If movies are to be believed, villains, by and large, do not thrive—morally bad people lead lives that are in some way marred and unhappy, precisely because they’re morally bad. The idea that virtue is necessary if not sufficient for happiness is, of course, one of the most persistent themes in European ethical thought from Socrates to Hume, at least. One surprising thing about this persistence is that it survived major changes in conceptions of both virtue and happiness itself. With the revival of ancient virtue ethics in the latter half of the twentieth century, the thought that happiness requires virtue has found at least tentative defenders even in our own day, such as Philippa Foot (2000), Julia Annas (2011), and Daniel Russell (2012). According to at least one study, it also resonates with many ordinary people’s conception of happiness (Philips, Nyholm, and Liao, 2014).

Here’s one way we can put the idea while taking into account the fact that both virtue and happiness are a matter of degree:

Virtue Necessity
Possessing the full set of ethical virtues of character to a high degree is necessary for a high level of happiness.

Given the way we tend to understand happiness these days, it’s easy to dismiss Virtue Necessity as some kind of wishful thinking—wouldn’t it be great if being good would be in everyone’s self-interest? Nevertheless, I propose to take it seriously enough to examine the case that contemporary Aristotelians make for it. I will argue, though, that they end up stretching the notion of happiness so far that it’s no longer recognizable as such. In consequence, there can be at best a causal connection between virtue and happiness. But perhaps this causal connection isn’t quite a matter of chance, given the best way to understand of each. Maybe the following kind of interesting generalization still holds:

Virtue Conduciveness
Other things being equal, possessing a full set of ethical virtues of character to a high degree and manifesting them in action increases happiness relative to acting from less complete or absent virtue.

In this paper, I’m going to argue that there are good a priori reasons to think that Virtue Conduciveness is true, at least for most people, and also that we can specify the ceteris paribus clause enough to say that it’s not trivially true.

I’ll make the case as follows. In the first section, I’ll introduce the notion of virtue as I understand it. The conception I’ll work with will be broadly Aristotelian: a virtue is a multitrack disposition to recognize reasons pertaining to a particular good and to respond to those reasons correctly in terms of emotion, deliberation, and action. It is crucial for my argument that virtue involve both (a) emotional concern (only) for what is genuinely valuable and (b) skill at rationally pursuing related aims. Further, different virtues form a mutually supportive system. Second, as I said, I’ll reject recent neoclassical defences of eudaimonistic conceptions of happiness. Their basic move is to start from a formal conception of our final end, which they identify with happiness, and argue that what best fulfills this happiness role is (or can be) a life of virtuous activity. I argue that no conception of happiness that meets reasonable minimal constraints is suitable to play the role of a unique final end in practical reasoning, which we could instead identify with something like “the good life” (Haybron) or the most choiceworthy life.

Third, I’ll introduce my version of a sentimentalist account of happiness, according to which we are happy, roughly, to the extent that the phenomenal character of our experiences of the things that are subjectively important to us is predominantly positive in valence. I argue that this view of happiness as positive significance-adjusted affective balance meets more of the reasonable criteria for an account of the nature of happiness than Dan Haybron’s (2008) industry-standard view. Assuming sentimentalism about happiness, I’ll argue that, given the nature of virtue and common human concerns, we have good reason to think that Virtue Conduciveness is true. Here’s a rough picture of the relevant mechanism:

The idea is that insofar as we’re virtuous we’ll skillfully pursue realistic, genuinely valuable aims—not only prospective ones of realizing future-directed goals, but also reflexive ones of performing worthwhile activities well—that take into account the interests and rights of people around us. At the same time, we lack desires for what doesn’t really matter, like having a newer car than our neighbour. Given these facts about our cares and practical reasoning, it’s likely that we’re successful and are in good standing with people around us, which in turns fosters positive affect, while at the same time many known sources of negative affect are absent. So, virtuous outlook and activity are apt to indirectly bring about happiness in affective terms. In the final section, I’ll compare and contrast this view with Mauro Rossi and Christine Tappolet’s (2016; forthcoming) recent sentimentalist defence of noncontingent links among virtue, happiness, and well-being that bears some important similarities to the view defended here.

To begin with, let’s briefly consider the nature of virtue and, more specifically, ethical virtues of character, which contrast with intellectual virtues like theoretical reasoning Let’s start with a tentative list of some central putative virtues: justice, honesty, trustworthiness, generosity, kindness, temperance, tenacity, and courage. These are all morally admirable traits when manifest to a great degree. (It may well be that traits like wit or innovativeness are virtues in a broader sense, as philosophers like Hume like to think—they are admirable in some nonmoral way.) But what is it to have such a trait? Suppose that Bernie is honest. Everyone agrees that it follows that he tends to tell the truth even when it’s inconvenient to him and that he can be trusted not to take what’s not his. But such behaviour is neither sufficient nor necessary for honesty, since people who lack the virtue might do the same things, and, in appropriate circumstances, even honest people may arguably lie. What is distinctive about honest people is rather to be found in their recognition of reasons to be truthful and their psychological response to them.

According to the standard Aristotelian view, a full virtue is a complex disposition to have integrated cognitive, conative, and affective responses that are situationally appropriate. Bernie’s honesty involves, first, the ability or skill to tell when telling the truth is the right thing to do. This skill requires a grasp of the point of telling the truth and, according to many, the ability to articulate reasons why something is the right thing to do (Annas, 2011, p. 19). As people like Alison Hills (2016) point out, it involves understanding why one should tell the truth rather than just knowing what one should do. Having to account for our choices by giving reasons forces us to reflect and go beyond what we learn from the behaviour of others around us with respect to, say, truth telling (Annas, 2011, p. 54).

In addition to this cognitive aspect, virtue has a conative part: Bernie has a desire to tell the truth when he considers it appropriate. Importantly, virtue also consists in affective dispositions, since it is a state or disposition (hexis) to feel pleasure, pain, anger, pity, and other emotions “at the right time, about the right things, towards the right people, for the right end, and in the right way” (NE 1106b: 30). So Bernie takes some pleasure in doing the right thing and would be pained and ashamed if he happened to mislead someone. Similarly, a generous person not only gives the right amount—neither excessively nor deficiently—but also does so without displeasure or resentment. Such feelings would manifest disproportionate concern for possessing the good thing in question instead of giving it to someone who needs it more. This makes for a difference between the virtuous and the merely continent or self-controlled, who do the right thing as a result of overcoming internal opposition. Fully honest people aren’t tempted to lie, so they don’t need to exercise self-control in order to be truthful (NE 1152a: 135). (To be sure, the picture is not so straightforward with virtues like temperance and courage, which are precisely virtues of self-control.) If an honest person is internally conflicted, for example, it’s because there are genuinely conflicting reasons for and against telling the truth, so there may be a moral residue even if one does what’s all-things-considered best.

The idea of the virtuous not being tempted by the bad ties in with one of the emphases of the Stoic tradition: the sage is distinguished by not caring about worldly things. The Stoics tended to take this to an extreme, holding that only virtue is good and vice bad, while everything else is strictly speaking indifferent (even if things like nourishment are “preferred indifferents”). I’m not convinced by the reasons they have offered for this extreme revisionism about value (see, e.g., Brennan, 2005, whose interpretation emphasizes that only virtuous activity actualizes human nature, which in turn determines our good). Nor am I even convinced of the coherence of the view (what makes justice a virtue if it doesn’t matter who has what?). But it is a real insight that an important part of being virtuous is not being too concerned with what doesn’t really matter.

According to the Aristotelian picture, affective dispositions are thus an important aspect of virtue. This is not in the first instance because of their motivational role: according to many, a correct perception of a reason to act is by itself sufficient to motivate the virtuous person to act. John McDowell, for example, argued that one can’t really conceive of a situation in virtue terms without also being motivated to act: “The relevant conceptions are not so much as possessed except by those whose wills are influenced appropriately” (McDowell, 1978, p. 87). Arguably, this point generalizes to appropriate emotional responses: ceteris paribus, you don’t really conceive of someone’s action as being despicable if you’re not at least disposed to despise or otherwise negatively react to it. Our emotions constitute, in part, our subjective perspective on the world; if we lack appropriate emotional responses when something of real value is at stake, we’re not sufficiently responsive to the value. As Thomas Hurka (2000) says, it is intrinsically good to love what is intrinsically good for itself and to hate what is intrinsically bad. While a part of what it is to value truth, for example, is to perceive a strong reason against lying for your own advantage, there is more to valuing truth than perceiving reasons. Your valuing truth is also manifest in your positive emotional response to truth telling for its own sake and in your disposition to be ashamed about lying. Someone who lacks such dispositions has a split or incomplete evaluative perspective, even if that person does the right thing. So, against Martha Nussbaum’s (2016) criticism of anger, I’ve argued that if we value justice, we must be angry about injustice when we stand in a suitable relation to it (Kauppinen, 2018). And since we do value justice if we’re virtuous, we are sometimes angry if we’re virtuous. For similar reasons, the courageous person who values the security of innocents takes pleasure in actions that promote security, even if these involve risk to herself or himself. Such person’s affective and cognitive perspectives are unified.

Having the right sort of emotions and desires means that one has the right sort of ends, but that doesn’t suffice for virtue yet—you still need to figure out what to actually do in your particular situation to promote the good. For Aristotle himself, virtues of character can be had without practical wisdom (phronesis), while practical wisdom can’t be had without virtues of character (NE 1144b: 118). Practical wisdom is thus something like skillful practical reasoning in pursuit of situationally appropriate aims, distinct from mere cleverness or intelligence in pursuit of any goal whatever. Consider generosity, for example. As debates around effective altruism have shown, it’s not a trivial question how to best give to others—should I, for example, work within the capitalist system to gain wealth that I can share with others, or work to change the system? It evidently takes practical reasoning to get at what to give (money, time, advice, attention, etc.), to whom, when, and how.

The debate about effective altruism also illustrates how virtues interact with each other: the values to which justice and generosity are responsive may be in tension with each other. Defenders of reciprocity of virtue from Aristotle to McDowell (1979) hold that virtues must come as a package, since in practical reasoning we need to take into account all sorts of ethically relevant considerations—it takes more than generosity to give in the right way, or, alternatively put, you can’t have the full virtue of generosity without other virtues. We do, of course, sometimes say that someone is honest without being generous, or that loyalty can conflict with justice. It’s even more natural to think that people can have executive virtues like courage and tenacity that bear on how they pursue whatever ends they may have. Defenders of reciprocity will thus maintain that the kind of honesty you can have without generosity is defective, even if most of the time it leads you to do the right thing, and that the kind of loyalty or courage you can have without justice is a simulacrum of virtue. I formulated Virtue Necessity and Virtue Conduciveness above in terms of a “full set of virtues,” but, given the mutual dependence among virtues, it’s not unreasonable to talk about “virtue” in the singular either.

Supposing that something like the above conception is correct, which character traits are virtues? As Philippa Foot (1978) emphasized, virtues of character can be seen as corrective of common human moral and prudential defects. That’s why they’re praiseworthy. For example, we tend to give too much weight to our own safety in comparison to other goods, and courage corrects for this defect. To say this is of course not to explain how much weight we should give to our own safety and why. That’s a matter for first-order ethical theory or perhaps more broadly a theory of the good life, all things considered (Haybron, Manuscript). I want to remain neutral on these substantive views, though my own take is that theories of value and right action are explanatorily prior to any account of what makes a character trait a virtue (cf. Hurka, 2000)—embracing the importance of the notion of virtue doesn’t entail accepting any form of virtue ethics, which would reverse the explanatory priority. But focusing on human defects is useful here, since it’s easy enough for people holding different first-order views to agree on our characteristic shortcomings: we tend to be too selfish and short-sighted, and to be biased in favour of our own kind, consumed by desire for approval, and tempted by pleasure or deterred by apparent danger. Modern (social) psychology has added considerably to the list of human defects: we’re influenced by framing, priming, and the status quo; seduced by choice architecture, and subtly biased against outgroups and those who are statistically abnormal. We are also sensitive to social comparison and surprisingly conformist, to mention just a few things (see, e.g., Thaler and Sunstein (2008) and Kahneman (2011) for influential overviews of these studies). I don’t think this means that there is no such thing as Aristotelian virtue (pace Doris, 2002). It’s just that it’s even harder to be fully virtuous than we may have thought.

For the purpose of discussing Virtue Necessity, I’ll simply take for granted the common-sense first-order view that the right thing to do sometimes involves putting ourselves in harm’s way, letting go of something we really want, risking our health, and even giving priority to the needs of a complete stranger over those of a friend or ourselves. This gives rise to a simple and familiar challenge: how can we be sure that virtue will make us happy (or happier) when it appears to common sense that it may on occasion lead to utter misery?

Given that Virtue Necessity is counterintuitive, there’s a temptation to qualify it by saying that virtue is necessary for true happiness, where “true” is meant to convey that the ordinary understanding is mistaken or at least ambiguous. Here is how the argument in these terms might go:

• The Eudaimonist Argument for Virtue Necessity

  1. True happiness consists in virtuous activity with which one is subjectively satisfied.

  2. Virtue is necessary for engaging in virtuous activity with which one is subjectively satisfied.

  3. So, virtue is necessary for true happiness.

The first premise evidently goes against the way most of us think of happiness these days, though eudaimonists like to point to some contemporary usages of “happiness” that require meeting some objective standards in addition to feeling subjectively satisfied (Kraut, 1979). Either way, eudaimonists are right to insist that this isn’t decisive. It’s certainly possible that most people are indeed mistaken about the nature of happiness. At the same time, there’s a risk of changing the topic if supposedly “true” happiness is insufficiently linked to our actual shared concept of happiness. To ensure this doesn’t happen, we need to identify some key conceptual connections between happiness and other things—in Michael Smith’s (1994) terms, some platitudes about happiness that had better come out true. This will give us a job description for happiness against which we can measure competing accounts without relying only on intuitions about who’s happy and who isn’t.

The job description that eudaimonists focus on highlights the putative deliberative role of happiness as the final end. Here is a sketch of the argument:

• The Final End Argument

  1. Deliberation requires a unique final end that is complete and self-sufficient.

  2. Of the ends that serve as starting points for deliberation, only happiness is complete and self-sufficient, provided that its initially indeterminate content is suitably specified—“happiness” is a placeholder for whatever the final end consists in.

  3. In order to serve as a complete and self-sufficient end, happiness must consist in virtuous activity with which one is subjectively satisfied.

  4. So, happiness consists in virtuous activity.

While this type of argument derives from Aristotle, I’m going to focus on contemporary formulations by Julia Annas and Daniel Russell. Let’s start by examining the first premise about the nature of deliberation. Julia Annas (1993; 2011) emphasizes that for ancient ethical theories, the key starting point is reflection on our lives as a whole: we stand back from our ongoing concerns and ask ourselves what the best way for us to go on living would be. As rational agents, we do things for the sake of some end, which we regard as good. These ends form a hierarchy, in which realizing one end is either a means to or a way of realizing another. Perhaps we exercise for the sake of health or study for the sake of getting a job or go to the movies for the sake of having fun. Here we take maintaining health, getting a job, and having fun to be higher-level ends for the sake of which we do many other things. But why be healthy, have a job, and have fun? And what to do if these ends conflict? According to Aristotelians, to answer such questions rationally, we need a further end for the sake of which we seek both to be healthy and have fun, say, and we need to consider whether it is better promoted or realized by the health-promoting or the fun-constituting option. If there is more than one end at this level of the hierarchy, we keep going in the same way until we reach the final end or good for the sake of which other goods are pursued, and which we don’t pursue for the sake of a yet further end. This will then be our highest good (NE 1094a).

As Aristotle says, this final end has certain formal characteristics: it’s complete, in the sense of being something pursued for its own sake and not also for the sake of something else, and self-sufficient, in the sense of needing nothing to be added to it for a good life, at least provided that certain background conditions, like sufficient health and wealth, are in place (NE 1097a-b: 10–11). Being satisfied with our lives is part of what’s required for self-sufficiency, since otherwise there would be something that could be added to the final good to make it better.

The next step is the claim that happiness is the final good for the sake of which other goods are pursued and which is itself never pursued for anything else. Aristotle himself says that everyone agrees that happiness (or eudaimonia) fits the bill (NE 1095a: 5), and Annas says that “once we see the need for a final end, an end we pursue in some way in every action, there is no better specification than happiness that we can come up with for it” (1993, p. 332). In a way, this doesn’t help much, since the nature of happiness is itself in dispute. Indeed, Annas maintains that “happiness” is in fact largely a placeholder for whatever has the formal characteristics of a final end: “Happiness in ancient theories is given its sense by the role it plays; and the most important role it plays is that of an obvious, but thin, specification of the final good” (1993, p. 46). So the hypothesis, if you will, is that happiness is whatever best answers the job description of a final good for beings like us who are asking the question. Methodologically, we might say that this is the conceptual role that we hold fixed in figuring out what happiness consists in.

Third, for happiness to be the organizing and self-sufficient goal for us, it can’t be a state like pleasure, because then it couldn’t serve as the focus of large-scale deliberation about our lives as a whole. Here Aristotelians draw on the endoxa, the reputable opinions of all, most, or the wise, to argue that something like pleasure can’t be the focus of the best kind of life for a human being, because pursuing it as the highest good would result in “a life fit only for cattle” (NE 1095b: 7). Given our nature, our final end must consist in rational activity of some sort. Eudaimonists take such activity to be virtuous activity, which involves the use of practical wisdom as well as virtues of character, possibly because virtue is just the name for the kind of excellence that helps something perform its function or distinctive job (ergon), and our job is to engage in rational activity (NE 1097b). The conclusion then follows: being our final end, happiness consists, at least in part, in virtuous activity. This then feeds into the main eudaimonist argument for Virtue Necessity.

What should we make of this picture? I think there are several hurdles in the way of identifying happiness as whatever plays the deliberative role of final end, even if we allow that our ordinary notion of happiness is “weak and flexible,” as Annas puts it (1993, p. 129), and we are willing to revise it. At the very least, our happiness is still something that is good for us. More precisely, my happiness is nonderivatively good for me, so anything that is constitutive of it must be nonderivatively good for me. But is it really the case that everything we have reason to pursue for its own sake and not for the sake of anything else, and for the sake of which we pursue other things, must be good for us? Doesn’t virtue, in particular, require pursuing things like justice for the sake of others, and perhaps some impersonal values, not for the sake of ourselves?

This is, of course, an ancient worry. Russell puts it very nicely: “In the one direction, we look for a good that could make sense of all the ends we have reason to pursue; but surely, the worry goes, many of these ends cannot plausibly be understood as being for the sake of one’s own good life” (2012, p. 24). Unsurprisingly, then, eudaimonists have standard responses. One basic idea is to distinguish between developmental stages or levels of explanation. We might initially cultivate virtue or, for that matter, friendship for the sake of leading the best kind of life for ourselves, but subsequently pursue the good of others for its own sake. By forming personal relationships, we make the good of others part of our own good. Russell says that “eudaimonism is a view about how we adopt our ends—for the sake of a good life—but it does not say that the ends we adopt should be self-serving” (2012, p. 26).

Even if we accept this developmental idea, a dilemma still remains. After deliberation and developing virtue and relationships, either we have reason to pursue the good of others its own sake and not for the sake of our own good or we continue to have reason to pursue the good of others for the sake of our own good. As one might put it, the problem isn’t just that we’re (perhaps initially) motivated to pursue all other ends for the sake of our own happiness, but also that our reasons for pursuing all other ends derive from their relation to our own happiness. The dilemma is then that if our reasons for acting for the sake of others don’t derive from our own happiness, happiness isn’t the final end of practical reasoning, but if they do, the justificatory structure of practical reasoning remains fundamentally egoistic, despite all the protestations of eudaimonists (such as Annas, 1993, p. 322).

Now, eudaimonists might argue that the first horn of the dilemma is based on an illusory opposition between the good of others and our own good. For the virtuous and the caring, they might say, the two are one and the same thing (cf. Foot, 2001, p. 102). I don’t deny that we can make the good of some others part of our own good. That is surely what happens in some personal relationships. It’s better for me to lose a limb than to lose a child, so that giving up a leg for a child would not be a genuine sacrifice any more than going to the dentist for a painful operation now for the sake of avoiding an even greater pain in the future is one. But not all virtues are exercised in that context, and, even there, genuine sacrifices are possible.

This leaves the eudaimonist stranded on the second horn: holding that virtuous people act ultimately for the sake of their own happiness after all, at least in the sense that their reasons for doing so derive from promoting or realizing their happiness. Somehow, our own good (which includes subjective satisfaction) continues to shape our final end. To make this palatable, the eudaimonist may insist that this doesn’t presuppose that virtuous people somehow have their own self-interest in mind while deliberating or acting. Being “for the sake of” their own eudaimonia can shape their pursuits otherwise. We can see some examples of this in the way Annas and Russell themselves describe the virtuous person in candid moments. Consider what Annas says about bravery:

The properly brave person grasps how acting bravely, while done for its own sake, also forms part of her overall good in her life as a whole. … Bravery involves a correct judgement of the good achieved in one’s life by acting bravely, rather than in a cowardly way.

Annas, 1993, p. 75, my emphasis

Notice the reference to good achieved in one’s life, not the good, period. This is what a coherent eudaimonist should indeed say. And I’m not persuaded that this is consistent with being genuinely virtuous or acting for the sake of others. While something like courage may promote the realization of a good life for me considered as a whole and associated positive feelings, the real thing won’t fit in the picture, even if self-interest is not part of the virtuous agent’s deliberation. If virtuous activity is part of our final end, then our final end isn’t our own happiness.

Fortunately, this dilemma has a simple resolution that allows us to accept most of Aristotle’s insights. We should say that virtuous agents pursue the good of others for the sake of leading a good life or the most choiceworthy or successful life, all things considered, and not for the sake of their own happiness, however conceived. That is to say, we should reject the job description that eudaimonists post as far too broad to get at happiness rather than something like the good life or the choiceworthy life. With that, the identification of true happiness with virtuous activity loses its basic rationale.

I’ve just argued that being the final end for the sake of which we pursue all other things is not the right job description for happiness. Instead, we should begin from a job description that is defined by platitudes like the following: happiness has an experiential quality in that it feels good to be happy; both experienced and expected happiness and unhappiness explain and justify people’s life choices; false beliefs can make you happy; different things make different people happy; we can fail to realize that we’re happy or unhappy (for discussion, see Kauppinen, Manuscript).

I’m not going to survey the many different accounts of happiness as a psychological state here, but I will jump right into the view that in my opinion best explains such platitudes, Dan Haybron’s (2008) emotional condition account. According to it, happiness consists in being disposed to have a favourable emotional response to our lives. Such emotional response has three aspects (2008, p. 111–121): Attunement consists in emotional states or moods like tranquility, confidence, and imperturbability. When one is attuned to one’s environment, one is relaxed and open, “at home” in the world. Engagement is manifest in emotions like exuberance and flow, which motivate energetic pursuit of goals. Finally, endorsement consists in feelings like joy and cheerfulness, which motivate broadening and building on the things one’s life contains. For Haybron, it is above all attunement and engagement that are states of the self and thus crucial to happiness. They are central in the sense that they dispose us to have other affects and moods—in Haybron’s slogan, they are productive, persistent, pervasive (permeating the whole of consciousness and setting its tone), and profound (felt as deep) (2008, p. 130–31). In short, then, being happy is psychically affirming one’s life in terms of positive central affective states in the three dimensions, as well as having a propensity for positive moods.

Haybron’s account explains nicely why happiness feels good (because positive moods and emotions do so), while we can nevertheless be mistaken about how happy we are (because moods, in particular, are in the background of our consciousness and are thus not self-intimating, and our dispositions are not transparent to us). It also makes sense of the explanatory and justificatory heft of happiness in virtue of emphasizing the centrality of many happiness-constituting states to the self, and de-emphasizing the role of peripheral states like sensory pleasures. As a psychological account, it’s also compatible with the modern platitude that we can be happy because of false beliefs.

Nevertheless, there are three challenges that can reasonably be raised against Haybron’s account. The first is explanatory depth: why does happiness consist in just these emotions and moods and not others? The second is what I’ll call the weighting problem: how do the different dispositions, moods, and emotions combine to determine someone’s level of happiness? What determines how much weight each of them has? And, finally, there is the experientiality question: how can our dispositions, which are themselves unexperienced, make a difference to our happiness? To see why this last issue is a problem, consider that it’s possible (in principle) for two people to have different emotional dispositions and propensities, yet actually experience exactly the same affective states for their whole lives due to, let’s say, lucky circumstances in the case of one of them. It seems clear to me that the two people with identical affective experiences over a lifetime must be equally happy, even if only one is disposed to be happy. It is, of course, true that one of them has a happier disposition and can thus be said to be a happier person. And it’s typically good for you to have a happy disposition—but only instrumentally good, since it means you’re more likely to be happy. But all of that is compatible with happiness itself being entirely nondispositional.

My own version of an emotional-condition view, which I develop more fully elsewhere (Kauppinen, Manuscript), is constructed to avoid these problems while building on Haybron’s insights. Let’s begin with explanation. What is in common to experiences that determine an individual’s level of happiness or unhappiness is that the phenomenal character of each has a valence, negative or positive, so that they feel good or bad.[1] The default position must surely be that all phenomenal states with a valence contribute to (un)happiness, because they contribute to experiential quality. But what about Haybron’s objection that superficial pleasures and pains, like the pleasure of eating a cracker, don’t make us any happier or unhappier? On my view, this is a question best answered at the second stage of the theory, which deals with the weighting of valence-bearing experiences in the accounting of one’s level of happiness. It is important to note at this point that Haybron is right in emphasizing that not everything we experience at a moment is at the foreground of our attention. A pleasure or a pain can be just the tip of the iceberg. Beneath the water, shaping our experience while easily slipping our attention, lie our emotions and moods. Someone who is sad and someone who is happy may laugh as much, but the totality of their experience will nevertheless be different. Our total affective state at any moment is likely to be complex, nontransparent, and hard to articulate (cf. Bramble, 2016).

How much weight do these various valence-bearing experiences have for someone’s happiness, then? As Fred Feldman argues, we should start with an account of happiness at a time, or momentary happiness, since we can easily define total and average happiness over time in terms of momentary happiness (2010, p. 118–123), so I’ll proceed accordingly. The first thing to note is that valence is a matter of degree, whether positive or negative. If we represent the intensity of negative valence with negative numbers, we get the following simple formula for what I’ll call momentary affect balance:

Momentary affect balance at t is the sum of the intensities of all of S’s valence-bearing phenomenal states at t divided by their number.

If at the moment you’re bored with reading the paper to degree –3, feel self-esteem as a result of comparing yourself to the present author to degree +7, and enjoy sitting on the sunny deck to degree +5, your momentary affect balance is –3+7+5/3 = 3. I think the average rather than the sum better captures the idea of affect balance—otherwise feeling positively about five things would amount to a better total affective state than feeling equally positively about three things.

Momentary affect balance, however, is not the same thing as happiness. For one thing, it is arguably too volatile: not every new feeling should make a noticeable difference to our level of happiness. Related to this, it lacks the heft that happiness can have: it’s not the kind of thing that explains and justifies life choices. It is also indiscriminate: all feelings count equally, regardless of how important to us their objects are. This observation gives rise to a central modification: an affective state’s impact on happiness depends not only on its intensity, but also on the subjective importance of its object—roughly, the extent to which we value it. For example, the joy you feel when your child recovers from an illness might be as intense as the joy you feel when your favourite team wins a game, but, on the assumption that you value your child’s health more than your team’s success, the former will make a bigger difference to your happiness.

But why is that, if the affective state feels just as good or bad? I think Haybron is on the right track here in emphasizing the link between happiness and the self: the states that count the most for happiness are states of the self, and not just peripheral ones. But unlike him, I don’t think that centrality is a matter of disposing agents to experience certain affects rather than others (2008, p. 130), or “output-centrality,” as Mauro Rossi and Christine Tappolet (forthcoming) put it. This would result in double-counting the contribution of an affect. Suppose in situation A my joy at the intensity of +5 gives rise to a +3 boost in self-confidence, while in B I experience the same joy, but it doesn’t dispose me to feel further positive states. Joy in A is then by definition more output-central than joy in B. But I don’t think we should say that the confidence-productive joy in A in itself contributes more to my happiness than the equal joy in B. Of course, it does causally contribute more (+3 more, to be specific), but it would be double-counting if the joy itself had extra weight in addition to its own intensity and causal contribution (say, it would contribute +6 to my happiness level instead of +5).

So, instead, I start from the notion that it is our values (or valuings) and concerns that define our perspective or “self.” By “valuing” an object or person or cause X or a state of affairs Y, I mean, as is fairly standard these days, a complex disposition to pay attention to X or Y, to be pleased when one believes things are going well for X and sad when they’re going badly (or when Y is realized or not), to want things to go well for X (or for Y to be realized), to judge that there is reason for us to have such responses to the fortunes of X or to Y’s reality, and to give weight to X’s interests or Y’s realization in deliberation (e.g., Tiberius, 2018). A “concern” I take to be something more primitive, an attachment to some X that may come apart from judgments about reasons—for example, we may have self-concern even if we don’t value ourselves. Such concern entails motivational and emotional responses to apparent threats to X and to their absence. Both values and concerns belong to the broader class of sentiments (Tappolet, 2016). Even among our sentiments, some are more important to who we are than others. We typically have concern for ourselves and our loved ones, and consequently we value our and their honour and health in a different way than we value our favourite television shows, for example. Affects that manifest such core concerns and values, such as worry for a child having trouble at school, can consequently be said to be states of our selves and carry a lot of weight in determining our level of happiness—the child’s troubles get to us, as Haybron (2008) says.

Taking this into account, we can define significance-adjusted momentary affect balance:

The significance-adjusted intensity of a phenomenal state for S is the product of the intensity of its valence and the subjective importance of its object for S. Significance-adjusted momentary affect balance at t is the sum of the significance-adjusted intensities of all of S’s valence-bearing phenomenal states at t divided by their number.

(Presumably subjective importance here should range from 0 to 1.) Significance-adjusted momentary-affect balance is, I claim, what we could call “momentary happiness.” The pleasure of eating a cracker doesn’t manifest self-concern or core values—after all, you could have the same pleasure even if your values were different. Of course, there may be people who really value little moments of pleasure in the midst of a busy day and consequently feel joy and delight when they encounter a nice cracker—but then it isn’t implausible that they are a little bit happier for a moment. In contrast, the joy, pride, and self-esteem boost that result from finally getting that big promotion are typically manifestations of valuing professional success and will consequently carry a lot of weight in the happiness of those invested in it.

Significance-adjusted momentary affect balance may still seem too lightweight to capture the deliberative role of happiness—after all, we aren’t, and shouldn’t be, too concerned about momentary happiness when making life choices. But, in my view, as in Feldman’s, we can compare which of two lives or portions of lives are happier by adding up or averaging momentary happiness at each relevant moment. And these results—say, how much total happiness I’m going to have in my life going forward—are certainly normatively and explanatorily weighty.

Call the view that identifies happiness with a sufficiently positive sum of significance-adjusted momentary-affect balances over time sentimentalism about happiness, or just “sentimentalism” for short (obviously, not to be confused with sentimentalism about moral judgment or metaphysics[2]). Given typical human concerns, sentimentalism explains many of Haybron’s apt observations about happiness and our emotional condition. Self-concern and concern for loved ones are core sentiments for just about everyone. They manifest themselves in anxiety and depression when our horizon of expectations is threatening or unpromising, and in moods of serenity, tranquility, and optimism when it isn’t. They are also manifest in feelings of self-worth and self-respect when we value ourselves, and in the dreadful feelings of worthlessness otherwise. Through such moods and emotions, we construe our environment as secure or threatening and ourselves as worthy or unworthy of love. According to sentimentalism, these moods and emotions will consequently have a lot of weight in determining our level of happiness—though they rarely take centre stage in our consciousness.

There is more variety in people’s core values, in the importance they give to career, family, causes, hobbies, and other projects. But since the central function of values is to guide our pursuits, emotions that are associated with exercising our agency, such as interest, flow, sense of purpose, and pride (or, conversely, disinterest, feelings of pointlessness, guilt, and shame) will typically have considerable weight in our happiness level in virtue of manifesting core values. We can, of course, value other things, such as the survival of coral reefs or national self-determination, making their fate subjectively important to us and increasing the weight of emotions related to them, including attitudinal pleasures and displeasures (Feldman, 2004). But typically, when it comes to core values in particular, we do tend to do what we can to promote their realization rather than just watch what happens from the sidelines.

Finally, merely sensory pleasure and pains seem to be quite independent of our sentiments, and thus they will as such have little weight in determining our happiness level, even if they’re experiences that have a strong valence. To be sure, it could be said that we have a concern for the felt quality of our experience as such, and thus we will typically have second-order responses to pleasant or painful experiences—I’m attitudinally displeased to be experiencing physical pain, to use Feldman’s language. But here it is those second-order responses that carry weight more significantly in our happiness or unhappiness.

What does a sentimentalist account of happiness mean for the place of virtue? Clearly, virtuous activity won’t be constitutive of happiness, as happiness is not an activity but a total experiential quality. But that doesn’t mean that the nature of our activities doesn’t make a crucial difference to our affective condition—as I said in the introduction, there’s good reason to think virtuous outlook and activities are likely to make us happier than their opposites, as Virtue Conduciveness says. I suggested the following mechanism for this:

Let’s now investigate this mechanism in more detail. Recall from section 1 that if we’re virtuous, we’re good at recognizing what our objectively best realistically achievable options are and at figuring out how to bring them about in an ethically sustainable fashion, because we’ve been lucky enough to have been habituated into ways of thinking that avoid common human defects like selfishness, shortsightedness, and various biases. Consider what this means for virtuous people’s choices. I think that we often end up underselling virtue by focusing on responses to individual situations. After all, practical wisdom isn’t most conspicuously exercised in the context of deciding whether to hold a particular position in the face of enemy attack, but rather in the context of choosing whether to go to war in the first place. So, we should think about how virtuous people make more impactful choices in their lives, such as deciding which career to pursue or which project to take on. In these contexts, too, human beings are easily led astray by short-term thinking, selfish focus, or superficial attractions. The practically wise will brush aside such distractions and consider what someone with their particular abilities and in their position can do.

Consequently, the virtuous will be concerned about serving justice and helping those whose flourishing is hindered by their circumstances, and will aim to increase external resources that can be put to good use for their friends and others. So, we can expect them to become pediatricians rather than portfolio managers, public defenders rather than corporate litigators. If they have suitable talent and opportunity, they will conduct research into malaria prevention rather than perform cosmetic surgery, or engineer environment-friendly dams rather than disposable toys. These choices will often mean foregoing money, fame, and comfort, but the virtuous aren’t pained by such losses, as long as these don’t stand in the way of realizing their aims. Their hearts are in what they do—that is, they emotionally endorse the choices that they make, and don’t regret what they’ve given up or haven’t had a chance to pursue. Since they have practical wisdom, the aims they’ve selected are realistically achievable and their pursuit of them is intelligent, so they are likely to be successful at realizing their aims, whether these are prospective ones that can be realized once and for all or reflexive ones of living up to ongoing commitments to, say, family or community (Setiya, 2014; Kauppinen, 2021b). They’re also likelier than the nonvirtuous to enjoy good relationships with trustworthy others, because, rather than “takers,” they are the kind of “givers” that most people like to be around.

So far, I’ve described the first two steps of the above diagram. The next and crucial step is that given the kind of values, aims, pursuits, and outcomes that the virtuous are likely to have, they are very likely to enjoy positive-valence-bearing experiences with regard to subjectively important matters. To begin with the attunement aspect of their emotional response, we see that since they’re likely to be loved and respected by others and to enjoy good friendships (in part because they are more likely to recognize those who lack virtue and avoid them), they’re consequently likely to have positive feelings of self-worth. In ordinary circumstances, they’re less likely to be anxious or depressed than others, since they don’t put as much weight on their own good as others do on theirs. (Evidently, you can find yourself surrounded by vague threats and stuck without appealing prospects even if you’re virtuous, so virtue isn’t a guarantee that negative moods are absent.) But probably the biggest advantage of engaging in virtuous activity is that you’re more likely than otherwise to enjoy the pleasures and feelings associated with merited success as well—you’ll be able to take pride and joy in what you do. You’ll probably enjoy a sense of purpose and healthy self-esteem, and you’ll be grateful for the luck you’ve had in terms of opportunities, upbringing, and external goods. All this means that the virtuous are likely to experience their lives as meaningful, which is an important element of happiness (Kauppinen, 2013). Of course, virtuous people may find themselves despised and fail at what they do. If you dedicate yourself to freeing Palestine, your prospects of success are not good. But even failure stings less if you know you’re doing the most anyone in your position can do for a good cause. It’s also important that the virtuous not be particularly anxious about what other people think of them or how they compare to others, because they don’t value success of that sort. Finally, because positive moods and success-related emotions are, as Haybron says, productive of further positive emotions, they make it easy for the virtuous to take joy in little things, like sunshine or a Knopfler song on the radio.

Finally, as I argued in the last section, this overall positive emotional state with respect to things that subjectively matter to the virtuous amounts to happiness. Unlike the eudaimonist picture, sentimentalism considers virtuous activity still to be only a likely cause of happiness. So, while I agree with Annas when she says that “our happiness comes at least in part from the way we do or don’t actively live our lives, doing something with them or acting in relation to them” (2011, p. 152), I demur when she concludes on this basis that “happiness is at least in part activity” (ibid.)—after all, there’s a big difference between being a source of happiness and being constitutive of happiness. In the debate about whether virtue is sufficient for happiness, sentimentalism takes the Aristotelian side, which says it’s not: virtuous people might lack resources, suffer from a chemical imbalance or other disease, and find themselves required to sacrifice their happiness for others. Nor is it necessary. Inside the Experience Machine, one might have the illusion of virtuous activity and be genuinely happy without actually doing anything. This simply follows from the fact that the connection between activity and happiness is causal rather than constitutive.

There are two more interesting ways in which the nonvirtuous, or even the vicious, might gain an equally positive affective condition as the virtuous in favourable circumstances. First, people who lack virtue will sometimes do the right thing and draw enjoyment from it. As a nonvirtuous person, I certainly hope this is the case. But those who lack virtue will not reliably track what they have most reason to do in their actions and emotions, and their emotional perspective and judgments may come apart, even if either gets things right. So, sometimes, and perhaps quite a lot of the time, the nonvirtuous will pursue aims that are not worth pursuing and will be likelier to fail at worthy pursuits. I think that in spite of our extraordinary capacity for rationalization, many people do recognize the low worth of their projects at an emotional level. If you choose a job solely on the basis of external rewards, don’t be surprised if you feel a lingering dissatisfaction in spite of the stellar sound system in your new Tesla. Many nonvirtuous, of course, succeed in deceiving themselves, in spite of such incipient recognition of worthlessness, and take pride in what they do. This is greatly assisted if they form a community with other nonvirtuous. I have no problem believing that there are happy investment bankers on Wall Street. This goes even for those who are not only nonvirtuous but actually vicious in some way, such as the greedy or the lazy.

I do think, however, that the virtuous have an advantage here. Their positive emotions are actually fitting and reliably so, which makes them robust in a way that the positive emotions of the nonvirtuous aren’t. The virtuous are also less vulnerable to luck when it comes to hitting on a genuinely valuable and thus rewarding activity. It’s not a matter of chance that their emotional and rational perspectives form a unified whole. Because they don’t pursue pointless or evil aims, they don’t need to suspend disbelief or deceive themselves or rely on the unpredictable support of bad company. Nor will gains in knowledge or understanding make them unhappy. So, the positive affective condition of the virtuous depends less on sheer luck than that of the nonvirtuous does.

The second way in which the nonvirtuous in the ethical sense can successfully reach happiness isn’t subject to such contingencies. There is, after all, genuine achievement in arts, science, and even business. Such achievements can be sources of pride and self-esteem and beget love and admiration, thus predicting a robustly positive affective condition. Notoriously, they are possible even for the mean, dishonest, and selfish, even if they involve a kind of self-transcendence. But observe that this source of happiness is available only for the few who can really pull it off. It takes a real and special talent to manage and enjoy artistic success, say, without treating others well, although we know it is strictly speaking possible. But then again, in the real world, moody geniuses are rarely paradigms of radiant happiness!

As I’ve made clear, my account of happiness owes much to Dan Haybron’s emotional-condition view in spite of my rejection of some of its main tenets. In recent years, Mauro Rossi and Christine Tappolet (2016; forthcoming) have also developed an affectivist view of happiness in a different direction from Haybron and have used their account to defend a noncontingent causal connection between virtue and happiness. As we’ll soon see, it’s fair to describe their account of happiness as a form of sentimentalism. So, it may initially seem that they are the People’s Front of Judea to my Judean People’s Front. But closer examination of the details of their view reveals significant differences.

Let’s start with happiness and its relation to well-being. Like me, and unlike Haybron, Rossi and Tappolet want to include all occurrent, valence-bearing affective states as constituents of happiness, including pleasures and pains. Drawing on Haybron, they then argue that intensity and centrality of an affective state are both relevant to its contribution to happiness. Unlike Haybron, however, they distinguish between two ways in which a state can be central: it can be the cause of many other affective states (output centrality) or it can result from, or be “grounded in,” central affective states or sentiments (source centrality). While Rossi and Tappolet are somewhat ambivalent about the importance of output centrality, I unambiguously reject it as a determinant of how much weight an affective state has for happiness. The role of source centrality, however, is a point of agreement between us.

Building on their past work (Tappolet, 2016; Rossi and Tappolet, forthcoming), Rossi and Tappolet emphasize that happiness-constituting affective states are “phenomenal evaluations” that represent their objects as having evaluative properties. Being composed of such states, happiness itself constitutes “a global affective evaluation.” Since such affective evaluations can represent the world correctly or incorrectly, it follows that happiness itself can be fitting (both in terms of its constituent affects and the underlying affective dispositions)—that is, represent the subject’s evaluative situation correctly.[3] The bold claim that Rossi and Tappolet next make is that well-being, or what is in itself good for a subject, just is fitting happiness. This, they argue, avoids the central objections made to earlier experientialist theories of well-being, such as hedonism—after all, while you may be happy in an experience machine (Lin, 2016), your happiness won’t be fitting, because things are not actually the way your positive feelings represent them as being.

Could well-being just be fitting happiness? Rossi and Tappolet’s view belongs to the class of experiential-objectivist hybrid theories of well-being, such as Shelly Kagan’s (2009) view that well-being is enjoying the (objectively) good. The basic problem with all such views, as I see it, is that they exclude both positive experiences that are not fitting and objective goods that are unexperienced, as well as corresponding bads. On the first point, consider a simple variation of the Experience Machine thought experiment. Bill and Ted are both in the machine. Bill has a lot of positive experiences, all of which are unfitting, while Ted has none. A strict fitting-happiness view entails that they’re equally well off, but they’re not—while both are deluded, Bill is at least happy, while Ted isn’t. A more relaxed fitting-happiness view (similar to one of Fred Feldman’s (2004) suggestions) says that while fitting positive feelings are better for you than unfitting ones, unfitting ones are also good. Such a view would explain the asymmetry between Bill and Ted, but it would lack the explanatory unity that Rossi and Tappolet claim as an advantage of their view—now we have two kinds of final good, fitting and unfitting happiness. What’s more, it’s not clear whether there’s any rationale for giving more weight to fitting than unfitting happiness that’s not ad hoc—after all, as an experience, unfitting happiness seems to be just as good as fitting happiness.

The other basic problem for experiential-objectivist hybrids is that just as they rule out enjoyment of the nonvaluable, they rule out unenjoyed values. Anything that is in itself good for us without affecting our happiness is a counterexample. Many people believe in posthumous benefits, which we may well not be fittingly happy about (though you could, in theory, correctly anticipate and take pride in posthumous success while you live!). Less controversially, while we live, our goals may be realized or our trust betrayed without our finding out about it. Indeed, such things may never make a difference to our experience. Nevertheless, they seem to make us better or worse off (e.g., Nagel, 1979)—they are the kind of things we’d intrinsically want for those we love for their sake (Darwall, 2002). Indeed, some valence-bearing affects seem to be fitting precisely because they correctly represent things as going nonexperientially well or badly for us, which puts the fitting-happiness view in a bind. Suppose I’m relieved to discover I wasn’t betrayed after all. Such relief is presumably fitting only if it’s bad for me to be betrayed independently of my feelings about betrayal (what precisely I’m relieved about is the fact that I wasn’t betrayed, not that I was spared the bad feelings that would have arisen upon my knowledge of events had I been betrayed). But if that’s the case, some instances of fitting happiness presuppose that things other than fitting happiness can be welfare goods.[4]

Rossi and Tappolet (forthcoming) may think that the above consequences of their experiential-objective hybrid view are worth taking on board, because their view offers explanatory unity that treating happiness and, say, agential success as potentially distinct welfare goods lacks. It is evidently not possible to discuss all the alternative ways of unifying welfare goods here. But just to give a hint of the view I prefer, on the version of perfectionism I’m developing (for an initial statement, see Kauppinen, 2015, p. 218–219), what unifies the things that are in themselves good for us is that they constitute realizations of the overarching aims that are implicit in the exercise of our essential capacities as the kind of well-being subjects that we are. In the adult human case, this means that our well-being consists in developing and exercising our capacity both for temporally extended rational agency and for positive experience, since we’re essentially both extended rational agents and subjects of valence-bearing experiences.[5] Such perfectionism, I believe, provides the desired explanatory unity while allowing the two dimensions of prudential value to vary independently, thus avoiding the problems of hybrid views while capturing their attractions.

Finally, let me turn to the link between virtue and happiness in Rossi and Tappolet’s work. Instead of the Aristotelian picture of virtue, they endorse a more minimal account, according to which a virtue just consists in disposition to experience fitting emotions, which further disposes one to act and judge correctly (2016, p. 116). For reasons noted in the first two sections, I think the Aristotelian view gives a better account of practical reasoning and moral motivation. But set that aside and focus on the noncontingent link between virtue and happiness they propose:

If virtue involves a disposition to experience fitting emotions, then, to the extent that the external circumstances are propitious, virtue disposes the individual to experience positive fitting emotions.

Rossi and Tappolet, 2016, p. 122

Someone’s external circumstances are said to be “propitious” “when the individual is confronted with positive evaluative properties or, more simply, to [i.e., with] positive values” (2016, p. 123). Note that this is a much weaker noncontingent connection between virtue and happiness than my view makes. Basically, what Rossi and Tappolet seem to say is that if you happen to find yourself in good circumstances, then, if you’re virtuous, you’re also happy. In contrast, what I say is that if you’re virtuous, you’re likely to bring about good circumstances (this is the Aristotelian part) and you’re likely not to mind most things that actually cause bad feelings (this is the Stoic part). (No doubt this issue is related to the passivity worry about Rossi and Tappolet’s view of well-being.) So the mechanism for the noncontingent link between virtue and sentimentalist happiness that they suggest is very different from mine, which highlights the role of virtuous activity and related emotions.

I don’t know about you, but I’d like to be happier (and, sure, fitter and more productive, too). I’ve argued that what this would amount to would be having more positive feelings—and fewer negative feelings—about what I most care about. I’ve also argued that since virtuous people not only have the right values and concerns but also have the skills to pursue these wisely, they’re likely to be successful and to be liked. At the same time, they don’t give a damn about things that don’t really matter. Consequently, virtue and virtuous activity are likely to be relatively conducive to happiness in comparison to nonvirtue, though this noncontingent tendency is still subject to luck. Moreover, a virtuous person’s happiness doesn’t depend on self-deception, mistake, or illusion, and it doesn’t require extraordinary talent. It is thus effortless to sustain and, in the real world, more likely to persist than happiness based on unfitting positive emotion. Being convinced of these things, I’d like to be more virtuous for purely selfish, borderline-hedonistic reasons, as well as for ethical ones. Alas, it’s easier said than done for an old dog. It’s doubtful whether even acting as the virtuous do would improve my character. But, as Aristotle says, there’s little else I can do, so I had better go and answer some student emails.[6]