Friday, August 20, 2021

Preferring to Act Wrongly

Deontologists hold that it's wrong to kill an innocent person, even to prevent five other such killings. Does it follow that they should prefer Five Killings over One Killing to Prevent Five?  If so, my previous argument kicks in to demonstrate that they care insufficiently about killing.  In this post, I want to argue that the alternative -- preferring to act wrongly -- results in an even worse theory.

First, consider the objection that deontologists needn't take any stand on matters of preferability.  Perhaps they are just concerned to elucidate our obligations, and remain silent on all else.  But that doesn't help, because so long as there are further questions to be asked here (as there clearly are), their incomplete moral view must (if it is to have any hope of being correct) be coherently completable. That is, if certain verdicts about the deontic statuses of actions cannot be coherently combined with any plausible further claims about the relative preferability of various possible worlds, then those initial verdicts cannot all be true.  So we might as well move on to the question of what the most plausible completion of a deontological view would be: what verdicts about preferability fit best with the claim that it's wrong to kill one to prevent five killings.

Okay. So, why not prefer One Killing to Prevent Five, while regarding it as impermissible to bring about? Here I think it's important to get clear on what kind of preference we're talking about here.  One could certainly feel a kind of wistful pull towards saving the greater number, like, "I wish there were some way that I could, permissibly, save the five, alas..."  (Like how someone on a diet might feel the pull of the chocolate cake.)  But our question is which outcome you should all things considered prefer. To regard the chocolate cake as truly preferable, one must regard one's restrictive diet as comparatively less important -- perhaps you should really just give it up and indulge!  But in the same way, to regard it as truly preferable to kill the one, you must regard the constraint against killing as comparatively unimportant -- perhaps you should really just disregard those deontological hang-ups and do what will best serve humanity!  The latter is a fine thing to think, but it's precisely to reject one's previous deontological view as mistaken.  If one is to maintain a commitment to deontic constraints, one must regard the constraints as worth respecting, and hence all-things-considered prefer not to violate them when the occasion arises.  Or so it seems to me.

Some deontologists disagree.  I'd be curious to hear more about how they regard the connection between all-things-considered preferability and normative importance.  Note that in general, when there's a conflict between warranted attitudes or responses and preferable ones, the latter are "more important" in the practical sense that we should try to bring them about.  Tautologically, they are what we ought to prefer, i.e. care more about -- they are what matter more.  For example, if an evil demon will torture humanity unless we take a pill that makes us vicious, we obviously ought to take the pill.  It would seem obscenely self-indulgent to prioritize our own fitting attitudes over the good of humanity (or indeed anything at all that warrants greater concern, or is more worth caring about).  So if the "warranted" or "obligatory" action was truly dispreferable, or such that we all-things-considered ought to most want (and hope) to do otherwise, wouldn't this likewise show that doing otherwise was more important than meeting our "obligations"?

Such a view would seem to rob constraints of their normative force.  That is, if deontic constraints are combined with utilitarian preferences, the upshot would seem to be that deontic constraints don't really matter.  Sure, the deontologist may maintain that there is an "obligation" not to kill.  But this would seem a merely verbal victory if it turns out that we shouldn't really want agents to fulfill such obligations, and that what's truly preferable is to kill one to save five. Put another way: if we're all agreed that maximizing happiness is what we should most want and care about, then any residual disagreements about "obligation" would seem no more threatening to the utilitarian than residual disagreements about what's "honourable" (when we all agree that we've no reason to care about "honour" as such).

Compare my old post on Nye, Plunkett & Ku's 'Non-Consequentialism Demystified'.  These authors hold that killing one to save five may be wrong (contrary to reason), even though we should prefer and hope to act wrongly, as it would be "monstrously narcissistic" to "look... fondly upon following the dictates of reason" rather than saving more lives.  Those sound like fine sentiments to me, but they don't seem compatible with remaining committed to deontic constraints.  If we really ought not to kill, not even to save five, then this constraint must matter, and indeed matter so much that it would not be "narcissistic" to prioritize respecting it over saving more lives.  I'm not a deontologist, so I'm not here concerned to defend the unconditional claim that constraints really do matter so much.  But I think it's clearly what deontologists have to be committed to, if their moral verdicts are to carry normative weight. If violating constraints matters, then the prospect of such a violation ought to generate strong (indeed, decisive) reasons to prefer that one instead acts permissibly.

The possibility of such deontic reasons determining what's ultimately preferable (in contrast to taking preferability to be settled by simple utilitarian considerations or the like) would also seem to undermine Chris Howard's argument (in 'Consequentialists Must Kill', p. 749) for incongruence between preferability and reasons for action.  Howard argues: "these reasons to act have their source in the value of particular people, and so aren't given by value-making features of outcomes, [hence] they're reasons to act which don't derive from, or correspond to, reasons for preferring some outcomes to others." But they may correspond to such reasons without deriving from them, as deontologists may instead reverse the order of explanation and take the reasons for preference to derive from the reasons for action.  

Presumably, we should prefer to abide by deontic constraints just if it is truly important that we do so.  If deontologists deny that it's truly important to abide by deontic constraints, they've surely given the game away.  So, serious deontologists must prefer that constraints be respected, even if this makes the world "worse" in utilitarian terms.  This makes sense, because their view is precisely that utilitarian considerations are not all that matter (in the sense of what we should care about in practice).  If you instead agree that utilitarian considerations are all that matter practically, then congratulations -- you are a utilitarian (in all but name)!

[See also my draft paper on the 'New Paradox of Deontology'.  Comments welcome!]


  1. Not trying to speak for the deontological position here - but its seems important that we presumably have other relevant positive obligations on the deontological view, i.e. to save lives and/or to not allow killings/preventable deaths. Those would clearly be at odds with the obligation not to kill here, and I suppose one solution is to prioritize/rank them, but that leads us to consequentialist territory I believe. A more appealing alternative to me is to admit a certain pessimism about the world we live in - that its not ideal and that we are sometimes forced to choose among choices which don't allow us to avoid wrong-doing. I think this has the benefit of not sweeping the killing part of killing one to save five under the rug, so to speak, yes five were saved, but also one was killed and that can't be simply summarized as a good/the best best decision - a wrong was done, even if most would agree that was the best decision. I think this tracks with real life as well, since I would expect most people in that position would feel regret/have ptsd due to the choice they were forced to make (and I think most would think a person in that position should have those feelings, or that there is something wrong if the person didn't).

  2. Preferability interacting w/ obligation is slippery, but two thoughts.

    if certain verdicts about the deontic statuses of actions cannot be coherently combined with any plausible further claims about the relative preferability of various possible worlds, then those initial verdicts cannot all be true.

    This direction of implication just seems like an insistence that deontologists should recognize that consequentialist concerns, if of any value at all, should override their deontological principles. I think generally it's possible to hold instead that preferability is just a different thing. If we take Kantianism as a clear case, Kant (who is sympathetic to happiness-based preferability) thinks that certain things contributing to preferability (the moral endowments -- conscience, moral sentiment, natural philanthropy, and self-esteem) will, if reasonably cultivated, *tend* to make it *easier* for us to fulfill our moral obligations in *many* (maybe even most) situations, but that's as close as the fit gets. Your evil demon case is a good one; what makes it "seem obscenely self-indulgent"? Well, that it goes against the grain of our natural love of others. But this has no normative authority; its 'ought', if we want to put it in those terms, is just one of feeling. It's just our normal feelings placed in an abnormal situation. They don't provide any moral 'ought' at all, just an 'ought' (if we think of it that way) in the same sense that, when someone is hungry, they might think "I ought to grab a bite to eat." What naturally appeals to our ordinary sense of what is important, is not necessarily what rationally is; they are distinct systems. That doesn't mean that the former is nothing (even Kant thinks it's important), but obligation is just what it is to matter morally; everything else is just circum-moral (if you'll forgive the barbarism), so in principle detachable. Thus, for instance, that there be less killing is preferable from natural love of others, but natural love of others is only relevant for morality to the extent that it leads you to recognize that the rational nature of others is an end in itself in general; and this latter is violated by murdering someone, regardless of how natural love of others might feel in that particular case.

    There's also an issue with the question of what you do immediately in the situation vs. what you do contextually with respect to it. I don't think there's any doubt that Kant himself would hold that you should absolutely not kill an innocent person, and you have a moral obligation to punish severely, and to build a justice system that punishes severely, anyone responsible for putting people in situations with horrible choices like this. I think this has some bite, because consequentialists simply can't ignore the idea that what gets counted as your response to a situation should be taken to include your contextual actions, not just your immediate one.


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.