Article body

1. INTRODUCTION[1]

Here’s a common enough thought about happiness: to be happy is just to feel happy. In such claims it is usually clear that “happy” is being used as a purely psychological term and not as an evaluative expression for well-being, or what Aristotle called “eudaimonia.” No value judgment seems to be at issue here, so it can be perfectly intelligible to say things like “happy people are pathetic fools.” We’re just describing someone’s state of mind. And yet, even in what’s called the long-term psychological sense of “happiness,” few philosophers conflate being happy with the acute emotion of feeling happy. At least, it is rare to find a philosophical theory of happiness, in any important sense, as amounting to nothing more than a particular emotion or mood. Rather, philosophers have focused on something to do with a person’s state of mind over a more or less extended period of time. If you are happy “these days,” that must be a matter of rather a lot of feelings, if it’s about feelings at all. And it probably includes feelings other than the smiley-face one. So perhaps happiness is better framed in terms of feeling good, where this might include all manner of positive affect besides feeling happy, like a pleasant sense of tranquility. Indeed, this is something of a commonplace.

More broadly, it is entirely standard to think of happiness as a wholly occurrent state of mind: to be happy is a matter of what actually transpires in one’s head and not, for example, how one is disposed to feel. Scholars sometimes frame this latter idea in terms of “dispositional happiness”: a trait, such as being a happy person.[2] But you could possess “trait happiness” without being happy in the present; though blessed with an upbeat and laid-back nature, maybe you’re having a bad week.

These are natural suppositions, but I think they are mistaken. Dispositionality is a crucial aspect of happiness, understood not as a trait but as the relatively lasting psychological state that animates most everyday concern with happiness, including the desire to be happier that fuels a multibillion-dollar self-help industry. I have argued at length for this idea in earlier work defending an “emotional state” theory of happiness, and the broad approach has gained some uptake in the philosophical literature.[3] But the basic schema of emotional state theory admits of various readings, including forms that take happiness to be entirely occurrent, or at least to be dispositional to a far lesser extent than I have claimed. For instance, perhaps happiness is purely a matter of occurrent emotions and moods, including feeling happy, serene, and energetic, and, on the negative side, feeling sad, anxious, lethargic, stressed, and so forth. (Note: I often use “happiness,” like “health,” as a generic term for a domain of concern that includes both positive and negative states. Accordingly, many of my examples will involve unhappiness or “unhappiness-constituting” states like anxiety.)

While I have suggested that even these states have dispositional elements, not all commentators have agreed, and there has been considerable resistance to my suggestion that some happiness-constituting states are not occurrent at all, such as a heightened propensity for anxiety or for feeling happy.[4] And, not surprisingly, emotional state theories have been rejected tout court by hedonists, who identify happiness with wholly occurrent states of pleasure, versus unpleasure.[5] So the form of emotional state theory of happiness I’ve defended faces skepticism from at least two directions owing to its focus on dispositionality.

This focus can indeed seem odd if the venerable doctrine of hedonism, on which happiness just is pleasure, serves as one’s reference point in thinking about happiness. There are better ones: for starters, the popular English locution, “happy and healthy.” (I do not know how prevalent equivalents are in other languages, though “heureux et en bonne santé” has some purchase in contemporary French.) Whatever the expression refers to, people tend to think it quite important, so that they’ll often say things like “I just want my kids to be happy and healthy.” This frequently reduces to “I just want my kids to be happy.” If happiness alone is that important, then happiness and health must be really important. It is doubtful that in making such assertions many of us mean to be committing ourselves to the idea that these things literally exhaust the constituents of a good life. It’s just that the other bits, like not being a serial killer, don’t require mentioning. Moreover, the fact that “and healthy” seems not to be redundant suggests that whatever happiness is thought to be, it isn’t necessarily the same thing as well-being, even if “happiness” can take on such a meaning in other contexts. What is being wished for, beyond health, is very likely something to do with mental states: happiness in the long-term psychological sense.

I think the parallel with health reveals a crucial feature of the way in which happiness is ordinarily conceived: namely, as something belonging to the category of what I’ll call functional conditions, or “conditions” for short. Just as ascribing health to someone is saying something about that person’s condition, so, I will argue, is ascribing happiness to someone. If this is right, then the dispositionality I have attributed to happiness should not be puzzling after all. For dispositionality is an essential feature of conditions; to be healthy, for instance, is at least partly to be disposed to function well, not to drop dead of a stroke, and so on. Likewise, to be happy is at least partly to be disposed to respond favourably to things in one’s life, and this licenses various predictions and explanations—for example, that one will likely handle annoyances with relative aplomb. Put another way, the concept of happiness, like the concept of health, is what we might call a condition assessment concept. (One might alternatively refer to it as a “functional assessment” concept, a framing I owe to John Doris.)

In what follows I will, first, rehearse the essentials of the theory of happiness in question and the puzzle it raises: why think happiness is dispositional and is not merely an occurrent state? To explain this, we shall need to introduce a distinction in the metaphysics of states—namely, the idea of a condition, which I am not sure has been elaborated previously, at least in the sense intended here. To motivate the distinction and illustrate its importance in the present context, we imagine an artificial system, a security robot that monitors and responds to threats in the environment. To characterize this system in relation to its purpose, we need concepts having to do with various aspects of its state, including its condition—for example, whether it is on high alert.

The upshot is that the emotional state theory of happiness, dispositions and all, is well motivated, capturing an important and natural aspect of a person’s psychological state: that person’s emotional condition. As with the security system, my emotional condition amounts to a summary evaluation of my circumstances relative to the goals that structure my makeup: are things going well for me? This underwrites the use of happiness as a convenient, nonevaluative proxy for well-being—a relatively simple way to tell how someone is doing.

The paper concludes with some reflections about the import of these points for the empirical study of personality and well-being, both of which have deficiencies owing to a failure to distinguish functional conditions from mere events in the emotional realm. A person’s emotional life is not just a series of transient responses to stimuli, as is commonly supposed; like life satisfaction attitudes, it concerns the broad character of one’s circumstances, and may indeed offer a better gauge of those than a person’s explicit judgments. Moreover, a state’s being dispositional does not suffice to make it a trait, as is widely supposed in social and personality psychology, with the result that personality measures are sometimes confounded with happiness measures. And “happiness” measures typically fail to attempt to measure key aspects of happiness, though to some extent this may be an inherent limitation of standard self-report methods.

At the core of this paper are several distinctions in the metaphysics of states and events that, to the best of my knowledge, are novel.[6] Presumably the arguments ramify for other questions in metaphysics, but the relevant literature is vast and intricate, with considerable dispute about even basic questions. As my aim is chiefly to clarify some issues in the psychology of affect, I will try to minimize entanglement in those debates. Where that can’t be avoided, it may be possible to translate my claims to accommodate different accounts of events, states, and so forth, so long as the basic relationships I’ve posited among different affective kinds are preserved. I regard the framework sketched below as a first approximation, not a finished product—but hopefully good enough to sustain my claims about the nature and significance of happiness.

2. BACKGROUND: THE EMOTIONAL STATE THEORY OF HAPPINESS

The theory in question, the emotional state theory of happiness, was introduced as an alternative to the then-dominant views of happiness in the long-term psychological sense, hedonistic and life satisfaction accounts. The former identifies happiness with a person’s balance of pleasant versus unpleasant experiences, while the latter identifies it with a judgment-like attitude of being satisfied with one’s life as a whole, or with one’s life as it is these days.[7]

In its basic or default form, the emotional state account reduces happiness to a person’s emotions and moods, taken together. The basic idea is that happiness is roughly the opposite of anxiety and depression, these being prominent forms of unhappiness. The view differs from hedonism, first, in that emotions and moods are rich, deep affective states with various aspects—some of them nonconscious—and are not themselves reducible to experiences. Second, while emotional pleasures tend to be the most important for our hedonic states, they do not exhaust the hedonic realm. Many sensory and “notional” or cognitive pleasures—e.g., approving of a pretty house one passes—seem not in any ordinary sense to be emotional states. Intuitively, they don’t impact our emotional conditions. Nor do they seem to affect how happy we are, as illustrated, for instance, by the intense pleasures of sex, which notoriously can be emotionless, leaving us cold even as they feel good. The distinction in question is quite familiar both from common sense and from philosophical and spiritual traditions, where it is widely taken to be an important sign of character what manner of things we allow to get to us, upset us, get us down, raise or lower our spirits, and so forth. Not so for the things that merely give us pleasure or pain, like potato chips and hangnails. This distinction is central to Stoic and Buddhist practices, which focus on the management of our emotional conditions, but expressly let us off the hook regarding mere pleasures and pains. To take a Buddhist example, when pierced by an arrow, the pain is inevitable, but the suffering is optional, and our task is to learn to avoid the emotional disturbance. Aristotle equally took our emotional conditions to be centrally important, though certain sorts of emotional disturbance, like fitting anger, are not to be avoided but actually cultivated. On the emotional state theory, these are all points about the things that make us happy or unhappy; they cannot plausibly be read as claims about pleasure. (Aristotle had a good deal to say about the role of “pleasure” in virtue and hence eudaimonia or well-being, but it is clear he meant emotional states; like most commentators, he simply conflates these categories. He certainly is not suggesting that virtuous activity is completed by a pleasing tingle in one’s belly. If the emotional state theory is correct, then what is ordinarily said about the role of pleasure in Aristotelian eudaimonia is really about the importance of happiness for well-being.)

The foregoing distinction is what I’ve called the central/peripheral distinction, where only central affective states—roughly, moods and emotions—constitute our emotional conditions and, in turn, happiness. I argued that central affective states are all “mood-constituting,” so that one’s mood is thereby altered by an emotion. Accordingly, we could just as well speak of mood-related affect. I’ve left it open whether the central/peripheral distinction is merely one of degree, so it is possible that even peripheral affects play some role in happiness and our emotional conditions, though centrality in any event is distinct from degree of pleasantness (e.g., orgasm).

As noted at the outset, some form of emotional state theory has been endorsed by a number of commentators. The variant I have defended, however, adds the idea that happiness is substantially a dispositional affair. This figures partly in how I conceive of central affective states, which I take to have dispositional elements that are important for their contribution to how happy one is. When feeling anxious, for example, one is thereby disposed to respond to events in ways characteristic of anxiety—with more fear, less enjoyment, greater vigilance, and so on. It is partly by virtue of these dispositions that one is less happy when anxious. To this picture I added a further element, mood propensities—that is, one’s (appropriately grounded) disposition to experience certain moods and emotions. When depressed, for instance, one may at a given moment be in a good mood, yet still prone to slide back into a flat or depressed mood. One remains depressed—unhappy—by virtue of that propensity alone. In a later paper I allowed that we might substitute the categorical bases of mood propensities for the dispositions themselves—for example, perhaps one is more or less happy by virtue of an unconscious mood state that manifests itself via one’s mood propensity (Haybron, 2010; Hill, 2007). Little hangs on this question, however, so I’ll stick with the original framing.

3. HAPPINESS AS A CONDITION ASSESSMENT CONCEPT

3.1. The problem of dispositionality

We’ve seen that the main source of resistance to the present theory of happiness, even among partisans of an emotional state approach, is the role it accords to dispositional states. Happiness is fundamentally an occurrent phenomenon, the thought goes, with dispositions either being irrelevant or counting only insofar as they are aspects of occurrent emotions and moods.

There is a simple response to such worries: namely, to observe that dispositionality is likewise a central characteristic of the most widely employed conception of happiness in research outside of philosophy, the life satisfaction view. Indeed, life satisfaction may be purely dispositional—essentially a cognitive counterpart to a pure “mood propensity” theory of happiness. It certainly cannot require that any occurrent states be taking place whilst one is happy or unhappy, since the occasions on which we are thinking about how our lives are going will likely be few and far between. To conceive of life satisfaction as an occurrent mental state would yield a theory of happiness with grave “attitude scarcity” problems, whereby most of us are rarely on the happiness-unhappiness scale at all—a result that both is deeply counterintuitive and vitiates the apparent significance of happiness. While life satisfaction theories have come in for strenuous criticism of late, leaving few defenders in the current philosophical literature, the problems have little to do with the dispositionality of life satisfaction.[8] As far as that goes, the approach is plenty intuitive: prima facie, the idea that happiness consists in being satisfied with your life—for instance, in being disposed to judge your life favourably—is quite plausible, to the extent that empirical researchers routinely assume such an account without argument.

While theorizing about the nature of emotional conditions such as depression is less prominent in philosophy, it is very plausible—and indeed is part of the argument for an emotional state view—that depression is substantially dispositional. To be depressed is not merely to feel bad, but for one’s psychological stance generally to be altered for the worse.[9] And no one supposes that your depression has resolved if, given the time of day or a particularly agreeable situation, you happen to feel good at the moment. You’re still depressed, even when you don’t feel it at all. On the emotional state theory, depression just is a pronounced form of unhappiness, and this is not an unintuitive result. Contrary intuitions regarding happiness, then, may owe something to the particular connotations of “happiness,” including residual influence from the well-known history of philosophical work using the term in a hedonistic vein. Perhaps it would help to reflect for a bit on life satisfaction views before assessing the emotional state theory.

It is surprising that the inclusion of dispositional states has gotten a mixed reception even among commentators sympathetic to the basic emotional state framework: the basic argument for such a view rests heavily on the failure of hedonism to take seriously the nonconscious aspects of happiness, arguably including dispositionality, in thinking about happiness. If one favours an emotional state theory over hedonism, then hedonism is liable to seem the odd man out, as both emotional state and life satisfaction theories conceive of happiness as a matter of one’s orientation or stance toward one’s life, whereas hedonism conspicuously does not, reducing happiness instead to a series of purely experiential events.

While dispositionality per se seems to me a positive feature of emotional state theories, and not an objection to them, there seem to me real worries in the neighbourhood. Specifically, the mixed character of the theory, whereby happiness has both occurrent and dispositional aspects, raises questions about how well the various constituents of happiness hang together. Is it just a grue-some assemblage of states rather than a genuine kind? Why not follow the lead of a life satisfaction theory and maintain that happiness is entirely dispositional? While I hope prior argument has made the naturalness of happiness as a kind seem plausible, greater clarity would be helpful.

3.2. Condition assessment concepts

The key lies in a better understanding of what it means to say that happiness is a matter of a person’s emotional condition, as the emotional state theory was partly defended by adverting to commonsense intuitions about what impacts a person’s emotional condition. (The view might thus be more aptly called an “emotional condition” theory of happiness, but that term has unhelpful connotations. It is plausible, at any rate, that a person’s emotional state is equivalent to that person’s emotional condition.) It would be desirable to have an established theory in metaphysics on the idea of a condition, as opposed, say, to that of an event or state. But as far as I know there isn’t one, so I will sketch such an account here. As I mean only to shed light on the nature of a particular sort of condition, happiness, the discussion does not engage very deeply with work in metaphysics and, again, is meant to be only a starting point. But it will, I hope, hold sufficient interest for specialists in that field to warrant further development in the metaphysics literature.

Let’s return to the locution “happy and healthy,” with which we began. In earlier work I noted that this conjunction signals a likeness of kind between happiness and health. In particular, both appear to involve what we might call condition assessment concepts (CACs). Such concepts serve to assess how things stand with functionally organized systems. Are they disposed to function properly? How are they presently configured to function? As noted above, the concept of health seems largely or wholly to concern the individual’s disposition: a healthy person is able to function well, is not prone to develop serious problems like a heart attack, and so forth. While the exact contours of the notion of health are much disputed, it is at least strained to regard, say, an episode of indigestion, or soreness or swelling from a stubbed toe as a decline in health. In fact, indigestion or inflammation may indicate good health, being healthy responses to the circumstances.[10] These points being noted, I am happy to allow that this notion of health might just be one among others given the diversity of practical concerns that arise in healthcare contexts. A healthy pain response might still merit treatment to ease the patient’s suffering.

Used-car buyers and sellers routinely speak explicitly of a car’s condition, here employing a CAC appropriate to automobiles: how do things stand with the car? Is it able to serve the purposes for which such vehicles are built? (At least, is it able to serve these purposes relative to the baseline set by nondefective new instances of the model? There’s only so much you could expect from a ’70s-era Peugeot.[11]) Can one depend on the engine, the transmission, and other parts to work properly? Cars also serve aesthetic purposes, so the body, paint, and interior should be relatively presentable if it is to be able to serve those ends. Note that not all car problems speak to the car’s condition: if the engine is knocking due to bad fuel, that in no way bears on whether the car is in excellent condition; if the engine is generally prone to knocking, that’s a reason to downgrade its condition.

With these initial examples in hand, I propose the following definition as a starting point doubtless needing refinement:

functional condition =df that part of the state of a system consisting of the variable dispositional properties that influence its functioning or the processes, parameter settings, or other states that ground those properties

Since “condition” can be used variously, it seems desirable to have a more precise term, hence “functional” condition, though I am not sure the modifier will prove entirely apt, say if there turns out to be a more general metaphysical category at issue. As the definition indicates, functional conditions are a kind of state. They may necessarily be composed of multiple states given the inherent complexity of functionally organized systems, just as a person’s health is. This is one reason it is often helpful to employ “emotional condition” rather than “emotional state”: the latter is ambiguous, varying between particular emotions, moods, and so forth and a person’s overall emotional condition.

Without getting too far into the deep waters of the event/state distinction, we might at least venture this much: conditions are not events; nor are they conjunctions of events. Intuitively, an event is a “happening” and essentially involves temporal structure. When a car engine knocks, that’s an event, a knocking, and these can naturally be counted (“It’s knocking right now, for the second time today”). When a car is in fair condition because its engine tends to knock, that’s not an event, and it isn’t readily dated or counted (“It’s fair-conditioning right now, for the second time today”).

Consider a more pertinent example: pleasures—that is, pleasant experiences—are events, and they can naturally be referred to in the imperfective aspect: “She was experiencing a thrill of pleasure.” (For how long, and when did it start and end?) To be in a certain condition—for instance, to be happy—is not merely for an event to be occurring: “She was happy” need have no temporal structure, such as extension over time or narrative form, and cannot similarly be rendered in the imperfective aspect without loss of meaning.[12] Note that “She was feeling happy” is not at all equivalent, since one can feel happy without being happy, and this is a core feature of the emotional state view. Again, I do not wish to venture too deeply into weeds best navigated by metaphysicians, but it should be plain enough that conditions are not events.[13]

In earlier work I suggested that hedonism about happiness involves a category mistake (Haybron, 2001, 2008b). Even without all the details sorted out, we can now see why: hedonism mistakes a functional condition (the condition of being happy) for a mere conjunction of events (a series of pleasant experiences). Emotional state and life satisfaction theories at least are talking, at some level, about the right sort of thing—roughly, one’s emotional versus cognitive orientation toward one’s life, which essentially includes how one is disposed to respond, emotionally or cognitively, to things.[14] If this is right, then hedonism isn’t merely false; it’s about the wrong kind of thing altogether.[15]

3.3. Kinds of functional conditions

The examples of health and cars illustrate one sort of condition: what we might call the functional soundness of a system—namely, whether it is disposed to function properly. But many functional systems introduce a further type of condition owing to the fact that they can reconfigure themselves to function differently depending on the circumstances. I will call this, for reasons that will soon be clearer, a system’s operational status. Happiness—more generally, a person’s emotional condition—has to do with a person’s operational status, emotionally speaking. That’s not entirely intuitive, so let’s work our way to that idea, starting with simpler examples.

Again, a car. Many vehicles with four-wheel drive can operate in two or more modes: 2- or 4-wheel drive. Depending on which drive mode you’ve selected, the engine will engage the wheels differently. The condition of the system varies depending on the drive mode: it is disposed to function differently in one mode versus another. Admittedly, the term “condition” is not typically used in this context, though that may reflect the well-entrenched employment of that term to talk about the basic condition of a vehicle—whether it is in good or poor condition, and so forth. As well, it is easy enough to speak directly of the car’s “drive mode.”

The language of conditions becomes more apt when we turn to systems that, like human beings, are equipped with what amounts to evaluative mechanisms, so that the system’s configuration varies with the quality of the conditions it faces: is it configured for good conditions, bad conditions, or…? To illustrate, let’s consider a simplified analogue of a person, a robotic security system whose “emotional” repertoire consists entirely of varying degrees of anxiety (not entirely unlike the robot, Robot, from the television series, Lost in Space; so, let’s name it Robot).[16] Its job is to protect a warehouse storing valuable goods from major threats like burglars and minor ones like pests. As it scans the environment over a typical night, it detects the occasional cockroach or other insect or sometimes a misidentified piece of litter, and scoops it up or dispatches it with a laser. A single insect is a trivial matter: an utterly pedestrian negative that gives no reason for concern about the overall situation. It thus merits nothing more than a quick, focused point response to deal with it; the robot does so and resumes its business as before, at its modest baseline level of vigilance (intense monitoring consumes limited battery power and raises the risk of false alarms). Call this a Type-1 response: a transient response to a specific item or event.[17]

Suppose Robot detects a break-in. This is a bad situation requiring a broad-based response: sound the alarm! Alert the police, lock down anything that isn’t already secured, and try to frighten off or capture the intruder or intruders. The robot shifts from its baseline low-alert (“relaxed”) to high-alert (“high anxiety”) mode, configured to bring all its resources to bear to protect the warehouse. It scans the environment intently, and small anomalies that might normally be ignored are now treated as threats. Robot remains on high alert for some time, even after police have come and gone. But gradually it shifts to lower levels of readiness until finally resuming its baseline low-alert mode, where it remains until another intrusion is detected. Call this a Type-2 response: a broad-based, sustained response to the quality of the present situation.

There is more to the story. (Let’s assume that pests can be controlled individually and never warrant high alert.) This robot monitors not only the immediate situation, but also the general threat level it is facing these days. If there’s been a rash of intrusions in recent weeks, that suggests that the environment is generally hostile these days. Remaining on elevated alert is costly and increases false alarms, so that isn’t practical. Instead, the robot can adjust the gain on its mechanisms for detecting threats—and/or determining how to respond to apparent threats—so that the mechanisms are more sensitive: it takes less to put the system into high-alert mode. Again, though, this is not an optimal configuration when risks are low, so if things have been quiet for some time, the robot dials down the gain, reducing its propensity to go into high-alert mode, to sound the alarm, and so forth. We’ll call this a Type-3 response: altering the system’s preparedness to deal with concerns that may not be live at the moment, but are apt to become so given the general circumstances facing the system.

Security personnel for the warehouse may periodically inquire about the robot’s status—in the above terminology, its operational status—to see whether anything is amiss. “What is its condition?” Is it on high alert—“condition red”—indicating a possible break-in? Is it on heightened readiness, suggesting that there may be a higher-than-usual risk of break-ins or a possible vulnerability needing attention?

This framework is quite general and might apply to a wide range of functionally organized systems. Military units, for instance, can instantiate all three types of response, including Type-3 responses, as in the “readiness condition” of a warship. When faced with peacetime conditions, a ship may operate in peacetime cruising condition; in wartime conditions, when hostilities are imminent, the condition may shift to general quarters, with all hands at battle stations, poised to engage in combat operations.

When the health or functional soundness of a system like Robot are not at issue, questions about its “condition” concern not its Type-1 responses, which are merely passing episodes, but its Type-2 and Type-3 responses. Both types of response involve the system’s disposition to function, but there is an important difference between them: Type-3 responses are purely dispositional, having to do with how the system is disposed to function, while Type-2 responses are also occurrent, involving its functioning at the time as well as its dispositions. It “feels” anxious, so to speak, and this anxiety primes it to respond appropriately to its environment, since it is more prone to assess things negatively, to notice threats, and so forth. It is operating in anxiety mode. As the participle “operating” signals, this aspect of its condition is event-like even if it essentially has dispositional components and so is not merely an event. The system’s operational status, then, has two aspects—regarding Type-3 responses, what we may call its readiness condition and, regarding Type-2 responses, its operating mode.[18] When the robot is sounding the alarm, then, we can say its operating mode is “high alert”; when it is set to high sensitivity to threats, its readiness condition is something along the lines of “primed for high alert.” In some cases, it may be operating in low-alert, relaxed mode, while primed for high alert. Let’s call this a “fragile” form of low-alert mode; whereas in “robust” low-alert mode, it is also set for low sensitivity to threats.

I have tried to keep terminology as close to familiar usage as possible, but as we are dealing with a number of distinctions that have not often been made explicit, the jargon is bound to be a chore to navigate. Where feasible, I will refer generically to a system’s “functional condition” to simplify the exposition. Before returning to more familiar shores—the case of human happiness—let’s briefly summarize the distinctions introduced so far:

  • Functional condition =df that part of the state of a system consisting of the variable dispositional properties that influence its functioning or the processes, parameter settings, or other states that ground those properties. This can take at least two forms:

    1. Functional soundness: the disposition of the system to function properly or otherwise

    2. Operational status: that aspect of a system’s condition that determines how it is presently configured to function

      1. Readiness condition: the system’s disposition to function in certain modes rather than in others

      2. Operating mode: the mode in which the system is currently operating, which (inter alia) disposes it to respond to situations in certain ways

Again, for Robot: its functional soundness amounts to its “health”—what someone looking to buy it would want to know about it; is it in good or poor condition? But when deploying it as a security device, using it partly to inform us about how things are going in its environment, we are typically interested in its operational status, which tells us what alert mode it is in (its Type-2 states or operating mode) or what alert mode it is disposed to assume (its Type-3 states or readiness condition). When it both is on low alert and is disposed to remain on low alert—when it is robustly on low alert—then the robot is, in its sad and uninspiring way, happy. Such is the life of Robot.

3.4. Applying the framework to happiness

The terminology is a bit complex, but Robot is not, particularly. No great feat of engineering is required to build such a device, and indeed many examples of similar designs must exist in fact, because this sort of functional organization is an obvious way to build systems that need to respond adaptively depending on the quality of the circumstances they face. It would be very odd were human psychology not able to operate with at least the sophistication of our humble robot.

Even casual observation of human life makes clear that it does. If we return to our primary concern, happiness, the contours of the emotional state theory should already be apparent in the robot example. But it will help to give them greater definition. The distinction between Type-1 and Type-2/3 responses in the robot corresponds to the distinction between “mere pleasures” and happiness:

  • Type-1 (point responses): peripheral affective states (mere sensory and notional pleasures and unpleasures)

  • Type-2 (operating mode): central affective states (moods and emotions)

  • Type-3 (readiness condition): mood propensities

To be angry, for instance, is to be operating in “angry mode,” whereas a different but related question is whether one is poised to switch into angry mode, even when not already angry (say, because one is irritable or prone to irritability), whereas to be in a serene mood is to be operating in “serene mode,” and so forth. To be enjoying a cracker, by contrast, is just to be experiencing pleasure in the eating of a cracker, without this having a direct bearing on your functional condition. To be happy, according to this way of putting things, is to be operating broadly enough in positive versus negative modes and to be disposed to operate that way; it is for one’s operational status, emotionally speaking—that is, one’s emotional condition—to be broadly enough favourable.[19] At least, that is what robust happiness amounts to; one can possess a lesser, fragile kind of happiness when one is configured in a way that favours negative operating modes—that is, moods and emotions—while being lucky enough not to have them triggered. For instance, this might happen during a period of generalized anxiety disorder when one is not, at the moment, anxious, but feels relaxed.

While dispositionality is a hallmark of Type-2/3 states, as well as of happiness, recall that Type-2 states—central affective states—are substantially occurrent states, at least in the typical case. And here we can see the answer to our original puzzle, which was about how happiness could be essentially dispositional, but also partly occurrent: dispositions are often implemented by occurrent states. The hypervigilance characteristic of fear, for instance, is part and parcel of the occurrent state of being afraid. The feeling is (or can be) what makes you quicker to notice threats. If I annoy my cat, he may shift quite visibly into angry mode; that is, he becomes angry. One more false move and I’m liable to get zapped. Dispositions are essential to this story, but the most salient dispositions can’t enter the picture if nothing is actually happening: the animal must be conscious, operating in some fashion such that he is ready at any moment to zap me if I provoke him further. If happiness were a purely dispositional affair, its effects on our mental states and behaviour would be far remoter and weaker.[20]

This fact is what gives rise to one of the objections to life satisfaction theories of happiness: as a purely dispositional state—or at least as a state that requires little or nothing occurrent to obtain—it is fairly “causally inert” (Haybron, 2008b). Whether you have a favourable or unfavourable opinion of your life, for instance, may have no bearing on anything if that comes to light only when you are prompted to offer a judgment. Emotions and moods, by contrast, involve a lot of internal activity, enabling them to play a profound and pervasive role in regulating one’s other inner states and outward behaviour. Likewise, when Robot sounds the alarm, there is far more going on than merely setting parameters. Robot need not even be functioning for its parameters to be changed, just as you might switch a truck into four-wheel-drive mode when the engine is off. This is why I have claimed that a person can be happy, in a reduced and fragile way, even with a negative mood propensity: one’s actual moods and emotions do the lion’s share of determining how one is presently disposed to function. Mood propensity is part of happiness, but a secondary part. Ironically, there is a way in which this view challenges the idea that happiness is a state of passivity: happiness is not just dispositional, nor just a passive response to one’s circumstances. It also consists in occurrent states or processes of emotional functioning, with the organism adaptively reconfiguring itself to better meet the demands of its situation. This emotional activity underwrites various dispositions, but is no less a form of activity for all that.

To see more clearly the distinctions in question, it may help to imagine a simplified state space representation of one dimension of a person’s emotional condition: the level of anxiety. The possible states range in a line from zero anxiety to extreme anxiety, and its current level of anxiety is represented by a point on that line—think of it as a rolling ball, resting atop the line—whereas its propensity to varying levels of anxiety is represented by valleys and hills along that line, the valleys being attractors: states toward which the system tends to gravitate; the deeper the valley, the stronger the disposition, and the wider it is, the larger the range of initial states that tend to lead to it. In a case of generalized anxiety disorder, there may be a very wide, deep valley in the region of high anxiety. But at the moment, let’s suppose, the person is feeling pretty relaxed, having just gotten a massage. The ball is resting in the low-anxiety region of the line; this is close to the attractor basin (valley) for anxiety, without much of a hill separating it from the attractor, and so this relaxed mood is not a very stable state. But the mood has a bit of stability, and for the time being one is in “relaxed mode,” with the according propensities, which we can represent by imagining that the ball has a bit of weight and thus makes a small depression in the line: a little valley. Were this single dimension of anxiety the only aspect of the person’s emotional condition, we could say that the individual’s happiness is defined by the shape of this state space—the hills and valleys of the line—and the position of the ball. That is, it is defined by the person’s mood propensity and current mood.

I have left it open whether peripheral affective states involve the same mechanisms as central affective states, and I have left it open as well whether the central/peripheral distinction is merely one of degree. But for illustrative purposes, let’s assume that this anxiety system implements peripheral affects as well—fleeting concerns, say, as when one has a slightly unpleasant experience of noticing an ordinary insect. If the central/peripheral distinction is sharp, then such an experience will register as an effectively weightless dot on the line, with no impact on the line’s shape—that is, on one’s emotional condition. The dot comes and goes, and that’s it, as far as one’s emotional state is concerned. But little hangs on whether the central/peripheral distinction is sharp or merely a matter of degree, for in any case the theory requires degrees of centrality, with, for instance, profound anxiety being more central than shallow. In terms of our simplified model, the centrality of an affective state corresponds at least partly to the size of the valley it creates.

In arguing that happiness has this sort of structure, I am of course committing to various empirical claims. In earlier work I noted that my claims about the distinctive functions of different types of affect such as mood do appear to be consonant with relevant empirical findings.[21] But the basic framework is highly abstract and theoretical, a conceptual mapping of a sensible and plausible way for many systems, including creatures like us, to be constructed. The conceptual scheme can be implemented in countless ways, in robots, battleships—and, I’m suggesting, sapient apes like ourselves. For the most part the relevant empirical claims—for instance, that our propensities for different moods vary according to the quality of our circumstances—are readily verified from common experience. You don’t need a controlled experiment to figure that out or to know that it will hurt a great deal if you drop an anvil on your foot. There is indeed a great deal of empirical investigation to be done here, if only to determine how exactly these distinctions are implemented in human psychology; the present discussion only circumscribes our subject matter in loose terms. And certain aspects of the view doubtless involve nontrivial predictions that should be testable through scientific methods. But at this stage I think the more pressing questions have to do with the conceptual framework: whether it is coherent, whether it makes the right distinctions, and so forth.

4. HAPPINESS AS A FUNCTIONAL CONDITION: WHAT HAVE WE GAINED?

I hope by this point that the metaphysical status of happiness understood as emotional well-being, in terms of a person’s emotional condition, as well as its difference from happiness as conceived on a hedonistic theory, is reasonably clear. Doubtless more clarity could be brought to the general idea of a functional condition and of its various aspects, as well as to the precise relation between conditions and other states, as well as events. But it makes sense for human affect to be structured along the lines of this emotional state theory, and the dual character of happiness as a hybrid of occurrent and dispositional states should no longer seem ad hoc or chimerical. Certain kinds of functional systems, including human beings, can more adaptively respond to their environments if they are structured in this way, reconfiguring themselves on the fly at multiple levels as conditions demand. It would certainly make no sense for a person to have a fixed trait-level propensity for anxiety, anger, or cheerfulness, no matter what the circumstances. If you’re struggling to get by in war-torn Syria, anxiety should probably come more easily than if you’ve been frolicking for months on a South Seas beach.

But what do we gain by distinguishing the idea of a functional condition or by thinking about happiness in terms of a person’s emotional condition? One benefit, noted earlier, is that happiness, thus understood, enables us to predict and explain many things, such as that a friend is likely to be a pleasanter companion this evening than usual. Our example of Robot illustrates another: such a system’s condition offers an efficient means of gauging how well or badly things are going in the domain the system is meant to evaluate. Security personnel wanting to assess the facility’s threat level might do no better than to check Robot’s condition or status, and indeed that’s an explicit purpose of the system: to summarize and communicate how secure the warehouse is. For human beings, our emotional conditions do the same, but they add to this many other kinds of information about how well or badly things are going for the individual. In earlier work I suggested that happiness serves as a proxy for well-being: a rough and defeasible indicator of how the person is doing. Rather than just being a security indicator, emotional well-being amounts to an “affective welfarometer” that offers a broad gauge of the individual’s well-being (Haybron, 2008b).[22]

It is insufficiently appreciated just how efficient an indicator this sort of condition can be. Suppose our emotional lives were exhausted by the flow of affective experience, so that we only had Type-1 affective responses to our lives. How would you assess how your child or partner is doing these days? You can’t watch them all the time, and you have limited information about their objective situation and, more importantly, the way in which it affects their experience as far as that’s revealed through smiles, frowns, and the like as the day goes on—likely to be an unrepresentative sample as well, since they may tend to feel better (or worse) than usual in your presence. If they get very angry about what seems like a minor event, you can infer nothing other than that they really don’t react well to that sort of thing, or else that their mind was really on something else. It was a bad moment, that’s all. To figure out how they’re doing these days on the basis of this sort of information would require a lot of surveillance—tiring for you, and annoyingly intrusive for them.

But that’s not at all how things work in real life. If your friend blows up over a small matter that appears to be the sort of thing that they normally shrug off with ease, that’s an informative observation: something is amiss. She’s in a bad mood and may thus be having a bad day. Or, if you’ve seen several episodes like this recently, there may be a broader problem: something is probably off in her life these days, and she’s gone into “hostile environment” or “DEFCON 2” mode, as it were, primed to deal with bad events. She’s not happy, and this signals that she’s likely not doing well. More to the point, you were able to make this inference based on very little, easily gleaned information. There is no need for intense scrutiny of her myriad, mostly unobservable feelings over days, nor to compute the integral of such observations. Indeed, only one observation might be needed, say, if she had just burst into tears out of the blue. Similarly, if instead she’d uncharacteristically broken out into song or played a practical joke a few times in the last couple of days, you’d have reason to suppose she’s happy, and that things are going pretty well for her. In essence, an evaluation of how her life is going is encoded in her emotional condition, which in turn disposes her to react to events in certain ways. From such reactions you are able, with little effort and without neuroimaging gear from the future, to make reasonably accurate judgments about her well-being.

“Reasonably accurate”: the claim is not that happiness perfectly tracks well-being.[23] I won’t rehash previous arguments for thinking happiness serves as a proxy for well-being, by, for instance, rebutting dubious “set point” and “happy slave” claims regarding extreme adaptation (Haybron, 2008b). To be sure, there’s no reason to expect our emotional conditions to reflect chronic background conditions in our lives that should have no bearing on our behaviour or functioning. From a biological standpoint, this would just be a waste of resources. The death of a partner, the failure of a career, or unjust societal conditions might plausibly be thought to affect well-being in lasting ways even when the misfortune has no emotional impact. Such examples help to motivate my own view that happiness is just one aspect of well-being, which also includes value fulfillment: in these cases, important values are being frustrated even if the harm is no longer salient on a daily basis (Haybron, 2022). But even if the loss of a spouse or career leaves a permanent mark, diminishing one’s well-being, it seems entirely compatible with doing well on the whole, indeed thriving, just as a disability might diminish well-being without at all precluding flourishing.

The fact that our emotional conditions should not be expected to track closely all aspects of well-being may not be any great concession. If the three-dimensional framework I’ve suggested for emotional well-being is correct, for instance, then emotional well-being responds to three broad sorts of welfare concerns: security (attunement), opportunity (engagement), and success (endorsement) (Haybron, 2008b). It is not clear that any major domain of well-being, at least of a sort that might be anything like a consensus point, is left out of this framework. Even if emotional well-being doesn’t perfectly track well-being in, for instance, ceasing to register long-ago successes and failures, it is not clear that a great deal is left out in most cases.

Here are some external conditions that do seem quite central to well-being, so that if one is doing badly in these areas, one is very likely doing badly, period, however one feels: problems with or relating to relationships, such as a bad marriage or a child in crisis, or problems with one’s main occupation in life, be it work or something else. Intuitively, things like a bad marriage or a bad job—or a lack of any meaningful way of passing one’s time at all—tend to be incompatible with thriving or well-being. But these are also things that are strongly associated with unhappiness (Haybron, 2013a). And if someone manages to be genuinely happy despite having a bad job, then perhaps the work issue isn’t such a big deal—the boss is a jerk, but it’s just a job and it pays the bills, and so on. Likewise for a marriage: if the partners are happy, then perhaps they aren’t actually doing badly in their lives. (There might well be a specifically moral problem, say, if the woman is subservient to the man. While there is no reason to think that people living in oppressive circumstances tend to lead particularly fulfilling lives, there’s also little reason to address such cases as being mainly welfare problems.[24])

It may be helpful to recall that welfare hedonism is not in fact a nonstarter.[25] Like the other major theories, it seems at least to get roughly the right verdict in most cases, which is why people take it seriously. In general, people don’t adapt so thoroughly to their life conditions that their enjoyment of life is only loosely connected to how well they’re doing. This is both intuitively obvious and amply supported by the data, which reveal vast differences in subjective well-being across different life circumstances.[26] The contrary view was only taken seriously during a brief period a couple of decades ago when a number of researchers went on sort of an intellectual bender, suggesting against all experience with things like marriages and jobs that happiness is largely immutable. Stranger things have happened in the academy, like behaviourism, but that one was up there. Note, by way of illustration, that it took a wild thought experiment involving an experience machine to pose a really serious counterexample to hedonism (Nozick, 1974). If happiness were really only loosely connected to well-being, one shouldn’t have to work so hard to come up with counterexamples.

If technology advances sufficiently somehow to allow us to thrive emotionally in lives that are completely decoupled from reality, and if enough people derive their happiness from such sources, then we might have a problem using happiness as a welfare gauge. Until that time it remains plausible that happiness generally tracks well-being well enough to be a rough proxy for it: if someone is happy, they’re probably doing well; if unhappy, badly.

While our emotional conditions serve important informational functions, that of course is not their sole or even main purpose, which is broadly to regulate our functioning so that it is appropriate to our goals and circumstances. I believe that this point has ramifications for our understanding of human agency, and that our emotional conditions reflect the character of the self, implementing a person-level, though not quite rational, form of regulation.[27] This suggestion requires considerable elaboration, however, and will need to be developed elsewhere.

5. CONCLUSION

5.1 Summary

We began with a puzzle for emotional state theories of happiness, at least of the sort I defend: what is the justification for including dispositional states, including purely dispositional states like mood propensities, in an account of happiness? Heightening the worry is the notion that happiness also includes occurrent states. These questions are all the more pressing when the view is contrasted with an older account, hedonism, that seems traditionally to have been confused with it. The significance of pleasure is precisely that it is not dispositional. But, to begin with, this is the wrong contrast: in contemporary thought, both commonsense and academic, the most prominent understanding of happiness has arguably been the life satisfaction theory, which has an entirely different structure from hedonism.[28] Indeed, life satisfaction may well be entirely dispositional. To be happy, on that view, is to take a certain stance toward one’s life—to be disposed toward it in certain ways.

Once the emotional state theory is distinguished from hedonism, so that we can see how happiness might both be a matter of affect and have dispositional aspects, it becomes evident that hedonism is the outlier: it is odd to call someone happy simply on the grounds that a pleasant series of experiences has passed through that person’s mind lately. That makes happiness too nearly something that happens to a person—a mere agglomeration of experiential events—rather than a genuine state of the person. On reflection, it seems that hedonism isn’t even a plausible candidate theory of happiness. As I suggested in earlier work, it is guilty of a category mistake. But the exact nature of that mistake was left unclear: just what categories are being confused?

This paper has ventured an answer: the categories of events and conditions, specifically functional conditions. Whereas hedonism reduces happiness to a series of events, the emotional state theory, like the life satisfaction account, identifies happiness with an aspect of a person’s functional condition. As such, the concept of happiness belongs to a family of condition assessment concepts that we use to assess how a functional system is configured to function, with a robotic security system employed to illustrate. While the notion of a functional condition is in great part dispositional, it very often—almost invariably in the case of happiness—consists substantially in occurrent states, as the relevant dispositions are so often implemented by occurrent states. When angry, for instance, you are functioning in “angry mode,” so that the way you feel grounds various dispositions to respond, usually not favourably, to things in your environment. Understood this way, happiness consists both in dispositions and in occurrent functioning.

A great deal is lost if we do not employ—at least tacitly—condition assessment concepts in thinking about well-being, and certainly in thinking about our emotional lives. To assess a person’s well-being, we would be left in the hedonist’s predicament, attempting to infer from a hopefully representative-enough sample of observed emotions and moods, taken as piecemeal responses to the flow of events in the person’s life, how it all adds up. Even if we master this computational task, we still are given little sense of the macro-level picture regarding how the person’s life is going—the sort of information one might hope to glean from a life satisfaction judgment, for instance. The emotional information is essentially robbed of context—which might be important in its own right, but is all the more crucial to know insofar as much of the emotional story is not readily observed, say, because the depressed friend usually hides her distress behind an obligatory smile. We are left with a misleadingly shallow and fragmented picture of persons’ emotional lives—just one damned thing after another—so that we are bound to underestimate the importance of emotional well-being for human functioning and flourishing.

Finally, I have suggested that our ontology needs supplementation with the category of functional conditions, which among other things allows us more clearly to differentiate an intuitive understanding of a system or person’s state from that of an event, even if in broad terms the event-state distinction is not entirely firm. Given how many of our concerns relate to the way things stand with functionally organized systems, not least ourselves and other people, I expect that the concept of a functional condition will prove fruitful in a variety of contexts.

5.2 Implications for psychological science

I want to close by emphasizing the significance of the gap in psychologists’ understanding of the emotional realm, which was noted in earlier work but should now be even more clearly problematic. Contemporary psychology relies on a crude and implausibly sharp distinction between traits and states, where states are wholly occurrent, along the lines of Type-1 responses. Insofar as happiness has dispositional aspects, then, it must be a trait construct. But that is false even to the standard means of operationalizing mental-health constructs like depression and anxiety, which are typically measured with surveys including dispositional items about one’s “ability to” laugh, concentrate, sleep, and so forth. Without a category of functional conditions, psychological science lacks the vocabulary even to state what depression, happiness, and other emotional conditions are.

One result of this omission is that emotional well-being measures in well-being research—indulging for the moment the surpassingly weird convention of treating mental-health research as about something other than well-being—invariably do nothing more than inventory the frequency and perhaps intensity of various feelings individuals have experienced during some time frame, as if Type-1 states exhausted the emotional realm. Outside the mental-health paradigm, one seems to find dispositional items only in trait measures, for instance, in personality scales. Yet some such items do not appear to concern traits or personality at all, but rather emotional well-being. The popular International Personality Item Pool (IPIP), for instance, has a suggested fifty-item variant, starting with the unpromising prompt: “Describe yourself as you generally are now, not as you wish to be in the future.”[29] Even before seeing item one, we can surmise that “as you generally are now” is liable to include your response to present conditions, which may well be unrepresentative (perhaps your home was recently incinerated by a wildfire, and this has gotten your goat for some weeks now). If we take just the first four items, the problem should be apparent enough:

  1. Am the life of the party

  2. Feel little concern for others

  3. Am always prepared

  4. Get stressed out easily

It takes little imagination to see how someone who has become uncharacteristically downhearted in the wake of her house burning down might score poorly on these items, though normally she would give the opposite sorts of responses. These items convey good information about one’s present emotional disposition and might well constitute part of a good happiness measure. But they tell us little about one’s personality.

A good deal of psychological research, in short, is hamstrung by a primitive state/trait distinction that assimilates dispositionality to the realm of traits. It can’t even handle a simple and obvious cognitive architecture like Robot’s, let alone a human being’s. The ability to reconfigure ourselves according to the conditions is crucial to how we adapt and navigate the world, and a psychology that can’t cope with this very basic aspect of human nature is missing quite a lot. Psychology can’t do its work without something like the notion of a functional condition. That it has gotten so far without one may owe to the likelihood that you can ignore the dispositional elements of emotional well-being and still get reasonable-looking correlations among self-reports, since one’s emotional dispositions will tend to be associated with occurrent emotions, and self-reports of dispositions are not trivial to render and would presumably rely on recalled introspections of occurrent emotions. Likewise, people’s recent emotional histories probably correlate fairly well with their personalities; on a typical day, you’re liable to feel as you typically do. Moreover, “reasonable-looking correlations” is a pretty elastic notion: in science, lots of findings can seem reasonable, which perhaps makes instruments easier to validate than they should be. So long as one is content with a psychology that merely summarizes correlations among self-reports without attempting to model the underlying processes—as if one were to do cognitive science entirely by word-cloud analysis—the problem may not look so bad.[30] But it would be better to understand what’s going on under the hood, among other things because this might help us better interpret what people’s fallible self-reports, which may uniformly share similar weaknesses, are really telling us.

A more pressing issue is that well-being research may be misconceived so long as it fails to distinguish people’s traits and occurrent states from their conditions. Emotional well-being metrics, for instance, are sometimes dismissed even by subjective well-being researchers as concerned merely with momentary feelings that reflect the flux of daily events, but not the global picture regarding the things that matter in life, which is thought to be the domain of life satisfaction metrics. The annual World Happiness Reports, for instance, sideline emotional measures in favour of global life satisfaction—strictly, life evaluation—measures as the chief indicator of well-being. This appears to reflect multiple factors, but at least in conversation the putative transience of emotional states is often mentioned as a major reason for according them secondary status.

In fairness, operationalizing the distinction between occurrent and dispositional emotional states is not trivial. As just noted, emotional-well-being measures may yield similar results in practice with or without that distinction. Asking people about their disposition to feel sad, say, may not get you very different answers in large-scale surveys from what you would find if you simply asked them how often they felt sad.[31] But this should be noted as a limitation of the measures, and efforts should be made to overcome it. To their credit, mental-health researchers frequently include dispositional items like “found it difficult to relax” in their scales, but the empirical difference made by that format has not, I believe, been very well explored. Other measurement techniques like mood induction—how individuals respond to stressors, say—may also be useful.[32] But new measures are not necessarily required: in one time-use study, the effects of unemployment were assessed using common instruments, with life satisfaction measures finding a plausible reduction in scores among the unemployed; but a hedonic instrument assessing experiences showed little difference between the employed and the unemployed.[33] At the same time, the unemployed derived less pleasure from given activities than the employed. The apparent discrepancy plausibly has a dispositional explanation: unemployed individuals appear to be less happy, with poorer emotional conditions, than their employed counterparts, but compensate by allocating their time to more enjoyable activities, like video gaming rather than sitting in meetings. The example suggests a further avenue for studying the dispositional aspects of emotional well-being, but also illustrates how pleasure and happiness can diverge in ordinary life: the less happy might still have relatively pleasant experiences through judicious time allocations. One could imagine French citizens being less happy than their U.S. counterparts while enjoying a comparable quality of life, because more time is spent on things like sharing meals in France.

Interestingly, what makes emotional conditions informationally useful in everyday life—namely, that we can efficiently assess each other’s emotional responses to our lives from very limited observations—may also pose significant obstacles for the scientific study of happiness. You and I can deduce a great deal about a loved one’s inner emotional life from a single episode, even drawing on information that no one explicitly notices. But such delicate exercises of social-emotional intelligence are no small matter to reproduce in any kind of scientific instrument, let alone a large-scale survey. Be that as it may, it would be reassuring to see some acknowledgement that there is a distinction, as well as some evidence about how best to manage it operationally.

One might ask whether life satisfaction is in fact a better proxy for well-being than emotional well-being, at least in empirical contexts like the World Happiness Reports. Nothing argued here is incompatible with that: there could be more than one useful proxy for well-being, and perhaps the cognitive and affective metrics are complementary. As well, simple measures from vast samples across diverse cultures are needed for these sorts of studies, and perhaps life satisfaction measures would serve better for those purposes. Moreover, problems that might vitiate life satisfaction as a proxy in daily life—for instance, the fact that they are so easily gamed and prone to rationalization and hence of dubious value in the individual case—may tend to wash out over large samples. The epistemic demands of scientific research and everyday human life are quite different. So, it is certainly an open question whether emotional-well-being measures should be treated as a proxy for well-being in data-driven policy contexts, say.

But I do not think we should be too concessive just yet. There is some evidence, for instance, that the affect items used in the Gallup World Poll, which informs the World Happiness Reports, may in fact track well-being better than the life satisfaction measures. For instance Diener and Ng found that the emotion questions were better predicted by “psychosocial prosperity” items, whereas life satisfaction more closely tracked material prosperity—things like income and possessions (Diener, 2010).[34] Psychosocial prosperity means things like good relationships and jobs, noted earlier as among the universally acknowledged nonnegotiables of well-being. The importance of money and stuff for well-being, by contrast, is eminently negotiable, and you might prefer well-being measures that track the obviously important stuff more closely than the dubiously important stuff. More recently, life satisfaction measures astonishingly showed little or no decline in most studies after the onset of the COVID-19 pandemic—possibly a textbook case of adaptive preference-like processes—though emotional-well-being instruments focused on mental health revealed a far grimmer, and more believable, picture of well-being trends, with skyrocketing symptoms of anxiety and depression.[35] These of course are just small slices of a very large and complex picture, and again, for current purposes, nothing hangs on whether emotional well-being measures offer the best snapshot of overall well-being. Measurement introduces all sorts of problems that don’t concern us here. The more important point is that our picture of human well-being needs a richer understanding of the how a person’s emotional life hangs together. It’s something deeper than a string of feeling episodes.