1. BENATAR'S THEORY OF PERSONAL GOOD

First, some preliminaries. Existence is a benefit when it is better than non-existence, and a harm when it is worse.Footnote ¹ We may think of this using possible worlds. Most of us, since we are mere contingent beings, exist in some but not all possible worlds. A person's existence in a world is a benefit to her if, and only if, this world is better for her than a world she doesn't exist in; and a harm if, and only if, it is worse.Footnote ² A theory of what makes worlds better or worse for a person we may call a theory of personal good.Footnote ³ Such a theory, on this picture, may have implications concerning whether existence is a benefit or harm.

Benatar's theory is a variant of familiar hedonism: a person's good is taken to supervene on the pleasure and pain she experiences. The innovation is that how these considerations are weighed in pairwise comparisons of worlds depends on the person's existence: if she exists in both worlds under comparison, then both her pleasure and pain are taken into account; otherwise only her pain is.

More exactly, Benatar's view may be stated as the conjunction of two principles:Footnote ⁴

1. (P1) If a person exists in both of two worlds, then which of these worlds is better for her depends on both the pleasure and pain she experiences in these worlds. Other things being equal, the more pleasure she experiences in a world, the better it is for her; and the more pain, the worse.

2. (P2) If a person exists in at most one of two worlds, then which of these worlds is better for her depends solely on the pain she experiences in these worlds. The more pain she experiences in a world, the worse it is for her.

A person cannot experience any pain without existing, and therefore (P2) entails that if a person experiences any pain at all in a world, then this world is worse for her than a world she doesn't exist in. Benatar also believes that, as a matter of contingent fact, every person who ever lives will experience at least some pain. It follows that existence is always a harm.

2. WHAT'S WRONG WITH IT

Benatar's theory has a familiar problem – familiar, at least, to those who know the relevant literature on population ethics, especially the work of John Broome, which Benatar, inexplicably, ignores.Footnote ⁵ To illustrate, consider three worlds, A, B and C, and a person, Jemima, such that:

It follows from (P2) that A and B are equally good for Jemima, and so are A and C.Footnote ⁶ But it follows from (P1) that C is better for Jemima than B. This seems simply incoherent. How can one thing be equal in value with both of two things one of which is better than the other? That seems too much like saying ‘Claire is taller than Bob, and they are both the same height as Anne’.

Sometimes the relations ‘better’ and ‘equal in value’ are each defined in terms of a single primitive relation ‘at least as good’, as follows (where ‘≻’, ‘~’, and ‘≿’ stand for respectively ‘better’, ‘equal in value’, and ‘at least as good’):

• • x ≻ y iff (x≿y and y x);

• • x~y iff (≿y and y≿x).

Given this approach, we may say that what the above example shows is that Benatar's view is incompatible with the transitivity of ‘at least as good’. It implies that B≿A and A≿C (because B~A and A~C), but that B C (because C≻B).Footnote ⁷

Purported counterexamples to transitivity are hardly unheard of. Here's one.Footnote ⁸ Imagine a sequence of painful experiences: the first is two years of excruciating torture; the last is an incredibly long period with a mildly painful ‘hangnail’; and each experience in the sequence is slightly less painful but much longer than its predecessor. You might think that a slight reduction in intensity of pain cannot compensate for a very large increase in duration, and so you judge that each experience in the sequence is at least as good as the next. Given transitivity, then, the first should be at least as good as the last. But you might doubt this; you might think that two years of torture is always worse than a hangnail, no matter how long the latter lasts.

Table 1

Examples like this, however, differ in an important way from the one given above. The pain example depends on the idea that sufficiently many non-compensating differences can combine to make a compensating one. The small reductions in intensity of pain between adjacent experiences in the sequence do not compensate for the corresponding increases in duration; yet the large reduction between the first and last does compensate for the corresponding increase in duration (or so it is claimed). But the example above is not much like this. It doesn't involve a long sequence of options with some parameter being decreased (or increased) in small increments from one to the next. Even if we were to accept intransitivity in these small-increment cases, we should remain reluctant to accept the quite different sort of intransitivity implied by Benatar's theory.

Further problems arise when we move from personal to general good – to what is good overall, taking into account the good of everyone. The following seems a plausible principle linking personal and general good.

1. (P3) If a world is at least as good as another for everyone, and it is better for someone, then it is better.Footnote ⁹

Now suppose we add two people, Kieren and Leanne, to our previous example, as shown in the table above.

As before, no person experiences pain in any of these worlds. The numbers represent quantities of pleasure experienced, and a dash indicates non-existence. For example, in world A, Jemima doesn't exist, Kieren exists but experiences no pleasure, and Leanne exists and experiences some pleasure. Compare first A and B. (P2) entails that B is at least as good as A for both Jemima and Leanne, and (P1) entails that B is better than A for Kieren. (P3) then entails that B is better than A, because it is better for someone and at least as good for everyone. By parallel reasoning, however, it follows that C is better than B, and that A is better than C. So we have a cycle: A≻C≻B≻A. Again, this is hard to make sense of.

Anyway, as is well known, such cycles present difficulties for rational choice. How might a person who subscribes to Benatar's theory make a rational choice between the three worlds, A, B and C? Suppose, for example, she would choose A. Then, assuming she is rational, we would expect her to stick with this choice if we were merely to remove one of the forgone options, C. But in fact she wouldn't: given a choice between only A and B, she would choose what she judges to be the better option, B. We could run the same argument no matter which option she would initially choose (e.g. if she'd choose B, we could ask why she wouldn't stick with this on removal of A). So it seems impossible for her to make a rational choice.Footnote ¹⁰ In technical terms, the problem is that Benatar's theory entails a ‘choice function’ that violates what Amartya Sen calls ‘Property α’ (a.k.a. ‘Independence of Irrelevant Alternatives’).Footnote ¹¹

3. THE ASYMMETRIES

Is there anything to be said in favour of Benatar's theory? The advantage he claims is that it gives the best explanation of several asymmetries (Benatar Reference Benatar2006: 31–36). The main one, which I'll call ‘the obligation asymmetry’ may be stated as follows:

This seems intuitively plausible. Suppose a man and woman know that, were they to have a child, it would have a very happy life. Nonetheless, we might think, they would do nothing wrong by deciding not to have the child. On the other hand, if they know the child would have a miserable life, they would do something wrong by having the child. This is the asymmetry that Benatar seeks to explain.

Here's how he explains it. First, his theory of personal good entails another asymmetry:

He then assumes (a) that it is obligatory to bring a person into existence only if this would benefit her and (b) that it is permitted (i.e. not forbidden) to bring a person into existence only if this would not harm her. Given these assumptions, the obligation asymmetry follows from the benefit asymmetry.

An explanation of the obligation asymmetry would, I grant, be a helpful thing. I want to suggest, however, that it is not well explained in the way Benatar tries to explain it, that is, via the benefit asymmetry. As I shall argue, we have good reasons to reject the latter asymmetry.

Let a J-world be any world in which Jemima exists, and let w ₀ be a non-J-world. Then consider the following ‘hexalemma’:

1. (1) Some J-world is worse than w ₀.

2. (2) Not every J-world is worse than w ₀.

3. (3) If some but not every J-world is worse than w ₀, then some J-world is at least as good as w ₀.

4. (4) There is no best J-world.Footnote ¹³

5. (5) If one world is better than another, and this second is at least as good as a third, then the first is better than the third.

6. (6) No J-world is better than w ₀.

(For neatness, here I've abbreviated ‘better for Jemima’ as simply ‘better’, and so on.)

These six statements are jointly inconsistent. First, (1), (2) and (3) together imply that some J-world is at least as good as w ₀. Call one such world w ₁. Then, (4) implies that some other J-world is better than w ₁. Call one such world w ₂. Then, (5) implies that w ₂, a J-world, is better than w ₀. But this contradicts (6).

Statements (1) and (6) are the two sides of the benefit asymmetry, and so if we want to keep that asymmetry, we must sacrifice one of the other four, (2)–(5).

As we've seen, Benatar is forced to give up (5). This seems the worst option.

Another philsopher, Christoph Fehige, has proposed a theory – before Benatar did so – which supports the benefit asymmetry, but gives up (4) instead of (5) (Fehige Reference Fehige, Fehige and Wessels1998). According to Fehige's ‘antifrustrationist’ proposal, a world in which a person has no frustrated preferences is as least as good for her as any world.Footnote ¹⁴ This may be preferable to Benatar's view, but it still seems implausible. Consider a world where Jemima has no desires, or only very few or very trivial ones which are easily satisfied. Can we really not imagine a better world for Jemima than this? More generally, it seems plausible that for any really good J-world, we can conceive an even better one. Just imagine Jemima having a longer life, or having more friends, or being more successful, or whatever.

The hedonistic analogue of Fehige's theory is the view that avoiding pain is all that matters; pleasure never adds any value to a life. According to this view, there is a best possible J-world, indeed many of them. Any world in which Jemima experiences no pain, e.g. any world in which she doesn't exist, is at least as good for her as every world. Adopting this view would enable Benatar to retain the benefit asymmetry (and perhaps also the thesis that existence is always a harm, though this depends on the claim that life always holds at least some pain) without denying transitivity. But the claim that pleasure never matters just seems grossly implausible.

In support of (2), imagine a world in which Jemima has what we would normally regard as a very good life, virtually overflowing with pleasure, knowledge, beauty, and all that good stuff. Again, it seems implausible to claim that Jemima's existence in this world is worse for her than non-existence.Footnote ¹⁵

Finally, we have (3). It may be helpful to explain this as follows. Say two worlds are commensurable, with respect to a particular person's good, if either (a) one is better for this person than the other, or (b) they are equally good. Then the one possibility ruled out by (3) is this: some J-worlds are worse for Jemima than w ₀ and all the rest are incommensurable with w ₀. You might think (as I do, incidentally) that no J-worlds are commensurable with w ₀. How good a world is for a person, you might think, depends on how things are for her in this world. But in w ₀, there is no way things are for Jemima, because she doesn't exist, and so there is no basis for comparing w ₀ with other worlds with respect to Jemima's good. Suppose, however, you accept the antecedent of (3): you think some but not all J-worlds are worse than w ₀. Then, since you think some J-worlds are commensurable with w ₀, viz. the worse ones, it is unclear what reason you could have to deny that all are. In any case, it seems arbitrary to claim that the only J-worlds which are commensurable with w ₀ are the ones which are worse than w ₀ (unless you think every J-world is worse, which by hypothesis you don't).

So there is a significant cost to retaining the benefit asymmetry, a cost which in my estimation exceeds any gain. I favour instead rejecting this asymmetry and finding some other explanation of the obligation asymmetry. The symmetrical view I prefer (for reasons given briefly above) is that existence is neither benefit nor harm. This position is, I concede, not entirely comfortable. It deprives one of perhaps the most natural explanation of so-called ‘wrongful life’ cases, viz. that the person wrongfully given life is harmed by her existence.Footnote ¹⁶ Instead, one must say that in these cases procreation is wrongful, though no one is wronged; the world is made worse, though not worse for anyone. I do not have an alternative explanation to offer here. My point is just that it would be worthwhile to seek out such an explanation, since the alternatives, especially denying transitivity, do not look good.