Friday, October 20, 2017

Iterating Badness in the Paradox of Deontology

In 'Must Consequentialists Kill?' (forthcoming in J Phil), Setiya convincingly argues against the "orthodox" view that commonsense verdicts about the ethics of killing entail agent-relativity.  Instead, he observes: "In general, when you should not cause harm to one in a way that will benefit others, you should not want others to do so either." (p.8 on pre-print version)  For example, it's not just the agent that should prefer to avoid themselves killing one to prevent five killings, but we should generally prefer that others likewise avoid killing one to prevent five other killings.  The preference here mandated by commonsense morality is thus agent-neutral in nature: it makes no essential reference to your role in the situation.

This seems right (I mean, correct as a claim about "commonsense morality", not actually right...), and it avoids one horn of the paradox of deontology, namely worries about how the reasons to act morally, if merely agent-relative, could have the authority to trump more personal / self-interested reasons for the agent.  But the core puzzle remains: if killing is so bad, why should we not be concerned to minimize its occurrence?  Setiya's answer seems to be, in effect, that killing per se is not so bad (being roughly as bad as accidental death).  What's very bad is instead a more specific kind of killing, such as killing as a means to preventing greater harms (or whatever the right generalization of deontological intuitions over cases turns out to be).

And the badness must iterate.  Consider: According to Setiya, you should prefer that agents not kill one even as means to thereby preventing five utilitarians from each doing the very bad thing of killing one to prevent five other killings.  So killing as a means to preventing a greater number of very bad killings must itself be worse than five such very bad killings (which are themselves worse than 25 ordinary killings).  That is, it must be very very bad.  And so on.  In general, the badness of killing someone (in an intuitively disapproved-of way) must be a function of how much instrumental good is achieved by the killing, in order to ensure that the killing remains bad/undesirable on net.

Such a view strikes me as not very substantively plausible.  It is not worse to kill someone for the sake of helping others than to kill them (let alone five...) for purely selfish ends.**  So I think we should reject the "commonsense" view about the ethics of killing (given that it is seen to be absurd upon reflection), and embrace a more traditional form of (welfarist) consequentialism which treats the "commonsense" intuition as something like a useful rule of thumb instead of a fundamental moral principle.

Setiya disputes this by suggesting that we focus on the moral stakes as the situation unfolds in time:
At the beginning of Five Killings, five people are going to be killed. In One Killing to Prevent Five, five people are going to be killed unless they are saved by the pushing of the button, which kills an innocent stranger by dropping him off a bridge into the path of the speeding trolley. The situation in which someone is going to be killed unless they are saved in this way is as bad as the situation in which they are going to be killed. Ethically speaking, the damage has been done. [...] It makes things worse, not better, that the button is pushed, so that the innocent stranger dies. That is why One Killing to Prevent Five is worse than Five Killings: it starts out worse and then declines. If we think of the temporal unfolding of events in One Killing to Prevent Five, we can make sense of why we should prefer Five Killings.

Of course, this is not an argument intended to persuade the traditional consequentialist, but rather an "internal" defense of deontological* thinking, explaining why it makes sense from their starting point to reach this conclusion.  And I agree that this explanation makes clear why deontologists are committed to the conclusion in question.  I just don't think that it makes sense of the conclusion itself, which remains (at least to my mind, but then I was never sympathetic to deontology to begin with...) seemingly completely bonkers.

What do you think?

*: Setiya classifies this differently, but see my next post on how I think the consequentialism/deontology distinction is most usefully drawn.

**: A test case: Compare Killing One to Prevent Five to an alternative, Killing One for the Sheer Joy of It While Others Kill Four (we may call this case 'Killing One for Sheer Joy (-4)' for short).  I take it that from an impersonal perspective, Killing One for Sheer Joy (-4) is equivalent to (indeed, an instance of) Five Killings.  And Killing One to Prevent Five is, we are told, a morally worse state of affairs than Five Killings.  Presumably adding one more background death is not enough to flip the moral scales here, so we may assume that it's also worse than Six Killings.  If so, it would seem that Setiya is committed to holding that when five others are threatened, it's morally better for you to murder a sixth for the sheer joy of it than it is to kill that one innocent in order to rescue the other five.  But that's absurd.  (Or am I missing something?)

Update: I probably was missing something -- see Angra's comment!

6 comments:

  1. I don't think you are understanding deontology at all. You are trying to analyze it in terms of consequentialist thinking. That is, you are asking about how bad killing is considered as "a thing that happens," that is, as though it were a consequence. But deontology considers killing as "what you are doing", that is, as an action.

    ReplyDelete
    Replies
    1. I'm not trying to analyze deontology in this post. I'm responding to Setiya's paper. A key question raised by his paper is how bad/undesirable it is for various sorts of killings to occur. That's a perfectly legitimate question to consider. (At least, the mere fact that deontologists typically don't like to think this way doesn't show that it isn't a legitimate question to consider, and one that may reveal problems with their view.)

      Delete
  2. Hi Richard,

    "And Killing One to Prevent Five is, we are told, a morally worse state of affairs than Five Killings. Presumably adding one more background death is not enough to flip the moral scales here, so we may assume that it's also worse than Six Killings."

    As I read it (but I just read it; I would need to study the paper further), the reason Killing One to Prevent Five is worse than Five Killings according to Setiya is that it adds another killing to the list (the immorality of the other five has already happened, as someone already intends the killing, and the result will happen save for some further immoral behavior to stop it), Killing One to Prevent Five would then be morally on par with Six Killings, if motive is not taken into consideration. But Setiya may not be committed to motive's not being a factor (barring inconsistency elsewhere, he could accept that some killings are worse than others), so he may hold that murder for fun is morally worse than murder to save five others.

    "So killing as a means to preventing a greater number of very bad killings must itself be worse than five such very bad killings (which are themselves worse than 25 ordinary killings)."
    That does not seem to follow, if my reading is correct. What happens (on Setiya's view) is that killing as a means to preventing a greater number of very bad killings adds another moral injury to the list, after the other immorality already happened. Then again, let's consider.

    S1: Five killings for pleasure happen.
    S2: One killing for pleasure that results (as an unintended side effect) in preventing the other five killings for pleasure happens.

    Going by Setiya's reasoning, it appears to me one also has a moral obligation to prefer S1, so that the intent is to save others seems to be a side issue: all the work is done by the 'adding one further injury to the list' condition.

    That aside, view is that there is no moral obligation (in many cases; one could find them intuitively) to prefer the scenario that contains less immoral behavior, or that does not add further immoral behavior after a certain point in time, so the matter has to be resolved on a case by case basis. Also, I don't think deontology in general is in trouble because of any of this, since (among other reasons) deontology isn't committed to the 'adds one more injury' condition.

    ReplyDelete
    Replies
    1. Thanks Angra, that's helpful. Your interpretation of Setiya, as treating "Killing One to Prevent Five [as] morally on par with Six Killings" seems plausible.

      But I'm not sure I follow why you think my earlier claims about iterative badness don't follow from Setiya's account. You write: "killing as a means to preventing a greater number of very bad killings adds another moral injury to the list, after the other immorality already happened"

      But there's something a bit figurative about the claim that the moral harm has "already happened". If the threatened innocents are somehow saved without harming anyone else, for example, then that's clearly a big improvement. Unsuccessfully attempted murder is a much better outcome than actual murder, as a general rule. The exception (on Setiya's account) is when the would-be victims are saved only by creating a new victim. Since the originally threatened harms no longer occur, and there would be no significant badness at all if it weren't for the newly harmful act, it seems like the only way to secure Setiya's verdict is for the newly harmful act to contain all the moral badness of the originally threatened harms that it prevents, in addition to the badness of the new harm that it causes. That is, it doesn't merely "add another moral injury to the list", it also retains the original list of "moral injuries" despite their threatened harms no longer eventuating.

      At least, that was how I was thinking of it. I guess your suggestion is that rather than relocating the moral badness of the original list of injuries into this new act of killing as a means, we just see the badness as remaining in the original bad act all along? It's harder to make sense of this in a case of killing one to save five from accidental deaths, as then there is no original bad act, just a bad state of affairs, namely the occurrence of the accidental deaths. But that occurrence never happens if they are saved, so where in the situation could the badness now be located, except in the new act of killing as a means?

      Another way to put my worry is to note that there is an obvious pro tanto good that occurs when the five lives are saved, even if one wants to hold that the event of killings as a means to save five is bad overall. To outweigh the immense good of saving five lives, it seems that the badness of the new act must be very great indeed. (As I put the point above, it must now contain all the original badness of the five deaths that it prevents.) Setiya certainly doesn't put it this way, but I'm not sure how he could avoid the datum that some pro tanto good occurs when lives are saved. Certainly the intuition that he appeals to is just that the act of killing to save five marks a turn for the worse (i.e., an overall moral verdict), not that there is no respect in which it makes things better.

      Delete
    2. On page 16, before saying "Ethically speaking, the damage has been done", Setiya says that "The situation in which someone is going to be killed unless they are
      saved in this way is as bad as the situation in which they are going to be killed.". As I read this, it applies to a situation in which more than one is going to be killed unless they're saved in that way, etc., and also would apply to the accidental deaths scenario as well. In other words, I think Setiya is committed to saying that the situation in which five people are going to accidentally die unless they are saved in that way is as bad as the situation in which they are going to accidentally die.
      So, where could the badness be located?
      As I understand it, the badness would be located in the situation in which which five people are going to accidentally die unless they are saved in this way - i.e., that situation already contains (in my interpretation of Setiya's position, that is) as much badness as the situation in which they will die accidentally, and also in the situation in which they do that accidentally.
      So, the new killing adds one further moral injury, which consists in an instance of immoral behavior - the previous moral injuries did not contain immoral behavior; they were injuries in the sense of being morally bad things (which Setiya defines in a way I disagree with, but leaving that aside).

      "Another way to put my worry is to note that there is an obvious pro tanto good that occurs when the five lives are saved, even if one wants to hold that the event of killings as a means to save five is bad overall."
      I think that Setiya (even if implicitly) denies that there is any pro-tanto good. He seems to imply that the evil that already happened is just as bad as the evil of the five people getting killed, and the fact that they are saved later by a killing does not make things any better. I think he is (probably) committed to holding that the same would happen if the killing were not meant to stop the other five deaths (or killings), but just happened to stop them (as long as a killing is the only way to stop them).

      But now, I'm thinking: what if five killings will happen unless they are prevented not by a killing, but by, say, two accidental deaths? (say, two fat men have heart attacks and fall on the tracks?)
      Given that Setiya thinks accidental deaths are not much better than killings, as long as two accidental deaths are no better than a killing (else, we can change the numbers a bit), then it seems to me that the situation in which five are going to be killed unless they're saved by two accidental deaths (of innocents, we could add) is not better than the situation in which five are going to be killed unless they're saved by a killing of an innocent. But then, if the accidents happen, it's worse (by Setiya's reasoning that it only adds another moral injury, even if one that does not involve immoral behavior).
      So, we end up with

      W1: Five killings>Two accidental deaths that prevent five killings.

      That's really weird. Then again, Setiya could say (as he said on page 10), that the situation is different, and in that case, we should prefer the accidents.
      But generally, his position that there is no pro tanto good at all (if that is his position) when a killing to prevent five happens, seems false to me, and also, even more difficult to keep if it's held that some pro tanto good happens when two accidental deaths prevent the five killings (assuming there was no other event that could prevent them).
      So, in short, at this point I think your objection seems pretty strong.

      Delete
  3. Thanks for the discussion! Without commenting on the specific cases, my general thought is this: other things equal, when it comes to killing and letting die, if you should not kill A in order to prevent outcome O, you should prefer [I do not kill A, outcome O happens] to [I kill A, preventing outcome O] and you should likewise prefer [N does not kill A, outcome O happens] to [N kills A, preventing outcome O].

    We can read off conclusions about better/worse outcomes if we interpret better/worse in terms of what you should prefer, considering outcomes specified without indexical reference to you. But I don't see how these conclusions give us any leverage on the plausibility of the initial claims. Put in terms of better/worse, some of these conclusions may sound odd. But that is only because we tacitly make false assumptions about how the goodness/badness of an outcome is a function of the goodness/badness of outcomes it contains as parts, etc. If we translate into what we should prefer, the claims in question will seem (I contend) as plausible as the claims about what we should do that motivate them.

    ReplyDelete

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.)