Thursday, October 26, 2017

A two-stage view of proportionality in the Principle of Double Effect

A question about Double Effect that hasn’t been sufficiently handled is in what way, if any, the good effects of bad effects are screened off when judging proportionality.

It seems that some sort of screening off is needed. Consider this case. An evildoer says that he’ll free five innocents unless you kill one innocent; otherwise, she’ll kill them. So you shoot at the innocent’s shirt covering his chest, intending to learn how the fabric is rent by the bullet (knowledge is a good thing!), while foreseeing without intending that the innocent should die, and also foreseeing without intending that the evildoer will free the five.

This is clearly a travesty of double effect reasoning. But the only condition that isn’t obviously satisfied is the proportionality condition. So let’s think about proportionality. Here are two ways to think here:

  1. All good and bad effects count for proportionality. Thus, both the death of the one and the saving of the five count, as does the trivial good of knowing how the shirt rips. Thus proportionality is satisfied: the goods are proportionate.

  2. The good effects that are causally downstream of the bad effects of one’s action don’t count. On this view, it is the intended effect that must be proportionate to the unintended bad effects. Thus, the death of the one counts, and the trivial good of knowing about how the fabric rips counts, but the saving of the five does not count, as it is not intended (if it were intended, the act would be impermissible, of course). But of course the good of knowing how the fabric rips is not proportionate to the death of the one innocent.

Option 2 fits better with the intuition that the initial case was a travesty of double effect reasoning.

But option 2 doesn’t seem to be the right one in all other cases. Suppose I am guarding five innocents sentenced to death by an evil dictator. If I free them, I will be killed. I also know that unless the innocents leave the country, they will be recaptured soon. The innocents are planning to bribe the border officials, which is quite likely to work. But it will be wrong for the border officials to let them escape, because the border officials will have the false belief that these people are justly sentenced, but are venal.

It seems permissible to free the innocents. Here, the unintended but foreseen bad effect is my own death. The good effect is the innocents’ being allowed out of prison. But it seems that if we don’t get to consider effects downstream of bad stuff, we don’t get to consider the fact that the innocents will escape the country, as that’s downstream of the border officials’ venal acceptance of bribes.

Here’s one theory I developed today in conversation with a graduate student. Proportionality is very complex. Perhaps there are two stages.

Stage I: Are the intended good effect and the foreseen bad effects are in the same ballpark? This is a very loose proportionality consideration. One life and ten lives are in the same ballpark, but knowing how the fabric rips is far out of that ballpart. If the intended good effect is so much less than the foreseen bad effects that they are not in the same ballpark, proportionality is not met. Here, the good effects that are downstream of the bad effects don’t count.

If the Stage I proportionality condition is violated, the act is wrong. If it’s met, I proceed to Stage II.

Stage II: Now I get to do a proportionality calculation taking into account all the foreseen goods and bads, regardless of how they are connected to intentions.

The proportionality condition now requires a positive evaluation by means of both stages.

On this two stage theory, shooting the innocent’s shirt in the initial case is wrong, as proportionality is violated at Stage I. On the other hand, the release of the prisoners may be permissible. For the freedom of the innocents is in the same ballpark as my life—it’s a big ballpark—even if they are going to be recaptured. It’s not a trivial good, like the taste of a mint.

I am not happy with this. It’s too complicated!


Heath White said...

FWIW, I think you are looking in the wrong place. It’s not the proportionality condition that needs finessing but the intention condition.

You can think of double-effect reasoning as placing further restrictions on consequentialist reasoning: everything forbidden by a consequentialist is forbidden by a double-effecter, but also forbidden are cases where your intention is evil. Thus it is key to have a useful concept of “intention” and your shirt-ripping case is one where the intention seems obviously engineered or gerrymandered. And this strategy would be available in pretty much any case of intended evil: just re-think it as intending some intrinsically okay bodily motion, with foreseen bad effects.

When Aquinas discusses action, he talks about the “object” which is distinct from the intention and is something like a species of action. I have never fully understood this concept but it doesn’t seem wholly determined by conscious mental states. It is more like “obvious moral category into which this action falls.”

When Anscombe discusses it in _Intention_, she is a Wittgensteinian (near behaviorist) about intention. She is clear that it is not the story you tell yourself in your head, but something like the inherent teleology of the action, that determines intention.

I think to avoid the gerrymandering problem you need a concept of intention which can’t be arbitrarily gerrymandered by conscious thought processes.

Alexander R Pruss said...

But we could imagine someone who really intends to see how the shirt rips. Imagine that they are aiming the gun at the shirt, and then notice there is someone wearing it. But noticing this need not change their intention.

Angra Mainyu said...


In the case in which you set the innocents free, it seems to me the intended effect is that the innocents manage to escape the country. One's actions (allowing them to escape prison) contributes to make that outcome likely, and contributes enough (all other things equal) to justify the unintended but even more likely bad effect - i.e., one's own death. It seems to me that that is indeed the intended effect.
So, if one removes from option 2 the sentence "The good effects that are causally downstream of the bad effects of one’s action don’t count.", but keeps the rest, that would seem to cover this case (I'm not saying that option 2 so modified is true, though; I'm saying it gets around this and similar objection).
One interesting (to me, anyway) fact about that scenario is that the promotion of the immoral behavior of the border officials (i.e., offering them bribes) by the escapees, is morally justified as a means of escaping the country. Similarly, when you let them go, you do intend that they escape the country, and are okay with their using bribery as a means - which you foresee. Moreover, I reckon that if they don't know how to escape the country, don't have enough info about the border officials to reckon that offering a bribe will probably succeed, etc., you may permissibly recommend them to bribe them (even let them know which officials are particularly likely to take bribes, if you know that), so in this case, the immoral behavior of the guards is intended as a means to an end.

I don't know whether you agree with my assessments about the permissibility (on your part or the innocents') of intending the immoral behavior of the guards as a means to a good end, though.

Angra Mainyu said...

One more thing: if the freedom of the innocents is in the same ballpark as your life even if they are going to be recaptured, then it seems to me that you don't need to consider the fact that they will escape the country, so it appears that even without the modification, option 2 survives the challenge (Side note: in my assessment, at least as long as you don't have family, close friends, etc., self-sacrifice even for a lesser good than your life is not immoral, but I'm leaving that aside).

Heath White said...

That's why I don't think "intention" in double effect can be quite the same thing as "intention" in contemporary folk psychology.

Alexander R Pruss said...


I have argued that a different, extensional rather than intensional, concept should be employed in double effect:

However, that particular solution doesn't help in this case. And I really think that in this case proportionality is what is at issue rather than intention. Compare a third case. The room is filling with poison gas that will kill all the innocents. One of the innocents is wearing a thick shirt whose chest pocket is made of fibers infused with an antidote. Everybody is tied down, and the only way to release the antidote is to shoot at the chest pocket---the gun is accessible. It seems that in this case double effect does justify shooting at the pocket in order to rend it in order to release the antidote.

But the intention to rip the shirt--both in the folk psychology sense and in the double effect sense--seems the same in this case. The difference seems to be with proportionality: in the first case, the good effect is downstream of the death of the shirt's wearer while in the present case, the good effect is independent of the death of the shirt's wearer.

Alexander R Pruss said...


The bribery case is trickier by my lights than by yours, since I think it is wrong to intentionally promote bad behavior.

I think in some cases it is permissible to bribe corrupt officials. For instance, it tends to be permissible to bribe corrupt officials to do something that they ought to do anyway. (I once saw something that said that the IRS allows you to write-off such bribes when they are necessary to doing business in a country with a lot of corruption.) In that case, you're just encouraging them to do the right thing rather than the wrong one.

The case at hand is a little bit more complicated, though, because in one sense the dutiful thing for the border guards to do is to release the innocents from the country, but I set up the case so that given the information available to the border people, it wasn't permissible for them to do it.

Angra Mainyu said...


But in that case, it seems to me one is still encouraging immoral behavior as a means to an end. For example, if Bob ought to allow Alice to cross the border with her suitcase because that's part of Bob's duty as a border guard, it's not the case that Bob ought to let her cross with her suitcase in exchange for a bribe. In fact, if he takes the bribe to let her cross, he's behaving immorally in taking the bribe. Isn't that still a case of promoting immoral behavior (by offering the bribe) as a means to an end (i.e., the just outcome that one is allowed to cross the border, etc.)?
Personally, I don't see a problem because I think encouraging immoral behavior is sometimes permissible (or even obligatory), and this would be one such case.

Anyway, going by your view that it's wrong to intentionally promote bad behavior, it seems that the innocents behave immorally by bribing the border guards, and so do you if you help them do that, advise them to do so, etc. I'm not sure whether it's also immoral in your view to intend that others behave immorally. If it is, then intending that the innocents flee the country by the stated means (which you did take into account in the scenario) is immoral.

At any rate, as I mentioned, as long as the freedom of the innocents is in the same ballpark as your life even if they are going to be recaptured, the behavior is permissible but compatible with Option 2.

As for the antidote, I don't think the intent is the same, at least if one does not restrict "intent" in a way that leaves aside precisely a morally relevant difference. In one case, the intent is to rip the shirt in order to learn more about the effects of the bullet on the fabric, whereas in the other case, the intent is to rip the shirt in order to save all of the innocents, except for one.
You could say the intent to rip the shirt is the same, but I'm not sure why that way of restricting intent would be preferable in this context.