwing the grenade, B notices C, an unlucky bystander, walk into the blast radius. Is B’s throwing the grenade morally justifiable if he foresees that C may also be killed? I think that many will likely consider this to be a case of justifiable self-defense. Of course, his action is hardly obligatory. Many people might even think more highly of B if he sacrifices himself to A and spares C. Nonetheless, I think most would acknowledge that B’s action is morally permissible as a form of self-defense.

But according to principle (III) it is not. For (III) seems not to allow taking C’s life unless C would die anyway in a relatively short period of time. But if B does not throw the grenade, there is no reason to think this condition holds. Once again, it seems that B can act legitimately because killing C, the feature of his action in question here, is an indirect killing. But why should this be morally relevant? An indirect killing, to be sure, is not consequentially different than a direct killing. That is, its immediate causal consequences are the same, although more remote psychological effects may not be. But an indirect killing is intentionally distinct and this can be morally significant. In particular, one’s intentions reveal one’s attitude toward the value and worth of human life, and it is this fundamental value that undergirds the ethics of killing.²⁷ In the case at hand, it seems clear that C’s death is neither B’s end in throwing the grenade nor, strictly speaking, the means to his end of saving his own life. In particular, B would rejoice if C escaped without harm or injury. So, B’s intention does not reveal any disrespect for the value of a human life.

But the case of the unlucky bystander, notably, does significantly differ in two ways from the case of the courageous soldiers. Here, C, unlike the soldier, does not volunteer his life. Moreover, here, B (indirectly) kills not to save another but rather to save himself from being killed and having his organs harvested. Because of these morally relevant differences, there does not seem to be anything heroic about what B does. Still, there is nothing illegitimate either. But despite their differences, both cases indicate that principle (III) may be too restrictive. Indeed, there seem to be cases of justifiable indirect killing that it excludes. Any serious alternative to (III) must incorporate one’s most widely held moral convictions and considered judgments about such cases. Because of this I propose to examine another principle (IV).

It is permissible, either directly or indirectly, to kill A to save B’s life if A is going to die anyway in a relatively short time, taking A’s life is the only way of saving B’s life and either:

• (i) taking B’s life (or doing anything else) will not save A’s life or

• (ii) taking B’s life will save A’s life but one has, by a fair random method, determined that one should save B’s life rather than A’s life, unless morally relevant differences between A and B make this unnecessary.

• (iii) It may be permissible, indirectly, to kill A if in doing so one saves B’s life even if A is not going to die anyway in a relatively short time.

Clearly, there is a need to explore the general import of what principle (IV) says about indirect killing. But this must wait for a time. I think it best to first consider, in a tentative way, the question of whether (IV) might justify abortion in some cases in which (III) would not. After all, it is in the hope of shedding light on difficult vital conflicts that I have come to formulate principle (IV) in the first place. It would seem that (IV) might justify abortion where (III) would not only if some abortions are indirect killings. To determine whether a particular abortion is an indirect killing, one needs to know whether the death of the fetus is the agent’s end or a means to his or her end. But one cannot know this unless one knows the agent’s intention.

In addition, knowing an agent’s intention requires insight into the context in which the agent acts. To some extent, while structuring acts, intentions are also structured by acts and the context and causal structures in which one acts.²⁸ Crucially, such causal structures constrain the reasonability and intelligibility of one’s choices and actions. So, while knowing another’s intentions can be difficult in some circumstances, one’s intentions are by no means wholly private and inaccessible to others.²⁹ Still, it seems that in principle an abortion to save a woman’s life can result in the death of the fetus without that death’s being intended. Even if the death is foreseen, it need not be the woman’s or physician’s means or end. If the unborn child were somehow to survive, they would both rejoice. In this respect, they are like the man acting in self-defense who will rejoice if the unlucky bystander is not killed and like the brave soldier who covers the bomb and who will rejoice if he survives the blast.

It seems then that principle (IV) could allow an abortion to save the mother if it were an indirect killing. It could allow for this even if, were it not aborted, the unborn baby would survive. But (III) requires that the fetus would shortly die anyway. So (IV) further relaxes Brody’s limitations on abortion that my own principle (III) reflects. Nonetheless, (IV) is still in accord with the standard argument’s insistence that homicide be limited to extreme circumstances and that one acknowledges and honors the value of a human life. To give some idea of how principle (IV) might be applied and what its limitations are, I think it would be useful to sketch three increasingly difficult obstetrical cases raised by Paul Ramsey.³⁰ Perhaps one could argue that one or all of them are indirect killings, but the arguments quickly become very difficult. For one thing, it is not obvious that, like the triggering mechanism case, each can be described as a single action having both a good and a bad effect.

In the first case a pregnant woman has advanced uterine cancer and will die unless her uterus is removed. In the second case a pregnant woman has a cancer that will yield only to a series of, say, a dozen chemotherapy treatments. If the therapy is undertaken, her fetus might perhaps survive three treatments; but will surely die before she is cured. In the third case a pregnant woman has an aneurysm that threatens her life. It is in the aorta, which has become weak and distended. To repair it the doctor must first cut through the uterus, killing the fetus.

Given principle (IV), how is one to treat such cases? If one construes killing the fetus in such cases as only an indirect killing, then (IV) allows one to kill the fetus even if he or she would otherwise survive. Brody, as I noted earlier, does not allow for this. Of course, if in any of these cases the fetus would shortly die anyway, then (IV) allows one to kill the fetus even if one’s action is a direct killing. Brody, too, would allow for this. But the worst dilemma arises when, given such a case, one supposes that the fetus would survive were it not killed in the course of saving the mother. If in such a case the killing is construed as direct and intentional, it is not permissible, even given (IV). In the last part of the paper, I will consider what might be said for the view that such killings are not direct and hence are morally permissible.

In employing, as I do, a distinction between direct and indirect killing, one introduces double effect reasoning in some form or another. This approach (hereafter DER) says that one may sometimes licitly perform a single act having both a good and a bad effect if the latter is merely a foreseen but unintended consequence or side effect of one’s act.³¹ Something like it seems necessary to supplement Brody’s principle (I") and accommodate some fairly ‘settled’ moral and legal judgments and practices.³² But there is another important task to do first.

A third alternative

In trying to formulate a principle for cases of justifiable homicide not assimilable to self-defense, I have examined a succession of candidates. So far principle (IV) seems the most promising. The other principles considered, except for (II), seem too restrictive. Each forbids killing in one or more cases where it seems morally permissible. But a principle can be both too restrictive and too lax. I argued earlier that Brody’s revised principle (I") suffers from just this two-fold flaw. Finally, while principle (IV) is an improvement over (I"), in that it is not too strong, it is unfortunately, like (I"), too weak.

This time there is no need to construct a series of counterexamples to support my argument as Brody himself has already provided a very helpful one. For my purposes here, I need only recall his original surrounded village case. The heart of the matter is this: Pietro will be killed anyway, but the villagers can save themselves if they kill Pietro.³³ As shown above, Brody claims that it is clearly permissible and perhaps even morally obligatory for them to do so. In response, I would contend, minimally, that it is not obligatory for them to kill Pietro and doubtfully permissible. Of course, both principles (I") and (IV) admit Brody’s view. That is, both seem to allow the villagers to intentionally kill Pietro. I would argue here, though, that this merely shows that each principle is flawed and needs to be amended. For my purposes, though, I need only consider principle (IV), since it is in other respects superior to (I"). I think the best approach, then, is to proceed by carefully examining Pietro’s case as a way of developing an alternative to (IV).

The easiest claim for me to establish should be that the villagers have no moral obligation to kill Pietro. In this regard, anyone who maintains that in a particular case there is an obligation to kill a human being must, it seems to me, accept the burden of proof. Brody, though, never really elaborates just why he thinks the villagers may have such a moral obligation. His argument, moreover, cannot be based on the utilitarian principle that the right action is always to maximize the number of lives saved insofar as this is a principle he explicitly rejects. Perhaps he is tacitly relying on principle (I") to indicate not only what is morally permissible but, at least in this case, what is obligatory. But this seems a questionable assumption, either with respect to (I") or (IV). In particular, it overlooks, in the case at hand, Pietro’s right to self-defense. Pietro, as far as one may know, is innocent of any crime and he is not violating anyone’s basic rights.

Specifically, it is not his fault that the brigands are attacking and terrorizing the village. Moreover, one cannot say that he ought to accept his punishment insofar as he is morally blameless and has not harmed anyone. Rather, Pietro has as much right to defend himself against the villagers as he does to defend himself against the brigands. In this regard, there is considerable debate about a strict logical correlativity of rights and duties.³⁴ In particular, some scholars have pointed to cases in which an individual has a right although no other individual has a corresponding duty. But I think the correlativity thesis does in fact hold in the case at hand. If Pietro has a right to self-defense, then the villagers, as well as the brigands, have a duty not to harm Pietro. It hardly seems then that they also have a moral obligation to kill him.

But the same argument from Pietro’s right to self-defense establishes a second and stronger claim. Not only is it not obligatory for the villagers to kill Pietro it is also not permissible to do so. Thus, Pietro’s prospects, while still bleak, seem slightly better. In addition, I think one can draw on a consideration Brody himself proposes to construct a second argument for both claims and give independent weight to Pietro’s appeal to self-defense. If the villagers do in fact kill Pietro, they might attempt to excuse their deed by arguing that he would have been killed in a relatively short time anyway and therefore their actions were not really unfair or unjust to him. But, as Brody recognizes, the time period in question is only a rough way of measuring the losses a person will suffer if he is killed before he would have otherwise died. If this time period is very short, it may seem plausible to suppose that the person will not suffer any significant losses.³⁵ But one must look carefully at individual cases. For the time period may be very short but the goods lost enormously significant.

Clearly a crucial question is who should decide whether the losses to be suffered are significant. The likely answer to this question is, it seems, that the person who would suffer the losses should decide, at least when a rational decision is possible. An infant or a comatose patient could not, for example, make this decision. But in this case Pietro clearly can. Therefore, I would argue it is ultimately a matter for Pietro to decide whether the time the villagers would deprive him of is significant. In fact, he might argue that it is very significant. Perhaps he might think that he has a chance to fend off his attackers and escape with his life. Or perhaps he might like to use this time to reconcile with estranged friends or family members. Or perhaps like the protagonist in Joseph Conrad’s Lord Jim, he might even see this as a chance at redemption for previous mistakes in his life and that it is a matter of integrity for him to struggle, even to the point of death, and defend the community against the brigands [128]. If the villagers were to betray and rob him of this test of integrity, he might see this as one of the greatest betrayals and disappointments of his life as well as a serious injustice.

So, given Brody’s own reading of the phrase ‘in a relatively short time’ that occurs in (I") and is retained in (IV), it is not clear that it is either obligatory or permissible to kill Pietro. Still, I think an alternative to (IV) is needed that explicitly prohibits killing Pietro. Contrasting Pietro’s case with the original triggering mechanism case might be helpful here. In this regard, I would like to argue that the differences between the two scenarios can serve as guides in constructing the alternative principle needed. In addition, the alternative to (IV) must still allow destroying the trigger mechanism even at the cost of an innocent life. I think, too, that a careful comparison of the two cases will again underscore the import of the distinction between direct and indirect killing.

Consider, then, the differences between the two cases: I find at least three that are morally significant. First, in the triggering mechanism case one is responsible only for an indirect killing. In the first case, one saves five people one by blowing up a mechanism and in doing so unfortunately kills one other person. But here I would argue that killing this person is neither one’s end nor means. The killing, strictly speaking, is not required to save the others. All that is required is disabling the triggering mechanism. Should the man in room R1 somehow survive, and whose death is not required to save B’s life, everyone would rejoice. But Pietro’s case is very different insofar as it is clearly a case of direct killing. Here it is Pietro’s death, strictly speaking, that is required. The brigands want him dead; and nothing less than this will do. While it is true that Pietro’s death is not the villagers’ ultimate end (saving their own lives is) Pietro’s death cannot but be construed as a means to their end. If Pietro somehow eludes them, they cannot rejoice; they must literally hunt him down and kill him.

In addition, the presence or absence of complicity also distinguishes between the two cases. Moreover, this factor of complicity is based on the agents’ intention. Pietro’s death is at least in part the brigands’ purpose. The villagers, although one may lament their tragic misfortune, choose to cooperate in this evil business. They are guilty of formal cooperation and are complicit in Pietro’s death, because they directly and intentionally kill him, albeit with reluctance. They do the bidding of the brigands and hope to save their own lives by killing an innocent man. But formal cooperation and complicity, as far as we know, is absent from the triggering mechanism case. In this case, the death of the trapped man serves no one’s evil purpose. He is no one’s victim, but a victim of tragic circumstances only. But to underscore the import of intention, one might suppose someone wanted him dead and contrived to bring about the dilemma being faced. Could one still blow up the mechanism or would one then be guilty of complicity? In this scenario, I would argue one could still blow up the mechanism because one does not directly or intentionally kill him, and so one does not join their intention with that of his murderer.³⁶

A third difference between the two cases, though doubtless related to the first two, is suggested by the question of self-defense. I have argued that Pietro can appeal to the basic right of self-defense to resist the villagers. But what of the man in room R1? To illustrate the difference between the two scenarios, suppose that the man in room R1 sees what is about to happen. Can he appeal to the principle of self-defense to stop someone from destroying the mechanism? If he can, how is one any more justified in killing him than the villagers would be justified in killing Pietro? In this scenario, it seems that if the man in room R1 knows that there is nothing that can be done to save his life, he (unlike Pietro) would have no reason to stop it. As Brody notes, he is experiencing a massive heart attack (presumably brought on by the intense stress of the moment) and will ‘certainly die’ in a matter of seconds. Therefore, one does not violate any basic right of his, given Brody’s scenario and his final principle (I").³⁷

There are, then, three important and related differences between the two cases. The first is that killing Pietro is a direct killing, whereas destroying the mechanism results in an indirect killing. The second is the factor of formal cooperation and complicity, present only in killing Pietro. Finally, since Pietro’s basic rights are being violated by the villagers, he can appeal to self-defense to resist them. But the man in room R1, whose life is shortened by seconds and who is a victim of tragic circumstances only, cannot claim his basic rights are being violated. What is, I suspect, the most basic difference between the two cases is that the one is a direct killing and the other an indirect killing.

But to draw this distinction, I must show that killing Pietro is a means to an end, and hence directly intended, but that disabling the trigger mechanism in the first scenario is not. Thus, my analysis of the two cases leads immediately to the question of determining, in a general sort of way, what is a means to an end and what is not. In this regard, I would argue that that the killing of the man in the room is an effect of destroying the mechanism rather than, strictly speaking, the means of saving five lives. But, it also seems, killing Pietro is the very means of the villagers’ saving their lives by complying with the brigands’ evil demands. If this was not the case, the question of their complicity in an evil deed would never arise in the first place.

Two factors especially merit our attention here. First, in Pietro’s case one cannot speak of a single act’s having two effects. Rather, there are at least two acts in question and two sets of agents as well. The first act is the villagers’ killing Pietro (Brody does not specify exactly how this comes about and this is morally significant). The second is the brigands’ sparing the lives of the villagers. This factor of intervening agency is important for the individuation of acts. As Edmund Pellegrino and Daniel Sulmasy have argued, if there are intervening agents or causes the proper moral category is cooperation or complicity.³⁸ Here killing Pietro is a distinct act and cannot but be construed as a means. In contrast, one can speak of blowing up the mechanism as a single act having two effects: the saving of five persons and the killing of one other person. In this scenario, the killing of the one seems to be an effect rather than a separate act that is the means to an end.

Second, we should focus on what the agent intends to do to carry out his intention in each case. Whether someone’s death is predictable and, given current medical or scientific technology, causally necessary is a separate question. In Pietro’s case the agents intend to save their lives by complying with the brigands’ evil demands. Thus, they must seek Pietro’s death. It is logically impossible for them to carry out their intention without killing Pietro. It is precisely his death that must be their means.³⁹ But in the second case, one intends to save lives by destroying a mechanism. In this scenario, it is possible to carry out the intention of disabling the mechanism and saving lives without anyone’s death being intentionally sought. Unfortunately, to be sure, one foresees that achieving one’s end causally involves a killing. But I think this killing, unlike killing Pietro, is not a means to an end. Hence it is not directly intended.

Enough has been said now, I think, to show the basic differences between Pietro’s case and the case of the triggering mechanism. I have faulted both Brody’s (I") and my own principle (IV) for not adequately distinguishing between the two cases; now it is time to offer a principle that does so. This new principle must rule out killing Pietro but must also permit destroying the triggering mechanism. I now propose the following:

• (V) It is permissible, indirectly, to kill A to save B if:

• (i) A is going to die anyway in a relatively short time;

• (ii) taking A’s life is the only way of saving B’s life;

• (iii) taking A’s life does not intentionally serve anyone’s evil purpose to kill A (e.g., scapegoating, exemplary killing, vengeance, personal enrichment, material gain, killing for the sake of pleasure or thrill, racist, sexist, ableist, and/or eugenicist reasons)⁴⁰; and either:

• (iv) taking B’s life (or doing anything else) will not save A’s life, or.

• (v) taking B’s life will save A’s life but one has, by a fair random method, determined that one should save B’s life rather than A’s life, unless the morally relevant differences between A and B make this unnecessary (here considerations of innocence either in the etymological sense of ‘not harming’ and/or moral blamelessness would be paramount).⁴¹

It may be permissible, moreover, to indirectly kill A if in doing so one saves B’s life even if (i) does not obtain.⁴² A good example of this would be turning the wheel of the trolley to the right and killing the one worker rather than the five. A strength of principle (V) is that it allows for indirect killing while ruling out scapegoat cases involving the intentional killing of the innocent that (IV) might permit. In this way it is more restrictive than the earlier (IV). But it is as flexible as (IV) with respect to indirect killings. And of particular importance, I know of no abortion dilemma that principle (V) would treat differently.

In this regard, some might find it tempting to try to reconstruct this principle to explicitly forbid any direct killing whatsoever.⁴³ In addition, my arguments might at some points hint at such a restriction. But I think it would be unwise for at least two reasons. First, the distinction between direct and indirect killing (or directly and “obliquely intended” to use Philippa Foot’s terminology) can at times be very hard to apply in specific cases [134]. The use of this principle, like the use of any specified moral principle, will require very close attention to the particularities of each case as well as practical wisdom and good judgment. What is an indirect killing in particular cases will sometimes be difficult to discern and it would be misleading and even dangerous to try to make the distinction bear more weight than it has to and there are legitimate concerns about abuse of the direct vs indirect and/or intending vs foreseeing distinctions.⁴⁴ In particular, I would regard the attempt to characterize the craniotomy as an indirect killing to be implausible.⁴⁵ In thinking through these issues, I find Elizabeth Anscombe’s example of transfixing a man with an arrow to be helpful and illuminating.⁴⁶ According to Anscombe, one cannot intend only to transfix the man but not to harm him, on the ground that it is precisely the transfixing and not the harming that realizes one’s end (for example, self-defense or the killing of an enemy combatant in a justified military conflict). In this scenario, the act of transfixing him is eo ipso (by that very fact) killing him.⁴⁷ Likewise, in the craniotomy case, the surgeon cannot reasonably claim that he intends only the “changing or reshaping of the dimensions of the baby’s head,” and not the baby’s death.⁴⁸ For to act in this way is eo ipso to kill, that is, the killing is in the very nature of the procedure and inseparable from it.⁴⁹ As Luke Gormally notes:

a surgeon giving a truthful, clear-headed account of what he is intentionally doing would not allow himself to be confused by the fact that he did not desire the death of the child—for he had chosen an intrinsically lethal way of proceeding in order to save the life of the mother—nor would he allow himself to be distracted by thoughts of possible worlds in which entities had different constitutions from the world in which he was obliged to act or in which it was open to him to employ different but, in reality, unavailable procedures [165, p. 101].

Finally, I would like to briefly address the much publicized case which took place at St. Joseph’s Hospital in Phoenix, Arizona in November 2009. A young pregnant woman with a history of moderate but well-controlled pulmonary arterial hypertension was admitted to the hospital with worsening symptoms. After testing, the medical team concluded that she had life threatening pulmonary hypertension. After consultation with the woman and her family, the ethics consult team argued that an abortion in such a scenario should be classified as indirect insofar as the goal would not be to end the pregnancy but rather to save the mother’s life. In response, I would argue that principle V does not allow for an abortion here because the exact medical procedure performed (a placentectomy) involved the removal of an integral vital organ, namely, the placenta, from an innocent human being in a manner that would necessarily cause his death and is thus a direct killing.⁵⁰ In addition, no one involved in the case disputed that the eleven week old child was healthy and would thus violate condition (i). It is possible that in the near future the use of artificial wombs could help resolve these very difficult vital conflicts.⁵¹

Second, a restriction against any direct killing may seem to prohibit killing in self-defense against an unjust material aggressor, killing an enemy combatant in a justified military conflict,⁵² or capital punishment.⁵³ In these instances, I would argue that direct killing can be sometimes justified because the (1) killing can be construed as self-defense or defense of the community against an unjustified material aggressor (2) individuals in question are no longer innocent in the sense of not harming and/or morally blameless and (3) such killings may be necessary to protect the innocent depending on the circumstances of the particular case. Of course, in the case of military conflict, other just war criteria such as just cause, right intention, and proportionality would have to be met as well. The further question of whether someone is a state actor such as a soldier or police officer with unique responsibility for the common good would be another important morally relevant factor to be considered. In this regard, Aquinas has been interpreted by some scholars as claiming that a killing in self-defense need not be direct.⁵⁴ Moreover, his justification of self-defense had a profound influence on early formulations of DER as well as subsequent developments.⁵⁵ Still, there are difficult questions about how to interpret Aquinas on this point and it seems better to keep the principle of self-defense and DER separate as each faces enough challenges without sharing the other’s.

Conclusion

This paper began by asking whether and how Brody’s argument that sometimes abortion is morally justifiable, even when (1) no appeal can be made to the principle of self-defense or (2) to diminished responsibility and (3) the abortion is a killing rather than a not saving. He has argued that homicide, and abortion in particular, is sometimes justifiable apart from self-defense in a sharply restricted range of cases. The intuitive idea that gives rise to his principle is this: sometimes one does not treat an individual unjustly or unfairly, even if they kill him or her; for sometimes a person has nothing to lose [9, p. 18]. Hence, Brody reasons, if tragic circumstances dictate that the only way to save a life is to take the life of another, we may sometimes do so if the person we kill would very soon die anyway.

After carefully refining this principle through the examples of the triggering mechanism and the surrounded village, Brody applies it to a number of difficult abortion cases. Ultimately, he concludes that his principle justifies abortion when it is necessary to save the mother’s life and the infant would die anyway or the infant could be saved but only if his or her mother were killed. In this last case, the person to be saved must be decided by a fair random method. A key requirement in every such case is that no injustice is done to the individual killed. Brody’s final principle (I") is his best statement of just when this key requirement is met. As I have shown, however, (I") is both too strong and too weak. It is too strong because there are cases of justifiable indirect homicide that it forbids such as the heroic solider throwing himself on the grenade to save his comrade. And it is too weak because there are also troubling cases of direct homicide that it might permit such as the killing of Pietro by the villagers. In particular, Brody seems to recognize no moral difference between scapegoat killings involving the intentional killing of the innocent and indirect killings. Because I find Brody’s principle (I") inadequate, I have introduced and examined three alternatives to it: principles (III), (IV), and (V), respectively.

Finally, my own principle (V) justifies an abortion whenever it is necessary to save a mother’s life, unless the abortion directly kills a healthy fetus that would otherwise survive. In exploring cases outside the context of abortion, I also acknowledged situations in which homicide is justifiable, even apart from self-defense. It is, for example, permissible in the triggering mechanism case to kill the one in saving the five. In emphasizing that such a killing is indirect, I again draw on DER and in particular the distinction between the intended and foreseen consequences or side effects of an action. But it is not just that my normative claims require something like DER. I have also argued that Brody ultimately goes astray partly because he slights the important distinctions which DER emphasizes and which play a key role in a consistent and defensible ethics of homicide.