Desire-satisfaction and Welfare as Temporal

Abstract

Welfare is at least occasionally a temporal phenomenon: welfare benefits befall me at certain times. But this fact seems to present a problem for a desire-satisfaction view. Assume that I desire, at 10am, January 12th, 2010, to climb Mount Everest sometime during 2012. Also assume, however, that during 2011, my desires undergo a shift: I no longer desire to climb Mount Everest during 2012. In fact, I develop an aversion to so doing. Imagine, however, that despite my aversion, I am forced to climb Mount Everest. Does climbing Mount Everest benefit me? If so, when? A natural answer seems to be that if in fact it does benefit me, it benefits me at no particular time, and hence the desire-satisfaction view cannot accommodate the phenomenon of temporal welfare. In this paper, I argue, first, that a desire-satisfaction view can accommodate the phenomenon of temporal welfare only by accepting what I call the “time-of-desire” view: that p benefits x at t only if x desires p at t. Second, I argue that this view can be defended from important objections.

Welfare is a temporal phenomenon. As Ben Bradley writes:

A person’s whole life, or a world, can go well or badly for her. But parts of lives, or times can go well or badly for a person too. Things go well for a person at some times and badly at others. When we say that someone is having a bad day, or the time of her life, we are not speaking metaphorically; what we say is, at least sometimes, literally true (Bradley 2009, 18).

Not only does welfare occur at a time in addition to occurring over an entire life, it is plausible to say that one’s lifetime well-being is at least in part a function of the levels of welfare one maintains at the individual times of one’s life.¹ Hence, it would appear, any acceptable theory of welfare must account for the phenomenon of welfare at a time. However, one common complaint about desire-satisfaction theories of welfare is that such views cannot adequately satisfy this relatively straightforward desideratum.

In this paper, I defend a desire-satisfaction view against the charge that it cannot accommodate the temporal nature of welfare. My argument proceeds in two stages. First, I argue that a desire-satisfaction view can in fact do so, but can do so only by accepting what I call the “time-of-desire” view, i.e., that ϕ is good for x at the time x desires ϕ, and at every time x desires ϕ. This is good news and bad news. The bad news is that the time-of-desire view has been thought totally unsuccessful; the fact that a desire-satisfaction view cannot plausibly avoid it might be thought dispositive reason to believe that a desire-satisfaction view cannot explain the phenomenon of temporal welfare. The good news is that the time-of-desire view can be defended; I show that popular arguments against the time-of-desire view either fail, or simply rely on a prior dismissal of a desire-satisfaction view. Hence we should not reject a desire-satisfaction view simply because it cannot provide an adequate account of the phenomenon of temporal welfare.

1 Desire-satisfaction: A Primer

A desire-satisfaction theory of welfare holds, roughly, that ϕ is good for x to the extent that ϕ satisfies x’s desires. This is a rough characterization in part because different desire-satisfaction views will offer very different interpretations of its main terms. First, different desire-satisfaction views will construe welfare-relevant “desires” differently. Some hold that x is benefited only to the extent that x’s fully-informed desires are satisfied.² Others hold that a person is made better-off only to the extent that her second-order desires are satisfied, perhaps under some idealized set of cognitive conditions. (Railton 1986, 54) For the sake of simplicity, I refer generically to “desires”. Proponents of any one particular understanding are welcome to fill in the details as they see fit; these details won’t matter for the arguments to come.

Second, different desire-satisfaction views will construe the bearers of intrinsic value—the ϕ that satisfies x’s desires—differently. The desire-satisfaction view as endorsed by Hobbes (1651, 28–29), Sidgwick (1907), Railton (1986), (Lewis 1989), and others (Perry 1926), holds that value bearers are the objects of a given individual’s desire. On this view, to satisfy a desire is to be the object of a desire; ϕ is intrinsically good for x if and only if ϕ is desired by x. Call this the “Hobbesian” desire-satisfaction view. However, other authors have identified a desire-satisfaction view differently. Bradley (2009) and Heathwood (2005) hold that a properly conceived desire-satisfaction view will treat a state of affairs that conjoins the desiring of a particular object ϕ with the occurrence of ϕ, as the proper bearer of value. Call this conjunctive state of affairs a “D-state”. On this view, to satisfy a desire is to be a D-state; ϕ is intrinsically good for x if and only if ϕ is an instance of a D-state. Call this the “Moorean” desire-satisfaction view.³ Except where noted, I assume a Hobbesian construction here.⁴ The arguments pro and con will apply (save a short discussion in Section 7.2) to either view.

One further point is worth mentioning before I move on. My discussion of the desire-satisfaction view is in many ways a crucible. The desire-satisfaction view is but one version of a general class of views (often called “subjective” views⁵) that hold that well-being is, at heart, the fulfillment of pro-attitudes on the part of a given individual, whether that pro-attitude is a desire or not.⁶ But each of these alternative views faces problems in accounting for the phenomenon of temporal welfare. Hence though I confine my discussion to desire-satisfaction views, the arguments for and against should be applicable to a wide range of non-desiderative views, as well.

2 The Problem Introduced

To see why a desire-satisfaction theory might have difficulty accounting for the temporality of well-being, note that a commonsense principle is that my welfare at t ₁ must be determined by the welfare benefits I receive at t ₁. For instance, to say that my wife had a good childhood is to say that she was benefited during childhood. Hence to account for the phenomenon of temporal welfare, it is necessary and sufficient to account for the time of welfare benefits. For most theories of welfare, accounting for the time of welfare benefits is trivial. As Bradley notes, “Pure hedonism can account for temporal facts about welfare in a straightforward way: the good times in a person’s life are the times when she is pleased; the bad times are the painful times,” (Bradley 2009, 18). But a desiderative view cannot offer a similarly straightforward approach. The problem here is a result of two general facts about desires. First, individuals can, and often do, desire things that occur at times other than the time of desire. For instance, I might desire, at 10am, January 12th, 2010 to have a very productive remainder of the working week. Assume that I do have a very productive remainder of the working week. When am I benefited? Am I benefited at the time of the occurrence of that which I desire? Or am I benefited at the time I desire that which occurs? Does my productive working week mean that I had a good January 12th? A good working week? Which is it?

Second, desires can change over time, and indeed can change from the time of desire to the time of fulfillment of that desire. Assume that I desire, at 10am, January 12th, 2010, to climb Mount Everest sometime in 2012. Also assume, however, that during 2011, my desires undergo a shift: I no longer desire to climb Mount Everest in 2012. In fact, I develop an aversion to so doing. Imagine, however, that despite my aversion, I am forced as a result of circumstance to climb Mount Everest. I certainly do not value climbing Mount Everest while I climb it. Does climbing Mount Everest benefit me? If so, when? If I do climb it, does that mean that the time at which I climb the mountain is actually good, given that I desired to climb it two years earlier? Alternatively, does it benefit me two years earlier, and hence mean that I had a good January 12th, 2010? Is there any principled reason to adopt one or the other answer? Is either answer acceptable?

In the face of such problems it may seem natural for a desire-satisfaction theorist to say that if in fact climbing Mount Everest benefits me, it does not benefit me at any particular time. Generally speaking, though the satisfaction of my desires benefits me, because the object and desire can diverge in time, there is no one particular time at which their satisfaction benefits me. And hence a desire-satisfaction theory cannot account for the phenomenon of temporal welfare. Insofar as welfare is at least some of the time a temporal phenomenon, however, this is a disastrous result.

3 Constraints

Of course, a proponent of a desiderative view need not stand idly by. The obvious solution is to offer a desiderative theory of the time of welfare benefits. Of course, such a view must retain the standard attractions of a desiderative approach. Two such attractions are worth discussing here.

The first is often referred to as a “resonance”, “non-alienation”, or “internalism” constraint on the personal good. Of this constraint, Railton (1986, 9) writes:

Is it true that all normative judgments must find an internal resonance in those to whom they are applied? While I do not find this thesis convincing as a claim about all species of normative assessment, it does seem to me to capture an important feature of the concept of intrinsic value to say that what is intrinsically valuable for a person must have a connection with what he would find in some degree compelling or attractive, at least if he were rational and aware. It would be an intolerably alienated conception of someone’s good to imagine that it might fail in any such way to engage him.⁷

A desire-satisfaction view can accommodate the resonance constraint by insisting that that which is good for x is responsive to what x cares about in a very specific way, i.e., is determined by what x desires.⁸ In that way, a desire-satisfaction view will never allow “alienating” goods to benefit x: x will, in all cases, desire that which benefits x. Accommodation of the resonance constraint is important for a desire-satisfaction view. Without it, a classic argument on behalf of a desire-satisfaction view against so-called “objective” views wilts. Objective views insist on a number of welfare goods that may or may not be cared about by the individuals for whom they are good. Desire-satisfaction theorists rightly complain that this feature of an objective view is problematic: welfare benefits should not alienate.

A further motivation is important to retain. A desire-satisfaction theory is not the only one that can satisfy the resonance constraint.⁹ One important such view is “preference hedonism”. According to Derek Parfit, preference hedonism holds that welfare goods are confined to experiences that one desires (Parfit 1984, app. I). (This view deserves the name “hedonism” insofar as pleasurable experiences just are those experiences that are desired. This view is controversial, of course. I assume it here for the sake of argument.¹⁰) How, then, is the desire-satisfaction view motivated against preference hedonism? A desire-satisfaction view, unlike preference hedonism, can accept the claim that the boundaries of a person’s good are not limited by the boundaries of that person’s experience. Call the view that a state of affairs, object, or event ϕ can be a welfare benefit to the person who desires ϕ, even if ϕ is never experienced, the “distance allowance”.¹¹ Preference hedonism denies the distance allowance: preference hedonism insists that only a given individual’s experiences can be beneficial for an individual. But, at least for those attracted to the desire-satisfaction view, the allowance of such distance is appealing. If I desire to be a good father, and part of what makes me a good father is the success of my children, it may be the case that my desire is satisfied, and my life made better, despite the fact that I may not realize, or come to know, that my children have been successful as a result of my efforts on their behalf.¹² Furthermore, I might desire to be told the truth about my medical conditions. Though I may never come to realize that I have been lied to, that I have been lied to seems to make me worse-off. In this way, liberating welfare goods from the boundaries of experience seems to tell in favor of a wider desire-satisfaction view in comparison to, say, preference hedonism.¹³

These motivations serve to constrain any potential answer a desire-satisfaction theory might give to questions concerning the temporality of welfare benefits. It must be the case that any attempt to provide a desiderative account of welfare at a time is compatible with the general motivations cited above. Violating the first appears to sap any motivation that a desiderative view might have against objective competitors. Violating the second appears to sap much of the motivation that a desire-satisfaction view has against preference hedonism.

However, these constraints also shed light on the dialectical context within which this objection arises. This objection, if successful, is powerful for the following reason: if a desire-satisfaction view cannot account for the temporality of welfare, this entails that even if a desire-satisfaction theory is otherwise plausible, i.e., well-motivated, responsive to considered judgment, etc., it fails insofar as it cannot adequately accommodate all the facts of human welfare, including facts of welfare at a time. But if this is correct, it is important that any argument against a desire-satisfaction view on grounds that the desire-satisfaction view provides an untenable account of welfare at a time does not simply assume, at the outset, that the main motivations for any desire-satisfaction view are false. For instance, one might argue against a desiderative account of welfare at a time on grounds that this approach allows too much distance between a person’s experiences and that which improves their lives. But this just assumes, at the outset, that a desire-satisfaction view is unmotivated against rivals. Doing so saps all the per se force of the objection from temporal welfare: on this assumption, no desiderative approach to welfare is tenable, and hence the objection in question simply devolves into a more general complaint about the desire-satisfaction view and its motivation. Of course, any of the major motivations for a desire-satisfaction view may very well fail. But insofar as we are interested in objections to the desire-satisfaction account on grounds of the temporality of welfare, we should ignore objections that rely on a prior commitment to the claim that a desire-satisfaction view is not properly motivated. Such assumptions sap all force from the objection from temporal welfare as an independent objection to desiderative approaches to prudential value.

4 The Solution(s)

For a desire-satisfaction view, there seem to be three potential accounts of the time of welfare benefits worth considering.¹⁴ A first, and quite natural, way is to suggest that a person is benefited by an object of desire at the time at which the object of desire occurs. Call this the “time-of-object” view. The time-of-object view could be used to account for welfare at a time: my wife had a good childhood only if the objects of her desires occurred during childhood. If I desire to have a productive remainder of the working week, my productivity benefits me during the remainder of my working week. This view seems plausible enough.

But on reflection the time-of-object view is unacceptable. The time-of-object view would leave it open that I can lack a desire for ϕ at a particular time, but that ϕ is nevertheless good for me at that time. If I desire ϕ at t ₁, and ϕ occurs at t ₂, the fact that I desire ϕ at t ₁, given the time-of-object view, renders ϕ a prudential benefit whenever it occurs, including at t ₂—even if, at t ₂, I lack a desire for ϕ. In saying this, the time-of-object view commits a cardinal sin: it allows welfare goods that alienate the person whose benefits they are. Recall that a desiderative view is motivated in part by the rejection of accounts of well-being that fail to resonate, or that alienate, the person in question. But accepting the results of the time-of-object view—that ϕ can be good for me at a time at which I do not desire ϕ, or that a particular D-state is good for me at a time at which the resonating element of the D-state does not hold—would seem to be alienating just in the way a resonance constraint rules out.

One might respond by claiming that ϕ still resonates with me in one sense: there is a time of my life (t ₁) during which I desired ϕ. But if we are assuming that welfare benefits are to be assigned to particular times, it would seem strange, in light of the stress desiderative views place on a rejection of alienating welfare goods, to assign those goods to times during which they actually alienate the person whose goods they are. If we are to assume that welfare benefits should be assigned to a time, the resonance constraint would seem to insist that they be assigned to times during which they fail to alienate.

For this problem, there is an ingenious fix. One can say that an object of desire benefits at the time of occurrence and respect the resonance constraint, if one holds that a desired object ϕ cannot be a benefit to an individual unless ϕ occurs at the same time as the desire. If the desire and object of desire are not concurrent, the object of desire fails to be a welfare benefit. In proposing this possibility, Heathwood (2005, 490) writes:

[I]n order for a state of affairs to count as a genuine instance of desire-satisfaction, the state of affairs desired must obtain at the same time that it is desired to obtain. If I desire fame today but get it tomorrow, when I no longer want it, my desire for fame was not satisfied. A desire of mine is satisfied only if [I] get the thing while I still desire it, and continue to have the desire while I’m getting it.

Heathwood deems this suggestion “concurrentism”. Concurrentism has some initial attractions. For one thing, it seems to adequately accommodate the resonance constraint. Concurrentism will never say that ϕ is good for x at a time at which x does not value ϕ. This seems both to satisfy a central motivation for a desire-satisfaction view against objective competitors, and also to provide a straightforward and satisfying account of the phenomenon of temporal welfare: the welfare value of t ₁ is determined by the desire-object pairs that occur at t ₁.

But there are two problems with concurrentism as stated. First, a concurrentist approach must say that all desires I had prior to 2012 for things that occur in 2012 are irrelevant to my welfare. But this seems problematic. As Bradley writes: “[On a concurrentist view] there will be far less momentary well-being than we might have thought. For if we adopt a concurrence version of desire-satisfactionism, desires about the past and future cannot affect our well-being. But why should this be? Surely a great many of our desires are future directed. None of these desires would affect our well-being given the concurrence thesis,” (Bradley 2009, 23).¹⁵ As Bradley rightly notes, insofar as many of our desires are future-directed, one would expect a desire-satisfaction theory to account for their welfare value.¹⁶

The second is more problematic. Recall that any desire-satisfaction view must accept the distance allowance. Heathwood’s proposal can accept this, insofar as I needn’t experience the fact that my desire is satisfied at the time it is satisfied for the satisfaction of that desire to be a benefit to me. But the rejection of future-directed desires adopts an unwarranted limitation in the way in which desired objects are unbound to experience. Because Heathwood’s view allows spatial distance between the desirer’s experiences and that which benefits the desirer (insofar as one needn’t experience, at the time it is good for one, the satisfaction of a desire), a refusal to allow temporal distance would appear to place unwarranted evaluative significance in the fact that the relevant “distance” is in the now. Surely this cannot matter.¹⁷ What possible motivation could there be to say that one needn’t experience the satisfaction of desires so long as the satisfaction occurs at the same time as the desire? If we needn’t experience the objects of desire, what difference does it make when these objects occur?

One might put this problem more pointedly. Insofar as there can be no principled reason to treat one form of distance (spatial) as more evaluatively significant than another form of distance (temporal), to accept that temporal distance would void the evaluative significance of ϕ would seem to require us to say that spatial distance would also void the evaluative significance of ϕ. But this would just be to reject the distance allowance. But insofar as this also puts pressure on the motivation for a desire-satisfaction view against, e.g., preference hedonism, concurrentism cannot be accepted by a desire-satisfaction theory.¹⁸ Absent convincing argument that spatial, but not temporal, distance is prudentially significant, it seems that concurrentism—for a desire-satisfaction view—is ill-fitting.¹⁹

There is a third open option. One might say that ϕ is good for x at the time at which x desires ϕ. Call this the “time-of-desire” view. The time-of-desire view, when combined with the Hobbesian desire-satisfaction theory of well-being, is stated thus:

The Hobbesian Time-of-Desire View: ϕ is good for x at t if and only if ϕ is desired by x at t.

When combined with the Moorean approach we get:

The Moorean Time-of-Desire View: ϕ is good for x at t if and only if ϕ is an instance of a D-state, the desiderative component of which is possessed by x at t.

Putting this all together, the time-of-desire view allows a desire-satisfaction theory to account for the phenomenon of temporal welfare: x has a good t to the extent that x’s t-desires are satisfied (whenever their objects occur).

The time-of-desire corrects defects of previous approaches. First, it respects the resonance constraint: because ϕ is good for me only when I maintain the appropriate pro-attitude, that ϕ is good for me at t can never alienate me at t. Second, the time-of-desire view would allow us to say that future-directed desires impact our well-being: if I desire now to climb Mount Everest in 2012, and I do it, that I do so makes me better-off now, and at all times at which I desire to climb Mount Everest in 2012. Third, the time-of-desire view respects the distance allowance for both spatial and temporal distance (and hence does not adopt an arbitrary limitation in the distance allowance), and hence can accommodate the second crucial motivation for a desire-satisfaction theory.

As far as I can see, any desire-satisfaction theory that appeals to the motivations detailed in the previous section must accept the time-of-desire view. Indeed, one might think that the most significant result of this conclusion is that a desire-satisfaction view clearly cannot provide an adequate account of the temporality of welfare. Three reasons for thinking so will be explored here. First, the time-of-desire view seems unable to accommodate the possibility of a sacrifice of present welfare for the sake of future welfare. Second, the time-of-desire view cannot explain, or so it would seem, the normativity of welfare. Third, the time-of-desire view appears to violate an important constraint on the welfare value of times: that the welfare value of times should be determined by facts that occur at those times.

I think these objections fail. I will take them in the above order.

5 Sacrifice and Payback

One feature of the time-of-desire view is that a person can be made better-off at an earlier time by events or states of affairs that occur at later times. But Ben Bradley notices a potential problem here:

Suppose Kate is a pianist, and will be giving a big performance at the end of September. She practices hard, making many sacrifices, during September; as a result, she gives a spectacular performance, and this is a very good thing for her.... During the concert we might well say that all her hard work is paying off now; we might say that it is good that she worked so hard before, or more stiltedly, that her previous hard work had instrumental value as a result of what is now happening. It is much stranger to say that her current performance is paying off her past self, in the sense that it is retroactively making her better off in the past. If Kate’s performance made it the case that she was well-off while she was practicing over the previous month, it would be hard to see her practicing as involving a sacrifice of current well-being for future well-being, since her ‘sacrifice’ would have been beneficial to her at the very time she was practicing (Bradley 2009, 20–21).

According to Bradley, the time-of-desire view cannot treat Kate’s practice as a sacrifice of her early-September welfare for the sake of her late-September welfare. But this is problematic. Surely, we should like to say, early September involves substantial sacrifice of Kate’s welfare.

Bradley’s objection takes two forms. The first suggests that the time-of-desire view cannot make meaningful sense of the idea that Kate sacrifices her earlier welfare for her later welfare. The second holds that it is implausible to believe that later events “pay back” earlier times. I take these in turn.

First, we can and should distinguish between two forms of sacrifice: pro tanto sacrifice, and all-things-considered sacrifice. A particular time may involve pro tanto sacrifice if that time involves the sacrifice of things that are prudential benefits. A time involves all-things-considered sacrifice if that time is, on the whole, worse than it would have been had the pro tanto sacrifice not occurred. How does the time-of-desire view stand with regard to each form of sacrifice?

The time-of-desire view can easily accommodate Kate’s pro tanto sacrifice. After all, Kate gives up a number of important individual welfare benefits during early September for the sake of other things she desired, including to put on a good performance in late September. She spent hours practicing rather than enjoying those lovely autumn evenings, hanging out with friends, seeing films, or spending time lingering over romantic dinners. The fact that she does not accomplish these objects of desire is a sacrifice of things that would be intrinsically good for her prior to the end of September. The time-of-desire view can accommodate these very commonsensical claims. Kate sacrifices many things she desires for something that she is actually rather averse to: practicing piano. In that way, early September is a pro tanto sacrifice of many things that would be prudentially beneficial for her.

All-things-considered sacrifice is more complex. Early September involved all-things-considered sacrifice if Kate’s practice made early September worse on the whole than it would have been had she not practiced. But the time-of-desire view is certainly not barred from saying that early September, for Kate, is an all-things-considered sacrifice. On the time-of-desire view, early September was an all-things-considered sacrifice if Kate’s early September desires, say, to hang out with friends were stronger than her early September desires to perform well later in September. If this fact holds, Kate’s early September is an all-things-considered sacrifice. However, it is also possible, on the time-of-desire view, that Kate’s early September is merely a pro tanto sacrifice. If during early September Kate desires to succeed in late September more than she desires not to sacrifice for the sake of this achievement, her success in late September renders her early September merely a pro tanto sacrifice. Is this a problem for the time-of-desire view? I don’t think so. Two points. First, one might wonder whether the time-of-desire view’s verdict is really all that unintuitive. After all, if Kate strongly desired, in early September, to perform well in late September, and her desires are stronger than any desires for anything else that might have obtained in early September, I find it difficult to see why we would or should say that her early September is worse for her than had she not engaged in such practice. Second, Bradley’s complaint, as I’ve shown, is off the mark. The time-of-desire view is not barred from describing Kate’s early September as involving sacrifice: it clearly involves pro tanto sacrifice, and—depending on the strength of Kate’s early September desires—potentially all-things-considered sacrifice as well.

Now to the second form of Bradley’s objection. Bradley seems to insist that no matter how strongly Kate wanted to succeed at her performance in early September, her success in late September cannot have any effect on her early September welfare. This form of welfare payback is simply implausible. In response, however, though this considered judgment may be robust, it must be offered a rationale. After all, the time-of-desire view offers a rationale to believe that her late-September success benefits her at an earlier time: she desired this success at an earlier time, and these desires were fulfilled. But perhaps there is a competing rationale available. Most obviously, one might consider Kate’s own perspective. From Kate’s perspective, though her desires were satisfied, early September was nothing but a painful slog through annoying practice, giving up many things she otherwise wanted. At the time of desire, nothing about Kate’s early September seemed better to her. Kate’s life seems better to her only in late September. But this rationale cannot establish that the time-of-desire view does not succeed. To reject the time-of-desire view on the basis of Kate’s experience of September is to simply pound the table against the distance allowance. And if this is correct, the objection at hand does not show that the desire-satisfaction view cannot accommodate the phenomenon of temporal welfare. Rather, it reflects (at most) a more basic difference between a desire-satisfaction view and its rivals. Recall that, insofar as we are interested in objections to the desire-satisfaction account on grounds of the temporality of welfare, we should ignore objections that rely on a prior rejection of the main motivations behind a desiderative approach to welfare. Hence we should ignore the present objection.

Before I conclude this section, one additional case is worth considering that, perhaps, cannot be dismissed as table-pounding so easily.²⁰ Imagine that, at age 30, I desire not to get Alzheimer’s disease. Now imagine two lives, life A, in which I get it at age 50, at which point my desire ceases, and die at age 80 of complications related to this disease; and life B, in which I get Alzheimer’s at age 79, at which point my desire ceases, and die at age 80. The time-of-desire view might seem to indicate that I’m better-off in life A rather than in life B, insofar as the negative effects of my frustrated desire—the negative “payback”, as it were—lasted an additional 29 years in life B. But surely we would say that the first is worse: I suffer from the disease I wanted to avoid for 29 extra years.

I think there are three responses to offer here. First, to say that A is better than B one must make something of an unusual claim about the content of my desires, viz., that though I desire not to get Alzheimer’s, the length of any period of Alzheimer’s is more or less insignificant to me. But we would ordinarily think that someone who preferred not to get Alzheimer’s would prefer a shorter-length duration of this illness to a longer-length duration. My preferences, therefore, would have a tendency to rank the second life better than the first life in both lives, with that preference frustrated in the first, satisfied in the second. Second, though there is of course one respect in which one’s life goes better in A, there is no guarantee that one’s life goes overall better. After all, Alzheimer’s disease is marked by a desire, once one has it, not to have it; desires to do things one cannot do because of it; experiences to which one is averse, etc. And if this is the case, it would appear that the longer duration of the disease is worse, even on time-of-desire view standards. Third, if we abstract from these considerations, it does not seem to me implausible to say that there is one respect in which B is worse when we focus precisely on the explanation of this respect, viz., the fact that a very longstanding desire, much more longstanding than the first life’s desire, was frustrated. We generally regard desires that are held for a longer period of time as more significant to a person’s welfare. And, indeed, the time-of-desire view can explain this perfectly: longer-held desires are more significant because they affect the welfare value of more times throughout a person’s life. If we assume, for instance, that I am neutral with regard to how long I have the disease (simply that I don’t want to get it), and that I’m not averse to the disease while I have it (which seem far-fetched and unusual assumptions), I find it not so implausible, given the variable duration of the desire, to assign A a higher welfare score.

6 Normativity and the Future

I think it is not so implausible to say that, at least under the right circumstances, a later object of desire can render an earlier sacrifice merely pro tanto rather than all things considered. However, a further objection runs as follows. Facts about welfare have normative consequences. We have reason to promote the welfare of ourselves or others. But the time-of-desire view seems to imply implausible verdicts concerning what we have reason to do. Richard Brandt, for instance, writes:

The desire theory holds, then, that greater welfare corresponds to greater satisfaction of desire, and that a benevolent person, in deciding what to do, does or at least ought to perform that act among the options open to him which will maximize desire-satisfaction. The idea seems to be that we consider all the desires a person has...at some time or other, or many times, over a lifetime, and what that person...should aim at is to maximize the satisfaction of these desires. This conception is unintelligible.

That there is a problem begins to appear when we reflect that some desires need not count. Suppose my six-year-old son has decided he would like to celebrate his fiftieth birthday by taking a roller-coaster ride. This desire now is hardly one we think we need to attend to in planning to maximize his lifetime well-being. Notice that we pay no attention to our own past desires (Brandt 1979, 249).

Brandt’s objection is powerful, but, like the previous objection, takes two forms that should be distinguished. The first concerns the nature of prudential choice on the part of a given individual in light of past desires. If Brandt is correct that “we pay no attention to past desires” when it comes to prudential planning, the time-of-desire view must fail in light of the normativity of welfare. The time-of-desire view seems to imply that we must pay attention to past desires insofar as their satisfaction, at later times, makes us better-off on the whole. The second concerns whether current future-directed desires should affect rational or benevolent planning for ourselves or others. According to Brandt, the answer is no: in planning for a six-year-old’s lifetime well-being, we can justifiably ignore his desire to ride a roller-coaster at age 50.

6.1 Prudential Rationality and Future Bias

If the time-of-desire view is correct, because later events, objects, or states of affairs can affect that which benefits one at earlier times, we may have prudential reason to act in ways that benefit ourselves only at earlier times. Consider an example. Imagine that at t ₁ I have a strong desire to play the organ at t ₂. Imagine that when t ₂ rolls around, I have a very slight preference not to play the organ rather than to play it. I am, in other words, almost indifferent. It would seem, at t ₂, that I have overriding prudential reason to play the organ despite my slight aversion to so doing. After all, my t ₁ desire was very strong. Playing the organ would thus be satisfying a very strong desire I maintained at t ₁, and hence would benefit me at t ₁ more than doing so would harm me at t ₂. But that I should have such a reason seems implausible. Why should I care about benefiting myself at t ₁, when my desires at t ₂ have changed? Surely at t ₂ it is all-things-considered prudentially rational not to play the organ, given my slight aversion to so doing at t ₂. (Assume that at all future times I am wholly indifferent to my playing of the organ at t ₂.) If so, we should reject the time-of-desire view.

This argument takes the form of a reductio, as follows:

• 1. I have a prudential reason to maximize the welfare value of my life.

• 2. Later events can affect the welfare value of earlier times (given the time-of-desire view).

• 3. If later events can affect the welfare value of earlier times, in playing the organ at t ₂, I improve the welfare value of my life more than refraining from playing the organ.

• 4. Hence, I have greater prudential reason to play the organ at t ₂ rather than to not play the organ at t ₂.²¹

• 5. But I do not have greater prudential reason to play the organ at t ₂.

• 6. Hence, later events cannot affect the welfare value of earlier times.

As with all reductios, the most important question concerns which premise is false. For the objection from prudential rationality to go through, it must be the case that (2) is false. Of course, one could also avoid the reductio by denying (5), and holding that one does have reason at t ₂ to play the organ. But (5) seems very difficult to deny. After all, why should I care, today, if I desired yesterday to play the organ today? Furthermore, (1) seems to be a standard claim about the normative importance of welfare, the denial of which would be to explicitly capitulate to the present objection. Given the case at hand (3) seems obvious.

I want to step back a moment and consider the overall structure of the present objection. This objection seeks to exploit a generally plausible thought about the nature of prudential rationality, viz., that prudential rationality is future-biased.²² (Note: I will neither accept nor deny the future bias here; I assume it, however, for the sake of argument.) In other words, we have greater reason to care about our future than our past. If future bias is correct, the time-of-desire view fails, because it would seem that in some cases the time-of-desire view allows that we have greater reason to care about our past than our future. If we are not future-biased, we should accept (4), reject (5), and hence reject the claim that the time-of-desire view is in any way problematic when it comes to the prudential normativity of welfare.

But putting the objection this way shows clearly that it is question-begging. There are two ways to accommodate the future bias. First, one could deny (2). But, second, one could insist that prudential rationality commands only an interest in present and future benefits. I don’t seek to argue for one or the other, but rather for the claim that if one accepts (2), and if one accepts the future bias, one should reject (1). To insist on (1) and the future bias is just to beg the question against any view that accepts (2).

To see this, let’s ignore desire-satisfaction views for the moment. Let’s assume that hedonism is true, and that, incidentally, time-travel is possible. Under these conditions, (2) is correct. But under these conditions, would we accept (1)? Not if we also accept the future bias. Imagine that I could send a book of knock-knock jokes back in time to my 12-year-old self, who would find the book very funny, or I could read it tomorrow (which would make me laugh, but not as hard as I would have were I 12 years old). If we accept the future bias, it is prudentially rational to read the book now. But this would be ruled out if we accept (1). According to (1), the prudentially rational thing is to send the book of knock-knock jokes back in time; this would generate the greatest amount of past-present-and-future well-being. If we accept the future bias, we should, under such conditions, accept an alternative to (1):

• 1*. I have a prudential reason to maximize the benefits I achieve at present and future times of my life.

I do not claim here to insist that (1) is false. But I do suggest that if we accept the future bias, whether we should accept (1) must be determined at least in part by whether we accept (2). Hence to insist both on (1) and the future bias to reject (2) is to beg the question against any view that accepts (2). (For instance, one certainly wouldn’t reject hedonism if we found out that time-travel is possible; we would simply alter our understanding of the general principles of prudential rationality—in light, of course, of the future bias.) If we accept (1*) rather than (1), we need not reject the time of desire view on the basis that we generally ignore past desires. The time-of-desire view, together with (1*), treats the fact that we do so as prudentially rational, and hence allows (5). This is precisely what one would expect given a commitment to the future bias and (2), whether or not one accepts the time-of-desire view.

6.2 Benevolence and Future-Directed Desires

Take now Brandt’s suggestion that his six-year-old son’s desire to celebrate his fiftieth birthday with a roller-coaster ride “is hardly one we think we need to attend to in planning to maximize his lifetime well-being.”²³ Of course, all the arguments in the previous section apply here: if I am prudentially required to maximize only my present and future well-being, it would seem that any other person can be required by norms of benevolence only to maximize my present and future well-being, rather than lifetime welfare. Fair enough. But the critique here is different: it is certainly possible that we could plan, now, to satisfy the six-year-old’s current desire to ride a roller-coaster at fifty; this would satisfy a present desire of the six-year-old, and hence improve that six-year-old’s present welfare. But this, according to Brandt, seems absurd.²⁴

Much of the force of the intuition noted by Brandt, I think, derives from the general assumption that many of our desires about future experiences of this kind are conditional on their own persistence.²⁵ (Indeed, this may be true even of my desire, at t ₁ to play the organ at t ₂.) In all but the rarest cases, a six-year-old child’s desire to ride a roller-coaster on his fiftieth birthday will be conditional on his continuing to desire to do so when he is fifty. And if this is correct, Brandt is of course right that we should not take this desire into account in planning for the child’s future. Given the general facts of human psychology, it is unlikely that this desire will persist, and hence there would be no benefit to be gained for the six-year-old in making it the case that his fifty-year-old self actually rides a roller coaster. But this is perfectly compatible with the time-of-desire view.

However, what if we imagine that the child’s desire is not conditional on its own persistence? In other words, what if we imagine that the child desires to ride a roller coaster at fifty no matter what he actually desires at fifty? It would appear, at first glance, that the time-of-desire view tells us that we should seek to make it the case, in planning to maximize this child’s lifetime well-being, that he rides a roller coaster at fifty, insofar as we can. Perhaps this remains an implausible consequence.

But this objection is not compelling. To see why, consider the difference between subjective and objective rationality. One’s objective reasons are those reasons one has “in light of all the facts”; subjective reasons are those reasons that one has in light of the information one has at the time of decision.²⁶ Given this, we should distinguish between two different potential objections. Brandt might be suggesting that the time-of-desire view implausibly suggests that it is subjectively rational to attempt to satisfy this child’s desire for a particular state of his future self. Alternatively, Brandt might be suggesting that the time-of-desire view implausibly suggests that it is objectively rational to do so.

Brandt’s objection seems explicitly tied to the suggestion that it is subjectively rational to seek to satisfy this desire, i.e., that we should “attend to” this desire in “planning to maximize” the six-year-old’s lifetime welfare. But the time-of-desire view need not hold that it is subjectively rational to put in motion a plan to satisfy the six-year-old’s future-directed desire. We have every reason in the world to believe that doing so will actually lead to more harm than good in terms of this person’s normatively relevant welfare. The six-year-old’s desire is likely to be fleeting, and is likely to be sharply outweighed by contrary future desires. Given our information, it is sensible to believe that as this child ages he will become less and less enamored of riding a roller coaster at 50, instead preferring to, say, relax with a titillating detective novel and a glass of bourbon. Hence to attempt to plan a roller-coaster ride forty-four years hence will—for all we know—do more harm than good. Brandt’s stated objection misses the mark.

Brandt might be interpreted slightly differently. Why not claim that it is not objectively rational to treat the child’s current preferences as normatively significant? We might imagine, for instance, that this child at 50 years old is simply indifferent (or almost indifferent) to roller-coaster rides, and hence setting in motion a plan to satisfy the six-year-old’s desire will in fact maximize welfare. In this case, it would be objectively rational—assuming the truth of the time-of-desire view—to do so. But stated as an objection to the objective rationality of taking this desire seriously, Brandt’s objection is not at all compelling. Consider a purely intrapersonal case. Let’s assume that I, at t ₁, desire—in a way that is unconditional on the desire’s persistence—that forty years hence I will still be teaching philosophy. Imagine that it’s also true that, forty years hence, I will be indifferent. It does not seem prudentially unmotivated to now take the steps necessary to ensure that I am teaching philosophy forty years hence given that I desire that I should do so now (and will be indifferent to doing so later). It would certainly be prudentially unmotivated (objectively speaking) if my future aversion to so doing is comparatively stronger than my present desire. But it does not seem at all unmotivated—assuming that my current desire is stronger than any potential future aversion—to treat this desire as objectively normative for me now. (Of course, there remain strong reasons to be subjectively averse to taking such steps.)

But if we accept that this purely intrapersonal desire is objectively prudentially normative for me now, we cannot fail to treat a six-year-old’s desire as objectively normative from the point of view of benevolence (assuming that respecting it will improve his welfare). Furthermore, to say that it is so normative is plausible. Assume that (for whatever reason) I know that my son, at age fifty, will be indifferent to roller-coaster rides on birthdays. But assume, also, that I know that my son strongly desires that forty-four years hence he will ride a roller-coaster on his fiftieth birthday. Imagine also that I can set in motion a sequence of events that will make this occur. Would doing so be beside the point? Unwise? Irrelevant from the perspective of what I, as a father, have reason to do? Surely not! Though my son will not experience the satisfaction of his desire, I have made a wish for his future self come true. And in so doing, I have made him better-off. From the perspective of a desire-satisfaction theory, one that accepts the resonance constraint and distance allowance, this is precisely the answer one would expect.

7 Internalism

The final objection to the time-of-desire view is stated by Bradley. This objection seeks to marshall a constraint on the welfare value of times to rule out the time-of-desire view. To see how the objection works, consider Bradley’s definition of a “value atom”. For Bradley, a value atom is a value bearer that is “intrinsically good or bad for us in the most fundamental way” or is “basically, non-derivatively, good for us,” (Bradley 2009, 5). In contrast, something is derivatively intrinsically good for x if it is intrinsically good only insofar as its proper parts that are value atoms, (Bradley 2009, 5). With this in mind, Bradley’s objection to the time-of-desire view makes use of a principle designed to constrain any acceptable account of temporal welfare:

Internalism. The intrinsic value of a time for a person is determined entirely by the value atoms obtaining at that time (Bradley 2009, 18).

According to Bradley, a time-of-desire view must reject Internalism. Bradley (2009, 21–22) writes (directing his attention to a Moorean view):

Remember that desire satisfactionism...[takes] the value atoms to be compound states: states of affairs that are conjunctions of someone having a propositional attitude (desire, belief, pleasure...) directed at a proposition P, and P being true. Suppose the person has the attitude at time t ₁ and the object obtains at later time t ₂. When does their conjunction obtain? It does not obtain at t ₁; only part of it does. The same goes for t ₂.

Bradley accurately notes the problem posed by Internalism for Moorean desire-satisfaction theories. Given that, on my view, benefits accrue to the individual at t₁, and given that the state of affairs in which I desire ϕ and ϕ does not obtain at t ₁, t ₁ cannot be an intrinsically valuable time. But Hobbesian desire satisfaction theories also face a problem, at least as I understand them. Because it is ϕ that is good for x, if ϕ occurs at t ₂, and x desires ϕ at t ₁, the value of t ₁ for x is not determined entirely by the value atoms obtaining at that time. The value of t ₁ is determined by value atoms occurring at t ₂.

I have two responses. First, I argue that there is no good reason to accept Internalism qua constraint on approaches to the phenomenon of temporal welfare. Hence even if the time-of-desire view violates this principle, we should think no less of the time-of-desire view. Second, I argue that the time-of-desire view, in either its Hobbesian or Moorean formulations, is—contrary to Bradley’s supposition—compatible with Internalism.

7.1 Why Believe Internalism?

Why should we believe Internalism? Bradley claims that Internalism is supported by a general Moorean claim about the nature of intrinsic value, identified as follows:

SUP. The intrinsic value of something depends solely on its intrinsic properties. (Bradley 2009, 19)

According to Bradley, if we accept the claim that the intrinsic value of ϕ can depend only on the intrinsic properties of ϕ, we must accept that the intrinsic value of a time can depend only on things that occur at that time.

Bradley’s argument here relies on a crucial assumption. Even if we accept that intrinsic value generally supervenes on intrinsic properties, it must be the case that SUP applies not just to value atoms, but also to times. But I think we should reject this claim. There is no good reason to believe that SUP applies to times. And if SUP does not apply to times, there is no pressure to accept Internalism, which applies to times, on the basis of SUP, which does not.

To see why SUP doesn’t apply to times, consider the argument in favor of SUP’s application to any potential bearers of value:

SUP is a requirement of any acceptable theory of well-being. This is because, as noted above, the value atoms should be instantiations of the fundamental good- or bad-making properties—the properties that are fundamentally and completely responsible for how well a world (or a life, or ...) goes. Suppose SUP were false. Then there could be two properties, F and G, such that the only intrinsically good states of affairs are those involving the instantiation of F alone, but whose values are determined by whether there are any instantiations of G. But if that were true, F would fail to be a fundamental good- or bad-making property, for instantiations of F would fail to completely determine what value there is. (Bradley 2009, 19)

In other words, we accept SUP because we are interested in the fundamental bearers of intrinsic value—i.e., value atoms—SUP must hold. Without conforming to SUP, value atoms would not be fundamentally, non-derivatively good.

The problem here should be obvious. Times are not value atoms. On no theory of welfare is a time a fundamental, basic intrinsic value bearer. (Or, at least, on no theory that I or Bradley accept.²⁷) Times can be “derivatively” or “non-fundamentally” valuable, insofar as they might be made up of, or have as proper parts, or be otherwise related to states of affairs that are, in fact, value atoms.²⁸ But because a time is not a value atom (a moment of pleasure, an object of desire, a D-state, etc.) any intrinsic value a time possesses must be derivative rather than “fundamental”.

But if times are not value atoms, there appears to be no pressure to insist that times must conform to SUP: the motivation to accept SUP for any potential value bearers applies only to value atoms, rather than derivatively intrinsically good things like times. But if this is correct, Bradley’s argument does not support Internalism. Internalism only holds if SUP holds of times. Instead, we may claim that SUP does not apply to this form of derivative intrinsic value, and hold that a particular time 1 is (derivatively) intrinsically good for x given that x desires ϕ at t ₁, and ϕ, at t _n obtains. Whatever reason we may or may not have for accepting SUP says nothing about whether we should accept this account of the (derivative) intrinsic value of times. If that is correct, there is no reason to believe the time-of-desire view fails in light of SUP.

This proposal is compatible with Bradley’s reasoning for SUP. But one might be wary of it nonetheless.²⁹ After all, as someone may say, we’re talking about intrinsic value! Intrinsic value, as a conceptual matter, supervenes on intrinsic properties. That’s just what intrinsic value means.³⁰ If so, Internalism is analytic and hence one cannot claim, as a substantive matter, that the intrinsic value of times can supervene on extrinsic properties, as the proposal I sketched above declares.³¹ I think there are very good reasons to be skeptical of a conceptual link between intrinsic value (if we assume that intrinsic value is the sort of value we should care about) and intrinsic properties (Dorsey 2011). But no matter. As I argue in the next section, even if Internalism is analytic, the time-of-desire view can accommodate it.

7.2 The Time-of-Desire View is Compatible with Internalism

I think there is much to dislike about Internalism as a principle, but let’s assume that it holds, just for the sake of argument. Bradley is mistaken in his claim that the time-of-desire view must violate it. The key is Bradley’s failure to distinguish between two very different evaluative questions when it comes to times: “what makes a time intrinsically valuable?” and “what determines someone’s welfare at a time?” Internalism, and perhaps other principles that are relevant to the intrinsic value of times, respond to the first question. The time-of-desire view, by contrast, seeks to answer the second question: it suggests that what makes someone’s welfare higher or lower at a time is the extent to which a person’s desires, at that time, are satisfied (no matter when the object of the desire occurs). Given that the time-of-desire view answers a different question than Internalism, the time-of-desire view needn’t be incompatible with Internalism.

To see why, take an example. One might say that the time at which I climb Mount Everest is derivatively good for me, and is derivatively good for me because a value atom (my scaling of Everest) occurs at that time, respecting Internalism. Insofar as a proper part of this time is an object of my desire (i.e., my climbing of Mount Everest), we can perfectly well say, even on the time-of-desire view, that this particular time is derivatively intrinsically good for me on the basis of value atoms occurring at that time. What the time-of-desire view must say—which is not incompatible with Internalism—is that the time at which I climb Mount Everest is intrinsically good for me at a different time. If, say, during 2008 I desire to climb Mount Everest before I die, and I accomplish this feat in 2010, 2010 is derivatively intrinsically good for me in 2008: one intrinsic property of 2010—one of the states of affairs that make up 2010—is intrinsically good for me in 2008, the time at which I desired to climb Mount Everest. Nothing about this is incompatible with Internalism. The time-of-desire view can hold that 2010 is good for me only on the basis of value atoms occurring during 2010, which is precisely what Internalism requires. But Internalism says nothing about when 2010 must be good for me. Internalism only offers a criterion: what could possibly render a given time (derivatively) intrinsically valuable. It says nothing about when such a time is a (derivative) intrinsic benefit.

Bradley will respond in the following way. I have so far assumed the Hobbesian desire-satsifaction view. Doing so allows me to claim that value bearers, i.e., the objects of desire, occur at a particular time and hence can account for the derivative intrinsic value of the time of occurrence. But the Moorean has a bigger problem: unless the desire and desired object are concurrent, the value-bearing D-state does not occur at any particular time. And hence, because a time can be good only if that time is made up of value atoms, a Moorean view must violate Internalism: value atoms do not occur at any time, leaving us unable to assign the value of times on the basis of value atoms that occur at those times.

But merely because the Moorean view cannot account for the value of times does not entail that it violates Internalism. Internalism is violated only if there exists a particular time the intrinsic value of which is dependent on facts other than value atoms that occur at that time. But the Moorean view does not say this. The Moorean view is perfectly (indeed trivially) compatible with Internalism insofar as it simply denies that times are intrinsically valuable (except, perhaps, if the desired object and desire are concurrent; but in this case the Moorean view would not violate Internalism). It is perfectly open to the Moorean view to claim that the compound state of affairs in which I desire (at t ₁) ϕ and (at t ₂) ϕ obtains is good for me at t ₁, but that neither t ₁ nor t ₂ is intrinsically good for me, insofar as times are never intrinsically good, derivatively or otherwise. Internalism is thus satisfied by a Moorean approach trivially.

This might be thought a pyrrhic victory. According to Bradley, every theory of well-being should account for the intrinsic value of times (Bradley 2009, 28–30). Without doing so, says Bradley, a welfare theory cannot account for the phenomenon of temporal welfare. But I find it difficult to see why this should be so (see note 31). A time-of-desire view can accommodate temporal well-being without insisting that times have intrinsic value. To say that a person is having a good day is just to say that this person is benefited (or burdened) during the day in question. To say these things, one need not say that the day itself is intrinsically valuable or possesses any other kind of value. One need only say that a particular welfare benefit is good for a person during that day. The time-of-desire view can do this, whether one accepts a Hobbesian, Moorean, or any other version of a desire-satisfaction approach. Hence there is no pressure whatever for a theory of welfare to provide an account of the derivative intrinsic value of times. If so, both the Hobbesian and Moorean desire-satisfaction theories can accept Internalism (as a constraint on what makes a given time derivatively intrinsically valuable) and account for the phenomenon of temporal welfare.

8 Dénoument

I have argued that a desire-satisfaction theory can adequately account for the phenomenon temporal welfare only if it accepts the time-of-desire view. I have also responded to three important objections to the time-of-desire view. These objections either turn on a mistaken understanding of the time-of-desire view and its resources (such as the argument from Internalism) or seem to beg the question against any plausible iteration of a desire-satisfaction view (such as the argument from sacrifice and from the normativity of welfare). Hence the objection from the phenomenon of temporal welfare can constitute no independent objection to a desire-satisfaction theory. A desire-satisfaction theory has a natural theory of temporal welfare, one that properly accords with its central motivations.

Of course, some will disagree with the considered judgments I attempt to elicit here. Some will hold, for instance, that a father has no reason whatsoever to attempt to satisfy his six-year-old’s future-directed desires. Some will hold that Kate’s good performance cannot affect her earlier welfare, no matter how strongly she desired at earlier times to perform well. However, that we have reached intuitive bedrock says less, I think, about the per se phenomenon of temporal well-being and the ability of a desire-satisfaction view to accommodate it, and more about the wholesale plausibility of a desire-satisfaction theory versus its most important competitors. As I have shown, desire-satisfaction theories can offer an account of the phenomenon of temporal welfare that is perfectly reasonable by their own lights. Whether their own lights are at all reasonable is a topic I humbly lay aside.