Monday, June 06, 2005

Sacrifice and Separate Persons

Rawls famously complained that "Utilitarianism does not take seriously the distinction between persons." The idea is that, just as we think that later benefits can compensate harms to an individual, so utilitarians believe that benefits to one person can somehow make up for harms to another. But there is no super-person who receives this compensation. Utilitarianism is "thus" grounded on an illusion.

This strikes me as a pretty poor argument. The problem, of course, is that utilitarianism does not assume that any such super-person exists. Rather, the theory rests on other grounds - namely, the notion that each person's interests matter equally. As Hare writes:
It is indeed rather mysterious that critics of utilitarianism, some of whom lay great weight on the 'right to equal concern and respect' which all people have, should object when utilitarians show this equal concern by giving equal weight to the equal interests of everybody, a precept which leads straight to Bentham's formula and to utilitarianism itself.

It should hardly be necessary to spell this out. To have concern for someone is to seek his good, or to seek to promote his interests; and to have equal concern for all people is to seek equally their good, or to give equal weight to their interests, which is exactly what utilitarianism requires. To do this is to treat others' interests in the same way as a prudent person treats his own interests, present and future... To do this is not to fail to 'insist on the separateness of persons'. (R.M. Hare, 'Rights, Utility, and Universalization: Reply to J.L. Mackie', in R. Frey (ed.) Utility and Rights, p.107)

Nozick's objection echoes Rawls:
To use a person [for another's benefit] does not sufficiently respect and take account of the fact that he is a separate person, that his is the only life he has. He does not get some overbalancing good from his sacrifice (Nozick, ASU, p.33.)

But this is foolishness. No-one is claiming that the demanded sacrifice is for his own good. Rather, it is to benefit someone else in greater need, another person for whom their life is the only life they have. Nozick's egoistic objection is patently question-begging. Further, as G.A. Cohen asks, "if such sacrifice and violation are so horrendous, why should we not be concerned to minimize their occurrence?" (S-O,F&E, p.32) (This echoes Parfit's argument that common-sense morality is self-defeating.)

There is, however, a more charitable interpretation of the objection. We begin by noting that utilitarianism makes the normative claim that boundaries between lives lack moral significance (rather than the absurd descriptive claim that no such boundaries exist!), and we ask why this is so. Rawls' objection, then, is that the utilitarian's methodology (appeal to the impartial spectator who sympathetically identifies with everyone) leads them to neglect the descriptive fact. If the ideal spectator imagines himself as everyone at once, he might forget that they are different people, and that benefits to one will not compensate another for the harms they suffer. If this were the explanation, it would undermine the utilitarian's normative position. But, as we have seen, this is not the explanation -- as is obvious from the fact that many utilitarians employ the methodology of a detached rather than identifying spectator, and there is no reason why the former should suffer such confusion.

So, cutting to the core of the issue: Can it be right to burden someone merely to benefit someone else? Derek Parfit clarifies the issue:
We can first distinguish two kinds of weighing. The claim that a certain burden factually outweighs another is the claim that it is greater. The claim that it morally outweighs another is the claim that we ought to relieve it even at the cost of failing to relieve the other. Similar remarks apply to the weighing of different burdens, and to weighings of burdens against benefits.

An obvious case of a benefit factually outweighing a burden would be if everyone would prefer to have both rather than neither, i.e. if they were willing to undergo the burden for the sake of the benefit.

Now, it's just plain silly to deny that we can make interpersonal comparisons here. If I get a papercut and you get your head chopped off, it is absurd to deny that you have suffered a (factually) greater harm. And it is similarly absurd to deny the moral counterpart, that it is more important to save your head than my finger. Egoistic complaints about "compensation" are irrelevant here -- of course saving your head will not compensate me for my papercut. But it can still be of greater moral weight. As Parfit rightly points out, these are two very different concepts, which proponents of the "separateness" objection mistakenly conflate.

As a final point, I think the issue of incomplete relativity also rears its head here. The "separateness" objection implicitly concedes that benefits can outweigh burdens for a single person. But why should me-now be forced to suffer for the sake of my future self? If I'm a bizarre 'aprudentialist', who cares only for my present well-being, then such impositions seem no more legitimate merely because I am also the (eventual) beneficiary. The separateness of persons thus seems irrelevant to the issue of utilitarian sacrifice. In the end, I think the objection really stems from an illegitimate attachment to the thesis of self-ownership. (And we all know where that leads!)

Update: I address a more sophisticated version of the objection here.


  1. "The first thing to notice is that your argument here does not fit neatly with your argument there."

    Ah, good spotting. The two are consistent though, as I use the two possibilities (normal and 'aprudential' persons) as particular counterexamples to universal claims. Neither requires that people are always normal or always aprudential. (Note also that the present argument doesn't deny standard views of identity. It just points out that there are divisions within time-slices of an individual person which can conflict in a relevantly similar way to divisions between persons.)

    "So what do we mean by 'needs' and how do they relate to 'utils?'"

    Yeah, 'needs' aren't a core utilitarian concept, I probably shouldn't have used the term here. Intuitively, the meeting of a need makes a very large difference to one's welfare, but as you say, further benefits are presumably also possible on top of that, and might (at least in principle, and perhaps through aggregation) even outweigh a 'need' in utilitarian terms.

    "By this rule, if I'm willing to work overtime so that I might enjoy a larger television, I ought to be equally willing to work overtime that others might enjoy larger televisions as well."

    No, humans can't (and shouldn't) act like utilitarians in their everyday lives. But perhaps we should set up societal institutions, or support practical moralities, that would lead to the sorts of utilitarian trade-offs discussed.

    "How different is it to treat the interests of everyone in the same way a prudent individual treats his own interests and to imagine oneself, or mankind, to be the sort of super-person that Rawls rejects?"

    Oh, I do think the imagined super-person can be a useful heuristic at times. My point is that utilitarianism doesn't depend on this, as explained in the main post, so this ignoring of the factual boundaries between people cannot be the explanation why utilitarians deny the ultimate moral significance of such boundaries.

  2. "If not, then super-person heuristic or not, utilitarianism seems morally blind to the separateness of persons."

    I think you're mistaking the shape of the dialectic here. We can all agree that utilitarianism denies any ultimate moral significance to the separateness of persons. The question is whether this is the correct position, or whether it is rather a mistake (or inappropriate "blindness", as you say). One way to answer this would be to look at why utilitarians hold this position. My "charitable Rawlsian" criticism hypothesizes that utilitarians overlook this (purported) moral difference because of a methodological flaw, namely, that their super-person heuristic leads them to not even notice the factual difference. But other heuristics (e.g. the dispassionate but benevolent 'ideal observer') yield the same results without overlooking such facts. So the "charitable Rawlsian hypothesis" is false.

    (Of course, even utilitarians will take note of the difference between people for practical reasons, e.g. to do with incentives, etc.)

  3. "the utilitarian has mistaken the importance of people's interests within their lives for the importance of people's interests tout court."

    That's a really good point, and not one I've considered before. Thanks for bringing it to my attention! I'll need some time to think about it before replying.

    You'd be right about the utilitarian kings, given the stipulation of no practical differences in consequence.

    I think commendation is part of our practical morality, so the two-level theory justifies differential responses to altruism vs. self-benefit in that respect. (Though, if pressed, the utilitarian would have to agree that there isn't anything intrinsically better about altruism.)

  4. It's true that utiltarianism seems to give the same results in normal decisions regardless of whether it considers the seperateness of persons or not. But it is when population ethics are brought into the picture that the problems with utilitarianism's failure to recognize the seperateness of persons really become obvious. Some examples:

    1. According to traditional utilitarianism it is a morally neutral act to painlessly kill someone and replace them with another person whose lifetime total of utility us equal to the total utility of the first person's remaining lifetime. According to traditional utilitarianism it is a morally good act to kill someone and replace them with someone whose total utility is slightly higher than the total utility of the first person's remaining lifetime. You might object that such an act would have side effects, but we are interested in evaluating the morality of the act in principle, not compounding factors like side effects.

    And this isn't an unrealistic scenario. The government could start a program to kill handicapped people, take their bank accounts, and use the money to pay a family to have a nonhandicapped child. This is monstrous.

    2. According to traditional utilitarianism building a machine that would create a new person, let them experience one happy moment, painlessly kill them, create another new person, and so on, for a million times total, would be morally equivalent to creating a person who will live a flourishing life containing a million happy moments.

    3. According to traditional utilitarianism if I am stuck in a time loop and repeat one of the better days of my life over and over again (like the movie "Groundhog's Day" except that I am unaware I am in a loop and my memories don't carry over when it resets) this is morally as good, or morally better, than me living a flourishing life where I have a wide variety of different experiences.

    I consider the main reason utilitarianism gives such insanely wrong answers in these scenarios is that it fails to take the separateness of persons into account, and instead just tries to maximize the total amount of experiences without regard for their context or who or what they happen to. You've written about value holism since you made this post, so maybe you have come around to this way of thinking.

    1. Hi Evan, thanks for your comment. These are much stronger objections than Nozick's crude "sacrifice" objection, but I think utilitarianism still has the resources to address them. See my recent paper, 'Value Receptacles', for full details. Briefly:

      (3) This is a problem with hedonistic theories of welfare, not utilitarianism (= welfarist consequentialism). We can accept the latter without the former.

      (1) and (2) are both helped by recognizing that death typically involves positive harms, insofar as it involving thwarting preferences, cutting short life projects, etc. This creates an extra justificatory barrier for "replacement" scenarios -- it's not enough for the new person to be marginally happier; they must have a significantly better life to outweigh the disvalue of the extra death.

      (2) is additionally helped by my 'token-pluralistic' view of utilitarianism, according to which benefits to different people can still be recognized as non-equivalent (calling for ambivalence rather than indifference) even if equal in value.

      You may still be worried about (1), even after factoring in the extra disvalue of death. This would suggest you prefer a "person-affecting" view in population ethics. I'm interested in looking into that issue further, but at this stage the best answer doesn't strike me as obvious either way.

    2. Hi, I'm really glad to hear your reply. I have some further thoughts on it:

      The reason I think (3) is relevant to the idea of separateness of persons is that I consider it to be a challenge to the idea you sort of expressed in "Incompletely Relative Rationality." (I'm sorry, I'm not sure how to insert hyperlinks, so I'm just quoting the title of the post). Basically, if it is truly irrational for a prudent person to show partiality towards one or the other of their time-slices, does that mean we shouldn't have any partiality towards a new time slice versus a repeated one, and therefore have no preference between a time loop and a normal life?

      In the time loop example whoever created the loop is essentially "killing" my future time slices and replacing them with copies of a previous time slice. I consider wholly rational for a prudent person to prefer new original time slices to repeated ones, even if some of the experiences in the new time slices are less good by themselves than the one in the looped slice. Similarly, I consider it rational for an altruistic person to oppose killing someone and replacing them with a new person, even if the new person's life would be better.

      I have heard the argument that death involves positive harms before, but I have trouble understanding it. Won't the new life projects and preferences of the new person "make up" for the thwarted ones of the first person in a total utilitarian calculus? If not, how are you altering the calculus to stop them from doing so?

      I'm not sure how the 'token pluralistic' view helps. If I understand it, it basically means that I am allowed to feel really bad when I sacrifice someone else's interests for the sake of another person. I already do that. I think killing someone and replacing them is bad because it's immoral, not because it makes me feel bad.

      I don't prefer a pure person-affecting view. I am well aware of its problems, for instance, in Parfit's classic example of the sick woman who delays her pregnancy to stop the illness from harming whatever baby she conceives, I think the woman would be acting wrongly to not delay her pregnancy.

      But I think jumping from that to "individual people don't matter, all that matters is the total quantity of utility" is a big stretch. For one thing, imagine a modified scenario where the woman would have conceived triplets if she got pregnant while sick, and only one child if she got pregnant after recovering. I think most people would think the woman would have acted just as badly, if not worse, if she had chosen to get pregnant with triplets while sick instead of delaying her pregnancy and having one healthy child later.

      It seems to me that there are two good reasons, one impersonal, one person-affecting, to prefer creating a new high-utility person vs. a new low-utility person, all other things being equal, without also preferring to create large amounts of low utility persons over a smaller amount of high-utility people. The impersonal reason is the value-holism you talk about, creating a person with low utility in a society full of flourishing people has a negative impact on the "shape of the world." The person-affecting reason is that creating a new person with low utility places an unwanted obligation on other people to help them, which I consider to count as a "harm."

      Not only do these strike me as two good reasons to prefer creating small amounts of high-utility people, they strike me as being the actual reasons most people prefer to do this in real life. As for "killing and replacing," I would say that the bad done by destroying an actual person with actual existing preferences is far worse than the bad done by not creating someone who doesn't have preferences yet. I don't know if this makes my views some hybrid of "person-affecting" and impersonalism or not, but I agree with you that pure person-affecting views are problematic.


Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)

Note: only a member of this blog may post a comment.