Nov 19, 2010

Follow-up on Double Effect and Consequentialism

My last post argued that consequentialists can accommodate the principle of double effect in their moral theory: contrary to popular wisdom, double effect is not an anti-consequentialist principle. In this post, I want to touch on some important points that I left out of the first post.

1. It's generally true that my argument cuts against popular wisdom, though there are some exceptions. Amartya Sen agrees that consequentialists can't rule out, as a matter of principle, including actions from the way states of affairs are characterized.

2. I've suggested that consequentialists can accommodate double effect by thinking about the comparative disvalue of killing in order to save. What, precisely, gives rise to this disvalue? Broadly speaking, we can think of the disvalue as deriving from facts about the agent (an agent-centered view) or facts about the victim(s) (a victim-centered view), or both (a hybrid view). 

One victim-centered explanation would be that the five who are saved in the organ harvest case are comparatively worse off than they are in the trolley case in virtue of the fact that the agent killed in order to save in the first case but not the second. We might think, for example, that the dignity that each of the five possess is degraded because the one person’s life is ended as a means to saving their lives, and that dignity is in fact a relevant value in a consequentialist framework; it is worse to live having been saved impermissibly than to live having been saved permissibly, we might suppose.[1] Alternatively, or perhaps in addition, we might think that the value of justice is better preserved when the five are saved justly rather than unjustly. If any victim-centered or agent-centered account holds, we can explain why consequentialists ought to care about double effect. Further, since explanations for the value differential need not be rooted solely in the character of the agent, we can preempt the sort of objection to double effect that has been made by T.M. Scanlon, to the effect that it conflates certain kinds of agent evaluations with evaluations of actions.[2] I won't take a stand here on which explanation of the disvalue consequentialists should endorse.

3. While Sen doesn't think we can rule out the kind of consequentialism I'm considering here in principle, he does argue against what he calls "broad consequentialism," i.e. a consequentialist theory that takes into account the way outcomes are brought about, as I have been proposing, at least as a subject of discussion. Sen thinks it is implausible that there could be sufficient disvalue in someone’s killing one person in order to save twenty so that it would be better, in terms of broad consequentialism, not to kill the one in order to save the twenty. However, Sen’s consideration of broad consequentialism in this case is strictly limited to a theory on which there are agent-centered but not victim-centered sources of disvalue. Provided that there are some victim-centered sources of disvalue when some are killed in order for others to be saved, it is much more plausible to think that broad consequentialism should rule out killing a few to save many.

4. The kind of consequentialist theory I'm discussing here has an interesting feature. In many cases, as the number of those who would be saved unjustly if the agent killed in order to save increases, it becomes increasingly tempting (as some see it) to have one’s moral theory permit killing in order to save. One virtue of my proposed account is that, as the number of those who would be saved by the killing increases, it is plausible that the disvalue associated with the degradation of justice or dignity also increases or at least remains constant, so that the temptation to render killing in these cases permissible does not increase as the number to be saved grows. This means that killing a small number to save a large number can be ruled out, even on a consequentialist theory, without introducing explicit side constraints such as inviolable rights. This is a neat result. While I am not a consequentialist, I think the version of consequentialism I've discussed here is much stronger than some competitors and not prone to some of the more obvious objections. It becomes much more difficult to reject act-consequentialism, for example.

[1] Anscombe suggests something along these lines: “… In an attenuated sense it can be said that something that belongs to, or concerns, one is attacked if anybody is unjustly attacked or maltreated.” Elizabeth Anscombe, “Mr. Truman’s Degree,” in Ethics, Religion and Politics (Minneapolis, MN: University of Minnesota Press, 1981): 69.

[2] Scanlon makes this argument in Moral Dimensions: Permissibility, Meaning and Blame. Judith Thomson also makes an argument along these lines.


  1. It might be important for you to specify exactly what you mean by 'consequentialism.' A key feature of the doctrine of double effect is the distinction between the intended and the merely foreseen. The term 'consequentialism' was originally coined to refer to ways of thinking about ethics that deny the moral significance of this distinction. So in that sense, consequentailists could not accommodate double effect. I'm not saying that you are wrong, though. No doubt the meaning of 'consequentialism' has changed somewhat over the years. But if it doesn't mean what it used to, it might be helpful to specify precisely what it does mean.

  2. Interesting post, Matthew. DR is right: to the extent that most consequentialists reject the doing/allowing distinction, I wonder where they would locate the disvalue of killing one to save five (at least in cases where one is not killing to save five people from being killed). You seem to suggest that one's life is worse if it's been saved by a killing. This involves controversial psychological claims regarding bad conscience, remorse, indebtedness, and so on (see for example how those who benefited from organ donation tend to feel indebted). But are these cases in which while being worse than simply continued, these lives are still better than not saved?

    I'm sympathetic with your attempt, and I tend to think having been saved thanks to a killing makes one's life worse; but I would agree with the consequentialist that it is still a life, hence a source of value. Finally, the consequentialist could further claim that the life was not 'unjustly' saved since, by hypothesis, the killing was, if not required, at least permitted.

  3. DR,

    Excellent question. I think you're right about consequentialism as understood by, say, Anscombe in "Modern Moral Philosophy." Indeed, she says this is what separates consequentialism from utilitarianism.

    Contemporary usage is somewhat different, as you suggest. My sense is that consequentialism tends to be seen as an umbrella under which utilitarianism falls. Roughly speaking, a consequentialist theory identifies some value and makes the criterion of right action a function of that value. Typically the function is maximizing, but that need not be the case.

    One of the reasons I think my suggestion to the consequentialist is useful is that it allows her to circumvent the most obvious objection to, say, act-utilitarianism: that it would sanction killing in order to save a greater number. This is an objection that typically inspires the move to more sophisticated theories like rule-utilitarianism. I think the act-utilitarian has a way to resist that move, along the lines I've suggested. If we want to avoid worrying about what a consequentialist theory is, I can recast the argument to be about utilitarianism. The basic thought is the same.

  4. Nicolas,
    Thanks for your comment. While I spoke of the disvalue as stemming from the agent or the victim(s), I meant to be neutral about whether the disvalue in question was badness simpliciter or badness for the individuals in question. Your worry, I think, is addressed to the latter. Some ethicists have worried about goodness and badness simpliciter. John Taurek, for instance, claimed not to see any sense in claims that something is good or bad without being good or bad for a person, group of persons, or purpose. (See "Should the Numbers Count?") You raise a good point, and one that I'll need to consider further.

  5. Matt,

    Yes, contemporary usage is different from Anscombe's. That's why your idea is interesting. I don't think it matters much what the theory in question is called, but if you wanted to develop this idea more I do think you would need to specify somehow or other what the theory is that you are talking about. This isn't in any way a criticism of your basic thought. I meant it as a constructive criticism, however obvious and tiny.