Saturday, November 22, 2008

Do Merely Possible People Matter?

Consider the claim that the moral status of an act in world w depends only on the w-people. So, in particular, the moral status of actual actions depends only on actual people. This 'indexical principle' may be motivated by the thought that we don't have any obligation -- or any non-instrumental reason at all -- to bring more (possible) people into (actual) existence. The indexical principle implies this desired conclusion. But it also has other, less palatable, implications, as Melinda Roberts (drawing on Casper Hare?) pointed out in comments to a recent talk. In particular, the indexical principle implies that if we refrain from bringing into existence a bunch of slaves that would boost our welfare, we have done the wrong thing. This is because all the actual people (the only people that matter, according to the principle) would - ex hypothesi - be better-off in the slave world. The slaves might live intolerably bad lives, but since they are non-actual that is apparently of no moral relevance. Absurd.

Curiously, it looks as though the indexical principle further implies that whether an action is wrong or not may depend on whether we actually perform it (i.e. which world we end up actualizing). That violates the principle that Moral Status is Modally Robust:

(MSMR): if it is wrong to φ, then it would have been wrong to φ. (Put another way: if φ-ing is wrong in a world w where one doesn't φ, then in the nearest world w* where one does φ, one acts wrongly in w* by φ-ing.)

On the supposition that we refrain from bringing the slaves into existence, we saw above that the indexical principle determines this to be 'wrong'. Yet if we had acted differently, i.e. if we had actually brought the slaves into existence, they would then be actual people who mattered too, and so it would have been wrong to do that (and instead would have been right to do what we actually did, namely, refrain from actualizing the slave world). More generally, in a situation where w1 people do better in w2, and w2 people do better in w1, then when faced with a choice to actualize either w1 or w2, the indexical principle implies that whatever we actually choose will thereby be 'wrong', and the unchosen action will be 'right' (although it would have been wrong had we chosen it). This is doubly absurd.

To avoid the first absurdity, we must at least grant that harms to merely possible people in the worlds where they exist are morally significant. It's good to refrain from actualizing the slave world, because in doing so we prevent some morally significant harms: harms to the slaves in the slave world. Must we go all the way to accepting that the benefits of existing give us reason to bring (well-off) merely possible people into existence? Roberts rejects this last step, hoping to instead secure a 'middle ground' view. If I understand her view correctly -- i.e. as a more moderate version of the indexical view -- it would still seem to lead to some violations of MSMR: for example, supposing we will bring happy Jill into existence, it is wrong for us not to (since that would remove benefits for an actual person - Jill), but it wouldn't have been wrong had we chosen not to (since then Jill wouldn't have been counted among the actually existing).

The MSMR principle reminds us that a plausible moral theory cannot assess harms and benefits differently depending on which world is actual (or which choice one actually makes). So one cannot appeal to the mere 'actual non-existence' of Jill as a reason for not counting the benefits that would accrue to her were she to be brought into existence. Instead, if one really wants to hold on to this conclusion, one must make the universal claim that no-one is benefited by existence.

(Personally, I think we often will have reason to bring more good lives into existence, but I guess I don't really think this is for the sake of the otherwise non-existent person. Instead, it is an impersonal reason -- a benefit for the world.)


  1. How are we using the word 'world' here? The first paragraph makes it sound as if we can't be talking about possible worlds: in any possible world in which the slaves would be brought about, they would be w-people, because they would be people in that possible world. (And, since we are putting it in terms of actual people, the actual world is not a possible world.) But MSMR seems difficult to phrase in any terms other than those of possible worlds.

    (I don't think this affects the argument, since I'm sure it could be run in a different way; I'm just not quite clear on how the concept of 'world' is being used here.)

  2. Hey Richard,

    About the last paragraph: Could this be explained by an asymmetry between pain and pleasure (or benefit and harm) - the kind David Benatar argues for?


  3. Anlam - I'm not sure exactly what you have in mind -- can you say a bit more? (Aside: I argue against Benatar here.)

    Brandon - I'm talking about 'possible worlds' in the standard sense (i.e. possible world-states). I don't see the problem. Here are two possible worlds, with identical histories up to the branching point:

    (w1) We refrain from bringing any slaves into existence. (Everyone has 50 utils.)

    (w2) We choose to bring slaves into existence. (They have -999 utils, everyone else has 55 utils.)

    Suppose we refrain from bringing the slaves into existence. So w1 is actualized. So, in assessing this act, only the w1 people matter. So (by this metric) w2 is a better world, and our act (failing to bring about the slaves) was wrong.

    But suppose we bring the slaves into existence. So w2 is actualized. So the slaves (as w2 people) matter in assessing this act. So our act (bringing the slaves into existence) was a heinous moral crime.

    This is just the situation I discuss in the paragraph after I introduce the MSMR principle. In w1, where you phi, you should (by the w1 standards) instead not-phi; and in w2, where you don't phi, you should (by the w2 standards) phi. Whatever you do turns out to be wrong, and whatever you don't do is right. (But if you had acted differently, then the moral statuses would have been switched in such a way that then your new action would be wrong, etc.)

  4. Thanks, Richard. I suspected as much from the further course of the argument, but was confused by the switch back and forth between what is actual in the first paragraph.

  5. It is wrong to murder my mother. Thankfully, I haven't. The nearest world in which I intentionally kill my mother is not, though, a world in which killing her is wrong. This is because I love her. The nearest world is a world in which I am threatened with some much greater harm than the loss of my mother, say the destruction of the world. So if I were to kill my mother, I would not be acting wrongly. Seems to me then that MSMR is false.

    Am I wrong or do you need to build in constraints on nearness?

  6. Ha, good point! Yes, I was presupposing certain constraints here. Something like, "the nearest possible world (in which the circumstances of the decision are relevantly similar) where one phis".

    The intuitive idea seems clear enough, I hope, though perhaps one could raise questions about whether this can really be spelled out in a precise and rigorous way?


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.)