Monday, October 11, 2010

On whether "ought" implies "can"

Might I be incapable of meeting my obligations? No, you might think, because obligations are the sorts of things that cannot
be unless they can be met. Or, better put, you might think that you must be able to meet your obligations if indeed they are your obligations because “ought” implies “can”:

Necesarily, if you ought to φ, you can φ (OIC).

OIC has come under attack (again). The latest attack due to Graham (forthcoming) is novel and not without its intuitive force, so it is worth taking a look to see if OIC has finally met its match.

According to Graham, there are cases where an agent has an obligation where the agent has not both the opportunity and ability to act in such a way as to fulfill that obligation. This example poses no threat to OIC, but it is
where we should start:
A surgeon has ten patients, each of whom will die of organ failure if he does not receive an organ transplant. The surgeon wants to save her patients and is convinced by philosophical arguments to the effect that it would be morally permissible to kill two people in order to save them. She notices that in another room of the hospital there are two innocent and unconscious tonsillectomy patients who are perfect organ matches for her patients. The only means by which the hospital janitor, who is aware of the situation, can stop the surgeon from chopping up the two and redistributing their organs among the ten is by shooting her with his pistol. He does so and thereby kills her (TRANSPLANT).

It is intuitive to say:
(1) It is morally permissible for the janitor to kill the surgeon.
Let’s assume this is so. Graham says that if this is so (which it is), we should also say:
(2) If the janitor had not killed the surgeon, the surgeon would have impermissibly killed the two patients.

This also seems correct and I agree with Graham that (2) is an important part of the explanation of (1).
Now, consider a variant on TRANSPLANT:
Everything is as it is in TRANSPLANT except that the surgeon cannot refrain from killing the two because the ten are her grand-children, and she is as compelled to save them as is the most severe kleptomaniac to steal (COMPULSION).

Concerning COMPULSION, Graham observes that (1) still seems true. He also observes that the doctor’s compulsion seems irrelevant to the fact that (1) is true concerning COMPULSION. So, he concludes that we should say that (2)
is true concerning COMPULSION as well. And now we can see why OIC is in trouble. The surgeon could not have met his (alleged) obligation not to kill the two patients.

Like Graham, I think the janitor is permitted to shoot in both cases. He thinks I should thus reject OIC:
As in TRANSPLANT, in ... [COMPULSION] (1) is true. But not only does the addition of the surgeon’s compulsion not change this fact, it also seems irrelevant to it. In other words, whatever explains the truth of (1) in TRANSPLANT, it seems, must also explain the truth of (1) in COMPULSION. But if this is right, then OIC must be false. For if, as I shall argue, what explains the truth of (1) in TRANSPLANT is (2), and what makes (1) true in COMPULSION is the same as that which makes (1) true in TRANSPLANT, then it must be the case that (2) is true in COMPULSION (forthcoming, pp. 7).

I worry about his inference from the (correct) observation that (2) explains (1) in TRANSPLANT to the further claim that (2) explains (1) in COMPULSION. First, while I agree that the surgeon’s compulsion seems irrelevant to the fact that (1) is true in COMPULSION, I do not see why this should lead us to conclude that the proper explanation of (1) is the same in COMPULSION and TRANSPLANT. We know, for example, that if you fiddle with the right details, you can construct pairs of cases where the effects to be explained are the same in both cases, the potential causes are the same in both cases, but differences between the cases determine which cause explains where the differences are irrelevant to the fact that the effect was produced in these cases. If the cause of some effect was sufficient but not needed because the effect would have occurred anyway had the cause been removed (e.g., there was a backup cause), modifying details of the case that seem intuitively irrelevant to the fact that the effect was produced might involve modifying the very details that determine which potential cause is the actual cause and so determines which causal explanation is the correct one. I see no reason to think that moral explanation will not
involve similar complications.

I want to do two things. First, offer an alternative explanation for (1) in COMPULSION. Second, argue that we have no good reason for preferring Graham’s explanation to mine even if we accept his explanation of TRANSPLANT. Our janitor has to decide whether to shoot or not to shoot. Suppose our janitor believes OIC and so believes the conditional that if the surgeon cannot avoid killing the two, he cannot act impermissibly in so doing. But, suppose our janitor does not initially believe that the doctor is compelled to act while removing his pistol from his utility belt. He would, let us assume, shoot the doctor if he thought that the doctor was acting freely in the way we might imagine he did in TRANSPLANT and do so from the same motive. I imagine
that this was a concern for the patients and not a pathological dislike of impermissible actions. But, now our janitor comes to learn (never mind how) that our surgeon is not acting freely and so has to decide whether to carry out his
intention. Perhaps matters are complicated for him because he thinks (rightly or wrongly) that the doctor is not acting impermissibly since the doctor cannot refrain from trying to kill the two patients. It would be strange, I think, for
the janitor to put his pistol away. What could his reason be? If he decided not to save them because he knows how their organs will be distributed, it is hard not to think of this case as one where the janitor is deliberately allowing some
to die in order that others might be saved. This seems very unKantian.

We might cite, then, the prohibition against acting on maxims that involve treating others as mere means as the explanation as to why (1) is true in COMPULSION. As for TRANSPLANT, the relevant moral facts can be explained in part by this, but we could also say that the janitor acted permissibly because he was preventing another from murdering two. We have the materials to explain the relevant moral data even if the doctor would have acted impermissibly in only one of the cases.
Is the explanation I have offered as to why (1) is true in COMPULSION fare better than Graham’s explanation? It has this much going for it. My explanation allows us to remain agnostic as to whether OIC is true. Since there is (presumably) a default presumption in its favor, conservatism gives us some reason for preferring my explanation to his. I would also note that there are further modifications of COMPULSION that involve the permissible killing of the doctor where it is clear that the doctor does not act impermissibly in the relevant scenario:
Everything is as it is in TRANSPLANT with two exceptions. First, the surgeon cannot refrain from killing the two because the surgeon is controlled by remote by a neurosurgeon who has wired our surgeon’s brain forcing the surgeon to act and reason as the surgeon did in TRANSPLANT. Second, this surgeon would not normally reason and act in this way were it not for the neurosurgeon. (CONTROL).

The neurosurgeon’s goal is to distribute the patient’s organs so that the greatest number can be saved. If the only way to save the patients is to kill the surgeon, See Scanlon for 2000 for discussion of cases that involve treating someone as a mere means
by allowing something to happen to them. my own view is that the janitor would be permitted to shoot knowing that this would eventuate in the surgeon’s death. This is so even though the janitor knew (never mind how) that it was not the surgeon who was acting. If the surgeon did not act, she did not act impermissibly. Still, I say, the intuition that it is permissible for the janitor to use force to intervene is as firm in this case as in Graham’s. So, even if Graham gives us the right explanation as
to why intervention is permissible in TRANSPLANT, I say this gives us little reason at all to think the same explanation holds true in COMPULSION since CONTROL illustrates that there are cases nearby where the permissibility of shooting the surgeon tells us nothing about the moral status of that surgeon’s actions.

Graham anticipates this sort of response and offers this challenge that I shall try to meet in closing the paper. Consider:
A bystander can redirect an out-of-control train away from ten trapped track workers and toward two other trapped track inspectors. If the bystander does nothing, the train will kill the ten, and if she redirects the train, it will kill the two. From a distance, a hunter sees that the bystander is about to redirect the train and realizes that he can prevent her from doing so only by shooting her with his rifle. He does so and thereby kills her (TRAIN).

Graham says that it is not permissible for the hunter to do this, and I agree. But, Graham asks, what is the difference between this case where the hunter acts impermissibly and my case where the janitor kills a “passive threat” (forthcoming, pp. 15)? It is what I suggested earlier. In COMPULSION as well as CONTROL, the decision to do nothing would involve letting some die as a
means to some end and so would violate the Kantian prohibition that tells us we must not act on maxims that involve treating others as mere means. This is not a feature of TRAIN and it is a mistake to think of the hunter’s situation and
the situation of our heroic janitor as the same. Two differences are immediately apparent. The first is that the Kantian reasoning that helps us understand why it is that the janitor acts permissibly seems not to apply TRAIN to help us
understand why someone in the hunter’s position might shoot. The second is that in TRAIN, we have an agent who is using violence to prevent someone from acting justifiably whereas this is not so in CONTROL, COMPULSION, or TRANSPLANT.

1. Graham, Peter. Forthcoming. “Ought” and Ability. Philosophical Review.
2. Scanlon, Thomas. 2000. Intention and Permissibility. Proceedings of the
Aristotelian Society (Supplementary Volume) 74: 301-17.


Richard said...

Loop cases pose problems for the idea that "treating others as mere means" is what's intuitively objectionable in these cases.

But whatever the moral details, something along the lines of your response to Graham seems right. That is: responding to 'impermissibility' as such would be weirdly fetishistic on the part of the janitor. Rather, we should understand him as responding to whatever the reasons are that would have rendered the surgeon's act impermissible had it been voluntary.

Clayton said...

Thanks, Richard.

I think I agree with you concerning Loop cases on both counts. I wonder if it's still possible to say that there's a p.f. duty that requires us not to treat others as mere means, but my aim here is more to point to potential explanations that leave OIC untouched.


Kimberly said...

According to Graham, there are cases where an agent has an obligation where the agent has not both the opportunity and ability to act in such a way as to fulfill that obligation.

Would this example suffice to show the above proposition?

1.) Student A ought to do well on test X.
2.) Test X is such that if one did not study a specified chapter, (let's call it C1) then one would not do well on test X.
3.) Student A did not study C1.
C.) Student A cannot do well on Test X yet student A ought to do well on Test X.

Clayton said...

I'm glad your argument wasn't:

1.) Teacher A ought to send comments to Student B.
2.) Comments are such that if A had managed to keep on top of things, A could have written comments.
3.) A did not send comments to B.
C.) A again fails to do what A ought to do.

Anyway, your argument _might_ do the trick, but this is where things get tricky. For certain sorts of "ought"s, I think there's not a presumption that "ought" only applies to the agent if the agent can do the specified thing (e.g., tenants ought to pay their rent on time). I can't recall what the right term for these are (role-oughts?), but I think some say that OIC doesn't hold for role-oughts but it does hold for moral oughts.

Having said that, I worry about:
(i) You ought to live a life where you commit no wrongs.
(ii) No one can live a life where they commit no wrongs.

That's why it's best not to have views apart from views about the views of others.

Kimberly said...

Well, if (ii.) No one can live a life where they commit no wrongs is true, then it seems that (i) You ought to live a life where you commit no wrongs. should be modified to (i) You ought to attempt to live a life where you commit no wrongs. Of course that's easier stated than proved...

Also, I was this (-) close to posting that exact argument! Out of decency I opted to spare you feeling, of course.

Alastair Norcross said...

I agree that something along the lines of your response seems correct. It is a pity, though, that Graham chose to use such a contentious example. It would be easy to concoct an example in which the surgeon is clearly about to do something such that it's a good thing that the janitor stops him. Then, consider the variation in which the surgeon suffers from an overwhelming psychological compulsion to do that same thing, but it's still good that the janitor stops him. The problem with Graham's example, of course, is that it only appeals to those who accept a false moral theory. Of course it's not good that the janitor stops the surgeon in his example. Ten innocent people are dead instead of two. Those who fetishize causal routes (such as kantians) can blather on about not treating (or allowing to be treated) others as means, but this only serves to conceal the kind of extreme lack of respect for the value of life which can applaud the janitor's appalling death-maximizing behavior.
It would have been far better if Graham had had the surgeon attempting to kill two to save one. Then we could have all agreed that the janitor acted well in preventing him, even in the case in which the one was the surgeon's grandchild.

Alastair Norcross said...

A slight correction to my previous post. Graham's example should have had the surgeon trying to kill three to save one. That way, the janitor's intervention would have resulted in fewer deaths than had the surgeon succeeded (I had omitted to count the surgeon's death).

Clayton said...

Hey Alastair,

I was going to tell you that I wrote a paper recently where the rhetoric was very Alistairotelian. Quite fun. It was hard to know how to formulate the response to Graham because I had to flip that "off my trolley" switch to think things through like a proper Ptolemaic deontologist. A better response would have been straightforwardly consequentialist. Janitor shouldn't intervene because 10 better than 2. Done! Still blocks the argument, but it blocks the argument in style!

It's strange because it's clear from Graham's discussion that he thinks there's a prohibition against killing passive threats in the defense of others. Even with my non-consequentialist hat on, I don't feel the pull of that intuition at all.