There is a temptation to think that the principle of double effect rules out the need for an “ends-justify-the-means” approach in ethics, and to think that double effect is always capable of adequately distinguishing between ethical and unethical practices. In this paper I argue that the principle of double effect cannot replace an “ends-justify-the-means” approach in all circumstances. This is because there are some cases in which failure to adopt an undesirable means will result in an unacceptable effect. This is seen mostly clearly in the issue of amputation, when a bodily evil is inflicted for the sake of a greater good. I will begin my argument by explaining how the principle of double effect is supposed to work, and proceed by critiquing what I take to be the problems with its application.
Double effect is often used to justify or explain actions which result in harm or have other negative consequences. Joseph Mangan argues that the principle has existed implicitly within the Judaeo-Christian tradition since the time of the Hebrew Scriptures, citing this example from I Maccabees 5:63,
The Syrians, in the war against him, fastened wooden towers on elephants’ backs, and each beast then charged into battle with a thousand armored warriors surrounding it. Eleazar, Judas’ second-youngest brother, lost his life in 163 BCE when he stabbed an elephant from underneath. In dying, the beast fell on top of him and crushed him.
Mangan explains that Eleazar’s act was courageous because he foresaw that he would be killed, but his act was not suicide, which would have been impermissible.
We might also consider another example of double effect in a different circumstance: A boat is sinking and one of the cabins is filling with water. There are some people in the cabin who are going to drown unless a passage is sealed off. It seems likely that sealing off the passage will block the escape route for other people and cause them to drown as a result. The principle supposedly says that the action of sealing the door is not impermissible because it aims at a good (saving life) while foreseeing that a likely or inevitable bad effect (people drowning) will follow. In this way, double effect purportedly makes a distinction between causing a bad effect, and intentionally doing a bad action as a means to a good end. While the former is sometimes permissible, the latter is never permissible. This allows some ethicists to coherently argue that while a foreseen bad effect (e.g., people dying) may be permissible in given circumstance, a bad action which is intentionally chosen as a means to some good end (e.g., murdering people) is never permissible. For example, someone could agree with the ethics of closing the ship’s door, in order to block the water, in order to save some passengers, even if the action was done with the recognition that some other passengers would probably not escape. Yet that same person could still coherently pose an ethical objection if the captain consented to intentionally drown a few passengers as a means of appeasing some pirates who were threatening to kill the rest.
Mangan has explicitly characterized four stipulations of double effect. These stipulations, if met, render an action with a bad effect permissible:
1. That the action in itself from its very object be good or at least indifferent.
2. That the good effect and not the evil effect be intended.
3. That the good effect be not produced by means of the evil effect.
4. That there be a proportionately grave reason for permitting the evil effect. 
I believe this characterization of double effect’s stipulations is correct. Moreover, each stipulation seems to have been fulfilled in the example cases mentioned above.
For Eleazar, killing the elephant was a good object. In achieving that object, his death was foreseen but not intended. His death was not a means to killing the elephant because his death played no essential role in killing the elephant; rather, it was a foreseen consequence. Provided that there was a proportionately grave reason for permitting Eleazar’s death, it seems the action could be rendered permissible by Mangan’s stipulations, even while still considering suicide to be impermissible. Similarly, but perhaps more controversially, blocking the passage in the sinking ship aimed at the good of saving some passengers. The deaths of the other passengers were not intended as a means to the good of saving life (as they would have been in the case of appeasing the pirates); however, they were a foreseen effect of saving some passengers by closing the door. Moreover, the risk of death and the need to preserve life seems to be a grave issue. While the pertinent concern of how many people are at risk in each course of action, as well as the likelihood that those people would escape from the ship still needs to be considered, I will not take up that discussion in this paper. Rather, having set the stage for discussion, I will now turn to the principle’s abuses.
In “Theology and the Four Principles of Bioethics,” John Finnis provides one account of double effect’s application in practical situations. Yet, in doing so, he has helpfully illustrated why the principle is sometimes problematic.
In the article Finnis explains his formulation of the principle of double effect by saying that it
Is reducible, on analysis, to two propositions: (i) neither as an end nor as a means may one choose to destroy, damage or impede any instance of a basic human good; (ii) a result of a choice is not a chosen means merely because it is foreseen as a probable or certain result but because it is part of the proposal which one shaped in deliberation and adopted in that choice.
Finnis argues that amputation can be “morally good” or “obligatory” in order to promote health. For this reason, amputation is permissible if the detriment to the limb’s function is not intended, and even a healthy limb may be amputated for the good of the whole. Yet he seems to contradict himself by maintaining that one cannot choose directly to impair a “function,” because, as he explicitly states, one would be intending to damage a basic human good. He says,
Even when well-motivated, every choice directly to impair a function involves an intention to damage a basic human good, and is always wrong. But it can be morally good and even obligatory to remove a part of the body when this removal, of itself, protects or promotes health, and one does not intend the detriment to function as a means to any end, but only accepts it as a side effect. For example, an infected limb or nonvital organ may rightly be amputated or excised when that is necessary to prevent the infection from doing great harm to the whole body. Even a healthy part may be removed if doing so has natural consequences which are necessary for the health of the whole body and cannot be effected in another, less detrimental way.
According to Finnis, directly impairing a function through amputation intentionally goes against a basic human good. But Finnis suggests that one may undergo amputation for the good of the whole body. So unless Finnis has poorly articulated his position, he seems to have suggested that one may choose to act against a lesser good (a limb’s function) for the sake of a higher good (the sake of the whole body). From this account, we might assume that the factor that determines whether or not the limb can ethically be amputated is what the amputation is done for. Yet both of these cases constitute an “ends-justify-the-means” approach. Regardless of how we characterize Finnis’ notion of amputation, it seems clear that the detriment to function which occurs in the limb cannot be considered a side effect (i.e., a consequence which is permissible by double effect). This is because, in Finnis’ words, it is a “part of the proposal which one shaped in deliberation and adopted in that choice.” In the example of amputation that Finnis has given, excision of the limb is clearly not a side effect, and neither is the limb’s excision separable from the action of impairing it. (This will become clearer as my argument develops in the following pages.)
Before I proceed, allow me to restate precisely what I take to be the problem with Finnis’ account. He has maintained that directly impairing a “function” goes against a human good when the impairment to function is either an end or a means to an end, and this is the case even when the end is well motivated. However, he has also said that one can remove a part of the body if it is done for the health of the body which cannot be effected in some other way. What seems clear is that removing a part of the body is impairing the function. Moreover, in amputation, removing part of the body serves as a means to the end of promoting a person’s holistic bodily health. So if we allow amputation, we are in some way acting against one human good in order to promote a more complete human good.
Now the problem here is not that amputation is somehow unethical. I think that most healthcare workers would refuse such an ethic and it is for that reason that I will leave all arguments against amputation alone. The problem is that we allow amputation but need to develop a thoughtful conception of how amputation is allowable and whether or not it necessitates an “ends-justify-the-means” approach. My argument is that Finnis’ version of double effect either falls into an “ends-justify-the-means” approach without admitting it, or else it effectively bans all amputation. Precisely why Finnis fails should become clear when I elucidate my response to possible defenses of his argument.
Finnis might be defended in a number of ways. First, it might be argued that removing part of the body is not equivalent to impairing function in every case. Second, someone might argue that the negative effect does not constitute a means to the end of health—rather, it constitutes an unintended side effect which occurs when one pursues the good of heath. Thirdly, someone might say that the morally impermissible thing is the intention to damage a human good, but amputation involves no such intention, so it is permissible.
I will alternate between treating each of these defenses of Finnis, starting with the third. This third defense of Finnis fails because the means to an intended end are themselves intended. This shows the relevance of intention when considering the principal of double effect. While Finnis might try to escape this criticism by saying that one must not intend “the detriment to function as a means to any end, but only [accept] it as a side effect,” this raises the following question: can one really say that detriment to function is not intended when one has a limb cut off? I would argue that one cannot. Someone who thought otherwise would also have to say she could smoke a cigarette without intending to light or burn it.
Since Finnis thinks amputation is permissible if it benefits health and if “one does not intend the detriment to function as a means to any end, but only accepts it as a side effect,” it seems he might be saying that the detriment is merely a side effect and not a means. This is the second defense mentioned above. Yet if the action of amputation is considered in minute increments of time, so as to separate the act of healing the person from the act which causes detriment to the arm’s function, then the order of events does not place the detriment to function in the realm of a side effect or unintended consequence; rather, impairment occurs prior to any benefit. This means that impairment of function is a means to the end of health, since one has to concede that the arm’s function was permanently impaired before the beneficial result was achieved—it was impaired as the knife passed through the muscle and bone, and prior to the moment that the patient was made well. On Finnis’ own terms, then, the impairment constitutes a means to the end of health. Since Finnis thinks that doing damage to the body’s function is “always wrong” and the damage to the limb is “part of the proposal which one shaped in deliberation and adopted in that choice,” it seems that Finnis implicitly accepts that the good end sanctifies the bad means; though he tries to hide this behind the guise of double effect.
Yet I still need to respond to the first defense of Finnis mentioned above. For someone might insist that saving a life, or promoting the end of health, is still the action which is chosen. And they might conclude that the detriment to function is not chosen, but merely accepted as a side effect. This is similar to the second defense but slightly different. Yet G. E. M. Anscombe has successfully articulated why one cannot intelligibly make such a claim. Because a single action has multiple descriptions under which it is intentional, she explains, “you cannot just pick one of these and claim to have excluded others by it.” What this means is that in cutting the arm for amputation, you are depriving it of its function. Again, one who thought otherwise would have to say that she could effectively use a cigarette to appease her nicotine craving without actually smoking it. While she may not want to smoke the cigarette, her idle wanting has nothing to do with her intentional action if she lights the cigarette, puts it to her lips, and inhales. This intentional action constitutes smoking, just as amputation constitutes an intentional deprivation of function.
Anscombe’s exploration of the amputation issue in her essay “Action, Intention and ‘Double Effect’” raises similar problems. She concedes that amputation is an evil and concludes that the evil is inflicted “for the sake of a greater good.” In this respect, she accepts the “ends-justify-the-means” approach that Finnis tried to reject. Yet Anscombe qualifies her acceptance by saying, “the evil is all in the particular thing that is purposely done, but the evil of it and the intentionalness do not determine the quality of the act. That is principally determined by what it is done for . . . what will give it its moral character (and maybe a lovely one) is what you do it for.” It is here that Finnis’ articulation of double effect, along with the principle’s ability to coherently account for the permissibility of amputation, seems to crumble. This is because Anscombe is right in her conclusion that cases like amputation necessitate an “ends-justify-the-means” approach and are not somehow defensible by double effect.
Anscombe herself concedes that “if deliberately cutting off a hand is so far forth only pre-moral evil, then can’t the same be said of killing your uncle, or aborting your baby or . . . ?” For this reason, she sees it as a problem that justified amputation is “a case in which an intrinsic evil is inflicted for the sake of a more valuable good.” Nevertheless, it is a problem that she thinks can be rectified, since “the evil and the good [of amputation] concern the bodily condition of a single human individual. [And] he occupies a pre-eminent position in questions about good and bad action.” This is to say that because only the good of one person is involved, the situation is less clouded.
Continuing this line of thought, Anscombe gives another reason why amputation might be justified while still maintaining some notion that the ends do not always justify the means. Her reason is that a man’s decision to have his arm amputated is an indeterminate action-description that cannot be applied to other kinds of action—presumably it differentiates actions like amputation from actions like murder, abortion, or euthanasia. To explain, she illuminates the distinction between an act of a human and a human action. An act of a human occurs when a human has no agency. (To see this, consider the amputation example, but imagine the surgeon was drugged into compliance, and his arm was moved by some external force in order to perform the amputation—this action occurs without agency.) On the other hand, a human action would occur if the surgeon himself performed the amputation under normal circumstances. From this, one will see that every human action is an act of a human, but not every act of a human is a human action. For this reason, considering the distinction between act of a human and human action, Anscombe says, “not all action-descriptions which . . . describe human action, are such that whatever acts fall under them are good or bad human actions.” In other words, they might be indeterminate action-descriptions.
By these criteria, it could be argued that actions like amputation are bad qua act of a human and not bad qua human action. Anscombe argues this point as follows:
There might be a good human action . . . which by sheer accident is an inadvertent cause of some evil. Here some might say that the act of a human being has something bad about it. . . . But the human act has not got something wrong with it qua human act from being an involuntary cause of some evil. That respect of badness does not belong to it as a human act. For being the involuntary cause of something does not belong to it as a human act.
Thus, it is possible that an act of a human can still have a moral action-description, though the human being in question is only responsible for the act in the sense that “responsibility” means “causality.” In this case, one might say “impairing function is bad,” but a human action like amputation is not bad per se. This makes the notion of ‘amputation’ qua human action become somewhat neutral, though it could be bad or good qua act of a human.
This account is interesting and probably helpful in circumstances where there is a question of agency; however, it ultimately is most plausible to say that when a doctor amputates, the doctor actually does have a kind of total agency. Therefore, she intends the evil of damaging the limb qua act of a human and qua human action (though to arrive at a greater good of course). It is for this reason that the doctor uses an ends-justify-the-means approach by doing surgery, which is necessary for the health and good of the patient.
Anscombe actually seems to concede this point, saying, “there are several kinds of action, which, if they are done intentionally, are evidently evil actions, no matter what they are done for. The good end only sanctifies such means as either are good considered in themselves, or would naturally fall merely under a neutral action-description (amputation, when justified, is an exception to this ‘only’).” The parenthetical statement is one which Anscombe penciled into the margin of her lecture notes in “Action, Intention and ‘Double Effect.’” It indicates that even Anscombe is unconvinced about double effect’s applicability when it comes to amputation, since the evil of damaging a limb cannot be construed as good or neutral. Thus, she thinks that amputation is an evil action which is acceptable based on the action’s end goal.
Anscombe’s conclusion is that double effect is applicable in situations where the question of agency is unclear, or debatable. She thinks that double effect may be formulated coherently if it includes three features:
- The principle of side effects: an action which produces a bad effect is not excluded from the list of possible actions merely because it produces the bad effect. It may still be permissible.
- A person may never do an action which has an intrinsic certainty or great likelihood of being morally bad (e.g., murder/killing an innocent person) for any reason.
- In Anscombe’s own words, “It will be apparent that this principle tells you rather what you can’t do than what you can. Also, it is particularly devised for the causing of death; causing other harm is not covered by it. I think these are not faults.”
When considering the different ways that double effect plays out in practice, it certainly seems as though Anscombe’s approach more reasonable than that of Finnis. Yet Anscombe does not escape a serious problem. How is it possible that the ethics change between issues of life or death and issues involving bodily harm? Is it sensible to hold a principle that forbids an “ends-justify-the-means” approach when considering issues of life and death, but not when considering amputation? How would one know ahead of time that a specific case is life or death? Moreover, if we consider Anscombe’s first reason for making amputation an exception to the principle of double effect (the fact that amputation involves someone seeking his own personal holistic good), it seems that this would mean euthanasia should be justified too. After all, in both amputation and euthanasia one person is seeking that holistic “good” for themselves and no one else. To say euthanasia does not contribute to the holistic good while simultaneously maintaining that amputation does contribute is to side-step the entire conversation about the permissibility of a bad means because of a good end.
In these and other similar cases double effect seems to collapse into an “ends-justify-the-means” mentality. As a result, the principle seems prone to being misapplied and confused. From these considerations it should be clear that double effect is useful in helping to consider the relation between what is intended and what is an unintended but foreseen effect. The principle, if adopted as a rule or maxim which excludes an “ends-justify-the-means” approach, will inevitability lead to inconsistency or absurdity. In some cases, one needs to evaluate the goodness or badness of the means in terms of the ends.
. J. T. Mangan, “An Historical Analysis of the Principle of Double Effect,” Theological Studies 10 (1949): 41–61.
. J. Finnis and A. Fisher, “Theology and the Four Principles of Bioethics: A Roman Catholic View,” in Principles of Health Care Ethics, ed. R. Gillon (Chichester: John Wiley & Sons, 1993), 31-44.
. G. E. M. Anscombe, “Action, Intention and ‘Double Effect’” in Human Life, Action and Ethics, ed. M. Geach and L. Gormally (UK: Imprint Academic, 2005), 223.
. Ibid., 218.
. Ibid., 217.
. Ibid., 218.
. Ibid., 217.
. Anscombe thinks that talk of pre-moral evil may avoided by instead speaking of indeterminate action-descriptions. (Amputation is considered by some to be a pre-moral evil. That is, the moral goodness or badness of an amputation is not implicit in the act itself. The act is an evil, but not morally evil. Instead, the moral quality of the act is given by what the act is done for. Anscombe dislikes the notion of pre-moral evil because it is too easily applied to other cases like killing, which could easily justify violations of what she holds to be absolute prohibitions like murder and abortion.)
. If this is unclear, consider that Anscombe’s example of the distinction between act of a human and human action is the contrast between tripping and scratching one’s head. Ibid., 208.
. Ibid., 210.
. Ibid., 212.
. Ibid., 218-219.
. Ibid., 225.