Pulling Apart Well-Being at a Time and the Goodness of a Life
Skip other details (including permanent urls, DOI, citation information)
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. Please contact firstname.lastname@example.org to use this work in a way not covered by the license. :
For more information, read Michigan Publishing's access and usage policy.
1. The Standard Assumption about Well-Being and the Goodness of a Life
Some assessments of what is good for a person focus on the state of the person at a moment or during a short period of time and how the person is faring then. These are assessments of the person’s well-being at that time. Other assessments take the person’s full life into view, saying how good that life is on the whole. These are assessments of life-goodness. Thus delineated, assessments of well-being and assessments life-goodness are different. That cannot be a controversial point. After all, the respective objects of assessment—the state of a person at a time and the series of events that compose a life—are of different kinds. But also uncontroversial is the assumption that these two kinds of assessment are so closely related that one can be understood in terms of the other via the principle that well-being makes a life good for the person who lives it. I will argue that, even on a weak interpretation of this principle, it is not true. Refuting that weak connection between well-being and life goodness is tantamount to accepting that a person’s well-being at a time and the goodness of her life as a whole are distinct values that should be theorized independently and (in at least some cases) pursued independently.
It is easy to see why the principle that well-being makes a person’s life good is uncontroversial. Mundane examples support thinking in terms of this principle. Imagine Elliot, driving from work to his daughter’s soccer match. In one possible life, the drive is pleasant, he arrives for the kickoff, and he sees his daughter score an early goal. Another possible life for Elliot is similar, except that a traffic jam delays his arrival, and he does not see the goal. We can imagine that, besides this difference, the lives are about the same. Elliot has a good relationship with his daughter, and missing this one minor event would not affect that at all. Still, for the period when the two possible lives differ, Elliot has more well-being—he is faring better—in the life without the traffic delay. And furthermore, on the whole, the life in which he sees his daughter score the goal seems to be a slightly better life than the otherwise similar life with the traffic.
However, as I hope closer inspection will reveal, we cannot count on increased well-being to coincide with improvement of a life on the whole. My thesis is that sometimes a period of elevated well-being fails to make a life better.
If correct, this result bears on two central issues in the philosophy of well-being. The first is the debate about whether a person’s well-being during a long period is simply the aggregate of her well-being over the shorter periods that compose it. If we understand the overall well-being of a life as the goodness of that life, then the thesis here entails that the answer is no. More importantly, the argument here shows that the connection between well-being at a time and the goodness of a life is even looser than the failure of aggregative conceptions of the good life might suggest. I will pick this up in Section 3.
The second is the challenge of demarcating well-being and other sorts of value. Some, notably Sumner (1996), have been willing to admit that an evaluation of a life may diverge from evaluations of a person’s well-being at a time. His rationale is that a person’s life may be evaluated in ways that do not reflect the person’s own good. For instance, lives may have perfectionist or aesthetic value, varieties of subject-independent value distinct from the good of the person herself. However, the argument here supports the view that a person’s own good is itself plural, that a person’s well-being at a time and the goodness of a life for the person living it are themselves distinct values. I will return to this point in the final section.
2. Making the Assumption Precise
Before we see the grounds for rejecting the standard assumption that well-being makes a person’s life good for her, we need a more precise interpretation of that assumption. As a preliminary step, we can be more specific about what well-being is. In accord with the gerundive terms we use to describe it—well-being or faring well—I will understand well-being as something a person has at moments or during periods of her life. To talk about a person’s well-being at a time is to talk about an aspect of her state at that time. Her well-being at a time is constituted by having some particular collection of (perhaps highly relational) properties then, and her level of well-being at that time will be determined by which properties are instantiated and to what degrees. Which properties are in the well-being-constituting collection is a substantive, normative question. So, for instance, according to a hedonistic theory, they will be the person’s mental properties, specifically those that constitute pleasure and displeasure. According to a list theory that includes friendship among the determinants of well-being, some of the properties constituting the person’s well-being at a time will be whatever intrinsic and relational properties of the person make it the case that she is friends with her friends.
Notably, it is common to theorize well-being at a level of abstraction that is not specifically focused on the state of a person at particular times. If the relevant question is posed as what is good for a person, and “well-being” is used as a label for things that are, then a focus on the state of a person may seem unnecessarily narrow. Daniel Haybron, for one, characterizes well-being broadly, without distinguishing assessments of states from assessments of lives: “The concept of well-being is a normative or evaluative concept that concerns what benefits a person, is in her interest, is good for her, or makes her life go well for her” (2008: 29). At this level of abstraction, substantive theories need not be formulated in terms of a person’s life or her state at a time at all. Haybron says, for instance, “the desire theory identifies well-being with the satisfaction of the individual’s desires” (2008: 34) and “we have list theories of well-being, which identify well-being with some brute list of goods, such as knowledge, friendship, accomplishment, pleasure, etc.” (2008: 36). One of my main goals is to show that the details obscured by this level of abstraction are significant.
Although much work on well-being does proceed at that abstract level, much other work invokes (explicitly or implicitly) a temporally sensitive notion aimed at the state of a person at a time. Indeed, the entire debate about whether well-being is aggregative (about which more below) presupposes a central theoretical role for well-being at a time. Otherwise it would not make sense to ask whether the total well-being across the whole span of a life was equal to the sum of the well-being in the life’s shorter parts.
Ultimately, though, the reason for focusing on well-being as an aspect of a person’s state at a time is our ordinary thinking about how things are going for ourselves and others. We worry about each other’s ups and downs. When I think of how poorly a friend was faring while in a dysfunctional relationship and how much better she is doing now that she has extricated herself, I am comparing her state at one time to her state at another. Similarly, we must be thinking of well-being as something that attaches to particular time periods when we devise or evaluate policies that are intended to make people better off than they are now. Hence, it is irresistible, in at least some quite serious contexts, to think of well-being as something that can increase or decrease over time, and thus as an aspect of a person’s state at a time.
With well-being thus circumscribed, we can begin to refine the assumption that well-being makes life good. The first step is to understand the assumption as a relation between two kinds of improvements: Increased well-being improves a life. But with these increases or improvements, what exactly are the comparisons? Often, we think of an increase in a person’s well-being in terms of a temporal comparison: When a person’s well-being has been increased, she is in a better state than she was before. The comparison is between earlier and later periods in the person’s life. In contrast, with life improvement the comparison is always counterfactual. If a person’s life is improved, her life is better than it would have been without the improvement. The comparison is between two possible lives of the person. Crucially, though, counterfactual comparisons are possible also for a person’s well-being at a time. For instance, I might say that some person was better off because of her good night’s sleep than she would have been otherwise. In that case, I would not be saying she is in a better state now than before she slept, but that she is better off at a particular time than she would have been at that time had she not slept.
The standard assumption about well-being and life-goodness has to be about counterfactual, not temporal, comparisons. Otherwise, it would be obviously false. For example, suppose I took you out for a nice meal in order to raise your well-being above its present level. If that outing would keep you from doing something else even better for you, then I have not improved your life. Thus, a temporal increase in well-being is not at all guaranteed to improve a life. So the assumption must be about counterfactual comparisons: Having greater well-being than she would have had during some period makes a person’s life better than it would have been (had there been no extra well-being).
Finally, we must be careful about cases where increased well-being at one time comes at the expense of decreased well-being at another time. It could be that some event makes my life better than it would have been during some particular period, but this causes my life to be worse during some other period. Examples are easy to come by; paradigm cases of imprudence are among them. Suppose I elevate my well-being by enjoying an expensive holiday trip, but, because of this, I cannot afford to repair my furnace later on. Then the trip effectively increased my well-being during one period but caused it to be reduced during another. So we should not expect the initial well-being increase to entail that the life, as a whole, has been improved. Hence, we need to formulate the intuitive connection between well-being and life-goodness in a way that holds outside factors fixed. We need an “all else equal” or “ceteris paribus” qualification. The result is this:
|(CP)||If possible lives A and B are just alike, except for a period in which the person has a higher level of well-being in A than in B, then A is a better life than B.|
In CP, talking about lives that are just alike, except for a period, expresses the ceteris paribus condition on the relation between increased well-being and increased life-goodness. Observe that CP correctly handles the example involving the expensive trip and the furnace repair. The trip does increase my well-being during some period of time; while enjoying my trip I am better off than I would have been at home. And, then, in a comparison of two lives that are just alike, except that one includes the trip and the other does not, the one with the trip is (presumably) better. This is what CP predicts. Note that, in this comparison, it is crucial that the two lives not differ regarding the subsequent furnace episode; either the repair occurs in both or it occurs in neither. If the two lives differed not just with regard to the period of increased well-being (during the trip), but also regarding the furnace situation, then the ceteris paribus condition would fail, and so CP would not apply.
Because the ceteris paribus condition is fairly strict, CP is fairly weak. Yet CP does express the core of the assumption that well-being makes a life good.
3. Pervasiveness of the Standard Assumption
Before criticizing CP, it is worth putting it in perspective. Across radically different views of well-being and even among philosophers who have challenged prevailing views about the relationship between well-being and life-goodness, CP remains a background assumption.
CP is a plausible background assumption precisely because of what it does not say. It does not provide any explanation of why increased well-being should improve a life. It does not say anything about how the magnitudes of increase are related, let alone specify a function relating them. Finally, the conditional runs in only one direction; it is neutral about the converse claim that life improvement requires increased well-being.
The most widely discussed claim about the relationship between well-being and life-goodness is aggregationism, the claim that the goodness of a life is the life’s aggregate well-being. Take the entire span of a person’s life and partition it into short periods. For each of these short periods, multiply the level of well-being at that time by the duration of the period. According to aggregationism, the sum of these products is the total goodness of the life. Aggregationism is perhaps the most literal way to make sense of the intuition that well-being is what makes life good. It says, more or less, that the goodness of a person’s life is constituted by (by being simply the aggregate of) the well-being in the life. Now, CP does not specify a function relating changes in well-being to changes in life-goodness, but aggregationism does entail CP. So, aggregationism is strictly stronger, and it can be rejected while accepting CP. This is just what we find among critics of aggregationism.
Michael Slote (1983) famously criticizes utilitarians (e.g., Sidgwick) and contractarians (e.g., Rawls) alike for assuming aggregationism. Slote accepts time preference, the principle that the value of some good in a person’s life may depend, in part, on when it occurs. For instance, well-being at the end of a life may be more valuable than well-being at the beginning. If so, then two lives could be equal in aggregate well-being but different in overall value because of how the well-being was distributed. Thus, time preference threatens aggregationism. But it does not pose any problem for CP. According to time preference, increased well-being improves a life to a greater or lesser degree depending on when the increase occurs. It is natural to accept this while also claiming that, all else equal, increased well-being always improves (at least somewhat) a life.
Putting aside time preference, there are other good reasons for rejecting aggregationism that do not threaten CP. For instance, aggregationism faces a problem analogous to Parfit’s Repugnant Conclusion: A sufficiently long life barely worth living will be superior to a shorter span of excellent living. One might attempt to improve on aggregationism by holding that life-goodness depends in part on the average level of well-being in the life, not just the sum. This would be a rejection of strict aggregationism. But, far from rejecting CP, it would require it. After all, increased well-being during some period, other things being equal, increases the lifetime average well-being.
Whether in terms of sums, averages, or a combination of these, the views just discussed understand the goodness of a life as a function of well-being during the life. Since the value of the relevant function increases with any increase in well-being, the truth of CP is ensured. Hence, we might think that denying that there is any function from well-being to life-goodness would amount to a rejection of CP. But this is not so. CP is compatible with—and, in some cases, required by—views that deny that there is any such function.
Consider compositionalism, a view advanced by Johan Brännmark (2001). On this view, an individual’s well-being over some stretch of time depends on how that episode fits into the person’s life as a whole. Whereas aggregationist views start with an independent valuation of the parts of a life and figure the value of the whole life on that basis, a compositionalist view starts with the value of the whole and evaluates parts with respect to their roles in the whole. Despite this difference, compositionalism is not just consistent with CP; it straightforwardly entails it. Suppose that some conditions increase, in the counterfactual sense, a particular person’s well-being. In other words, if things had been otherwise, the person would have had less well-being during the time period in question. How are we to understand this increase, according to compositionalism? Well, according to compositionalism, a person’s well-being during some period of time depends on how the facts at that time fit into the rest of the person’s life. So, for facts to constitute an increase in well-being, those facts must fit into a life that is better than the otherwise similar life where those facts did not obtain. Hence, according to compositionalism, facts increase well-being only if the person’s life as a whole is better than it would have been had the facts been otherwise. This is just CP. Perhaps some variant of compositionalism will have a more complicated story about how the values of the parts of a life are determined from the value of the whole, but it is hard to see how a plausibly developed compositionalist view would not entail CP in the way just described.
Other views part ways with both aggregationism and compositionalism by denying that either well-being or the goodness of a life is derivative of the other. Shelly Kagan (1994) holds that the facts that determine a person’s well-being must be just the non-relational facts about that very person, but additional facts may bear on the goodness of the person’s life. For instance, according to this view, that I am having a pleasant experience is relevant to my well-being, but the fact that my pleasure is due to some false beliefs is not relevant, since the falsity of the beliefs is a relational matter. The falsity of the beliefs could, however, bear on the goodness of my life. Thus, the well-being in a person’s life does not fully determine the goodness of her life, and so Kagan rejects aggregationism. Furthermore, since the factors affecting well-being are specified without regard to their contribution to overall life-goodness, Kagan’s view is not compositionalist.
In the same vein, David Velleman (2000) argues that the contribution of some period of well-being to the goodness of a person’s life is affected by how that episode fits in the person’s life story. What Kagan’s and Velleman’s views share is the thought that factors beyond well-being may affect the goodness of a person’s life. Because they allow for differences in the goodness of lives that do not depend on differences in well-being, they reject the converse of CP. As Kagan puts it, “certain changes might constitute changes in the quality of a person’s life without constituting changes in the person’s level of well-being” (1994: 319). Still, these views are perfectly consistent with CP itself. As long as everything that improves well-being also makes a life better overall, CP holds. Accordingly, CP remains a seeming truism, which neither Kagan nor Velleman disavows.
4. Counterexamples to the Standard Assumption
Although CP is a pervasive assumption in philosophical discussions of well-being, we have reason to reject it. There are cases where increased well-being, even all else equal, makes a person’s life worse on the whole. But, even if I am right about this, exactly which cases are indeed counterexamples to CP will be controversial. Because of the many different conceptions of well-being and life-goodness which might be brought to bear on the evaluations, there may exist no single counterexample that everyone finds compelling. In this section, I present two cases that I consider counterexamples. For readers who share my evaluations of these cases, this section will suffice to refute CP. But, for those who do not find them persuasive, these cases will have to play a more modest role. First, they exemplify the structure that a counterexample to CP must have. Second, they will help us identify the assumptions about well-being and life-goodness on which any purported counterexample’s status as a counterexample depends.
In a counterexample to CP, we will have a comparison of two possible lives which are almost exactly alike. They will be different just during a particular period, differing just with respect to the facts make the person’s level of well-being greater in one life than in the other. And the life with the period of greater well-being will be worse overall. Consider this scenario:
L1 and L2 are two possible lives for Frances. In both L1 and L2, Frances graduates from college with a bachelor’s degree in fine arts during an economic recession. In both lives she is unable to support herself financially as an artist. In L1 but not L2, Frances’s parents serve as her safety net, and she moves back home. At home she is comfortable and happy. For a year, she devotes herself to researching the weak economy’s effects on people, from factory workers to other artists. During that time she creates some excellent paintings that deal with struggles of people in the poor economy, and, at the end of the year, she is recognized for her excellent artistic achievements. In L2, Frances does not rely on her parents as safety net, and she is miserable for a year. She works two low-paying jobs. She is overworked, sleep-deprived, uncomfortable, and generally dissatisfied with her life. However, during that time she manages to produce some excellent paintings that deal with struggles of people in the poor economy, and, at the end of the year, she is recognized for her excellent artistic achievements. In both L1 and L2, after gaining notoriety for her paintings, Frances becomes able to support herself as an artist, and she continues producing paintings with themes relating to poverty. After the year in question, L1 and L2 are the same. Frances dies fairly young and never reflects on the conditions under which she created her first acclaimed paintings.
During their only period of difference, L1 includes more well-being than L2. So, if we take CP for granted, then it should not even be plausible that L2 is better than L1. But consider just how plausible it is. We may think that, because of the way it is organized, L2 is a better life for Frances than is L1. In light of Frances’s values and the themes that animate her work, a life in which success grows out of a period of hardship, out of a period of low well-being, has a structure that suits Frances particularly well. Moreover, we can stipulate that if (contrary to what actually happened) Frances herself had reflected on her reduced well-being in L2, she would have valued it as an element of her life. We can imagine Frances elevating her assessment of her entire life because of it, and I think we should agree that it is indeed a better life for her.
I have been suggesting that Frances, because of increased well-being, missed out on a better life, but it would be an exaggeration to say that the increased well-being was a defect in her life. It seems better to say just that, because of the increased well-being during that one period, her life on the whole was not as good as it could have been. We can, though, imagine other cases where increased well-being constitutes a defect in view of the person’s whole life.
L1 and L2 are two possible lives for Wade. In both L1 and L2, on a particular afternoon Wade is driving on the highway where, off to the side, there is a motorist stranded with a flat tire. In L2, a butterfly splats on Wade’s windshield and distracts him. He does not see the motorist, and he continues driving. In L1, there is no butterfly. Wade sees the motorist, pulls over, and helps him change the flat tire. The motorist thanks Wade, and this makes Wade happy as he drives away. In both L1 and L2, Wade then becomes engrossed by a story on the radio. As a result, in neither life does he ever consider the motorist again. After that episode, L1 and L2 are the same. Several years later, in both lives, it comes to light that Wade’s son has been molested by a neighbor. Unbeknownst to Wade and the neighbor, the neighbor is the motorist whom he aided years before.
During that brief time when he was helping the motorist, Wade was faring better in L1 than in the corresponding period of L2. After all, in that episode, Wade achieved something worthwhile, and it was a (briefly) gratifying experience. If so, then, in light of CP, L1 should be better than L2. But there are reasons to think it may not be. The abuse of his son is a great tragedy in Wade’s life. In view of the life as a whole, the earlier aid to the perpetrator taints that part of Wade’s life with the badness that was to come. Now, in L1, after the day the motorist was stranded, Wade never thinks of him again. However, if in a slightly altered version of L1, he later thought about him and also became aware of the man’s misdeeds, he would regret having had the benefit of the increased well-being that came from his interaction with him. Assuming it would not be unnatural or inappropriate for Wade to respond this way, we should be inclined to think that L1 is a worse life for him than L2. So, then, since it is the difference in well-being between L1 and L2 that accounts for the difference in the overall goodness of the lives, Wade’s case, like Frances’s, looks like a counterexample to CP.
In both of these examples, I have tried to make my proposed evaluations of the lives compelling by stipulating that each protagonist would, if she or he reflected on it in retrospect, consider the increased well-being a demerit in the evaluation of her or his life as a whole. The assessments implicit in such reflections are strong, if not decisive, evidence about how the past episodes figure in overall evaluations of the lives. Furthermore, a person’s reflective endorsement buttresses the intuition that the value at stake is genuine goodness for the person, as opposed to some other sort of value that may be distinct, like perfectionist value or aesthetic value. However, it is not essential to counterexamples to CP that the protagonist, were she to reflect, would have any particular take on the relevant events. It is just that this sort of hypothetical retrospection makes for compelling cases. Crucially, though, if retrospective assessment is to play a role in a counterexample to CP, the retrospection must be merely counterfactual, not actual, within both of the possible lives compared. If the person actually, at some later point, appreciated the time period in question—the time period in which the two lives differed—that appreciation would amount to an additional relevant difference between the two lives, and so it would not be the case that all else was equal.
Because these examples are bound to be contentious, it is worth emphasizing again the role they play here. For the overall argument against CP to be persuasive, it is not necessary that these particular purported counterexamples be persuasive. For either example, and perhaps for both at once, it may be possible to identify conceptions of well-being and life-goodness that make the example consistent with CP after all. However, the question is not whether any particular example can serve as a counterexample to CP in the context of all theories of well-being and life-goodness. Rather, the question is whether each plausible combination of theories leaves CP susceptible to some carefully constructed counterexample. I think that reflection on the structure of the examples in this section will show that the answer to this latter questions is yes. Section 6 will examine the structure of the examples and the presuppositions about the nature of well-being and life-goodness on which they hinge. But, before that discussion, it makes sense to forestall the objection that these examples could not possibly be counterexamples to CP at all because the comparisons involved do not satisfy the ceteris paribus condition.
5. Is All Else Really Equal?
In a genuine counterexample to CP, the two lives compared must be just alike, except during the delimited period of well-being difference. What is it for the two lives to be just alike in this way? In the first place, put aside metaphysical worries about whether two distinct lives can unfold with, strictly speaking, exact similarity. For a genuine counterexample (or any other comparison to which the principle applies), we need not require that, after a period of difference, miraculously, the two possible lives realign to unfold in lockstep, second-by-second, molecule-for-molecule, or anything like that. That is not the point of saying that, all else equal, increased well-being improves a life. Instead the point is that, besides the difference in well-being, all the other factors that may bear on the goodness of a life (whatever those turn out to be) are the same across the two lives being compared. But, even so, a worry remains.
Regarding Frances’s case, I said that a life in which her success follows a period of low well-being has a structure that suits her. So should we say, then, that the difference in well-being was not the only difference in the two lives we are comparing? After all, in addition to the well-being difference, there was a difference in the organization of Frances’s life. To elaborate slightly, suppose that difference in life organization makes it the case that her later artistic achievements have more authenticity (or some other suitably rich positive value property), and, because of that difference, the life with the period of reduced well-being is better. Now it may appear that we do not have a counterexample to CP after all: In the two lives compared, besides the difference in well-being, there is also a difference in the way the lives are organized and a difference in the authenticity of some of the life achievements, and so it seems not true after all that all else is equal. If that is right, it appears to be a serious objection.
There is indeed an organizational or structural difference in the two lives for Frances, and perhaps also a difference in the authenticity of her achievements. And it is these factors, or factors like these, that explain how one of the lives could be worse despite containing more well-being. But these differences should not be taken to violate the ceteris paribus condition of CP. The reason is that these extra differences—the differences in structure and authenticity—are not extra differences at all. They involve nothing more than the difference in well-being in relation to the background conditions which are constant across both of the possible lives. Note that since there is a difference in one feature (the well-being level) of the lives being compared, there must be differences also in the lives’ relational properties for which that feature is one relatum. To take a trivial example, since Frances’s level of well-being during an early period of her life differs across the two lives being compared, then it must also be that the ratio of her well-being level during that period to her well-being level on her fortieth birthday differs also. This ratio of two well-being levels cannot but change when the well-being level changes. In Frances’s case, the life organization and authenticity are relational properties just like that ratio.
To see why we cannot take a ceteris paribus condition in a principle like CP to bar differences in these sorts of relational properties, consider an analogous case. Consider the (obviously silly) principle that says that, all else equal, the more muscular a man is, the more attractive he is. Call this the muscularity principle. Although there may be some association between muscularity and a man’s attractiveness, the muscularity principle is surely false. And it is false precisely because muscles can be too big to be attractive. But when muscles are too big, this is not because of some absolute standard of bigness. Muscles that are too big for a typical human might be fine for a giant. Rather, muscles are too big when they are disproportionately large, in relation to the other parts of the individual’s body. Hence, we can construct a counterexample to the muscularity principle like this: Consider Man A who is of average build, and compare him to Man B who is just like Man A except that his calf muscles are five times as massive. Man B will not turn out to be more attractive than Man A. And this shows that the muscularity principle is false. If there is a true principle in the vicinity of this one, then the example of the massively calved man shows one way in which the muscularity principle falls short and needs refinement.
This criticism of the muscularity principle is reasonable. But note that the counterexample could not get off the ground if the ceteris paribus clause barred all relational differences between the men being compared. Man B differs from Man A primarily in that his calves are larger, but he also differs in calf-to-thigh ratio. Given the first difference, the second difference is inevitable. And so, if we say differences such as the second one—i.e., differences of proportions—violate the ceteris paribus constraint and make the muscularity principle inapplicable to a particular example, then we have rendered the muscularity principle almost entirely vacuous. But the problem with the muscularity principle is not that it is vacuous; the problem is that it is false.
The same holds of CP. In a comparison of two possible lives to which CP applies, not only will there be a difference in the levels of well-being during some particular period, there will also be differences that are entailed by this primary difference as it stands in relation to the background features that are common to both lives. If we insisted that the ceteris paribus condition disallowed these “additional” relational differences, we would have replaced the substantive principle we were scrutinizing with a vacuous one. Assuming, as we have been, that there is a genuine, substantive issue about the truth of CP, we must not interpret CP such that the relational differences between the two possible lives for Frances violate the ceteris paribus condition.
6. Structure and Presuppositions of Counterexamples to the Standard Assumption
In general, the structure of a counterexample to CP is as follows: The well-being of the person involved is increased during some particular period of time. Then some subsequent events stand in evaluatively significant relationships to the elevated well-being. These newer events relate to the earlier facts in such a way that the person’s life is worse than it would have been had there been no increase in well-being. In light of this general structure, we can note the characteristics of well-being and life-goodness on which the counterexamples depend. First, an assessment of the goodness of a life may take into account the life’s broad, macroscopic features, specifically the relationship between the levels of well-being at particular times and events and states of the person at other times. Second, an assessment of well-being at a time is not sensitive to all of these broad, macroscopic features.
But should we believe that well-being and life-goodness have the characteristics just described? We do indeed have strong reasons for doing so. The case for holding that the goodness of a life depends on the macroscopic, relational features of the life—not just the facts that affect well-being at each moment—is clear-cut. After all, the properties of a life include the relations among the temporally localized features of the life. And evaluation of a life, as with evaluation of any complex entity, may depend on the relations among its localized parts. So no one should simply assume that the macroscopic, relational features were not relevant to the evaluation of a life. The default assumption should be that they are relevant, and it is hard to imagine what sort of argument could justify overturning this assumption.
So, we should accept a fairly unrestrictive view about what can affect the goodness of a life. In contrast, we should accept a fairly restrictive view of the factors on which well-being at a time depends. A defender of CP will likely resist this claim, because a broad and encompassing conception of well-being—one that allowed that well-being levels could be affected by all the factors on which life-goodness depends—might avert the sort of counterexamples I have sketched. However, it turns out that a conception of well-being broad enough to preserve CP cannot play the evaluative role that we expect well-being to play.
To see this, a few distinctions are helpful. First, we can distinguish theories of well-being with regard to diachronic sensitivity. If well-being is diachronically sensitive, a person’s well-being at a time may depend on facts that obtain at other times. According to some theories, well-being is not diachronically sensitive at all. A simple hedonistic theory of well-being is a paradigm case of a view according to which well-being is not diachronically sensitive. Similarly, most other theories that satisfy the so-called Experience Requirement—the requirement that the things that can affect a person’s well-being are things the person experiences—are committed to a diachronically insensitive conception of well-being. In contrast, varieties of desire satisfactionism, provided they do not count only desires that are contemporaneous with their satisfaction, exemplify diachronic sensitivity.
We can further distinguish theories that admit diachronic sensitivity according to whether present well-being registers facts about the past or about the future (or both). If well-being is past-sensitive, then it can be affected by events that have already taken place. And if it is future-sensitive, then it can be affected by events that will happen subsequently. So, for instance, if well-being is past-sensitive, hard work in the past may affect a person’s present well-being when the work pays off. And, if it is future-sensitive, the recognition and praise that the work will later attract may affect her present well-being as she is doing the work.
These distinctions are important for present purposes, because it is to future-sensitivity that an advocate of CP will likely turn in attempting to avert counterexamples. As an initial move, a proponent of CP may appeal to the point with which I began: that CP is a standard assumption in the philosophy of well-being. The proponent might further hold that CP is more credible than evaluations of well-being that are inconsistent with it. And, hence, if a period of apparently elevated well-being in a person’s life does not contribute to the goodness of that life as a whole, then we must conclude that her well-being was not actually elevated during that period after all. Now, for this response to be more than just a resolute insistence on CP, we need some explanation of why, contrary to the initial appearance, the person’s well-being during the period in question was not elevated. If well-being were future-sensitive, then such an explanation would be available: New events can alter the evaluative past, as it were, to make it the case that the original would-be well-being-increasing facts did not increase the person’s well-being after all. Accordingly, it might be that Frances really did not have less well-being during her year of poverty, precisely because the coherence of her life story would end up depending on the relation between her later career and how she was faring that year. And, similarly, perhaps Wade’s well-being was not actually increased by helping the motorist, precisely because a future fact about that motorist stands in a troubling relation to future facts about Wade’s life.
However, without future-sensitivity, this strategy of reading an evaluation of life-goodness back into an evaluation of well-being in accord with CP is not available. Without future-sensitivity, there would be nothing relevant to well-being that could redeem Frances’s misery, and nothing relevant to well-being that could tarnish Wade’s happy encounter. But even if future facts cannot affect present well-being, the ways that future facts stand in relation to present facts may still affect the goodness of a life on the whole. This is, in essence, the reason that assessments of well-being and life-goodness come apart. And, so, if well-being is not future-sensitive, then purported counterexamples to CP, like those I offered, cannot be blocked by revising the operative theory of well-being.
But why should we deny that well-being is future-sensitive? Well, if a person’s well-being at the present depends on future facts, then there is a sense in which her present well-being is not entirely present. And, so, the problem for theories countenancing future-sensitivity is that, in a situation where a person’s level of well-being changes over time, these theories reach unacceptably equivocal conclusions about when that change occurs.
To see the problem, suppose a person had a baseline level of well-being for a while, until her well-being became higher at time t1. And suppose it is the case (as future-sensitivity makes possible) that the elevation of well-being at t1 was due to facts that obtained at t1 together with facts that will hold at some subsequent time tn. Then we must say that although the person’s well-being changed at t1, the change was not fully realized until tn. Not only is this odd on its face, it results in a revisionist interpretation of our ordinary judgments about a person’s well-being at a time. Imagine someone at t2, where t2 falls between t1 and tn, wishes to assess and report the person’s change of well-being that occurred at t1. But not all the facts bearing on the subject’s well-being at t1 have yet been realized by t2. So, when the assessor reports the subject’s well-being, how are we to understand this? Because the truth of her assessment depends on future events, it will be, at best, a projection or forecast of the person’s past well-being.
Now, of course, almost any assessment of well-being will be prone to certain sorts of errors in observation and inference. But the predicament of the assessor at t2 is worse than that. Note just how unusual the assessor’s projection must be. It is not an ordinary projection about the eventual result of an ongoing process; it is a future projection of the past qualities of a person. The assessor might have to say something odd like this, “If everything works out next month like I expect, then she will have been faring well yesterday.” Whatever it is we are concerned about when we assess a person’s well-being at a moment, it is not something so sketchy. Questions about the goodness of a person’s present state are questions it makes sense to ask and answer in the present. “We’ll see in a few months,” is not an appropriate response to “How are you?” or “How was your day?” If someone did respond this way, then she would be most plausibly interpreted as having changed the subject, away from the topic of her present well-being to another topic that mattered to her more, or so I contend.
This case against future-sensitivity in well-being is, so far, little more than an exhibition of some of its counterintuitive consequences. So, of course, it is unlikely to be persuasive to anyone already committed to the possibility that future events may affect present well-being. However, a more thorough argument against future-sensitivity is beyond the ambitions of this article. For those not prepared to give up future-sensitivity, the ultimate conclusion, instead of a refutation of CP, will be conditional: If a person’s present well-being cannot be affected by future events, then CP is false.
Again, the two premises that underwrite counterexamples to CP are, first, that the goodness of a life is sensitive to the macroscopic, relational features of that life, and, second, that well-being is not future-sensitive. Notably, these premises are no stronger than the premises on which some cases against aggregationism have been built. In particular, very similar claims are the basis for Velleman’s argument against aggregationism. Velleman explains, “Intuitively speaking, the reason why well-being isn’t additive is that how a person is faring at a particular moment is a temporally local matter, whereas the welfare of a period in his life depends on the global features of that period” (2000: 58). Though Velleman is distinguishing between well-being during short and long periods, and I locate the relevant distinction as between well-being and life-goodness, the basic observations about the distinctive characteristics of the two kinds of evaluation are essentially the same. But Velleman applies these observations just to the sort of examples other opponents of aggregationism consider. He examines comparisons of possible lives in which, “the one life is better than the other even though they may include equal amounts of momentary well-being” (Velleman 2000: 68). If instead we apply the observations to a different class of examples, as I have been urging, we reach the conclusion that we should reject CP, not just aggregationism.
7. Reconsidering the Connection
Beyond the rejection of aggregationism, the rejection of CP is significant step. As long as we accept CP, it is still reasonable for us to continue thinking of a person’s well-being at a time and the goodness of her life on the whole as two aspects of a single evaluative concern. CP assures us that the two are bound together: Barring interfering practical contingencies, increasing well-being is sure to improve a life, and improving a life is guaranteed not to reduce well-being. But if it is instead the case that a period of increased well-being may make a life worse, then we have two divergent concerns. The possibility of such divergence is sufficient for concluding that we are dealing with two different values.
Though distinct, the two values in question are each species of a person’s own good. A person’s well-being is her good at a time, and we have been assessing lives as good lives for those who live them. Now we still might wish to follow Sumner in distinguishing prudential value (i.e., good for a person) from aesthetic value and perfectionist value. However, if we do, then that will be a further division, neither encompassed by nor encompassing the main distinction at issue in this article. The upshot here is greater disunity in the evaluative domain.
So, once we give up CP, how do we proceed? One predictable impulse is to adopt a mitigated variant of CP. We could hold that to have well-being is simply to be in those states that tend to occur in good lives. Such a view would posit a positive statistical correlation between the things that increase well-being and the things that improve a life, and define well-being in terms of this relationship. Call this the correlation view of well-being. If we accept the correlation view, our investigation of well-being ought to uncover those general regularities that hold between the state of a person during a limited period of time and the goodness of her life as a whole. Such an investigation would be largely empirical, especially if we already have a credible way of evaluating life-goodness.
At least one aspect of the correlation view is almost definitely right. It would be shocking if it turned out that there were not a very strong correlation between increased well-being and life improvement. But the correlation view suffers from essentially the same defect that we find in those views that satisfy CP by privileging the goodness of lives and relegating well-being to a derivative status. Like those views, the correlation view understands well-being at a time fundamentally in terms of the goodness of a life on the whole. The link is merely probabilistic, but, still, according to the correlation view, the significance of well-being is totally subordinate.
The temptation to fall back from views that accept CP to the correlation view is notable. Doubting CP, we might cast around for another way to tether well-being to the goodness of lives. But such a response would be misguided. Our concern for how things are going for a person at the moment is not exhausted by how the moment fits into the larger whole of the person’s life. And vice versa: Our concern for the goodness of our lives is not exhausted by our concern for faring well during all of life’s short periods.
A better approach is to think of well-being as an ingredient in a good life. An analogy to the relationship between sugar and cake is instructive. Although sugar is an essential ingredient in cake, this fact does not give us the full story on the goodness of either sugar or cake. Sugar is good in a lot of ways, and, indeed, it can be good by itself. And, though a cake cannot be good without sugar, a cake that is simply a vehicle for sugar—though enjoyed by some—is not an especially good cake.
The ingredient analogy provides an alternative, though tempered, interpretation of the thought that well-being makes a life good. After all, there is a sense in which sugar makes a cake good. But the ingredient analogy does not involve the mistake of privileging either well-being or life-goodness in such a way that the other is to be understood derivatively. Instead, it posits related but distinct modes of evaluation. This does not yet tell us much about the two evaluative categories and the relationship between them. Most of those questions await substantive accounts of the two values. The ingredient analogy provides, at most, a framework for such accounts. But this, in itself, is significant, because this new framework does not depend on a specious assumption.
I am grateful to Roger Crisp, Dale Dorsey, Donald Hubin, Robert Kraut, Tristram McPherson, and Sigrún Svavarsdóttir for feedback on earlier drafts of this paper. I am especially grateful to Justin D’Arms, who provided extensive criticism and guidance on this paper, at its inception and during its development. I also thank two anonymous referees from Ergo for their careful, nuanced comments. This paper was completed with generous support from the Netherlands Organisation for Scientific Research (NWO), under project number 652.001.003.
- Bigelow, John, John Campbell, and Robert Pargetter (1990). Death and Well-Being. Pacific Philosophical Quarterly, 71(2), 119–140. https://doi.org/10.1111/j.1468-0114.1990.tb00395.x
- Bradley, Ben (2009). Well-Being and Death. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199557967.001.1
- Brännmark, Johan (2001). Good Lives: Parts and Wholes. American Philosophical Quarterly, 38(2), 221–231.
- Darwall, Stephen (2002). Welfare and Rational Care. Princeton University Press. https://doi.org/10.1515/9781400825325
- Dorsey, Dale (2013). Desire-Satisfaction and Welfare as Temporal. Ethical Theory and Moral Practice, 16(1), 151–171. https://doi.org/10.1007/s10677-011-9315-6
- Dorsey, Dale (2015). The Significance of a Life’s Shape. Ethics, 125(2), 303–330. https://doi.org/10.1086/678373
- Feldman, Fred (2004). Pleasure and the Good Life: Concerning the Nature, Varieties and Plausibility of Hedonism. Oxford University Press. https://doi.org/10.1093/019926516X.001.0001
- Griffin, James (1986). Well-Being: Its Meaning, Measurement, and Moral Importance. Oxford University Press.
- Haybron, Daniel (2008). The Pursuit of Unhappiness: The Elusive Psychology of Well-Being. Oxford University Press.
- Kagan, Shelly (1992). The Limits of Well-Being. Social Philosophy and Policy, 9(2), 169–189. https://doi.org/10.1017/S0265052500001461
- Kagan, Shelly (1994). Me and My Life. Proceedings of the Aristotelian Society, 94, 309–324. https://doi.org/10.1093/aristotelian/94.1.309
- Kauppinen, Antti (2012). Meaningfulness and Time. Philosophy and Phenomenological Research, 84(2), 345–377. https://doi.org/10.1111/j.1933-1592.2010.00490.x
- Parfit, Derek (1984). Reasons and Persons. Oxford University Press.
- Railton, Peter (2003). Moral Realism. In Facts, Values, and Norms: Essays Toward a Morality of Consequence (3–42). Cambridge University Press.
- Rosati, Connie (2006). Personal Good. In Mark Timmons and Terence Horgan (Eds.), Metaethics After Moore (107–131). Oxford University Press.
- Rønnow-Rasmussen, Toni (2011). Personal Value. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199603787.001.0001
- Schurz, Gerhard (2002). Ceteris Paribus Laws: Classification and Deconstruction. Erkenntnis, 57(3), 351–372. https://doi.org/10.1023/A:1021582327947
- Schurz, Gerhard (2014). Ceteris Paribus and Ceteris Rectis Laws: Content and Causal Role. Erkenntnis, 79(10), 1801–1817. https://doi.org/10.1007/s10670-014-9643-8
- Sarch, Alexander (2013). Desire Satisfactionism and Time. Utilitas, 25(02), 221–245. https://doi.org/10.1017/S0953820812000477
- Sidgwick, Henry (1907). The Methods of Ethics (7th ed.). Hackett.
- Slote, Michael (1983). Goods and Lives. In Goods and Virtues (9–37). Oxford University Press.
- Sobel, David (1994). Full Information Accounts of Well-Being. Ethics, 104(4), 784–810. https://doi.org/10.1086/293655
- Stocker, Michael (1990). Plural and Conflicting Values. Oxford University Press.
- Sumner, L. Wayne (1996). Welfare, Happiness and Ethics. Oxford University Press.
- Velleman, J. David (2000). Well-Being and Time. In The Possibility of Practical Reason (56–84). Oxford University Press.
- Wolf, Susan (2010). Meaning in Life and Why it Matters. Princeton University Press.
Assumptions along these lines are so pervasive that philosophers considering well-being and life-goodness do not usually treat the two notions separately. Derek Parfit’s Appendix I to Reasons and Persons, which has been, since its publication, the standard starting point for discussions of well-being, introduces his topic by asking, “What would be best for someone, or would be most in this person’s interests, or would make this person’s life go, for him, as well as possible?” (1984: 493). At the beginning of Welfare, Happiness, and Ethics, Wayne Sumner states that welfare “attaches pre-eminently to the lives of individuals,” and he goes on to say “a person’s welfare is more or less the same as her well-being or interest” (1996: 1). In the introduction to Pleasure and the Good Life, Fred Feldman identifies “the Good Life” as “a life that is good in itself for the one who lives it,” and he notes, “Some philosophers speak here of ‘personal welfare’ or ‘well-being’. A good life, in this sense, would be a life that is outstanding in terms of welfare, or well-being” (2004: 9). Additional examples abound.
Talk of levels here might suggest a more robust scale than is plausible. That states of a person can be partially ordered with respect to well-being is enough for purposes of this article. This is consistent with some pairs of states of a person being incommensurable with respect to well-being.
What about a person’s well-being during the long period equal to the span of her whole life? On one reading, this is a question about the goodness of the person’s life on the whole. On another reading, it is asking for some kind of summary of the states of well-being during her life. For an analogy, consider how we might try to describe a city’s weather over a decade. Unless the weather was unchanging, there is no short description that captures the weather for that decade. There are only various kinds of summary: the mean temperature, the mean highs and lows, the temperature range, the total rainfall, etc. Similarly, we could give various summaries of the well-being within a life: the aggregate well-being, the average well-being, etc. Ultimately, one goal of this article is to draw a sharp distinction between such summaries of well-being and assessments of the goodness of a life on the whole.
Ceteris paribus clauses are commonly understood in two different ways. On one interpretation, ceteris paribus clauses exclude abnormal factors. The addition of this sort of ceteris paribus clause makes the resulting generalization a rough one: it holds for the most part, just not when abnormal factors prevent it. On the other hand, a ceteris paribus clause can mean something like, “all else equal”. This is the more literal interpretation, and it is the one that is relevant here. For an elucidating discussion of these two interpretations of ceteris paribus clauses, specifically with regard to the formulation of scientific laws, see Schurz (2002) and (2014).
One striking exception to this—in fact, the only one I am aware of—is Michael Stocker, who does not assume that CP is true, but rather embraces the assumption that it is false. He asserts, “increases in the goodness of the constituent goods of a life need not make the life a better life” (Stocker 1990: 323), but he does not provide an argument.
Perhaps according to an extreme version of the time preference, there could be periods in a person’s life for which the well-being counts not at all toward the overall goodness. But in the cases of most interest for this paper, increased well-being reduces the overall goodness of the life.
Views that assess the goodness of a life primarily in terms of average well-being are subject to broadly Parfitian criticisms as well. See Parfit’s example of the world that ends with Adam and Eve (1984: 401–402).
The term “compositionalism” may be confusing. It is not supposed to suggest that the goodness of a life is simply composed from the goodness of the parts. Rather the view proposes the opposite priority: The good of the parts is to be understood in terms of the good of the whole. Brännmark uses the term “compositionalism” because of what this view says about evaluation of lives: that the composition (i.e., the arrangement) of the constituents matters.
Brännmark does not explain how the goodness of some period in a person’s life is to be determined from the good of the whole. Depending on those details, compositionalism could be consistent with aggregationism. Still, the view would be very different from typical aggregationist views which tell us how to get the value of the whole from the independently determined values of the parts. Brännmark holds that no independent valuation of the parts is possible.
See also Kagan (1992). Kagan’s restriction of the domain of facts relevant to well-being is contentious. We find opposition in both in supporters of additivity, like Feldman (2004: 109–114), and opponents of additivity, like Bigelow, Campbell, and Pargetter (1990: 127–134).
Though Velleman does not deny CP, the sorts of cases and considerations he relies upon to undermine aggregationism would make it natural for him to do so. I return to this point at the end of Section 6.
To make this sort of reflective endorsement persuasive evidence that L2 is better than L1, we should also stipulate that, if in L1 (contrary to what actually happened) Frances had reflected on her early period of comfort, she would not have valued it.
Here I have in mind Sumner’s contention that, “Subject-relativity is a key ingredient in our concept of welfare, the feature which differentiates prudential value from the other modes of value applicable to lives” (1996: 20). That said, the argument here does not require that the good for a person depends on her actual or hypothetical attitudes. The present point is simply that we can construct these cases so that the assessment of life goodness is clearly an assessment of (one dimension of) the person’s own good.
That the correct theory of well-being is not diachronically sensitive is the claim Ben Bradley calls internalism about well-being at a time (2009: 18). This sort of internalism is a consequence of Bradley’s preferred variety of hedonism. But Bradley holds also that internalism is independently compelling, and he invokes it to criticize a broad class of diachronically sensitive views of well-being (2009: 21–28).
Several philosophers recently have endorsed views of well-being that allow future-sensitivity. Antti Kauppinen (2012) argues that present well-being can depend on future events because future events can affect the meaning of present events. Dale Dorsey (2013) takes up the question of what desire satisfaction theories should say about when a person with a satisfied desire is well-off. Dorsey argues that a satisfied desire makes a person well-off at the time she has the desire, even if the facts that satisfy the desire obtain in the future. Alexander Sarch (2013) defends a similar view. Dorsey (2015) argues that future-sensitivity is plausible and is a promising strategy for defending aggregationism.
One might object that the situation here is no different than with act consequentialist theories of moral rightness, and so this is not so implausible after all. According to act consequentialism, the rightness of an action will depend on all the downstream consequences of that action, and so the rightness of the action at the time it took place is only as determinate as those future consequences. Even if we suppose that this is not a serious problem for act consequentialism, the problem for future-sensitivity in well-being is worse. It is more out-of-step with our ordinary thinking to think that the goodness of the present state of a person depends on the future than to think that the moral significance of some event depends on the future. The response, “I should know by tomorrow,” makes more sense in response to, “Did you do make the right choice?” than to “How are you today?”
Among the things that might matter to her are the temporally extended, macroscopic features of her life, e.g., parts of her life narrative, including those features that may bear distinctively on the goodness of her life as a whole.
We may wish to draw further distinctions among types of goodness for a person. Susan Wolf (2010) argues that meaningfulness is a separate type of value for a person, distinct from well-being. As construed by Wolf, meaningfulness would be distinct from both a person’s well-being at a time and the goodness of a person’s life.
There are at least a couple distinct ways to develop the correlation view. Are we looking for states correlated with good lives for any person? Or do we for each person come up with the set of states that are correlated with good possible lives for her? The former seems closer to the way people think about well-being in public policy contexts, and it is the view that offers the relatively straightforward empirical epistemology. The latter view comes with a difficult epistemology since it makes well-being depend on evaluations of possible complete lives for a particular individual.
The sugar and cake analogy is extensive, but it is not perfect. A low degree of well-being over long spans of a person’s life greatly diminishes the overall quality of that life. Similarly, a cake with little sugar is not as scrumptious as it could be. So far, so good. But recall the core (correct) thought of correlation views, that increases in well-being tend to be life improvements. In most cases an increase in well-being will, ceteris paribus, improve a person’s life. Here the analogous culinary correlation is much weaker. Although many cakes will be improved by some additional sugar, many will not. It is much easier to have too much sugar in a cake than it is to have too much well-being in a life.