Monday, May 7, 2012

Mailbag Monday: The Double Effect

Gabe writes,

Can you do a writeup of the Doctrine of Double Effect? In particular, I'd like to know what you think the best arguments in favor of it are since I haven't been able to find one that I don't think is complete horse shit (the distinction between a bad effect that's used as a means to a desired end and a foreseen bad effect that's "merely" an unavoidable side effect of producing that end seems completely bogus and I don't see how it could ever be morally relevant). Thanks!

Holy shit, the ol' Doctrine of the Double Effect, huh? It starts all the way back with Thomas Aquinas, who was talking about what things we can and can't do - what is and is not permissible. His discussion was actually pretty short; he's like, "What if a guy is going to kill you and you have to really injure him to stop you? Uh, well, you don't mean to hurt him because you're not a dick; you're just trying to save yourself, so as long as you're not excessive, you're all good brah!" And then he's off to other topics and you're all, "WAIT COME BACK THAT SOUNDS SUPER IMPORTANT I HAVE SO MANY QUESTIONS" but it's too late. Fortunately, plenty of others have taken up the Doctrine since then, and we've got a pretty robust discussion going.

Sometimes an act has two different effects - one good and one bad. You might even say it has a... Double Effect. For example, hitting that guy in self-defense has a good effect (saving you) and a bad effect (harming another guy). There are lots of situations like that: sometimes I don't go out partying, and I have a shitty night but I'm really productive the next day. (That actually never happens, but roll with me here.) Sometimes you bomb an enemy base, which brings a war to an end but also some civilians also get killed. (I admit, that got dark quick. Welcome to moral philosophy!) 

 According to the Doctrine of Double Effect (DDE), you can only do an act that has both a good and a bad consequence if the following four conditions are all met at the same time:

1. The actual thing you're doing has to be good, or at least not evil. So you can't, say, drown a baby, even if that means you win a billion dollars, because baby-drowning is bad. BAD. 

2. The thing you're trying to achieve can't be the evil thing. Like, if there was a way to get the good effect without the bad effect, you totally would. But you can't. So you have to suck it up and accept the bad thing, but it's absolutely not what you're going for.

3. The evil thing can't be a means to the good thing. You can't do something that has an evil effect that in turn has a good effect - the good effect has to come from your action, not from the evil effect.

4. The good effect has to outweigh the bad effect. You can't, like, accidentally allow a city to burn down in order to save a kitten. Not that I don't love kittens; I am from the Internet after all. There just has to be proportionality there, is all I'm saying.

Let's say there's a general who is a pretty awful dude. Real dickface of a general, in fact. We'll call him General Dickface. General Dickface, general dickface that he is, is planning to assassinate the President, President Reallygreatguy, who is (as you'd imagine) just a super nice guy. You also know that General Dickface has an estranged daughter named Delightfulgirl (she goes by "Dee" for short) who lives with his ex-wife, the former Mrs. Dickface. Dee is a delightful girl - hard worker, volunteers at a children's hospital, can quote Monty Python extensively, just everything you'd hope for in a decent human being. If anything happened to his daughter, General Dickface would be so stricken with grief he would kill himself instead of going through with his plan.

Now, you have to save President Reallygreatguy. Let's say that Dee has to walk over a particular bridge to get to the children's hospital, and General Dickface has to drive a tank over the same bridge to kill the President. He's almost there - your last chance to stop him is on this bridge. Unfortunately, Dee is crossing over at the exact same time. But you have to save President Reallygreatguy. WHAT DO YOU DO?

Maybe you could shoot Dee right in front of General Dickface, and he'd be so shocked and upset he'd stop what he's doing long enough to be pulled out of the tank and subdued. You've saved President Reallygreatguy! But according to the DDE, can't just shoot Dee. Why? Because shooting an innocent person is itself an evil thing, so you fail condition 1. That seems straightforward, right? Here's a morally important difference for you. "How'd you save the President?" "I shot an innocent girl on her way to a children's hospital, which shocked the assassin just long enough to arrest him!" "...Oh. Well... oh. Uh... great?"

And you can't get out of it by saying, "Man, I was just pulling a trigger! And it had this weird effect of killing Dee, which caused General Dickface to kill himself." Even if we agree that "pulling a trigger" is not intrinsically bad, that still uses the death of Dee, which is a bad, bad effect (remember we like Dee, she's seriously wonderful) that causes the good effect of saving President Reallygreatguy. The bad effect is causally linked to the good one. So that's condition 3.

Also, you can't just nuke the bridge and accidentally wipe out the entire fucking city. Sure, you save President Reallygreatguy, but the cost is way too high. So that's condition 4.

Conditions 1, 3, and 4 seem more or less straightforward to anyone who isn't a utilitarian. If you're a utlitarian and you try to think about the double effect, you're gonna have a bad time. No, it's condition 2 that is really tricky - why does it matter what you intend?

So let's say you plan to blow up the bridge, halting the tank's progress and (accidentally) killing Dee. Blowing up bridges isn't necessarily evil - sometimes you need to demolish them to build a better bridge, after all, so you meet condition 1. Killing Dee isn't the means to stopping the General here, it's just an unfortunate accident, so you meet condition 3. And this seems proportional - saving the nicest President ever probably outweighs blowing up this bridge and hurting this awesome girl. It sucks that Dee will die, but that's the only way to save the President. There's condition 4, clear. So, we can blow up the bridge, right?

Hold your fucking horses. What do you intend? If you're trying to save the President, then by all means, blow that fucker right up. But what if the guy holding the button went on a couple dates with Dee, and then she dumped him, and now he just hates her? The Double Effect says he cannot blow up the bridge, because he intends to kill Dee, which is not okay. Which is fucking weird, right? You mean he can't, save the President, just because he doesn't like this girl? If there was another guy there with him, a guy who doesn't know Dee, new guy would be allowed to push the button, but this guy can't? That's really strange, that something like intention can prevent something as important as saving the President.

As usual, some people bite the bullet and say something like, "well, that sucks, but he shouldn't really blow up that bridge." But that isn't the approach most people take. Several philosophers have suggested that the Double Effect isn't particularly helpful in deciding hard cases - it just tells you what the hard cases are. T.M. Scanlon spent a chapter of his recent Moral Dimensions discussing (and critiquing) the double effect; ultimately, he accounts for our intuitions by suggesting that an evil intention doesn't affect the permissibility of the act, it only affects the blameworthiness of the act. So maybe this guy is allowed to blow up the bridge (because of other considerations) but he doesn't get any credit for being a hero because he's actually kind of a jerk.

I don't really have room for a thorough treatment of the entire history of the Double Effect, but in general, that's what it looks like. At bottom, it doesn't seem like the Double Effect helps us determine rights and wrongs in a broad sense -  it doesn't tell us what sorts of acts are right or wrong. Instead, it helps us weigh rights and wrongs already established elsewhere, in a larger moral framework.

On the other hand, it seems like there's a difference between accidentally killing civilians in an attempt to, say, end a war by bombing a factory, and killing civilians who live around a factory in order to terrorize them into ending the war. Even if either of those ends up destroying the factory and killing the same number of civilians, one of them makes you a much bigger douchebag in a morally relevant way. So one way to go about not being a douchebag is, when you do things, don't aim for evil shit to happen. I think we can all agree on that.


You can read Aquinas' original formulation of the Double Effect here.
T.M. Scanlon's Moral Dimensions has a good treatment of some of the problems which have been raised against the Double Effect in the first chapter.

Both the SEP and Wikipedia have pages on the Double Effect.


  1. Fuck this! Honestly all I can think of right now is "OMG PHILOSOPHY BRO IS BACK!!!!!"

  2. Holy shit, that's one hell of a response. Thanks bro!

    Yeah, I guess one reason I have a hard time seeing how the logic works is because I tend to lean pretty consequentialist. That's an interesting point about how condition 2 could be render an action impermissible just because of how you feel about each of the effects, but I was actually thinking more of condition 3 when I said I don't see how DDE makes a morally relevant distinction.

    I mean, imagine these two iterations of your scenario: in one, the only way to save the President is to shoot Dee, which you know for certain will cause General Dickface to kill himself. Alternatively, the only way to save the President is to blow up the bridge, killing both Dee and General Dickface. I'm inclined to stop right here and say wait a minute, these two actions have the exact same consequences (both of which you anticipate). Even speaking strictly from Dee's perspective, should it really matter to her whether she dies directly from your bullet or as a result of the bridge collapsing? In both cases it's a necessary and anticipated consequence of your plan to save the President. I just don't see how you can coherently say one action is okay and the other isn't.

    Even better, couldn't we argue that in an important sense, you're using Dee's death as a means to kill General Dickface? Either way, you're taking an action that you anticipate will result in her death. The reason you're doing it is because you can't save the President without killing her, and I'm pretty sure she doesn't care whether you kill her with a bullet to the head or a collapsing bridge.

    If somebody asks you why you killed Dee, in both cases you'd answer "I had to in order to save the President." That sounds an awful lot like using her death as a means, doesn't it? I get that the order of the causal chain is different in the two cases, but who cares?

    Finally, it seems to me that this formulation of DDE wouldn't even allow you to severely injure someone who was trying to kill you. Even though you may not intend to injure the person, you're definitely using the injury as a means to your end of preventing harm to yourself (since the reason why he can't kill you is because you've injured him beyond the ability to do so).

    Thanks again man, glad you're back.

    1. Yeah, condition 3 is why you're going to have a bad time trying to think about the Double Effect as a consequentialist. Who cares how we get to the consequences? We just ned to get there.

      As to whether you can severely injure someone, that depends on what you call an "action." Hitting a guy? Hitting a guy and injuring him? Moving your fist in a particular direction?

      That's why I think the Double Effect only comes into play VERY late in the reasoning process - you've got to have a ton of other shit already figured out.

    2. Well, the action's got to be hitting a guy and injuring him, doesn't it? Since the guy getting injured is the means by which your end (not getting killed) is achieved. That seems like a pretty clear violation of condition 3.

  3. I gathered that the anticipated action of killing dee was only relevant if the Dbag knew her, otherwise she was merely a consequence. Which is why it was ok for guy B to push the button but not guy A. What's important to remember is what action are you doin? are you killing Dee, or are you blowing up the bridge. If someone asks me 'why did you kill dee?' my DDE answer would be, whoa bro don't put a face to the name of the innocent girl i just killed to save mr.awesome president, i already feel bad about it you dick.

    1. (cuz my action was blowing up the bridge)

    2. Nah, you definitely kill her when you blow up the bridge because you know that blowing up the bridge is going to result in her death. It's just as specious to say "hey man, all I did was blow up a bridge" as it is to say "all I did was pull a trigger and then suddenly WHOA IT WENT THROUGH HER HEAD." The distinction that's being made is that the way that Dee's death and Dickface's death are causally related to each other is different. I just don't see why that should matter because either way, you're using an action that kills Dee to save the President.

  4. Glad you're back. I've been missing this.

  5. This comment has been removed by the author.

  6. Now that I think about it, this is very similar to some of the trolley problems. Blowing up the bridge is like pulling a switch that diverts the trolley away from the track with 5 people and onto the track with one person. Shooting Dee in the head is like pushing the fat man into the path of the trolley to save the 5. (And if you're like me, you'll conclude that there's no morally relevant distinction there either.)

    1. "the distinction between a bad effect that's used as a means to a desired end and a foreseen bad effect that's "merely" an unavoidable side effect of producing that end seems completely bogus and I don't see how it could ever be morally relevant"

      If your a consequentialist then intentions don't matter and DDE is of no use to you which is why the principle seems to lack all moral relevance to you.

      For self defense DDE doesn't prevent you from severely injuring someone because your action isn't injuring the person, your action is defending thyself. One of these actions is permissible(self defense) and the other is not (hurting another human being) even though, yes the consequences are the same.

  7. I'm interested in how you can justify punishing people for crimes under the doctrine of double effect. It seems to me that, at a minimum, most acts of punishment would be evil acts if they were not acts of punishment. They are, after all, the infliction of suffering on another human being.

    What justification can be given for punishing people that doesn't fall afoul of rule #3? If my goal in punishing you is to change your behaviour, it will be changed as a consequence of the punishment. In fact, any desirable consequence of chastisement is irrelevant under rule 3.

    You could say that their suffering is an unintended consequence of the punishment, but that doesn't seem right - if they didn't suffer, they wouldn't have been punished at all, and would have no reason to change their ways.

    The only possible escape I can see is to say that it's not bad to harm someone who deserves it... But that opens a whole nest of much nastier worms as we discuss who exactly deserves to be punished and what kind of harms it's non-evil for me to inflict upon them. If a person cuts me off in traffic, and I punish him by breaking both his legs, surely that's okay?

  8. According to this doctrine, there is a relevant difference morally between intending to commit something wrong and foreseeing that harm will occur as an unintended side effect of permissible acts morally. Its purpose is to justify an action that has good results, but can have harmful effects. The U.S. seems to have adopted a variant of the doctrine of double effect, so that it is possible to undertake military operations or objectives legitimate targets even if it is expected that these operations will have some consequences "adverse ". Such operations become permissible if they meet the following mandatory criteria: the bad effect is unintended; the bad effect is proportional to the military objective; the adverse effect is not a direct means to the good effect (eg., bomb cities to encourage peace talks); measures are taken to minimize the foreseeable bad effects even if it means for the fighters to accept a higher risk. The dilemma here is a problematic situation for which there are at least two possible solutions, each being regarded as equally valid and desirable or undesirable. In addition, there appears to be no other criteria for choosing between the options. It is useful to distinguish between three types of ethical dilemmas: the dilemma of uncertainty, the dilemma of conflicting values the dilemma harmful The dilemma of uncertainty is the most general type of ethical dilemma. He alludes to a problematic situation in which it is difficult to determine what to do because there are reasons equally valid as each other to support at least two of the best options to resolve identified the problem. The dilemma harmful actions is a special case and the ethical dilemma involves a situation in which each of the options identified in the process of ethical decision making will hurt or harm. For example, in military operations that require the destruction of strategic points, it is often unavoidable collateral damage. The doctrine of double effect * has been used on occasion to justify the moral taking certain actions based on one of the options. The dilemma of conflicting values is a special case and the ethical dilemma involves a situation of ethical decision making in which there are at least two ethical values in support of competitive options. For example, the options for loyalty to compete with other options that affect the professional integrity. So I m curious to have your answer :-)