II. THE EPICUREAN CHALLENGE

Epicurus argued, or at least seems to have argued, that we should not fear death because death is not a misfortune for the one who dies. Ben Bradley considers the following broadly Epicurean line of reasoning:

Some contemporary philosophers have found this line of reasoning to be persuasive.Footnote ² As we shall see, however, a plausible, independently motivated theory of value can accommodate the common-sense view that death harms the victim. Although there might be things that are bad for us but not at particular times, I will not call premise (1) into question. Ordinary cases of bad things seem to confirm premise (1). For example, stubbing my toe is bad for me when I feel the resulting pain and my activities are hampered. The badness of death might seem special. But as far as possible, we should account for the badness of death in the same way we account for the badness of other, more ordinary, bad things. In the next section, I will argue that how bad something is for a person, all things considered, depends on how bad it is for her at relevant times. If we are subsequentists about the time of the badness of ordinary bad things, then we should prefer subsequentism about the time of death's badness. So, I will focus my critical attention on premise (2).

III. VALUE MATTERS

If we reject premise (2), we need to provide a satisfactory answer to the question: at which times is a person's death bad for that person? Before I discuss what I take to be the best answer to this question, I would like briefly to consider the kind of badness at issue.

Death is an evil of deprivation. When it is bad for a person, it is bad in virtue of what it prevents the person from receiving. So death is not intrinsically bad, but instead it is extrinsically bad. A person's well-being, at a given time, is a matter of the things that have intrinsic value for that person at that time.Footnote ³ Extrinsically valuable things are good or bad for a person – sometimes in very important ways – but typically have no, or zero, intrinsic value for the person. The overall value of a thing for a person is the combination of its intrinsic value, if any, and its extrinsic value, if any, for the person. The view of overall value that I favour is the counterfactual comparative account. On this view, roughly, something is overall good for me if I have more well-being with it than without it, and something is overall bad for me if I have less well-being with it than without it. As Bradley describes it, ‘the overall value of something for a person is the difference it makes to how things go for that person.’Footnote ⁴ The kind of value at issue is all-things-considered value, not prima facie or pro tanto value. For example, a flu shot might be all-things-considered good for young Ben even if it brings about short-term pain. In what follows, when I discuss value (good or bad), I will be discussing all-things-considered overall value unless I say otherwise.

It might be useful to consider a few more examples. Suppose that Kay punches Jay in the stomach. This is bad for Jay, because he would have been better off had Kay not punched him. The pain that Jay suffers is part of what makes the punch bad for him, but it isn't the whole story. We cannot say that something is bad for a person if it results in pain (or any intrinsically bad state), as the flu shot example shows. We cannot even say that something is bad for a person if it results in pain and no pleasure. Suppose that a doctor treats a dying cancer patient and thereby reduces, but does not eliminate, the patient's suffering. Surely the treatment is a good thing for the patient in the ordinary sense of the term ‘good’. It is good because the patient would have been even worse off without it. Finally, something can be bad for a person even if it does not result in anything intrinsically bad for the person (and arguably even if it has no effect, causally speaking, on the person). Bradley describes an example of this kind of preventive badness: ‘unbeknownst to me, Ned has left me tickets to a baseball game, and Hud has subsequently stolen them from my mailbox. I would have been much happier going to the game than staying at home, but I never find out about the tickets.’Footnote ⁵ It is intuitively clear that Hud's theft harms Bradley, or that it is bad for him. It is bad for him because he would have been better off had Hud not stolen the tickets – the theft makes a difference to how things go for Bradley.

A theory of value needs to accommodate all these cases, and the counterfactual comparative account far surpasses competing theories. On this account, something is good for me to a certain degree if I would have had that much less well-being had it not happened, and something is bad for me to a certain degree if I would have that much more well-being had it not happened. We have good reason, I think, to maintain that something harms a person if and only if it is bad for her, and that something benefits a person if and only if it is good for her. (The sort of harm at issue here is overall, all-things-considered harm.) So, for example, Hud's stealing the baseball tickets harmed Bradley, and the doctor's treatment benefited the cancer patient.

I take it that the things that harm and benefit us are particular, concrete events. A person's death is one example of such an event. Below are two versions of the counterfactual comparative account, one stated in ordinary language and one in terms of possible worlds (the accounts will presuppose that the given event is one that actually occurs).

The Comparative Account, as formulated above, is very similar to views defended in the literature.Footnote ⁶ I will make use of the ordinary language version as much as I can, but the possible worlds version will facilitate discussion of certain points. If the value of an event for a person is positive – that is, if the amount of well-being she receives is greater than the amount she would have received if the event had not occurred – we can say that the event is good for her and benefits her; if the value is negative, we can say that the event is bad for her and harms her.

Let us now return to the question about time. Given that an event is bad for a person, when is it bad for that person? It is important to note that this is not a question about when the bad event takes place. Instead, it is a question about when the event is bad for its victim. Let's consider the earlier examples of events that are bad for people: my stubbing my toe, Kay's punching Jay, and Hud's stealing the tickets that Ned had left for Bradley. It seems clear that each of these events is bad for its victim after it occurs. The toe stubbing is bad for me when I feel the pain and cannot do things that I would have enjoyed, the punch is bad for Jay when he feels the resulting pain, and the theft is bad for Bradley during the game he would have enjoyed watching. We should say the same thing about death. If somebody's death is bad for her, it is bad for her when she would otherwise have been well off. So here, I think, is the most plausible view that will let us reject premise (2) of the Epicurean argument presented earlier (below I provide only the possible worlds version):

On this account, the relevant time might be an instant or a duration of time. (When applied to a duration of time, the term ‘well-being level at t’ picks out the amount of well-being received over the duration.) As before, positive value implies goodness or benefit, and negative value implies badness or harm. Suppose, for example, that Kay punched Jay moments before noon. The value of the punch for Jay at noon, then, will be well below zero since his actual well-being level at noon is much lower than it would have been, had Kay not punched him. At some point in time a few minutes later, the value of the punch for Jay might still be negative, but closer to zero. The punch is very bad for Jay at noon, but not quite so bad a bit later. Let's suppose that as of 1 p.m., Jay's life continues on just as it would have if Kay had not punched him at all. Then, the punch was bad for Jay during the one-hour interval from noon until 1 p.m. We could also tell how bad it was by comparing Jay's well-being during that interval in the actual world with his well-being then in the nearest world where Kay did not punch him.

These considerations suggest that the relevant facts about badness are determined by facts about badness at times. The same goes for goodness and neutral value. So, in an important way, Comparative Value at a Time is more basic than the Comparative Account. (I will not explore here how facts about value for people are grounded in facts about value for people at times.)

I want to defend subsequentism – the view that death is bad for the victim after the victim has died – and I also think that Comparative Value at a Time is the correct account of when things are bad for us. If I am right, then we should be able to make sense, in some way or another, of a dead person having a well-being level. Many philosophers have argued, however, that people cannot have any level of well-being at times at which they are not located, and at possible worlds in which they do not exist at all.

Before turning to the problem of the subject, I should note that all by itself, Comparative Value at a Time does not entail subsequentism (even given that death can be bad for the victim at certain times).Footnote ⁸ For example, it is compatible with priorism, the view that death is bad for the victim before she dies.Footnote ⁹ Those who think that a frustrated desire, for example, is intrinsically bad for the one who has it might be able to use a view like Comparative Value at a Time to conclude that death is bad for the victim before she dies – in particular, at those times when she has the desires that are thwarted by her death.Footnote ¹⁰ However, priorism faces a host of difficulties. As we saw earlier, subsequentism seems true of ordinary, non-death-related, harms. It seems clearly wrong to say that stubbing my toe was bad for me before I stubbed it. So, priorists seem unable to give a unified theory of overall harm. I hope to remain as neutral as possible with respect to the notion of well-being while trying to resolve the problem of the subject for subsequentism.

IV. THE PROBLEM OF THE SUBJECT

Epicurus remarked that ‘when death comes, then we do not exist.’Footnote ¹¹ This remark raises a problem that Joel Feinberg characterized as follows: ‘there cannot be harm without a subject to be harmed, and when death occurs it obliterates the subject, and thus excludes the possibility of harm.’Footnote ¹² Epicurus and Feinberg accepted the Termination Thesis, the claim that when one dies, one simply ceases to exist.Footnote ¹³ I will assume this is true, even though I don't really believe it. The main reason for this assumption is that if death can be bad for a person who has died but somehow still exists (as a corpse), then surely it can also be bad for a person who has died and ceased to exist (for example, someone near ground zero of a large nuclear explosion).

Feinberg supposes that a person can be harmed at a time only if the person exists at that time. This sort of claim about harm or badness is sometimes called the Existence Requirement or the Existence Condition.Footnote ¹⁴ Without additional argument, however, there is no good reason to believe this. Being harmed is a relational property – you are harmed provided that there exists something that harms you. People can be eulogized after they cease to exist, people can be remembered after they cease to exist, and so on. If we can have properties like being eulogized and being remembered at times when we do not exist, then there is no purely metaphysical reason why we cannot have the property of being harmed at such times. As long as we exist simpliciter, or at some time or other, we can have properties like these – we do not need to be present, or as I shall continue to say, located, at the times when we have them.Footnote ¹⁵ Even if the only things that exist simpliciter are things that presently exist, as Presentists claim, there must be a good way to accommodate the truth (or at least the apparent truth) of claims that seem to be directly about past objects: for example, ‘Jones, who died last week, is being eulogized today’. If this cannot be done (as I think), then so much the worse for Presentism; if it can be done, then Eternalists about time (those who think that past, present, and future objects and times are equally real) can simply use the relevant strategy to accommodate claims about harm to past people.Footnote ¹⁶

The challenging problem arises when we consider how subsequentists have to apply a view like Comparative Value at a Time. Suppose that Jones died prematurely last week, and that had Jones not died, she would have been quite well off right now. In order to say that her death is bad for her at the present time, when she is dead, subsequentists need to compare Jones’ present level of well-being with the positive level that she would have had if she had not died. But this seems problematic, since Jones is dead and does not seem to have a present level of well-being.Footnote ¹⁷ Given the comparative framework, the general argument against subsequentism goes something like this. For any x, if x is bad for a person at a time, the person's level of well-being at that time must be lower than what the level would have been without x. However, one must exist at a time to have any level of well-being at that time, even zero. So, nothing can be bad for a person at a time unless the person exists at that time. But given the Termination Thesis, if death is bad for the dead when they are dead, then it is bad for them when they do not exist. The rejection of subsequentism now follows: death is not bad for the dead when they are dead.

If this line of reasoning is correct, then we need to claim that the badness of death is located at some other time (or at no time at all), or we need to accept the Epicurean conclusion that death is not bad for the one who dies. As I see it, the first of these strategies leaves us with implausible and unwieldy accounts of overall value, and the second leaves us with uncomfortable positions on the wrongness of killing and related matters.Footnote ¹⁸ A much more attractive option is to attack the argument's key premise, that is, the claim that a person must exist at a time to have any level of well-being at that time. (I will try to show in section 7 that even if there is a sense in which the premise is true, this does not threaten subsequentism since the threatening reading is false.)

In population ethics, a similar line of reasoning has been used to argue that it is impossible to harm or benefit someone by bringing her into existence. In order to harm (or benefit) somebody, the argument goes, the procreative act must make her intrinsically worse (or better) off than she would have been had it not occurred. But no world in which a given person fails to exist can be intrinsically better for that person than a world in which she exists. A judgement of betterness like this would seem to require comparing the created individual's actual well-being level with the well-being level that she would have had, if she had not been created. One must exist in a world, however, in order to have any well-being level at that world, and so no such comparison can be made. The conclusion – that it is impossible to benefit or harm by creating – is striking and might have serious implications for wrongful life and related claims.Footnote ¹⁹ I think the best option is to reject the key premise, that is, the claim that one must exist in a world in order to have any level of well-being at the world. We can do this while accommodating its apparent truth.

Why think that a person can have a well-being level at a time, or at a world, only if she is located at that time or exists in that world? Let's focus on time, since any reason for lacking a well-being level at a time when one is not located is thereby a reason for lacking well-being at a world where one does not exist. Steven Luper argues that a dead person has no well-being level, and cannot be harmed while dead, because the dead are not responsive. He stipulates that an individual is responsive at time t ‘if and only if its well-being may be affected at t – rising if certain conditions are met, and falling if certain other conditions are met’.Footnote ²⁰ Luper does not defend a view of well-being, but he sees responsiveness as a matter of an individual's capacities. For example, if hedonism is true, then responsiveness ‘consists in the capacity to experience pleasure or pain’.Footnote ²¹ Luper's remarks suggest an argument like this. A person has a well-being level at time t only if at t, the person can – that is, has the capacity to – gain or lose well-being. But having this capacity at a time requires being located then. So, a person has a well-being level at only those times at which she is located.

I see no good reason to accept the first premise. We might imagine someone with a traumatic brain injury who lacks the capacity to experience pleasure or pain. The hedonist should say that this individual's well-being level is zero. This sort of criticism is developed by Bradley:

The critical point does not depend on the truth of hedonism, since the example could be adapted to fit other theories of well-being. So, Luper's argument appears to prove too much.

I should say that I am quite comfortable with saying that a shoe has zero well-being. After all, the shoe, like Greg and Marsha, never feels any pleasure or pain. There are no differences among their relevant, current hedonic states, since there are no such states. More generally, the shoe lacks the properties in virtue of which it would accrue any positive or negative well-being. (More on all of this later.) The challenge for Bradley is that if a dead person has zero well-being and a shoe has no well-being level at all, there must be something that grounds the difference. What could it be? Bradley suggests that the answer is this: there is some possible world and time at which the person has a positive or negative well-being level, but this is not the case for the shoe.Footnote ²³ Here we have a notion of responsiveness that is based in metaphysical possibility instead of capacities, which seems to get the case of Greg and Marsha right. I have some sympathy with it, but again, I do not think it is necessary.

Perhaps the best reason to think that one cannot have a well-being level at a time unless one is located at that time has to do with intrinsic properties more generally. It seems right that an individual cannot have an intrinsic property at a time unless it is located at that time. But the claim that a property of the form having a well-being level of n is intrinsic (however plausible this might be) begs the question against certain widely held views about well-being. On desire satisfaction theories a person's well-being level can rise or fall with things that happen outside of her. If I have a desire to be famous, for example, my level of well-being will rise (other things being equal) if many people come to know about me. Those who wish to argue that the dead cannot have a level of well-being should avoid assuming that desire satisfaction theories and related views are false. A good way to do this is to claim that having a well-being level requires having some intrinsic properties, whether or not the property of having a well-being level of n is itself intrinsic. This claim is attractive. (An anonymous reviewer points out that even desire satisfactionists should agree that having some well-being level is intrinsic, since having a desire seems intrinsic.) If it is correct, and if having an intrinsic property at a time requires being located at that time, then a person cannot have a well-being level at a time (not even zero) without being located at that time. I take this to be the most serious problem of the subject for subsequentists, and for those of us who think it is possible to harm or benefit by creating. We can state the two fundamental claims of the problem of the subject, then, like this:

If we were interested only in cases where a given person never exists, we would not need to bring in intrinsic properties – we could simply talk about well-being requiring properties, and properties requiring existence. But we are also interested (at least I am) in cases where a person goes out of existence. So, PS1 and PS2 constitute a fully general problem for any comparative account of harm and benefit. We need a good way to reject one of them. In what follows, I will consider and discuss several ways in which we might attempt to solve the general problem.

V. COMPARATIVE HARM WITH NO WELL-BEING LEVEL

There is a way to claim that death can harm the victim, and that procreative acts can harm or benefit the person who is created, without giving up on the comparative approach to harm and without rejecting PS1 or PS2. We can revise Comparative Value at a Time. Let's reconsider it:

It is easy enough to achieve the effect of the dead and the non-existent having a well-being level of zero while accepting the claim that they in fact have no level of well-being at all. For example, consider the revised version of Comparative Value at a Time below.

Let's apply this account to a couple of examples.Footnote ²⁴ Suppose that event E is the creation of a happy person, S, and that t is a time at which S has a positive level of well-being, let's say + 10. Suppose that S would never have existed – and so, we are presently supposing, would never have any level of well-being – if E had not occurred. Here, condition (b) applies. The value of E for S at t, then, is + 10, which is the difference between S's actual level of well-being and zero. In this case, E benefits S, or is good for S, at t. Now suppose that E is the premature death of an already existing person, S, and that t is one week after S's death. Although S does not exist at t – and so, we are presently supposing, does not have any level of well-being at t – S would have been doing well at t if E had not occurred. Let's suppose that at the nearest world where E does not occur, S's level of well-being at t is +10. Here, condition (c) applies. The value of E for S at t is −10, which is the difference between zero and S's well-being level at the nearest world where E does not occur. In this case, E harms S, or is bad for S, at t. Both of these results are very plausible, and the modified account generates them without any commitment to the claim that a person can have a level of well-being at a time even if the person is not located at that time.

Modified Comparative Value at a Time will strike many as ad hoc. Why insert a level of zero into the calculations if there is really no well-being level at all? John Bigelow, John Campbell and Robert Pargetter, for example, argue that the dead have no well-being at all in part because attributing a level of zero to them would be ‘arbitrary’.Footnote ²⁵Modified Comparative Value at a Time allows that dead people have no well-being level, but it does assign a zero value to any time at which a person has no well-being at all. However, this assignment does not seem arbitrary. Here is one reason. It is very plausible to think that a person who is not located at a given time is not, at that time, receiving anything that would make her either well off or badly off then. Arguably, a theory of well-being should assign a zero value when this is the case. Another reason has to do with rational indifference to alternatives in which, on the one hand, we exist but nothing has value for us, and on the other hand, we do not exist at all.

Bradley makes this point with the following fanciful example:

Bradley stipulates that comatose people have zero well-being levels (and notes that those who disagree may simply change the example). Given that Kris will have a well-being level (of zero) in F₂, and that it is rational for him to be indifferent between F₁ and F₂, it is quite plausible to think that he will have the very same well-being level (that is, zero) in F₁. We will return to this argument in the next section. For now, let's note that Modified Comparative Value at a Time is consistent with a softer conclusion. It allows that Kris has no well-being level at any time in F₁ – because he does not exist – but would then assign a zero level for those times. Given the facts about rational indifference, this does not strike me as arbitrary.

Having said all this, I think that we can give a more illuminating account of overall value than the one formulated in Modified Comparative Value at a Time. I am inclined to think that this account is true. I am also inclined to think that, at least on one natural way of understanding the locution ‘well-being level at t’, its conditions (b) and (c) are not redundant. However, I also think there is a more basic treatment of overall value at a time from which these facts flow. In the next section and especially in section VII, I hope to clarify and defend these claims.

VI. THE ZERO VIEW

Suppose that we have an actual, already existing person, who is well off to a certain degree at a certain time. It would be very nice to be able to compare how well off the person is at this time with how well off she is at any time at any possible world. A plausible way to do this is to assign a well-being level of zero to people at times at which they are not located. Here is a version of the needed claim formulated in terms of possible worlds:

Defenders of the Zero View will reject the problem of the subject's key premise, the claim that a person must exist at a time to have any level of well-being at that time.

Bradley uses the considerations about rational indifference to argue for the Zero View. Remember that in future F₁, Kris dies instantly, and in F₂ he goes into a comatose state and dies in ten years, never having regained consciousness. Bradley maintains that in so far as Kris cares about his own well-being, he should be indifferent between F₁ and F₂.Footnote ²⁸ If this is the case, Bradley argues, then Kris's level of well-being is the same in each possible future. This shows that a well-being level (of zero) can be assigned to a person at a time when he is dead, since Kris has a well-being level (of zero) at times when he is comatose.

David Hershenov notes that a person can be indifferent in the relevant way between two possible worlds in which she never exists at all. He concludes that ‘people can be indifferent to a pair of scenarios without the indifference being due to there being levels of well-being that are equal.’Footnote ²⁹ If this is right, then Bradley's argument is unsound. In particular, Kris's indifference between F₁ and F₂ would not imply that he has the same level of well-being in each future. Bradley points out that Hershenov has assumed that a person does not have a well-being level at a world in which she never exists – even an actual person, who might be considering scenarios in which she does not exist. Bradley, focusing on the idea of well-being at times after death, notes that Hershenov's assumption about worlds is no more or less plausible than the claim that a person who is not located at a time has no well-being level at that time.Footnote ³⁰ Arguably, then, Hershenov begs the question since this claim is the very one at issue.

We might think that the two claims are not quite on a par. Perhaps it is more plausible to claim that one who does not exist in a world has no well-being level there than it is to claim that one who is not located at a time has no well-being level then. Jens Johansson argues along these lines.Footnote ³¹ The idea is that to have a well-being level in a world or at a time, a person must at least have properties in the world or at the time. Having properties in a world requires existing there. This has been called The No Properties of the Non-Existent Principle, and also Actualism.Footnote ³² It seems to be correct. I do not wish to deny it and my own view is consistent with it. Having properties at a time, however, does not seem to require existing then (that is, being located at the time). As we have seen, people can have properties, like being eulogized or being remembered, when they do not exist. Arguably, this impairs Bradley's objection to Hershenov's point that one's indifference between a pair of scenarios need not be due to one's having equal levels of well-being in them. It also emphasizes the challenge to the Zero View as it applies to worlds where a given actual person never exists at all. How can my level of well-being be zero at any time in a world where I do not exist, if having a well-being level requires having properties and I have properties only where I exist?

What if a person's level of well-being at a time could be grounded in the fact that she lacks certain properties at that time? This might show that PS1 – that a person has a well-being level at time t only if she has intrinsic properties at t – is vulnerable. Several defenders of the Zero View have pursued this line of reasoning. Nils Holtug, for example, argues as follows:

Holtug accepts that an individual cannot have any properties unless it exists. His argument might provide a plausible way to reject PS1 that is consistent with this principle. Krister Bykvist objects by claiming that ‘having value for’ expresses a relation that holds between states of affairs and people. If a person does not exist in a given world, Bykvist argues, she cannot have any properties or stand in any relations there. So, if a person does not exist, then nothing can have zero value for her.Footnote ³⁴ I suspect that Bykvist's response, even if it is correct, would have no force against a view that dispenses with talk of (the state of affairs of) a person's non-existence having zero value for her, and focuses instead on a person's well-being level.

So, let's take a look at another argument for the Zero View that lets us reject PS1. Melinda Roberts considers a case in which an actually existing person, Nora, is born with a disease that leads to an anguished existence for her. She could not have existed without this disease – her parents could have avoided creating a flawed child only by having no child or a numerically different child. Nora's overall level of well-being is negative. Roberts argues that it would have been better for Nora never to have existed at all than to have the anguished life she actually has. Roberts does this by arguing for the Zero View on the basis of lacking properties. In particular, she argues that Nora's level of well-being is zero at any world at which Nora never exists at all,

Consider a possible world, w, where Nora does not exist. Roberts's argument, if sound, lets us say that Nora has a well-being level of zero at w, without giving up on the principle that an individual cannot have any properties unless it exists. We might consider an objection that is similar to Bykvist's response to Holtug. If Nora has a well-being level of zero at w, she must have at least one property there – namely, having zero well-being – and so Roberts's account is not, after all, consistent with the claim that an individual cannot have any properties unless it exists. Roberts considers this argument and she responds, in part, as follows: ‘when we say that Nora has zero wellbeing at β, we should eschew the idea that we are attributing to her the property of having at β some level of wellbeing . . . We are, rather, denying that she has at β any properties at all.’Footnote ³⁶

In further support of this, Roberts notes that not every predicate expresses a genuine property (for example, ‘is not a member of itself’) and also that when we say something like ‘George W. Bush would never have been president if he had not existed’, we do not take the predicate ‘would never have been president’ to attribute a property to Bush at a world in which he does not exist. Johansson argues that these points do not apply to well-being. After correctly pointing out that, in the context of the present debate about well-being, we need not understand talk of properties and relations to imply a commitment to universals, he says: ‘the important thing is not whether there are properties, but that Nora has to exist in order for the predicate “has zero well-being” to apply to her. She clearly has to, because of the logical equivalence of “Fx” and “(∃y)(y = x & Fy)”.’Footnote ³⁷

These formulas are equivalent in classical predicate logic, but not in free logic. Moreover, our semantics must be free if we take names to be rigid designators and we quantify over the domains of possible worlds where their referents fail to exist (as all sides in the debate seem to do). I think that if we take the right perspective, we can make sense of Roberts's view and also explain away the apparent tension between having zero well-being at a world, and having no properties at all at that world. I will try to show this in the next section.

Bradley also uses the idea of lacking certain properties to defend the Zero View. Bradley is a hedonist but the following argument could be recast to fit other views of well-being:

Again, we have a compelling reason to reject PS1 and solve the problem. Bradley, however, worries about negative properties like the properties of not being pleased and not being in pain. Since the corresponding positive properties are intrinsic, these seem intrinsic too. He continues:

This seems to be a big problem. I do not see how it follows, from the fact that Bradley has the properties of lacking pleasure and lacking pain when sitting in his chair, that he also has them at times when he is not located. It seems more plausible that when he is in his chair, he has the negative properties in virtue of lacking the positive ones, but when he is not located, he simply lacks them all. If we consider a world in which Bradley never exists, this becomes even more plausible – we should say that he has no properties whatsoever there. However, this line of reasoning leads Bradley to reject PS2 – that a person has an intrinsic property at time t only if she is located at t – instead of PS1. He notes that thinking of paradigmatic cases of intrinsic properties (for example, shapes) gives support to PS2. But, he claims, other cases (for example, trivial or necessary intrinsic properties such as being round or not round) make it far less obvious that PS2 is true. Since negative intrinsic properties are not paradigmatic cases, he concludes: ‘So we might as well say, of negative properties such as not being pleased or not being in pain, that a person can have such properties at times at which the person is not located.’Footnote ⁴⁰

This doesn't seem very plausible for times at which an actual person is not located, and it is far less plausible for times at worlds where an actual person does not exist. We should admit, for example, that there is a possible world in which Bradley neither exists nor has any properties. So it seems to me that the defender of the Zero View would do better to deny PS1 than PS2. Unless Bradley wants to understand the truth of something like S has the property of not being pleased at t (where t is a time at which S is not located) as consisting merely in the truth of the proposition that S is not pleased at t – or, if you prefer, merely in the truth at t of the proposition that S is not pleased – we should refrain from accepting his claims about negative properties.

More recently, Bradley has called PS1 into question by giving up on the idea that well-being is a property that a person has. Bradley instead makes the following suggestion: ‘I propose that well-being is not an intrinsic property an individual has, nor a relation an individual has to something else in the world, but rather a relation between individuals and worlds (and times), that can be represented as a function from individuals and worlds to numbers.’Footnote ⁴¹ This is reminiscent of Fred Feldman's notion of the value of a world for a person, which he sometimes calls the person's ‘welfare level’.Footnote ⁴² Bradley argues that the function is complete – and thereby assigns values for worlds (times) at which a person does not exist – on the basis of considerations we have seen earlier: we can have preferences, for our own sakes, for or against alternatives in which we do not exist.

I think this gets things backwards. I agree that we can consider a given person, a possible world, and a time, and come up with a number that we can identify with that person's level of well-being at that world at that time. However, it seems to me that the concept of well-being is prior to any concept of a relation between person and world (and time) that determines such a number. In other words, to talk about the value of a world (and time) for a person is really just to talk about how well off the person is at that world (and time), that is, the person's well-being level. So, an independent solution to the problem of the subject is needed.

VII. ZERO WELL-BEING IN ABSENTIA

It seems right that lacking certain properties (for example, pleasure and pain) makes it the case that a person has zero well-being. And alternatives in which we do not exist can clearly figure in the preference rankings we make, for our own sakes. Considerations like these make it plausible to say that if a person does not exist in a world or is not located at a time, then the level of well-being she has at that world or time is zero. As we have seen, certain philosophers have accepted the Zero View on the basis of these considerations. On the other hand, the view that a person must exist in order to have any level of well-being is compelling. If an anguished person (for example, Nora) had never existed, there would have been no person for whom this is better than a miserable existence.Footnote ⁴³ As we have seen, certain philosophers have denied the Zero View on the basis of these considerations. In this section, I want to argue that for each position in this debate, there is a sense in which it is correct. However, as I will show, the sense in which it is correct to deny the Zero View does not threaten our ability to make comparative judgements between levels of well-being for living, dead and non-existent people. In particular, it does not threaten views such as Comparative Value at a Time.

I will begin by making use of some ideas from the last section. Facts about a person's level of well-being are grounded by facts that are sometimes called value atoms or basic intrinsic value states. For purposes of illustration, let's assume that hedonism is true. The value atoms, then, include facts of the form S is pleased to degree n at t. If somebody experiences equal amounts of pain and pleasure at (or during) a given time, her well-being level at that time is zero. The good and bad value atoms balance each other out. As we have seen, there is another way for a person to have a well-being level of zero at a time – that is, by failing to experience any pleasure or pain at all. So, facts about a person's well-being can also be grounded in (negative) facts about her lacking certain properties. If there is some time, t, such that it's not the case that Greg is pleased at t, and it's not the case that Greg is in pain at t, then Greg's well-being level at t is zero. To put it another way, the propositions that it is not the case that Greg is pleased at t, and that it is not the case that Greg is in pain at t, together entail that Greg's well-being level at t is zero. This is uncontroversial if Greg is an actual person and t is a time at which he is located.

But what about times at which the person is not located? Let's reconsider the case of Nora, an actual person whose life is so full of anguish that her lifetime level of well-being is negative. We might ask: did being brought into existence harm Nora, either all-things-considered, or, as I will consider here, at a given time, t? I argued that the way to answer this question is to make use of Comparative Value at a Time. To do this, we need to compare Nora's actual well-being level at t with her well-being level at t at the nearest possible world where she is not brought into existence. Let's suppose that this world is w. I want to show that the considerations from the previous paragraph imply that Nora's well-being level at t at w is indeed zero.

Given the apparatus of possible worlds, it is hard to deny that a proposition is possibly true if and only if it is true at some world. Accordingly, it is hard to deny that the proposition that Nora does not exist is true at, or relative to, w. Likewise, the proposition that Nora is not pleased at t is true at w.Footnote ⁴⁴ The proposition that Nora is not in pain at t is true at w. These two propositions, as we have seen, together entail that Nora's well-being level at t is zero. As a result, this proposition is also true at w. So, Nora's well-being level at t at the nearest possible world where she is not created, namely w, is zero. Since her actual well-being level at t is negative, Comparative Value at a Time implies that being brought into existence harmed Nora. This is surely a plausible result.

I want to stress that I am not arguing that individuals can be harmed by failing to exist or by being prevented from existing. The Loch Ness monster and Barack Obama's (merely possible) third child have not been harmed. This follows from my commitment to Actualism. Only actually existing individuals can be harmed or benefited.

In order to get clearer on just how this account solves the problem of the subject, we need to appreciate a well-known and, I think, independently motivated distinction. Let's start with an example. Consider an actually existing person, say, Barack Obama. Obama might not have existed, and so the proposition that Obama does not exist is possibly true. Now consider a world at which the proposition is true. We might ask how the proposition can be true at this world even if it does not exist there – since this proposition exists only if Obama does – and even if Obama does not exist there to have the property of non-existence, or to have the predicate ‘does not exist’ apply to him.Footnote ⁴⁵ The answer has to do with a distinction made in several places by Kit Fine.Footnote ⁴⁶ The idea is that there is an inner and an outer notion of truth for propositions. Relative to a given possible world, a proposition is true according to the outer notion provided that it represents the world correctly; the proposition need not exist in the world. On the other hand, a proposition is true according to the inner notion provided that it correctly represents and exists in the given world.

Let w₁ be the world where Obama does not exist. The proposition that Obama does not exist is not true relative to w₁ according to the inner notion, but it is true relative to w₁ according to the outer notion. Following Robert Adams, let's say that the proposition is not true in w₁, but is true at w₁.Footnote ⁴⁷ We can make sense of world w, where Nora does not exist, in this way. The proposition that Nora is not pleased at t is not true in w, and the proposition that Nora is not in pain at t is not true in w. However, both of these propositions are true at w, that is, they are true according to the outer notion of truth. Accordingly, the proposition that Nora's well-being level at t is zero is also true at w. By focusing on the outer notion of truth, that is, truth at a world, we can maintain the counterfactual comparative account and dissolve the problem of the subject. Not everyone is sympathetic to the inner vs. outer distinction, but I find it irresistible. We need outer truth, it seems, to make sense of counterfactual possibility: we can describe a counterfactual situation by saying that Nora does not exist in it; we can say truly that if Nora's parents had never met, Nora would not have existed; and so on. And with outer truth, ordinary counterfactuals (about actual persons) like ‘If Nora had never existed, then she would have had a well-being level of zero’ come out true.

My account respects the principle that if an individual has a property in a given world then it exists in that world.Footnote ⁴⁸ Nora, for example, neither exists nor has any properties in w. In particular, she does not have the property of having zero well-being. This is strict and literal. Of course, we might want to loosen up a little bit and say that Nora has the property of having zero well-being at w, being careful to keep in mind that all we mean by this is that the proposition that Nora's well-being level is zero is true at (not in) w. Fine makes this point as follows:

In this way we might say that Nora has (negative) properties like lacking pleasure and pain at w, even though, strictly and literally, she has no properties in w at all. When I talk about properties in what follows, however, I will be talking strictly and literally.

I suggest that the account presented here provides a plausible way to compare the well-being levels of actual persons at alternatives in which they are living, dead, or non-existent. But how exactly does it solve the problem of the subject? Earlier, I argued that the problem turned on something like this key premise: a person must exist at a time to have any level of well-being at that time. Ideally, we would like to accommodate the apparent truth of this premise, but also – to solve the problem – find a way to reject it. We can do exactly this.

If this premise is going to imply that an actual person, such as Nora, cannot have any level of well-being at a world where she does not exist, then we need to understand it along these lines: For every person S and time t, necessarily, if S has a well-being level at t, then S is located at t. Let's consider an instance of this: Necessarily, if Nora has a well-being level at noon, then Nora is located at noon. As we have seen, the proposition that Nora's well-being level at noon is zero is true at world w, where Nora does not exist. So the antecedent Nora has a well-being level at noon is true at w, while the consequent Nora is located at noon is false at w. This makes w a world at which the entire conditional is false, and so there is a clear sense in which it does not hold necessarily. This is really all we need to solve the problem of the subject.

We can even admit that there is a sense in which the claim about Nora is true. We might read it like this: For every world w, if Nora has a well-being level at noon in w, then Nora is located at noon in w. If we take Nora has a well-being level at noon in w to hold if and only if the proposition that Nora has a well-being level at noon is true in (not just at) w, then the whole claim turns out to be true (there is a complication that I will address in the next two paragraphs). So, there is a way to accommodate the apparent truth of the key premise. As long as we can replace ‘in’ with ‘at’, however, we have no serious problem of the subject.

I have been concentrating on alternatives in which a person never exists. Let's turn to the times after a person's death, at which the person is not located. Let's consider an actual person, Michael Jackson, and suppose here that t is a time shortly after his death. Now, the relevant instance of the key premise is this: Necessarily, if Jackson has a well-being level at t, then Jackson is located at t. Here, the antecedent is true at the actual world, for the same reasons as the corresponding claim about Nora (Jackson feels no pleasure or pain at t, which entails that his well-being level is zero then). Since Jackson is not located at t, the conditional is false at the actual world, and so there is a clear sense in which it is not a necessary truth. This is really all we need to solve the problem of the subject. Ideally, however, and as noted above, it would be nice to admit a way in which the premise is true. The trouble is that the claim about Jackson seems false even if we read it like this: For every world w, if Jackson has a well-being level at t in w, then Jackson is located at t in w. Since Jackson exists in the actual world, so does the proposition that Jackson has a well-being level at t. So, the antecedent seems true in (not just at) the actual world. Since Jackson is not located at t, the whole claim seems false even if it is read as above.

This is a problem only if we index truth to worlds conceived as entire universes. But if we index truth to times, or world-time pairs, the complication disappears. Let @ be the actual world and t the relevant time at which Jackson is not located. Jackson exists in @, but since he is not located at t, he does not exist in the world-time pair <@, t>. So, given the earlier considerations, we can say that Jackson has a well-being level (of zero) at, but not in, <@, t>. This would give us a sense in which the premise about Jackson is true (the ‘in @’ sense), but this sense would not give rise to any problems (since the ‘at @’ sense is false).

These considerations also apply to PS1, which I claimed was the best reason to believe that one must exist in order to have a well-being level. Here is PS1 again: a person has a well-being level at time t only if she has intrinsic properties at t. If this means that a person has a well-being level in a world-time only if she has intrinsic properties in that world-time, then it is true. But if it means that a person has a well-being level at a world-time only if she has intrinsic properties at that world-time, then it is false. So, the support for the problem of the subject's key premise can be subjected to the same critical analysis as the premise itself.

Let's take one last look at the counterfactual comparative theorist's basic account:

The preceding discussion gives us an acceptable way to understand the locution ‘S's well-being level at t’ as it occurs in the account: S's well-being level at t is the number n such that the proposition that S's well-being level is n at t is true at (not in) the relevant alternative. (The relevant alternatives are the actual world, in which event E occurs, and the nearest world in which E does not occur. If we wished, we could use world-times and evaluate the (temporal rather than eternal) proposition that S's well-being level is n at (not in) world-time pairs.)

I began this section by claiming that for each side of the debate over the Zero View, there is a sense in which it is correct. In section VI, I formulated the Zero View like this: For any person S, any time t, and any world w, if t is a time at which S is not located in w, then S's well-being level at t, in w, is zero. I deliberately used the phrase ‘in w’ to distance my account from more extreme versions of the view, but I think it is possible to understand the view in two different ways. Let's say that S's well-being level at t, in w, is zero if and only if the proposition that S's well-being level is zero at t is true in w, that is, true according to the inner notion of truth. And let's say that S's well-being level at t, at w, is zero if and only if the proposition that S's well-being level is zero at t is true at w, that is, according to the outer notion of truth. (Again, we might index truth to world-times instead of worlds if we wished.) We may now distinguish between two views:

I think the Inner Zero View is false and that those who reject it are correct to do so. However, we do not need this view in order to maintain that people have zero well-being at times when they are not located. The Outer Zero View lets us do this, and it is true.

My account is compatible with the view that there are genuine properties such as having a well-being level, having zero well-being, not being pleased, not being in pain, and so on. There are people who, at some time or other, have all these properties (for example, Greg and Marsha, supposing they are actual). The job of a theory of well-being is to identify certain value-giving properties, so that an individual who has some of them at a time is well (or badly) off to a certain extent then, and an individual who lacks all of them has zero well-being. My account is also compatible with the claim that these value-giving properties are intrinsic. If a person is located at a given time and lacks the value-giving properties, it makes sense to say that the person has the corresponding negative properties and hence the property having zero well-being at that time. If a person is not located at a given time, whether in the actual world or a possible alternative, it makes sense to say that the person has no intrinsic properties, positive or negative, and hence (plausibly) fails to have the property having zero well-being, at that time. However, as we have seen, a proposition of the form S's well-being level is zero might be true at the world (time) in question. (Compare: Nora neither exists nor has any properties in world w, and so she does not have the property not existing. But the proposition that Nora does not exist is true at w.)

I will finish this section with a brief discussion of the claim that some objects – for example, shoes – fail to have any well-being level at all. Near the end of section IV, I mentioned that I was comfortable with a shoe having zero well-being, since an existing shoe will lack value-giving properties. I think this is at most a small cost, committing us to some truths – such as the claim that my shoe is better off than Nora – that might intuitively seem false.Footnote ⁵⁰ For those who are not comfortable with this perspective, I will note that my account can be incorporated into a theory of value that avoids it. For example, following Bradley we might say that a thing has a well-being level only if there is some world and time at which it has positive or negative well-being.Footnote ⁵¹ This (we should hope!) would imply that my shoe has no well-being level at all, but it would not rule out that Michael Jackson, for instance, has some level of well-being now.

Johansson criticizes this proposal on the grounds that it is inconsistent with something that a hedonist – such as Bradley – should say about one's present well-being level, namely, that it is completely determined by the intrinsic properties of one's current mental states. Instead, on Bradley's proposal it seems that one's present well-being level can depend on what is happening at other possible worlds and times. Johansson writes as follows about this view: ‘It implies that Michael Jackson, unlike the shoe, has a well-being level now – despite the fact that there is no intrinsic difference between their current mental states: there are no such states.’Footnote ⁵² Johansson's criticism is that since Jackson and the shoe are indistinguishable with respect to their current mental states, the hedonist should say that they are also indistinguishable with respect to well-being. Bradley, however, seems inclined to say that they differ in this respect.

I think that the hedonist, or, much more generally, anybody who thinks that one's well-being at a time is determined by what is going on at that time, can respond to Johansson by invoking the distinction that I have used in this section. The hedonist, for example, can maintain that the well-being of an individual, S, is determined by S's having or lacking the value-giving properties – being pleased and being in pain – as well as the truth of the proposition that possibly, S has positive or negative well-being at some time. (Since the value-giving properties are already picked out, this is not circular.) A theory along these lines might generate a principle like this: S's well-being level is zero if and only if (1) S has equal amounts of pleasure and pain, or S lacks both pleasure and pain; and (2) Possibly, S is pleased or in pain.

Let's reconsider our world-time <@, t>, consisting of the actual world and a time after Jackson's death. This principle allows us to say, just as we did earlier, that Jackson's well-being level is zero at <@, t>, but not in <@, t>. (This is in virtue of the temporal proposition that Jackson's well-being level is zero being true at, but not in, <@, t>. This, in turn, is in virtue of the propositions that Jackson lacks both pleasure and pain, and that possibly Jackson is pleased or in pain, being true at, but not in, <@, t>.) With respect to the shoe, since it is not possible that it is pleased or in pain, the principle implies that its well-being level is zero neither at, nor in, <@, t>. More general principles would imply that Jackson has no level of well-being in <@, t>, and that the shoe has no level of well-being either at or in <@, t>. Those who wish to distinguish between Jackson and the shoe could do so along these lines. Jackson will have a well-being level of zero at world-times where he does not exist, but the shoe will not.

What about Johansson's criticism? On the view being considered, Jackson has a well-being level at, but not in, the current world-time. This means that he does not (strictly and literally) currently have the property having zero well-being. The hedonist can say that one's well-being properties are determined by the intrinsic properties of one's mental states. If two individuals are indistinguishable with respect to their current mental properties, they will be indistinguishable with respect to their well-being properties. This view implies that Jackson and the shoe are now alike with respect to their well-being properties, which seems to be a good enough response to Johansson. But since the proposition that Jackson's well-being level is zero is true at the current world-time, there can be a fact of the matter about his being harmed by his premature death.