“Scheffler’s paradox” is a puzzling feature of the moral beliefs of most deontologists. According to these beliefs, it is wrong for you to kill an innocent person even if your killing the innocent person is the only way to prevent five killings of innocent persons from being perpetrated by someone else. What could possibly explain this?
This feature of deontologists’ beliefs cannot be explained purely by their commitment to the moral importance of such distinctions as doing vs. allowing, intending vs. foreseeing, or the like. Most deontologists think that there are stronger reasons against doing harm to non-human animals than against allowing harm to animals; many also think that there are stronger reasons against acting with the intention of harming an animal than against causing such harm to an animal without intending it. But surely you could permissibly kill one bear if that is the only way for you to save five other bears from being killed by someone else. In general, it seems that Scheffler’s paradox does not really arise for killing non-human animals.
So what is the special feature of persons that lies behind Scheffler’s paradox?
When no persons are involved, if you take the “interventionist” option of killing one bear to save five, you both actively do and intend harm to one bear; but you also both actively do and intend good to the five other bears. So it seems that in the case involving bears, the reason for the interventionist option and the reason against it are in the end fairly evenly balanced.
What is different in the case involving persons? The pattern of doing vs. allowing and intending vs. foreseeing is just the same as in the case involving bears. So I propose that the relevant difference is that when persons are involved, the values of the relevant consequences are different, on the grounds that these consequences involve relationships between persons.
I believe that there are certain intrinsic values and disvalues exemplified by relationships between persons. For example, other things equal, when one person saves another person’s life, he puts himself into a good relationship with that person; when one person fails to save another person’s life, he puts himself into a fairly bad relationship with that person; and when one person kills another, he puts himself into an unspeakably terrible relationship with that person.
However, if you take the interventionist option of killing one person in order to save five, your relationship with the five is affected by the fact that you saved them by means of sacrificing the one. In effect, you have put them into a relationship with the one that is almost as horrible as the unspeakably terrible relationship into which you have put yourself with the one. So overall the relationship with the five into which you put yourself is not clearly a good relationship – you save them, but only by making them the beneficiaries and cause of the involuntary sacrifice of the one.
Thus, I propose, the way in which the interventionist option saves the five is sufficiently tarnished that it does not count as a clearly “good consequence” of the option at all. On the other hand, the bad consequences are intensified – they involve not just a person’s losing his life, but also your putting yourself and the five into various horrible relationships with him. So the reasons against the interventionist option are now significantly more powerful than the reasons in favour.
Of course, the “non-interventionist” option has some pretty bad consequences too. By not saving the five, you put yourself into a fairly bad relationship with them; while by not killing the one you put yourself in only a very modestly good relationship with him. Still, you did not actively bring about these bad consequences; you only allowed them to happen. For a deontologist, the fact that the act involves only foreseeing and allowing harm (not intending or actively doing harm) significantly weakens the reason against the act that arises from these bad consequences.
Overall, then, the reasons against the non-interventionist option are markedly less strong than the reasons against the interventionist option; the balance of reasons tells decisively against the interventionist option. The special feature of persons that lie behind Scheffler's paradox is the special range of values and disvalues exemplified by relationships between persons.