2. The Molecular Model of Well-Being

One of the most natural ways to express the idea lurking behind the term ‘well-being’ involves the ‘good for X’ construction. If something is good for you, then having it increases your well-being. And yet this ‘good for X’ construction makes it almost irresistible to presume the molecular model of well-being. The molecular model works like this. Certain specific things are good for you: perhaps these things are pleasure, or friendship, or knowledge, or virtue, or the realization of your desires or your values. That’s the first step. The second step is that your well-being is the result of the contributions of the items in the first step (adjusting, of course, for the contribution of the disvalue of the bad things in your life). Well-being is thus a resultant property, a property of a person that arises from the presence of independently good things possessed by that person. How well you are doing is thus some kind of summary of all the particular welfare goods and bads in your life.

This way of assessing a thing’s value often makes sense. An accountant might calculate the value of a company in several different ways. Using the book value method, the accountant would (1) assess the value of each of its many assets, (2) assess the value of each of its liabilities, and finally (3) subtract the value of all the liabilities from the value of all the assets. The company’s value is constituted from the value of its various parts. Or an accountant might instead use the market capitalization method. Then, the value of the company is assessed by multiplying its market share price by the total number of shares. Again, the value of the company is the result of a calculation involving the value of some other thing (here, the market value of a share). There are other methods for assessing the value of a company, and I will not bore you further by describing them. My only point is that assessing a company’s value takes a particular form: its value is the result of a function whose inputs are the value of something else. Sometimes this is the correct way to assess the value of something.

Philosophers usually assess a person’s well-being using the same methodology. Suppose welfare hedonism is true. Suppose Hedwig is doing very well. Hedwig’s well-being is understood to be the result of the metaphysically prior fact that Hedwig is experiencing a lot of pleasure and very little pain, and that each and every little pleasure independently counts for a quantum of well-being. Someone who is experiencing a lot of these quanta is doing well. Things do not work the other way around: we would not say that Hedwig’s pleasure is good for her in virtue of the fact that she has a lot of well-being. Rather, well-being is a resultant property, a property that arises from independently valuable features like pleasure and pain. That is just the structure of how hedonism works. It is a specification of the molecular model.

But almost every other theory of well-being is formally similar to hedonism, likewise presupposing the molecular model. Suppose all of Désirée’s desires are satisfied. A desire-satisfaction theory of well-being entails that she is doing very well, this in virtue of the fact that her desires are satisfied. Her well-being is grounded in the satisfaction of her desires, each of which contributes some quantum of well-being to her overall or total well-being. Well-being is a resultant property, a property that arises from independently valuable features like the satisfaction of desire. That is just the structure of how the desire-satisfaction theory works. It too embodies the molecular model.

The same holds for more objective theories of well-being. An objective list theory contains multiple items, each of which provides some benefit to you. One example of such a theory comes from Fletcher (Reference Fletcher2016). He thinks that the items on the list include achievement, friendship, happiness, pleasure, self-respect, and virtue. And he thinks that if you acquire (say) a virtue, you are thus benefited – your well-being increases. Likewise, he thinks that if you lose a particular friendship, you are harmed – your well-being decreases. Your well-being is grounded in your acquisition of the items on the list. Again, well-being is a resultant property arising from independently valuable items.

The molecular model is thus compatible with many ways of understanding what the prior independently valuable items are. Central features of the molecular model are explicitly described and endorsed by Eden Lin:

Ben Bradley articulates a similar idea. He writes that ‘value atoms’ are the states of affairs that are ‘intrinsically good or bad for us in the most fundamental way, or are basically, non-derivatively, good for us ... The value atoms are what fundamentally and completely determine how well things go for us’ (2009, pp. 5--6).

There are two different claims here worth distinguishing. First, there is the weaker idea that for each particular collection of basic welfare goods and bads, there is only one level of well-being that can correspond to that collection. A person’s well-being can’t change unless at least one of their welfare goods and bads change. Then there is the stronger idea that basic welfare goods and bads fix well-being and is determined by them. This second idea describes an asymmetric dependence relation between welfare goods and well-being. Note that one could accept the first but not the second idea. That is, it’s at least conceptually possible that well-being and welfare goods covary in the way described, without it being true that the former is determined by the latter. This would be so either if both were fixed by some third thing, or if there were some top-down relation between the two, where well-being at least partially fixes the welfare value of particular goods and bads. The molecular model embraces both claims.

3. Moorean Organic Unities

On the molecular model, how well a person is doing is determined by the accrual of atomically valuable items in their life. And a person does better only by gaining some positive welfare atom or losing some negative welfare atom. If well-being is like this, then to calculate a person’s well-being, perhaps we need to figure out what the relevant welfare atoms are and then how to add them together.

It may be objected that there can be organic unities of welfare value. In Principia Ethica, Moore plausibly claimed that the value of a whole must not be assumed to be the same as the sum of the values of its parts (1993 [1903], p. 79). This might also be true for the relation between a person’s overall well-being and the welfare value of the particular things in their life.Footnote ¹ How well you are doing might not be fixed by the sum of the welfare goods and bads you have.

But is Moore’s idea really an alternative to the molecular model or is it instead a specific instance of it? If there is some generally specifiable (but, of course, non-summative) formula that takes welfare atoms as inputs and yields a person’s welfare level as output, then it seems that the spirit of the molecular model is thereby preserved. If a person’s welfare level is some algebraic function of the various welfare atoms in their life, then it seems that their welfare level is still ultimately grounded in these welfare atoms.

For example, suppose that well-being were to positively covary both with virtue and with pleasure: each additional virtue you acquire makes you better off, and each new pleasant experience likewise makes you better off. Even so, the amount by which each is good for you might depend in part upon matters other than the amount of the particular good in question. For example, it is possible that at least some kinds of pleasure benefit a virtuous person more (or less) than they benefit a vicious person. The degree of benefit of the addition of the pleasure, then, would depend upon something other than the nature of the pleasant experience itself. It would also somewhat depend upon other facts, such as whether the person having the pleasant experience is virtuous or vicious. If so, we would not be able to calculate how well off a person is by first figuring out how pleasant their life is, then separately figuring out how virtuous they are, and then finally combining these two results together. Rather, we would also have to consider how pleasure and virtue interact with each other. Well-being would be some function of pleasure and of virtue, taking both as inputs, but it would not be a merely additive function. Still, there might be some way to formulate generally how virtue and pleasure interact to yield overall well-being.Footnote ² So, even though the function from the base goods to the resultant good is not additive, it could at least be expressible generally, and that’s good enough for the molecular model. The molecular model of well-being can thus accommodate Moore’s point about organic unities.

4. Well-Being Enablers

Is there any reason to think that the molecular model is incorrect? In a moment, I will articulate and begin to defend a plausible alternative to it. But first, I just want to raise an initial worry. On the molecular model, your well-being is determined only by the basic welfare values of the welfare atoms in your life. Nothing else plays any role. But it seems plausible that things other than the welfare atoms in your life partially determine how well you are doing. Some things that are not themselves good or bad for you might nevertheless shape how well you are doing.

For example, consider Wayne Sumner’s (Reference Sumner1996) life-satisfaction theory of welfare. On his view, in order to be living well, you have to judge that you are happy. Judging that you are happy is a necessary condition for actually thriving. A person who otherwise has a lot of typically good things going for them, but judges that they are not happy, isn’t actually doing well.

Even so, Sumner does not claim that your judgment that you are happy is itself good for you. This judgment is not itself a positive welfare atom. Rather, it is what Jonathan Dancy (Reference Dancy2004) might call an enabler. Suppose that by Sumner’s lights you are living well. This means that there are various things in your life that are good for you, and shape how well you are doing. But these things wouldn’t make you well off in the absence of judging that you are happy. Your judgment that you are happy enables these things to shape how well off you are. But that means that your welfare is not determined solely by the things that are independently good or bad for you; it is also partially determined by whether you think that you are happy.

Sumner also lists other necessary conditions for living well. He claims that you are living well only if your happiness is authentic. If you are happy merely because you are a victim of ideology, then you are not living well. But even though authenticity plays this enabling role, authenticity is not itself one of the things that is good for you. Authenticity is not a welfare atom. And yet the authenticity of your happiness partially determines how well off you are, at least by Sumner’s lights. If that is correct, then the molecular model of welfare fails to describe the nature of welfare.

I don’t exactly endorse these two features of Sumner’s picture, but I do think they point to something important. Lots of things go into whether we are living well, and only some of them are the welfare value of the particular items in our lives. Our welfare can also partially depend upon more global or structural features of our lives and of the world we inhabit. Not everything that affects how well we are doing is itself good or bad for us.

Now one might object: yes, global and structural features of our lives can shape how well off we are, but they do so only through the welfare value of the particular welfare goods and bads in our lives. Let us grant that you are doing well only if you judge that you are happy. Still, the molecular model can explain why this is so. In the absence of such a judgment, all the particular things that otherwise would have welfare value lack that value. For example, in the presence of judging that you are happy, your relationship with your parents (say) is good for you; but in the absence of judging that you are happy, this relationship isn’t good for you. It turns out that ‘your relationship with your parents’ does not constitute a welfare atom; rather, ‘your relationship with your parents while judging that you are happy’ is a welfare atom. And how well you are doing is determined only by welfare atoms. Or so one might object.

I myself find this manoeuvre ad hoc, but I won’t pursue a specific response. Rather, here I merely intend to weaken the grip the molecular model has on our thinking. There might be some way to reconfigure what counts as a welfare atom in order to hang on to the molecular model, but there is a danger that such adjustments will begin to look objectionably gratuitous, in much the same way an epicyclic model of the heavens overfits the data it seeks to accommodate. Once the molecular model’s grip on us is loosened, we may be better positioned to consider whether a subject’s welfare can partially depend not only on their welfare atoms, and not only on other features of their lives (such as whether they consider themselves to be happy), but even on considerations seemingly external to their lives.

5. Approximation: An Aristotelian Alternative

I can now articulate and defend a picture of well-being at odds with the molecular model, one rooted in certain ideas to be found in Aristotle. Here I won’t be able to wrestle with all the different ways scholars have characterized what Aristotle actually thought, so I will have to settle with formulating a merely Aristotelian picture of welfare, even if it’s not the only possible way to understand what he was up to. I think this picture is both faithful to Aristotle’s views and is independently plausible. Most relevantly, it shows that the molecular model of welfare is far from mandatory or analytic.

Aristotle tells us that well-being (eudaimonia) is the actualization (energeia) of excellence (arete), or virtuous activity (1098a15). Moreover, one can’t act virtuously unless one is virtuous; for example, you can’t act generously unless you are generous. A generous person has a stable character, a character that is reliably actualized in certain actions, motivations, emotions, and pleasures. And since Aristotle holds that a person is generous only if they are practically wise (a phronimos), a generous person also has a disposition to form certain judgments about what they are doing. So, if you aren’t generous, then at best you can do something much like a generous person would do, even if you don’t thereby act generously.

There are important differences between the logic of the concepts of well-being and of virtue. Unlike the Stoics, Aristotle thinks that well-being can come in degrees. For him, the person leading the ideal contemplative life is best off, while the political life offers a merely secondary form of well-being. Even so, they are both ways of living well, but some who are living well are doing better than others who are living well. Further, Aristotle notes that various bodily goods and external goods directly shape how much a genuinely virtuous person can actualize virtue. A virtuous person who is in a coma, or poor, or powerless won’t be able to actualize their virtues to the same degree that a more fortunate virtuous person can. This too shapes how well off virtuous people can be. Again, some people who are living well are doing better than some other people who are living well.

Although well-being comes in degrees, virtue itself does not come in degrees. Either you are virtuous, or you’re not – there is no such thing as being a little bit virtuous. So, if generosity is indeed a virtue, then one person cannot be more generous than another. Rather, generosity lies in a mean, indicating that generosity is itself already ideal, and any deviations from this ideal are thus not (quite) virtuous. While we might colloquially speak of one person being more generous than another, we thereby really mean that one person is less ungenerous than another. So, although virtues don’t come in degrees, people can be closer to or further from possessing virtue.

As a result of this difference between the concept of well-being, which can be degreed, and the concept of virtue, which cannot be degreed, a problem emerges. If virtue doesn’t come in degrees, but well-being in some sense just is the actualization of virtue, then since almost no one is truly virtuous, it looks like no one is truly doing well. For even if you are only a little bit unvirtuous, you can’t act virtuously. Just as only Picasso can paint a Picasso, and only the United States government can issue a U.S. Federal Reserve note, only a virtuous person can actualize virtue. Actualizing something like what Aristotle’s translators call ‘continence’ is not the same thing as actualizing virtue. So, the problem is: how can Aristotle make sense of the idea that ordinarily decent people can be doing at least somewhat well, and better than the vicious are doing, if well-being is restricted to the truly virtuous? We don’t want to be stuck with the highly counterintuitive Stoic picture according to which all unvirtuous people are doing badly (and equally so).

Aristotle’s solution, I believe, lies in the idea that it is still good for you to approximate the activity of the ideal phronimos – the more closely, the better. Even if you aren’t virtuous, and even if none of your activity actualizes virtue, you still might be doing fairly well because your activity closely approximates the actualization of virtue. This suggests a very different model for understanding well-being. Instead of claiming your well-being is determined solely by the welfare value of all the welfare atoms you have, the approximation model of well-being claims that your welfare is determined by how closely you resemble the relevant ideal way to live. Different substantive theories of well-being that embody the approximation model will tout different substantive ideals, but they all hold that how well you are doing depends upon how closely you approximate the ideal way to be living, whatever that is.

It will help to articulate this notion of approximation a bit. Roughly, one thing approximates another when the first closely resembles the second, where the second is in some sense a standard or a paradigm for the first. Consider an archery target, itself an Aristotelian metaphor (1094a23). Hitting the bull’s eye of the target is the ideal, and it yields ten points. But you don’t need to hit the bull’s eye in order to score points. If your arrow misses the bull’s eye but lands in the innermost ring, you still score nine points. And your score decreases the further your arrow lands away from the centre, with arrows landing in the outermost ring counting for only one point. Your score, then, is better the closer you get to the ideal. If you closely approximate the ideal, you will score well. The more closely you approximate the ideal, the better your shot. Frank Robinson quipped ‘Close don’t count in baseball. Close only counts in horseshoes and grenades’.Footnote ³ We can add archery to his list. The value of your archery shot depends upon how closely it approximates an ideal, not upon how many distinct good-making features it embodies or collects. Similarly, then, according to the approximation model, there is a way (or ways) of living that is (are) ideally good for you. Chances are that you are not living your best life, alas. But the more closely you resemble this ideal, the better you are doing. Even if you are not instantiating the ideal, you can still be doing fairly well by merely resembling or approximating what would be best.Footnote ⁴

The idea of approximating well-being shows up in several different parts of Aristotle’s work, and together they form a compelling picture of what it is to live well. Here are five different aspects of Aristotle’s thought that express his commitment to the approximation model of well-being.

First, Lear (Reference Richardson Lear2005) argues that the final book of Aristotle’s Nicomachean Ethics is structured by the idea of approximation: while contemplative activity is best for us, political activity—because it too uses reason—approximates contemplation, and thus also is good for us (although not as good as contemplation itself). She contends that Aristotle thinks that ‘middle-level goods, and in particular morally virtuous actions, are worth choosing for the sake of contemplation because they approximate contemplation’ (p. 88). And, according to Lear, Aristotle holds that ‘if a good is an object of imitation or approximation, then it is the paradigm by which the subordinate good is to be judged’ (p. 83).

It would be a mistake to think that the value of the paradigm or model always stems entirely from its possession of traits that are antecedently good. Echoing some of Nietzsche’s thoughts about the nature of master morality, Lear writes:

A paradigm is not a paradigm simply because it embodies good-making features, such that whatever else embodies these features would have been likewise valuable regardless. Rather, a paradigm makes (some of) its features valuable, and thus makes them worth having.

Second, more evidence for Aristotle’s fondness for the approximation model comes from his discussion of the vices. When discussing the relative badness of two vices, each of which is contrary to some virtue, he writes:

Although it is best for you to act virtuously (say, act courageously), it is better to act viciously in one way (act rashly) than to act viciously in the other ‘more erroneous’ way (act cowardly) (1109a10ff.). Presumably, the one who acts rashly is doing better than the one who acts cowardly, because acting rashly more closely resembles acting courageously. Acting cowardly is ‘more contrary’ to acting courageously than acting rashly is. It’s not as bad for you to act viciously in one way rather than another, because one vice more closely approximates the relevant virtue.

Third, Aristotle thought that ‘continent’ people are doing better than vicious people. A continent person acts and judges much like a virtuous person does, but they are not pleased and even pained by so acting. For example, imagine a person who successfully struggles to eat a fairly (although not perfectly) well-balanced diet. A continent person approximates the ideally temperate person more closely than the vicious person does. The continent person’s behaviour thus closely resembles the behaviour of the virtuous person. And their judgements about what’s good resemble the judgements of the virtuous person. Their emotions and feelings, however, do not closely resemble those of the virtuous sage. Even so, their life overall is much more similar to the life of the virtuous person than the life of the vicious person is. The approximation model of well-being implies that due to their resemblance to the virtuous person, the continent person is doing better than the vicious person is, even though the continent person is not at all virtuous.

One might object, however, that while the continent person is doing better than the temperate person, this is not due to approximation, but entirely due to the welfare value of the continent person’s behaviour. After all, we said the continent person basically does what the virtuous person does. And the virtuous person’s behaviour must be good for them, since their behaviour is the manifestation of their virtue. So, if the continent person’s behaviour is the same as the virtuous person’s behaviour, then the continent person’s behaviour is likewise good for them, but this has nothing to do with approximation. Rather the behaviour of both the virtuous and the continent person is an intrinsic welfare good and perhaps even a welfare atom. Of course, the virtuous person is doing better overall than the continent person, but this is solely because of the remaining differences between the two.

But this objection rests upon a misunderstanding. The Aristotelian view does not hold that, say, eating a well-balanced diet is good for you. It says that virtuous activity – or, more technically, the actualization of virtue – is good for you. Any activity that isn’t the actualization of virtue is not directly good for you, at least not on those grounds. Acting merely in accordance with virtue is instead how virtue is eventually produced, but it is not how well-being is directly constituted. And since a continent person is not virtuous, nothing that they do is itself an actualization of virtue, including eating a well-balanced diet. So, if the continent person is doing better than the vicious person, then it’s for some other reason.

Fourth, and relatedly, the range of possible character-types with respect to a particular topic (e.g., the pleasures of eating and drinking) is much wider than the few options we have just seen, and, to make sense of this entire range, we need something like the approximation model. The virtues of character famously lie between two vices – in the case of temperance, the opposing vices are intemperance and insensitivity (anaisthesia, or lack of sensation). Now, both the standard continent person and the standard incontinent person lie toward the vice of intemperance, not insensitivity, which is rather rare. They both are excessively tempted by bodily pleasure, even though they think they should refrain from overindulging.

But there are still more possible character traits with respect to the bodily pleasures. Corresponding to the continent person, we can imagine someone whose bodily appetites are deficient, but they generally overcome their reluctance to eat, drink, have sex, etc. This would be someone who routinely lacks much of an appetite, but eats a good diet anyway, this because they know doing so is wise. Perhaps they often consciously remind themselves to attend to their bodily needs. I don’t know of a standard term for such a person, so I will just call this person the laboured eater. Likewise, corresponding to the incontinent person, we can imagine someone whose bodily appetites are deficient, but they do not generally overcome their disinclination to eat or drink, despite knowing that they should. They behave much like the insensitive person, but they regret doing so. I will call them the conflicted ascetic. Unlike someone who has the vice of insensitivity, then, these two types of characters do realize that they should enjoy the pleasures of the body more than they do. The full spectrum of possible states of character having to do with bodily pleasures, then, looks like:

Now, only the temperate person can be actualizing temperance, and so is doing better than any of the other six characters. But both the continent person and the laboured eater closely approximate the temperate person, albeit from opposite directions. Now if the continent person and the laboured eater are about equally well off, and are better off than the remaining four characters, the approximation model cleanly explains why: only continence and laboured eating closely approximate temperance. More generally, the approximation model cleanly explains why the two vicious people aren’t doing as well as the incontinent person and the conflicted ascetic are, who aren’t doing as well as the continent person and the laboured eater are, who aren’t doing as well as the temperate person is.Footnote ⁵ It is difficult to see how the molecular model could explain this range as simply as the approximation model can.

Fifth and finally, there are conceptual reasons that well-being, for Aristotle, is not to be understood as the accumulation of welfare goods. For he thinks we will know that we have fully identified what it is to be living well when we articulate a way to live that cannot be improved in any way, when our nominee for the nature of well-being is ‘complete’ (1097a24--1097a35). Aristotle treats well-being in much the same way we usually treat health. Being healthy isn’t merely the absence of disease and disorder; if so, a corpse or a rock might be healthy. But one cannot indefinitely become more healthy by the addition of yet another ‘health good’ or removal of a ‘health bad’. There are various ways my body could be different such that I would then be healthier. Maybe that’s true for everyone. But after enough changes, one would become the model of health. And then one could not be any healthier. The notion of perfect health makes more sense than the notion of infinite health.

Human welfare, according to the approximation model, is like that. No actual person is doing as well as one could. Various changes would make your life go better than it now is. But once enough of these changes have been made – and it would be a lot of them – you would be perfectly well off. Your welfare would be ‘complete’. This is opposed to the molecular model of well-being, which says that you can always make a person’s condition better by adding more welfare goods. The molecular model thus excludes the idea that a person’s life could ever be complete.

One might object: no matter how complete one’s welfare seemingly is, wouldn’t the addition of yet another, say, pleasure make one even better off? Couldn’t the Aristotelian sage do even better than they are doing? Now one can find the notion of ‘infinite happiness’ in Aquinas and other writers already prone to ascribe various infinite qualities to a divine being. Perhaps for any finite level of welfare, a god could be doing even better than that. But human beings—our subject here—have limited bandwidth. Our would-be virtuous activities are hindered by pleasures arising from other sources: Aristotle notes that people fond of the flute are incapable of attending to arguments if they overhear a flute, since they enjoy flute-playing more than the arguing (1175b23ff.). We are not built to do one thing well and benefit from the addition of things that would indeed be good for us under other conditions. Perhaps unlike for the gods, human welfare has a limit; once a person is doing as well as that, they could not be doing any better (while remaining a human being). Aristotle, then, articulates a vision of human welfare according to which there is a highest good (for us), one that cannot be surpassed.Footnote ⁶

Of course, Aristotle’s view of well-being is just a particular illustration of the approximation model, and, while I do think it is very attractive, I have not here argued that it is superior to all its rivals. But I hope we now see what a minimally plausible substantive view of well-being that embodies the approximation model looks like.

6. Kinds of Approximation

Of course, Aristotle’s view of well-being may have many specific flaws it would be better to leave behind. Even so, the approximation model of well-being is worth preserving, perhaps married to a more plausible substantive view about what makes for a good life. On the approximation model, something is good for you just when it moves you closer to the ideal. Joy, for example, is often good for you, just because being more joyful than you now are would more closely resemble the ideal life. Something is a welfare good in virtue of the role it plays in contributing to the way your life as a (synchronic) whole resembles the life of the paragon – whether that’s the Aristotelian phronimos or something else entirely.

When comparing the approximation model to the molecular model, the results often come out the same way, even when the reasons for the results differ. For example, if the ideally well-off person is experiencing a lot of pleasure, then an ordinary person who experiences a new additional pleasure will probably be better off, because then they will more closely approximate the ideal. And this is also what standard hedonists conclude: they too think that experiencing a new additional pleasure improves the quality of your life. But their reasons are different. On the hedonist view, the person benefits because the new pleasure is independently valuable: it’s one more mark on the well-being tally. On the approximation model, however, the person benefits because they more closely approximate the ideal.

The approximation model is compatible with the thought that not everyone shares the same ideal. Perhaps the kind of life that is ideal for you is different from the one that is ideal for me. We vary somewhat with respect to our talents, experiences, dispositions, and interpersonal relations, and plausibly such things can affect what would constitute the ideal life. For now, however, this is an open question, with different substantive theories embodying the model answering that question differently. But if there are different ideals for different people, then you are doing better the more closely you approximate the ideal for you, whereas I am doing better the more closely I approximate the ideal for me. Although the approximation model makes use of the notion of an ideal or a paradigm, there need not be a one-size-fits-all approach. It is even possible that there are different ideals for the same person at different times in their life. There is nothing mandating that the ideal cannot change over time. The fact that there is an ideal or a paradigm that fixes what it is for you to now live well does not entail that what is ideal cannot change.

Finally, there is nothing mandating that there be only one ideal or paradigm, each for one person at one time. How well off you are now might depend upon how closely you approximate the closest of several relevant ideals. Suppose there are two distinct ideal lives for you. Perhaps the contemplative life and the political life are equally good lives; if so, how well off you are may just depend upon how closely you approximate whichever ideal is in fact closer to your current life (you are already a philosophy professor/student, so…).

The notion of an ideal and the associated notion of approximation need not be bound to any narrow (e.g., Aristotelian) understanding of what is good for a person. Perhaps there are many possible ways to flourish equally well. Different substantive embodiments of the approximation model will need to take a stand on these matters, but the approximation model itself does not force a choice.

7. The Second Best

I want to consider a natural objection to the approximation model of well-being. This objection can be illustrated using Aristotle’s own views. Suppose the ideally well-off human being has various characteristics: they are courageous, temperate, friendly, witty, generous, etc., and they are well-off because they actualize these virtues: they act and feel and think courageously, temperately, and so on. Few if any fully instantiate the ideal. Suppose that Luisa has most of these characteristics, but she is not temperate. She struggles to eat and drink only a modest amount. And let’s just stipulate there is no way for her to improve in this respect. Now, is she best off if she spends a lot of time with her friends in the way the ideally well-off person is? Even though she is not temperate, should she still act virtuously in every other way? Perhaps not! For if she were to socialize with others in the way a friendly person does, she might very well overindulge even more frequently. That is, if she were otherwise-ideally friendly, she would act even more intemperately than she already is. Given that she is less than ideal in one way (lacking temperance), she may be best off if she is less than ideal in other ways (keeping to herself a lot, avoiding throwing magnificent parties, etc.) Isn’t this a problem for the approximation model, which seems to recommend that she fully instantiate friendliness (or come as close as she can) if possible?

This objection is based upon what’s known as the theory of the second-best, a concept better known in economics, law, and political theory (Lipsey and Lancaster, Reference Lipsey and Lancaster1956). This theory concerns itself with what kind of system is best given certain constraints. When one characteristic (or ‘variable’) of the system is constrained so that it cannot be ideal, the system might be best overall if some of its other characteristics likewise deviate from the ideal. If one market failure is inevitable (an unrepealable tax on the purchase of cars), introducing a second market failure (a tax on the purchase of trucks) may enhance overall economic efficiency. If one form of political inequality is inevitable (racially unequal school systems), introducing a second form of political inequality (affirmative action) may enhance overall justice. Now, the approximation model of well-being says that the more closely you approximate the ideal, the better off you are. But if you are inevitably less than ideal in one respect, couldn’t it be true that the second-best way to live would be to deviate even more from the ideal? Shouldn’t Luisa socialize less than the ideal person would?

The answer isn’t obvious. Compare two different ways Luisa can be.

1. 1. Luisa acts continently (successfully struggles to eat and drink a modest amount) and acts not friendlily but somewhat unfriendlily (stays at home a lot, avoids parties, etc.) She isn’t an unfriendly person; she just does not actualize her friendly character.

2. 2. Luisa acts incontinently (eats and drinks more than she should) and acts out of friendliness.

In which scenario does Luisa more closely approximate the ideal life? In the first scenario, she does not act virtuously. She neither acts temperately nor does she often act in a friendly manner. In the second scenario, she acts virtuously in one way, but quite badly in a second way.Footnote ⁷ But what does this really amount to? On the one hand, the ‘error’ that she makes in the second scenario (incontinent behaviour) seems like a larger deviation from the ideal than either of the deviations in the first scenario: it seems less ideal to act incontinently than either to act continently or to live semi-reclusively. But is acting incontinently a larger deviation than the combination of acting continently + semi-reclusively? Things could go either way! It can be difficult to discern which of two very different ways of living more closely approximate an ideal way to live.

I think that there is room to say the following: Luisa is better off in the first scenario, because overall the first scenario more closely approximates the ideal way to live than the second scenario does. A defender of the approximation model is not forced to say that Luisa is better off if she continues to actualize her friendliness, not if doing so must involve large deviations from the ideal with respect to the activities of eating and drinking. That is, Luisa might more closely approximate the ideal way to live overall by erring a little with respect to both friendliness and temperance. If some alternative model of well-being were to hold that the welfare value of a person’s life is based upon how it scores across multiple dimensions,Footnote ⁸ and that a person is well off just to the extent that they approximate the ideal in each dimension (somehow summing the scores generated by each), then the general theory of the second best might pose a genuine problem for such a model. But the approximation model itself, as I see it, does not tell us to approximate across multiple separate dimensions; it is instead concerned with overall approximation. This is ultimately compatible with the general theory of the second-best.

Fittingly, the notion of approximation here is not an exact notion. We cannot always determine which of two suboptimal lives more closely approximates some ideal. But it is a fantasy to hold that philosophy could ever equip one with the tools to cleanly settle each such comparison. Bentham aspired to affix quantities to every evaluationally relevant feature of what matters, quantities that could then be summed up and compared. But the moral sciences aren’t that kind of science. While the notion of approximation can give us some guidance about how well people are doing, it doesn’t obviate the need for good perception and judgment.