All's well that ends well? A new holism about lifetime well-Being

Abstract

Is there more to how well a life goes overall (its lifetime well-being) than simply the aggregate goodness and badness of its moments (its momentary well-being)? Atomists about lifetime well-being say ‘no’. Holists hold that there is more to lifetime well-being than aggregate momentary well-being (with different holists offering different candidates for why this might be). This paper presents and defends a novel form of holism about lifetime well-being, which I call ‘End of Life’. This is the view that the momentary goodness or badness of the end of lives has a disproportionate effect upon lifetime well-being. More precisely: a welfare subject's level of momentary well-being at the end of their life contributes more to their lifetime well-being than their level of momentary well-being at other points.

When we think about how well someone is doing at a particular time we are interested in their momentary well-being and when we think about how well someone's life went as a whole we are interested in their lifetime well-being.¹ Lifetime well-being seems to be at least somewhat sensitive to momentary well-being. But is there more to how well a life goes than simply the aggregate goodness and badness of its moments? Atomists about lifetime well-being answer ‘no’. Holists think that there is more to lifetime well-being than simply aggregate momentary well-being.

This paper presents a novel form of holism about lifetime well-being, the view that the goodness or badness of our lives at their end has a disproportionate effect upon our lifetime well-being:

End of Life: A welfare subject's level of momentary well-being at the end of their life contributes more to their lifetime well-being than their level of momentary well-being at each other point.

My case for End of Life comes in two parts. The indirect case consists of providing two objections against existing forms of holism before showing how End of Life avoids those objections and accommodates intuitions about lifetime well-being that stem from them. I also provide a direct case by considering the intrinsic merits of End of Life and replying to objections to it.

In Section I, I introduce some preliminary ideas and distinctions that serve as scaffolding for debates about lifetime well-being. I then (Section II), introduce the debate about the ‘shape’ of a life: the question of whether and how distributions of momentary well-being over time can impact upon lifetime well-being and the observation that there seems to be something better about ‘uphill’ distributions. In Section III, I examine the leading, holist, explanations for the superiority of uphill distributions. I then (Sections IV–V) give two novel objections to them. In Section VI, I show how these objections give an indirect case for End of Life. This is because End of Life simultaneously avoids the objections whilst also being able to accommodate the seemingly-conflicting lessons from them. In Section VII, I provide the direct case for the view, arguing for it and considering objections. I close (Section VIII) with some observations about the case I have made and highlighting some choice-points for End of Life.

I. Preliminaries: the shape of a life, atomism, and holism

Raibley provides an invaluable overview of atomism and holism about lifetime well-being thus:

[A]tomistic theories endorse (a) the explanatory priority of momentary well-being, (b) momentary well-being internalism, and (c) neutrality about the order of episodes when aggregating well-being over time. Theories are holistic in one respect or another if—and to the extent that—they depart from these theses.²

The explanatory priority of momentary well-being, Raibley explains, is the claim that ‘momentary…well-being is a more fundamental or basic evaluation than well-being over longer intervals of time, up to and including a whole life. Momentary well-being, on this view, does not derive from diachronic or global features of lives.³’ This rules out holding that the primary unit of evaluation is the whole life and that the momentary well-being of each moment derives from the value of the whole.

The second part of atomism—momentary well-being internalism—is ‘that well-being at individual times depends exclusively on the intrinsic properties of those times.’ This rules out the possibility of events later in life (non-instrumentally) affecting the momentary well-being of earlier moments, such as if hard-won success in later life affected the value of a difficult earlier period. To allow such a possibility would be to embrace momentary well-being externalism.

The third, neutrality, claim is that ‘an agent's well-being over intervals of time—up to and including the agent's life—is a simple function from his or her well-being at the smallest well-being-evaluable intervals, whether these turn out to be times, moments, or episodes of longer duration. This function might be summative or averaging, but it pays no attention to the temporal or narrative order of the smallest intervals.’⁴ This excludes the possibility that the pattern or ‘shape’ of how one's momentary well-being is distributed might make a difference to lifetime well-being.

From this, we see that if atomism is true then any two lives with equal total momentary well-being will, necessarily, have equal lifetime well-being (and vice versa). One would not need to know more than facts about aggregate momentary well-being to know facts about lifetime well-being.

To be a holist is simply to deny at least one of these claims. Herein, we can ignore possible denials of the explanatory priority of momentary well-being; none of the theories to be discussed deny that momentary well-being contributes to lifetime well-being (even if other factors play a role too). The theories I focus upon are atomism along with forms of holism that subscribe to momentary well-being externalism or non-neutrality about the order of episodes:

Atomism

Explanatory priority of momentary well-being

+ momentary well-being internalism

+ neutrality about the order of episodes.

Non-Neutralist Holism

Explanatory priority of momentary well-being

+ momentary well-being internalism

+  non-neutrality about the order of episodes

Moment-Externalist Holism

Explanatory priority of momentary well-being

+ momentary well-being externalism

+ neutrality about the order of episodes

II. The shape of a life

The contemporary debate between atomists and holists is centred on the question of whether distributions of momentary well-being across a life, or facts underpinning such distributions, can themselves affect lifetime well-being. The typical starting point is Velleman's influential discussion:

One life begins in the depths but takes an upward trend: a childhood of deprivation, a troubled youth, struggles and setbacks in early adulthood, followed finally by success and satisfaction in middle age and a peaceful retirement. Another life begins at the heights but slides downhill: a blissful childhood and youth, precocious triumphs and rewards in early adulthood, followed by a midlife strewn with disasters that lead to misery in old age. Surely we can imagine two such lives as containing equal sums of momentary well-being. Your retirement is as blessed in one life as your childhood is in the other; your nonage is as blighted in one life as your dotage is in the other.⁵

A common reaction is that the ‘uphill’ life seems better than the ‘downhill’ one. As Slote puts it: ‘Without hearing anything more, I think our natural, immediate reaction to these examples would be that the first man was the more fortunate[.⁶]’ In response, the atomist holds that, so long as Velleman's stipulation of equal momentary well-being holds, the two lives must have equal lifetime well-being. They will then try to explain away the apparent superiority of the uphill life by arguing that it stems from failing to hold total momentary well-being constant across the pair.⁷

Holists, by contrast, aim to vindicate our reaction that the uphill life is better. The holist who rejects neutrality about the order of episodes can say that the uphill life is better by allowing that there is more to lifetime well-being than simply aggregate momentary well-being. Specifically, the same quantity of momentary well-being distributed in an ‘uphill’ manner can yield more lifetime well-being than the same quantity distributed in a ‘downhill’ manner. This kind of holism—defended by Velleman himself, Glasgow, Kamm, and others—is thus able to take seriously the stipulation of equal momentary well-being whilst maintaining that the uphill life is better.⁸

The moment-externalist form of holism offers a different explanation of the superiority of the uphill life. On this view, there is more lifetime well-being but this is because there is more momentary well-being. And this is, in turn, because they hold that the distribution of the good and bad parts of a life, or the features that underpin them, affect momentary well-being. This kind of view has been defended by Dorsey, Kauppinen, Portmore, and others.⁹ Dorsey puts the point as follows (using ‘synchronic welfare’ for momentary well-being):

[T]he contribution to synchronic welfare of temporally discrete events or other goods in a life can be affected by the relations these temporally discrete goods and bads bear to other events, and so forth.¹⁰

Crucially, for this moment-externalist holist, the uphill life is not better for the kinds of prosaic, instrumental, reasons typically offered by the atomist. Rather, the improvement can be intrinsic and, crucially, retrospective.¹¹

So far, we have seen what the atomist says about such paradigmatic uphill vs downhill life cases and how the two kinds of holism can reach the verdict that the uphill life is better (I return to this below). For the rest of this paper, my focus is this claim about such uphill and downhill lives:

Uphill-is-Better: An uphill distribution of well-being has, or can have, a positive non-instrumental impact on lifetime well-being.

The ‘non-instrumental’ qualifier here rules out the kind of instrumental explanation that atomists can offer such as that uphill distributions themselves cause extra (e.g.) happiness and subsequently extra momentary and lifetime well-being. (Indeed, I mostly leave atomism aside for the remainder of the paper.) The ‘has or can have’ is intentional because whilst some holists endorse:

Uphill-is-Better (strong): An uphill distribution of well-being has a positive non-instrumental impact on lifetime well-being.

others endorse only a weaker claim:

Uphill-is-Better (weak): An uphill distribution of well-being can have a positive non-instrumental impact on lifetime well-being.

I now consider holist explanations of each thesis.

III. Holist explanations of uphill-is-better

The most widely-held holist view is that greater lifetime well-being can be generated by the narrative relations between events within uphill lives.¹² This can be combined with either the non-neutralist form of holism defended by Velleman or the moment-externalist form of holism championed by Kauppinen and Dorsey.¹³ On these forms of holism, Uphill is Better (weak) is true because narrative relations between life events can generate extra lifetime well-being. Velleman illustrates the idea thus:

Marriages

In one life your first ten years of marriage are troubled and end in divorce, but you immediately remarry happily; in another life the troubled years of your first marriage lead to eventual happiness as the relationship matures. Both lives contain ten years of marital strife followed by contentment; but let us suppose that in the former, you regard your first ten years of marriage as a dead loss, whereas in the latter you regard them as the foundation of your happiness. The bad times are just as bad in both lives, but in one they are cast off and in the other they are redeemed. Surely, these two decades can affect the value of your life differently, even if you are equally well off at each moment of their duration.¹⁴

Successes and windfalls

Slote's politician would have experienced an improvement in his well-being whether his years of toil were capped by electoral victory or merely cut short by his winning the lottery and retiring young. But the contribution of these alternative benefits to the overall value of his life wouldn't be determined entirely by how well-off each would make him from one moment to the next. Their contribution to his life's value would also be determined by the fact that the former would be a well-earned reward, and would prove his struggles to have been a good investment, whereas the latter would be a windfall in relation to which his struggles were superfluous. Thus, benefits that would effect equal improvements in his momentary well-being might contribute differently to the value of his life, by virtue of lending and borrowing different meanings in exchange with preceding events.¹⁵

In the first example, Velleman suggests, if you work on a relationship and turn it around then you redeem the bad times and you thereby generate an extra contributor to your lifetime well-being, as compared with simply starting a new equally happy relationship (which merely offsets the bad times rather than redeeming them). In the second example, redeeming early struggles by securing electoral success contributes more to your well-being, Velleman suggests, than simply receiving some windfall, even if each moment in the latter part has the same amount of momentary well-being. Thus, for Velleman, the uphill distribution that contains such goods of redemption has an additional higher-order good that factors into lifetime well-being.¹⁶

Because Velleman's view is a non-neutralist holism, it is not that the earlier bad times are (revealed to be) good by what comes later. He emphasizes repeatedly that the redemption of the bad by the good later times does not affect or reveal the value of the earlier periods:

[M]omentary well-being is ordinarily conceived as a temporally local matter, determined by a person's current circumstances, whether experienced or unexperienced. We think of a person's current well-being as a fact intrinsic to the present, not as a relation that he currently bears to his future.¹⁷

On Velleman's view, the later good times leave the badness of earlier periods unaffected, but the two together generates a higher order good that contributes to lifetime well-being.¹⁸

Dorsey and Kauppinen, by contrast, endorse a narrative relations view with a momentary externalist flavour: narrative relations generate extra lifetime well-being by generating extra momentary well-being. The earlier times in the marriage, and the politician's career, are, by virtue of later success, revealed not to have been bad moments, though they seemed it at the time and would have been bad moments had the later success not obtained. Dorsey makes this clear thus (using ‘synchronic well-being’ (etc) for momentary well-being):

For instance, consider a typically valuable event in a life, such as winning the Heisman Trophy. This event, surely important for synchronic well-being, could be a stepping stone, or merely a teaser: early success followed by a disastrous and shameful downfall. If it is a stepping stone to future success, one might say that the synchronic value of this event is greater than the synchronic value of an identical event which is just a teaser. For O. J. Simpson, his Heisman victory was, in fact, less synchronically valuable given what we now know: that it helped to frame his ever-so-public downfall.¹⁹

Thus, despite their differences, Velleman, Dorsey, and Kauppinen all hold that it is not the uphill distribution that really matters. The shape results from the narrative arrangement of good and bad moments that is the source of additional value. As Kauppinen puts it, ‘The narrativist thus offers a kind of debunking account of why improvement seems to matter: it is typically, but not necessarily, associated with a better underlying story.’²⁰ And Dorsey claims that ‘An upward-sloping life is signatorily valuable: a sign that the events of one's life are narratively related in valuable ways.’²¹ This is why these narrative explanations support only Uphill is Better (weak).

The Prime of Life Account

Slote offers a different holist explanation for Uphill is Better (weak).²² He claims that the contribution of momentary well-being to lifetime well-being is affected by the life-stage of the welfare subject, where momentary well-being in childhood and old age counts for less than the momentary well-being we accrue in what Slote calls the prime of life which he regards as the biological stage of full development (as opposed to just the arithmetic middle of a life).²³ On this view, an uphill life might fare better than a downhill life, so long as it accrues more momentary well-being within the prime of life period.

Improvementism

Having seen the main, narrative, explanations of Uphill is Better (weak), I turn to the explanation of Uphill is Better (strong). This view, which, following Bradley, I will call ‘improvementism’,²⁴ is put forward by a number of philosophers, most recently Glasgow and Kamm.²⁵ For Glasgow and Kamm, it is improvement per se that explains why the uphill life has extra value:

Glasgow

[T]here is also badness in the losing: when I go from a high level of momentary well-being to a low level of momentary well-being, that itself is bad for me. Losing is no fun at all. And what goes for losses goes for gains, too. When my well-being improves, this transition is itself good for me […] This view, which I will call the Loss/Gain Account of well-being, explains why the uphill life is better than the downhill life: the former contains an extra good (improvement in momentary well-being) and the latter contains an extra bad (diminution in momentary well-being).²⁶

Kamm

[I]t is better to start off badly in life and head toward improvement than to start off well and head toward decline, even when we hold constant all the goods and bads that are distributed in the two different patterns.²⁷

Velleman suggests that a life on an incline is better than one on a decline only if the good is caused by, and so in some way redeems, the bad. […] I disagree with Velleman. […] [I]nclines are better than declines even when redeeming the bad is held constant.²⁸

As is clear from these formulations, neither narrative relations, nor any other property of an uphill distribution, play any fundamental role in explaining the superiority of the uphill lives on this improvementist view. It is the uphill distribution as such, that makes uphill lives better. This is why this form of holism supports Uphill is Better (strong).

I now turn to objections to these respective explanations of Uphill is Better strong and weak, objections which pave the way for my preferred form of holism, End of Life.

IV. Objection 1: Short lives

Many of the cases discussed in connection with Uphill is Better can be used to support either improvementism, narrative relations views, or the prime of life view and it can be difficult to tease apart the three views. If a life has the kinds of narrative relations appealed to by Velleman, Dorsey, and Kauppinen, then it will have an uphill shape and will also fare well in the prime of life. (The reverse relationship does not hold, however. Lives can do well in improvementist terms without doing well on the narrative criterion.) Similarly, any life with an uphill shape will do well on the prime of life view. It is helpful, therefore, to consider lives where we can tease these three elements apart.²⁹

Against narrative and prime of life views we can find uphill lives that are better than downhill lives where this cannot be explained by these views. We see this by thinking about lives that do not span the normal human range. Here is one such case, adapting Velleman's case:

Short Uphill and Downhill: One life begins in the depths but takes an upward trend: a childhood of deprivation, an ok youthful period, precocious triumphs and rewards in early adulthood. Another life begins at the heights but slides downhill: a blissful childhood, an ok youthful period, and struggles and deprivation in early adulthood. Each life is ended abruptly by some piece of sheer misfortune.

The uphill life seems better despite its length. Someone might reply that this case is too permissive because it still allows for short uphill lives to have the right kind of narrative relations to explain their additional value or to cover just enough of Slote's prime of life period.

To combat this, let us take a more extreme and tragic case: children with life-limiting conditions who live very short lives.³⁰ For such short lives, it still seems true that it is better than those lives are uphill rather than downhill. A short life that ascends (in respect of momentary well-being) seems better than one that descends. But this cannot be explained by either the narrative relations view (or the prime of life view) given the nature of these lives.

Note that my argument here is not that narrative relations, or prime of life, cannot explain the value of uphill lives in at least some cases. I do not need to deny that. My point is only that there are cases where the uphill life seems better but narrative relations cannot do the explanatory work.

Uphill lives seem better than downhill lives even where this cannot be explained by narrative relations or by the extra significance of the prime of life period. This might lead us to conclude that an uphill shape matters in its own right (i.e. that Uphill is Better (strong) is correct). However, I now move to my second objection, which targets both forms of Uphill is Better.

V. Objection 2: worsening the beginnings

If uphill lives have higher levels of lifetime well-being fundamentally or even derivatively (by instantiating narrative relations) this would make a difference to the prudential and moral reasons that we have. This can be easily overlooked if we think about and compare only completed lives considered third-personally where there is no possibility of choice by, or on behalf of, the welfare subject.

I will present a range of cases designed to circumvent that issue. These examples have a common structure: reduced momentary well-being that gives a life an uphill shape. They stem from the observation that the shape of a life is a relational property of its elements and, although one way to give a life an uphill shape is to improve its later points, one can also give a life an uphill shape by worsening its earlier points. And this same point will apply across various explanations of why uphill distributions are better (pure distributive shape or narrative relations).

Here is the first case:

Phone call: You're approaching the end of your life. The second half of your life has been very successful. However, an accident in middle age left you with no memory of the first half of your life. You're rich enough to pay someone to track down information about your early life and one day the phone rings. “Good news”, they say, “your early life was horrible. Your life took an uphill shape!”

It does not seem to me that you have a reason, in such a case, to hope that your early life was horrible, such that your life had an uphill shape overall. Nor do people lament that they had a good childhood, on account of the uphill shape that was thereby prevented.

One might worry that this is an unfair example because the uphill shape comes with a much lower amount of momentary well-being! So, our judgement that the phone call does not give you good news stems from the fact that your total level of momentary well-being will be much lower if your early life was horrible than if it was good.

Note, though, that if we only compare uphill and downhill lives that are stipulated to be equal in total momentary well-being, and thus consider whether uphill can be a tie-breaker, we are closing off the question of whether an uphill shape can compensate for reduced momentary well-being.

Another point to make about this reply is that it still seems like we should be able to ask whether there is some outweighed reason to hope that the early life was horrible. But even that seems implausible.

Nevertheless, let us move to a second case, where we get an uphill shape with less loss of total momentary well-being:

A Town Called…Malice?: You're going to have a child. You have a good idea of the range of possibilities for your child's adult life. It's going to be pretty good, in terms of momentary well-being. You can't directly affect how well your child's life will go in its later stages. But you can affect the beginning. Convinced by the arguments of holists, you decide to leave Nice Town and move over to Underwhelming Town, specifically in order to give your child's life a somewhat less good beginning and so an uphill shape.

Suppose that the soon-to-be parent knows that the initial lowering of momentary well-being will cease to take effect at the end of childhood, so that they are choosing between giving their child a life with a flat distribution or a life with an uphill distribution (but which meets the line of the flat life at the beginning of adulthood).

Worsening the beginning would give this life an uphill shape. There is thus nothing in Uphill is Better, and the holist explanations we are considering, to rule this out as a way of promoting lifetime well-being. But the prospective parent does not plausibly have any reason to move to underwhelming town, to give their child's life an uphill shape. (Worsening the early years can enhance lifetime well-being in instrumental ways, of course but this is not relevant here.)

One might reply: ‘it is wrong to move to Underwhelming Town because the child will have a lower total amount of momentary well-being.’ I bracket this reply for now. Another reply is that a different consideration makes it is wrong to move to Underwhelming Town—that it involves interference in someone else's life. But this seems unpromising. It is permissible for parents to make beneficial interferences in their children's lives. If Uphill is Better were true, then this would be a way of benefitting the child. What is at issue is whether moving your child to Underwhelming Town and giving their life an uphill shape is a beneficial interference. To the extent that we think that the parent is not benefitting the child, we should doubt Uphill is Better.

Let us try a third case, to filter out worries about interference:

(Value) Gap Years: You have a good idea of the range of possibilities for your adult life. It's going to be pretty good, in terms of momentary well-being. Convinced by holism, you decide therefore to make the first half of your life a bit worse than it would otherwise be, in terms of momentary well-being, to give your life an uphill shape. You go home from parties early, you read slightly less enjoyable books etc. This gives your life an uphill shape.

Again, I do not think that we have reason to behave this way. Here is the objection, as derived from the latter two cases, in precise form:

• (1) If an uphill shape contributes to lifetime well-being, then there is some reason to worsen the beginning of a life, to make it an uphill life.

• (2) There is no such reason.

Therefore,

• (3) An uphill shape does not contribute to lifetime well-being.

Which premise should the defender of Uphill is Better seek to undermine? Premise 1 relies upon Uphill is Better plus the thesis that there are moral and prudential reasons to promote lifetime well-being. The defender of Uphill is Better cannot reject Uphill is Better and rejecting the other assumption seems like a non-starter. I will thus assume that the defender of Uphill is Better accepts premise 1 and will seek to reject the argument some other way.

Can the Uphill is Better proponent reject premise 2? The best reply that I think they can offer is the one that we saw earlier; it is easy to overlook reasons and we must be careful to ensure that we are not being misled into thinking that there are no reasons for the prospective parent to worsen their child's early years.³¹ Starting from this observation, the Uphill is Better defender might claim that premise 2 is false because there is a reason in the prospective parent case (but we are overestimating the consequences of recognizing such a reason). Specifically, they might claim that we are mistaking the question of whether we have some reason to worsen the beginning of a life with the issue of whether that is what we ought to do all-things-considered. Perhaps, so the Uphill is Better defender argues, we are misled by our confidence that the parent ought not to worsen the beginning of the life into thinking that there was absolutely no reason for them to do so. Presumably, it would be the loss of momentary well-being that provided the outweighing reason. Thus, this full reply would be that an uphill distribution really is valuable but the momentary well-being that would be lost outweighs the value of a life being uphill. In considering this kind of objection, applied to improvementism, Glasgow suggests some sympathy for this response:

And a plausible weighting is that the disvalue of the extra pain will always outstrip the benefit of the slope. […] a plausible principle might be that the dis/value carried by the shape of a sequence of moments is always less than the difference, in momentary value, between the terminal moments of that sequence.³²

This is the Uphill is Better defender's best reply. Yet, I doubt that we should be persuaded by it. After all, we thought, in the face of the original Velleman and Slote cases, that an uphill distribution contributes to lifetime well-being. Yet, once we examine worsen-the-beginning cases, where an uphill shape requires sacrifices of momentary well-being, the proponent of Uphill is Better is pushed to conclude that the value of an uphill shape is swamped by any loss of momentary well-being.

This is coherent but seems ad hoc. To avoid the possibility of any case where there is a reason to lower momentary well-being, it will have to be true that the value of having an uphill shape is outweighed by fractional losses of momentary well-being. But this seems like a striking coincidence. Each of these things are valuable but one of them lexically dominates the other. But why is that? Why such harmony between the value of shapes and total momentary well-being? Once having an uphill shape has such little value, it is tempting to think that the atomist may have been right all along; we were misled into thinking that uphill shape matters by features of the comparison cases, or a failure to think clearly about them (and that this is why total momentary well-being always wins out over having an uphill shape.)

From the two objections, it looks like we draw the following, apparently conflicting, lessons:

1. Uphill shape is valuable, even aside from narrative relations/prime of life (shown by: considering very short uphill and downhill lives).

2. Uphill shape isn't valuable or is, at most, only slightly valuable (shown by: implausibility of reasons to bring an uphill shape about by reducing momentary well-being at the beginning of lives).

One response is to just accept (2) on the stronger reading and reject (1). This is the stock atomist reply. Another route is to accept (1) and accept (2) on the ‘slightly valuable’ reading and try to make that combination plausible.

My interest is in a third response, of showing how my proposed holist view—End of Life—accounts for the plausibility of (1) and (2), as well as being generally plausible.

VI. End of life: the indirect case

Importantly, the uphill lives introduced by Velleman and others differ from downhill lives in ways aside from being uphill. The key difference here is that the uphill life ends well whereas the downhill life ends badly.³³ From this observation, we can see the start of the case for End of Life. This view is as follows:

End of Life: A welfare subject's level of momentary well-being at the end of a life contributes more to their lifetime well-being than their level of momentary well-being at each other point.

Put another way: there's a multiplier effect on how momentary well-being generates lifetime well-being at the end of life.

By denying the neutralist claim that partly comprises atomism, End of Life is a form of non-neutralist holism. This is evident from the fact that two people whose overall momentary well-being totals are identical, but who have different momentary well-being at the end of their lives, will end up with different lifetime well-being values, according to End of Life.

End of life falls within a set of views claiming that some points in life have greater in determining our lifetime well-being. I focus here only on the thesis about endings, due to constraints of space and because I think it most plausible. Let me also note that there is scope for variation within formulations of End of Life; I do not pretend to provide the only plausible way of spelling out the idea.

It is important to point out that End of Life is non-comparative. It does not concern the difference between momentary well-being at the beginning and the end of life. It is simply that the end of a life has greater weight for the generation of lifetime well-being from momentary well-being. In this way, it resembles Slote's prime of life view, by giving special weight to a particular period, but differs in the period identified and by not tying that period to biological maturation.

The classic cases used to motivate Uphill is Better—uphill and downhill lives with stipulated equal momentary well-being—equally well support End of Life. The uphill lives have a much higher level of momentary well-being at their end than the downhill lives. We can thus equally well explain the superiority of the uphill life as stemming from how it is doing at its end as from it shape. End of life also enjoys two advantages over the kinds of Uphill is Better views considered so far, in how it fares with the objections outlined above.

First, End of Life explains why uphill lives are better in the fuller range of cases that includes very short lives. Thinking back to the short lives that were a problem for Uphill is Better (weak), End of Life offers a natural explanation: the uphill lives are better than the downhill lives due to differences in momentary well-being at their ends.

Second, because End of Life does not afford significance to a relational property, the view does not licence worsening-the-beginning. End of Life would not say that the prospective parent has a reason to worsen her child's momentary well-being to give their life an uphill shape, nor that the student should have a value gap year.³⁴ This generates a dominance argument for End of Life over Uphill is Better. Cases that support Uphill is Better support End of Life but End of Life avoids the objections to Uphill is Better that we saw earlier, based on short lives and reasons to worsen the beginning of lives.

Still, one might think these two points insufficient if End of Life is otherwise implausible. Let me therefore now move to providing the direct case for End of Life, by arguing for it non-comparatively and by considering replies.

VII. End of life: the direct case

The end of a life seems to have a disproportionate impact in our assessments of how well someone's life goes as a whole. For example, suppose that you must decide when an episode of excruciating pain would happen either to you or to a loved one and that your memory of your choosing will be immediately wiped. It seems plausible that one would not choose to have it happen at the very end of life, for one's last moments to be spent in excruciating pain. It seems preferable instead for it to happen at some earlier point, after childhood, and there are large sections of a normal human lifespan where it seems roughly equally bad for it to occur (e.g. there seems no reason, other things equal, to choose your 31st year over your 32nd). This suggests that the end of our lives matters at least somewhat more than other points, in determining lifetime well-being and this is why we do not want our last moments to be spent in excruciating pain. This is surprising, given that its happening right at the end of life would be better in one way, by insulating us from any instrumental effects. Yet, in spite of that reason to prefer that it occur at the end, it seems especially bad for it to happen at the end of life, in line with the End of Life view.

The extra significance of the end of our lives is arguably reflected in the decisions that we make about the end of life. In various ways, we place significant emphasis in end-of-life care on dignity and avoiding pain or suffering at the end of lives, to an even greater degree than at earlier points. We care about people dying with no suffering involved, or dying at home surrounded by loved ones, or dying peacefully in their sleep, or doing what they loved.

Someone might reasonably object that this observation merely reflects the fact that we want people not to suffer—at all—and so we do not gain support for End of Life from the fact that we do not want people to suffer at the end of their lives.

I think that this is mistaken. We care that the end of peoples’ lives be as free from suffering as possible. But this is not explained just by a general aversion to suffering. To see this, consider the difference between learning that a loved one was in significant pain in their last moments as opposed to discovering that, previously unbeknownst to you, they suffered an equally bad period of pain at some point in the middle of their life. The first revelation is much worse than the second. Why is that? Because it came at the end of the person's life, where there is extra significance for their life as a whole.

Consider a different case. You have one dose of a painkiller and know that two strangers are each in excruciating pain. As stated, you have no reason to give it to one over the other. But suppose you learn that, unbeknownst to them, one stranger is at the end of their life and the other is in the middle of their life (and there are no other relevant differences between them). This makes a difference to the reasons you have, I suggest. You have at least some extra reason—enough to break the tie—to give it to the person at the end of their life to ensure that their last moments are not ones of suffering. (Again, in spite of the lack of instrumental effects of pain at the end of life.)

Here is another case, involving non-human welfare subjects. Compare two veterinarians who provide care to dogs. Each provides a vaccination half-way through the dog's life and euthanasia for dogs at the end of life. One veterinarian uses a painful vaccination technique (the pain is immediately forgotten) and a painless technique for euthanizing. The other uses a painless vaccination technique and a painful euthanising technique. On the assumption that one cannot use different vets for different treatments, should an owner be indifferent when picking a vet for their beloved dog? It seems to me that they should not be. There is at least some reason to pick the vet who provides painless euthanasia, to give the dog a painless end.

Another argument for the intrinsic merits of End of Life is more theoretical. We can support the claim that endings have disproportionate significance to lifetime well-being through an analogy between lifetime well-being and aesthetic evaluation. That is because for many sub-domains of aesthetic evaluation, endings have special significance to the aesthetic value of the whole. Here are some examples of this phenomenon in the aesthetic domain.

A bad ending tarnishes a novel, no matter how good the rest of it is, in a way that a bad intermediate part does not (or nowhere near as easily). A great novel can have a poor part, perhaps, but it is harder for a novel to be great and have a poor ending. Take some great novel and add a bad ending to it (it was all a dream, a ridiculous deus ex machina, an inexplicable decision by the protagonist, or the destruction of the emotional connection to the characters). These serve to ruin novels, much more than bad earlier parts. The same plausibly holds for symphonies, movies, plays, and other aesthetic objects that are intended to be appreciated as wholes. An ending that is sufficiently poor casts a shadow over the work, undermining the value that it (would have) had.³⁵

From these observations, we can formulate the following claim:

Endings Matter (Aesthetic): For aesthetic objects with endings, the ending has a disproportionate weight for the aesthetic value of the whole.

This claim can explain the claims adduced above but is weak enough to allow that there could be, for example, great novels with bad endings. For my purposes, it does not matter if exactly this principle is true, only that something like it is plausible.

The question now, is whether the plausibility of Endings Matter (Aesthetic), or something like it can lend support to End of Life in the prudential domain, on the grounds that if it is true in one subdomain of the evaluative that endings matter more this makes it likelier that the same is true in other subdomains of the evaluative (at least for those where it makes sense to talk of endings).

This argument will likely face the following objection: ‘this conflates the prudential and the aesthetic evaluation of lives’. I do not think so. For one thing, the story of a downhill life can be aesthetically just as good or better than an uphill life. They are both great stories, after all. As Rosati puts it: ‘many lives that make for excellent stories are lives none of us would rationally want to live—at least not insofar as our concern lies with our good rather than, say, our fame or notoriety.’³⁶ In general, it does not seem like we conflate the aesthetic and the prudential evaluation of lives (as evidenced by the popularity of tragedy as a genre). We can separate these two kinds of evaluation. And more generally the argument is only supposed to identify a feature of one evaluative domain and use that as a licence for optimism that we should find the equivalent feature, where applicable, in other evaluative domains. It is not to slip between prudential and aesthetic evaluation.

The parallel between aesthetic and prudential evaluation, coupled with some points made earlier, helps to fend off an objection to End of Life constructed from an example given by Dale Dorsey when discussing explanations of Uphill is Better (strong):

I think there is very little reason to suggest that this [Uphill is Better (strong)] holds across an entire life but not across temporally extended segments within a life. If we wish to say that a life with an upward trajectory is better than a life with a downward trajectory, other things equal, then we should say, for instance, that a year with an upward trajectory is also better than a year with a downward trajectory, and a weekend with an upward trajectory is better than a weekend with a downward trajectory, other things equal.³⁷

The first thing to note is that there is substantial disagreement about whether the superiority of uphill distributions holds for periods shorter than whole lives. Though offering different explanations, Glasgow (2013: 668) and Kauppinen (2015: 199) each hold that, faced with a choice between an uphill and downhill year, the uphill year can be preferable.³⁸

However, suppose that one agreed with Dorsey that it would be bad to be committed to the view that the uphill weekend is better than the downhill weekend. What then could the End of Life proponent say to block this conclusion?

Dorsey's objection relies on the idea that, if the thesis is supposed to hold for whole lives, then it would hold of the parts too. But there is no reason to make this assumption. We see this by noting, for example, that we do not give special significance to the end of each chapter of a novel, though we do give special significance to the ending of novels as a whole. My reply, then, is that, similarly, End of Life can be true of wholes without the analogous claim about parts being true. (This is compatible with endorsing the analogous claim about parts, if one finds that plausible.)

Dorsey considers something close to this response, and rejects it thus:

[someone] might suggest that there is no reason to believe that the significance of a life's shape must also extend to the significance of a weekend's shape. But this response is implausible and ad hoc. In particular, we could imagine a life that simply consisted of a weekend of this kind. Why not think about the cases above simply as lives that “pop into existence” on Friday, and then pop out of existence on Sunday? Surely this is at least coherent. In this case a week-end's shape just is a life's shape. But even so, it seems wrong to think that the life consisting in the found weekend is better than the life consisting in the lost weekend.³⁹

Here, Dorsey switches from small parts of whole lives to very short whole lives. But these are different in the way that is relevant here: whether they are wholes or parts. End of Life has a perfectly plausible rationale for not applying to a small segment of a whole life: wholes and parts are relevantly different.

Dorsey's objection does, of course, bring out that End of Life is committed to thinking that it is better for a short whole life to have an uphill distribution (so that it ends at a higher point than if it had a downhill distribution, other things equal). But, as argued earlier, this enables it to reach the right conclusions about short lives. For many reasons, it is hard to think about lives that are 3 days in length but, I suggest, if we do so, we would think that it would be better for such tragic lives to be uphill rather than downhill.

One reason why it is difficult to think about such short lives in this context is that, unless we think about recherché cases of adults who pop into existence suddenly and briefly, lives that are extremely short can only have a very limited range of prudential goods and bads: just pleasure and pain, presumably. This connects with another objection that might be made to End of Life, either in general terms or specifically in connection with such short lives. Again, this objection comes from a point made by Dorsey in support of the narrative relations view:

But we might now compare two different experiments. The first person is hooked up to a pleasure-stimulating computer with software designed to start his life at a neutral level; no pleasure, no pain. Gradually, say, twice a year, the pleasure is increased in a linear fashion. The second is precisely the opposite: this person's machine starts out with quite a lot of pleasure. Gradually, also twice per year, the software decreases the pleasure in a linear fashion, such that both people, over the course of their lives, will generate the same amount of sensory pleasure. [For End of Life] the first person lives a better life than the second. But this seems to me, once again, quite implausible. If I live my entire life simply being sensorily stimulated by some computer, simply undergoing the barest sensory pleasure, how could it possibly matter when this pleasure occurs?⁴⁰

Does this pose a problem for End of Life? Note, first, that our intuitions here risk being clouded by the fact that the lives are lived within the experience machine, a feature not central to Dorsey's point. More significant (because it would survive the removal of the experience machinery) is that although Dorsey's point is meant to be general, both lives are filled only with bare sensory pleasures of various degrees. But to test the intuition, we should, I think, consider cases where there is pleasure and pain (otherwise the highly remarkable nature of such lives, in containing only positive states, might cloud our judgement about such cases). Once we do so, Dorsey's point is less powerful, I think. Ending one's life in pain seems worse than having an equally painful experience in such a machine at the beginning of one's life (other things equal). At least, it seems capable of breaking a tie between uphill and downhill distributions.

As a final objection, let me address an implication of End of Life in the form of a dilemma. Either:

Horn One: According to End of Life, lifetime well-being can be maximized by forgoing additional, positive, momentary well-being (by having one's life end earlier than it otherwise would). It might be better to have one's life end at a higher point of momentary well-being than for it to continue at a lower, yet positive, level.

Or

Horn Two: According to End of Life, lifetime well-being cannot be maximized by forgoing additional momentary well-being. It could never be true that lifetime well-being would be maximized by the ending of a life that would, otherwise, remain above a neutral level.

Horn two might seem the lesser evil. But whilst there is something attractive about it, taking this horn generates the same kind of dubiousness I pressed earlier against Uphill is Better. We would be forced to conclude that endings matter more but that the multiplier effect is so weak that any quantity of momentary well-being will outweigh it, such that End of Life could only ever be a tie-breaker. Such a position is coherent but seems terribly ad hoc.

I think that horn one is more plausible and that we see this in testimony from people who wish for their lives to end before they decline. Whilst in many cases people are, of course, aiming to avoid periods of negative momentary well-being, at least some are willing to forgo additional periods of life at low, yet positive, levels of well-being for the sake of ending their lives at a higher point. This seems right to me. Ending one's life with full awareness, for example, seems better than an alternative in which one becomes unresponsive and eeks out a few more days experiencing nothing more than basic sensory pleasure with no other awareness. Thus, I think that horn one is not a threat to the view but a conclusion to embrace. The greater importance of the end of a life is sufficient to outweigh at least some amounts of minimally positive momentary well-being; to some degree it is, in the words of the famous lyric, ‘better to burn out than fade away’. Here, we find a parallel with aesthetic evaluation, by thinking about cases such as the novel, film, or television series that carries on too long, not to the point where the extra parts are bad in themselves but where the body of work is worse for including them.

VIII. Final notes on my argument and End of Life

Although my case for End of Life was forged from objections to Uphill is Better strong and weak, it is worth noting a more concessive position: that End of Life is one factor that contributes to lifetime well-being, without being the only factor. Thus, whilst it is tempting to give a parsimony case for End of Life alone over Uphill is Better, there is nothing to say that both views could not be true. Either way, I hope to have shown that by separating out distributive shapes from endings, we can make progress in working out what is true in this area.⁴¹

A natural question to ask about End of Life is: what constitutes the end of a life? The first defensive thing to note is that there can be a range of plausible answers here and people might think that the view's plausibility is greater on one understanding of the end of a life than another. I do want to make a few specific commitments though. First, the case for End of Life suggests that it is the end of conscious existence that has special significance. To see this, compare two ways a person's life might end. In the first, their last moments are of excruciating pain, followed by death (with no cessation of the pain). In an alternate scenario, they experience the same excruciating pain, followed by a period where they are alive but with no conscious awareness at all, followed by death. Here, I think that, if we hold everything else equal (particularly effects on loved ones), there is no reason to choose one scenario over the other. In each of them, the person's conscious experience ends with great suffering and that seems to be the relevant test of how their life ended. Thus, I suggest, the most plausible candidate for the ending of a life, for the purposes of the End of Life view, is the permanent end of conscious experience, not the end of mere biological survival or something more minimal like that.

One might ask, for whatever constitutes the ending, exactly how long is the end of a life? Here, it is worth noting that talk of the end of people's lives is something with which we are pre-theoretically acquainted, rather than some purely theoretical notion, which imposes some conceptual limits on what can constitute the end of life (e.g. it cannot be 40 years). Moreover, there cannot be anything after the end of a life, so our inquiry is bounded in one direction. The question then is how far back into a person's life counts as the end of their life. Again, there will be disagreement here and no unique answer is entailed by the general view. For my part, I think the most plausible version of End of Life will include a very minimal conception of end of life—perhaps as short as one's very last hours, rather than anything of significant duration (e.g. one's last weeks or months)—as the period that has disproportionate impact in determining our lifetime well-being.⁴²

Footnotes

Acknowledgements

I must thank an enormous number of people who have discussed these topics with me over a prolonged period. These include: Gwen Bradford, Michael Cholbi, Dale Dorsey, Stacie Friend, Michael Gill, Antti Kauppinen, Filipa Melo Lopes, Mike Ridge, Debbie Roberts, Luca Stroppa, Joey Van Weelden, along with audiences in Edinburgh, Konstanz, Leeds, Saarland, Southampton, St Andrews, York as well as an online session organised organized by Anthony Kelley. For written comments I thank James Brown, Lorenza D'Angelo, Nikki Fortier, Jason Raibley, and Debbie Roberts. Finally, I thank the two reviewers for this journal for extremely generous assistance.

References