Agency and the Paradox of Deontology

Lest our readers begin to think that this blog will turn out to be some sort of consequentialist love-fest, here’s something from the deontology corner of PEA Soup.

I recently gave a paper at UNC-Greensboro, in which I defend deontology against Samuel Scheffler’s objection that its restrictions are paradoxical. Among the many helpful comments offered by the audience, one point that received some healthy discussion was whether the case that supposedly generates the putative paradox is actually possible. The case is this (there are other versions of it, but this is the one I focus on): either Agent 1 kills Victim 1, or Agents 2-6 will kill Victims 2-6. Deontology (let’s grant) obligates Agent 1 to not kill Victim 1, so this means that five other killings will occur, and as such deontology fails to minimize the overall number of killings. But if killing is so wrong, it seems paradoxical to require more, rather than fewer, overall violations of the duty to not kill.

Now early on, I was tempted to say that the case itself could not occur, but given that it doesn’t seem logically impossible, and it’s otherwise hard to prove a universal negative, I instead worked up a solution (I think) that renders deontology non-paradoxical given that such a case could arise. So I simply grant the case. Some of the comments from folks at UNC-G, however, got me reconsidering the idea that maybe the case itself is problematic.

The core of the concern – which should be credited to Michael Zimmerman – is that Agents 2-6 need to truly be agents for the paradox to even get off the ground. If they are not agents, then they won’t be violating any duties (any more than a hurricane violates a duty not to kill), and so deontology would not be requiring, paradoxically, more rather than fewer restriction violations when it holds that Agent 1 should not kill Victim 1 at the cost of Agents 2-6 “killing” Victims 2-6.

So, here’s why Agents 2-6 might not be agents, at least in the relevant sense. First, we might say that they are agents only if they have some choice as to whether to kill Victims 2-6. However, if there are only the two live options that either Agent 1 kills Victim 1, or Agent 1 does not, in which case Agents 2-6 kill Victims 2-6, then this seems to neutralize their agency. (It’s important here as a concession to proponents of the paradox that these be the only two options, for if there were others, such as none of Agents 1-6 killing anybody, then deontology would make that the obligatory thing for all of them to do, which does not paradoxically require more rather than fewer duty violations.) That is, Agent 1 is given a choice, but if she fails to kill, then (by hypothesis) Agents 2-6 must kill their respective Victims.

So if we suppose that agency involves having some sort of alternative possible actions from which the agent can choose, then clearly “Agents” 2-6 cannot be agents, since their “actions” are necessitated by the choice that Agent 1 makes.

Of course, the alternative possibilities approach to agency is controversial. So we might go with some sort of Frankfurt-style, hierarchically-ordered will version of agency, such that one is free if it is the case that one’s desire to kill is endorsed by one’s higher-order desires (about that desire to kill), in such a manner that one desires that first-order desire to actually be motivationally effective. (Yes, I’m sloppily running freedom and agency together here.)

But in what kind of cases could the agents each have such a will? That seems to depend on how we spell out the details of the origins of the case. If our parties ended up in the dilemma because of some outside force such as a natural disaster or some 13th person, then Agents 2-6 will not, presumably, endorse their desires to kill, since even if they realize that it is the only thing to do given Agent 1’s omission of killing Victim 1, it does not comport with their deepest commitments. In which case, they wouldn’t be willing agents.

By contrast, if Agents 2-6 are themselves responsible for putting Agent 1 (and all six Victims) in this dilemma, then they will be acting in a (Frankfurtian) willing manner. But in this case deontologists could just say that the restriction violations entered when Agents 2-6 set up this now-inevitable case (what I think Doug Portmore calls a ‘pre-murder’), and so they are responsible for whichever of the killings happens down the causally fixed road. In which case, deontology can still say that the right thing to do would have been for none of the six agents to kill anybody. That is, deontology can just locate the restriction violations in the setting of the unstoppable chain of events in motion, and say that Agents 2-6 should not have committed the original, dilemma-inducing actions that resulted, perhaps days later, in the killings of Victims 2-6. (The actual pulling of the trigger, because inevitable, is now morally irrelevant, just as it is morally irrelevant that I can’t stop the bullet of a gun whose trigger I’ve pulled.) But now there is a third live option, that nobody kills (or pre-kills) anyone. In which case, deontology can prescribe that third option, non-paradoxically requiring fewer (zero) rather than more violations.

So, while I think if we grant the case’s possibility, deontology can avoid the air of paradox, I’m now unsure whether the case can get off the ground to begin with, in which case deontology is even better off.

4 Replies to “Agency and the Paradox of Deontology

  1. Right now I’m transitioning between jobs, so I don’t have anywhere to post papers on-line at the moment. Hopefully at some point soon I’ll be setting up a new webpage. In the meantime, I’d be happy to email you a copy – email me if that works for you.

  2. Example first, then a suggestion.
    It’s World War II. During a bloody battle for a city, say Stalingrad, a macabre if by now all too familiar ceremony is taking place: five members of a Nazi firing squad are about to execute five prisoners, innocent civilians all. The only thing they are waiting for is a sign from a man in an officer’s uniform. I’m a sniper, and as it happens, I’ve got the officer’s heart in my sights – fortunately, since my gun is fixed in place so that I can’t change the aim. This “officer”, incidentally, is not really German at all, but a Russian civilian who put on the uniform just to keep from getting cold. Though he’s nothing like a friend, I know him, and I know that he believes that by raising his arm, he will stop the soldiers from shooting, while in fact that is precisely the sign they’re waiting for. I know the Germans’ aim is steady and their guns are working, since I just saw them execute five other prisoners. I also know that without authorization, they simply will not shoot but instead report to company headquarters for further instructions, during which time I will be able to release the prisoners and take them to safety. I see that the “officer” begins to lift his hand. Should I pull the trigger?
    In any real life case – if this counts as one – there are epistemological uncertainties involved, but let’s waive them aside here. Is the agency of Agents 2 to 6 (the soldiers) neutralized? I don’t see that. A better description of the case might go like this: the choices that Agents 2-6 have made necessitate that they will each x unless y happens. They’ve made a conditional decision (read this compatibilistically or incompatibilistically), indeed formed a conditional intention. The paradox is that I can bring about y and thus push them to another path (that they’ve chosen), but only by x-ing myself. How would this hamper their agency? They could have decided to fire when the moon comes out, instead. Would the moon staying behind the clouds remove their responsibility for not firing – or its coming out for their firing?
    I hope I understood your point correctly. I would, by the way, like to have your paper by mail as well – I’m presenting my own solution in a couple of weeks at a conference.
    Antti

  3. No, I don’t think the agency of the soldiers is neutralized in a case like this. Rather, it’s the kind of case I was imagining in the last long paragraph of my original post, where Agents 2-6 are, in fact, willing. So far, I think we agree.
    What I want to say, at this point, though, is that there is no paradox in this kind of case. The reason is that deontology doesn’t recommend (in this case) more rather than fewer violations of the duty not to kill. For, it recommends that none of the six Agents kill anyone. Those duties were violated, in turn, not when the trigger is actually pulled, but at what you’ve identified as the decision-making time, i.e., when Agents 2-6 set up the dilemma (the firing squad) in the first place. Of course, either Agent 1 or Agents 2-6 will have to violate their duties not to kill, but the point of wrongdoing lies not in the actual pulling of the trigger, but, rather, in the “pre-murder,” i.e., the setting up of this situation to begin with. This is what allows deontology to say that none of 1-6 should kill.
    Proponents of the paradox, though, need a case where deontology says something else, namely that Agent 1 should not kill, but Agents 2-6 therefore must kill, *and* that the wrongness of the duty violation(s) enters in *exactly* when those killings take place. So, in this kind of case, I don’t think we’ve got a paradox-generating scenario. I could be talked out of it – I’m still not sure on this one – but that’s the worry, anyway.

Comments are closed.