Journal of Practical Ethics

Measurements of self-reported happiness are taken increasingly seriously by psychologists, sociologists and (more recently) by economists. They form part of the official statistics of many nations. Yet they remain beset by methodological problems. Some of these are superficial. For instance, people’s satisfaction with their life as a whole can be significantly influenced by trivial recent events, such as finding a dime (See Schwarz and Strack, 1999, p. 62). This kind of “noise” can be eliminated by good survey design. Other problems are more intractable. It has been said, for instance, that studies claiming to show that English people are happier than Poles reveal nothing more than the fact that the English word “happy” is used more freely and lightly than its Polish counterpart (See Wierzbicka, 2004). Disentangling such semantic effects from real differences of happiness is on-going problem for compilers of international happiness statistics.

Interesting and important though these quandaries are, I am here going to set them aside in favour of a philosophically more basic question: do people know how happy they are? If the answer to this question is negative, the whole project of measuring happiness by means of self-reports is in jeopardy.

“Do people know how happy they are?” could mean one of two things. It could mean, “Are people infallible judges of their own happiness?” Or it could mean, “Are people in general reliable judges of their own happiness?” Taken in the first sense, I argue that the answer to the question is “no”. Taken in the second sense, I argue that it is probably “yes”. However, this is not the unequivocal vindication of the survey method that it might at first appear, for the very claim that people are reliable judges of their own happiness implies that we already have a measure of how happy they are, independent of what they tell us. Happiness surveys may help us extend and refine this prior measure, but they cannot, on pain of unintelligibility, supplant it altogether.

Are people authoritative judges of their own happiness?

It is often asserted that individuals are the ultimate arbiters of whether or not they are happy. “If people say they are happy then they are happy,” writes Michael Argyle in a foundational textbook on the psychology of happiness. “If people say they are depressed then they are depressed.” (Argyle, 1987, p. 2) In a similar vein, psychologist David G. Myers has written, “by definition, the final judge of someone’s subjective well-being is whomever lives inside that person’s skin. ‘If you feel happy,’ noted Jonathan Freedman … ‘you are happy – that’s all we mean by the term’”(Myers, 2000, p. 57).

Not all psychologists are so confident. Daniel Kahneman has famously claimed that I can be mistaken about my overall happiness level. But this is only because he believes that my overall happiness in a period is a function of my happiness at each moment in that period, and that I can fail to recall this latter accurately. His doubt, in other words, concerns memory. He does not think that I can be mistaken about how happy I am right now. Hence Kahneman has suggested that we can get a more accurate measure of people’s overall happiness level by asking them how happy they are hour-by-hour over a several week period and integrating the results (Kahnemann, 1999).

Whence this trust in first-person happiness reports? The answer, I suspect, is a certain picture of the mind familiar to philosophers from Descartes and many others following him. Mental states, on this picture, are indubitable, meaning that if one experiences them one cannot doubt that one experiences them. The mind is transparent to itself. This picture is a thoroughly misleading one. Indubitability is characteristic of some, but not all mental states. One cannot doubt that one is in pain or seeing red. But one can, notoriously, doubt whether one is in love or believes in God. One can also, I shall argue, doubt (and be mistaken about) whether or not one is happy.

In a sense, we all know this already. It is a point of common knowledge that many people, particularly young people, fancy themselves to be deeply unhappy when they are not really unhappy at all. The difficulty is making sense of this commonplace thought. What kind of thing must happiness be if it is possible to be mistaken about whether one is happy?

A useful starting point, sanctioned by both philosophical tradition and ordinary usage, is to think about happiness as an emotion. Happiness and cognate states such as joy and gladness have been reckoned among the “passions” by philosophers since Aquinas; today we would naturally call them “emotions” or “feelings”. Some philosophers use the word “happiness” to translate eudaimonia, which is a condition of life rather than an emotion, but this semi-technical usage is presumably not what the designers and subjects of happiness surveys have in mind. Other philosophers define happiness as “satisfaction with one’s life as a whole”. This definition comes closer to ordinary usage and is embedded in the design of many happiness surveys. However, it is not inconsistent with thinking about happiness as an emotion. Emotions can, as I argue below, involve judgements and persist over many months or years. Besides, mere satisfaction with one’s life, without any accompanying feelings – a “cold-blooded and dispassionate judicial sentence”, as William James put it – is unrecognisable as happiness (James, 1884, p. 194). So in what follows, I shall respect ordinary usage and treat happiness as an emotion.

Emotions have three features that make them vulnerable to introspective error: they are dispositional, they are intentional, and they are publically manifest. Let me take these in turn.

Emotions are dispositional. To say that John is in love with Mary or jealous of his boss is not usually to say that he is currently feeling a certain way about Mary or his boss but rather that he is generally disposed to feel that way, and to act accordingly. An emotion, writes Peter Goldie, “involves dispositions, including dispositions to experience further emotional episodes, to have further thoughts and feelings, and to behave in certain ways” (Goldie, 2000, pp. 12-13). This obvious point is sometimes overlooked in the psychological literature on the emotions, where (perhaps for reasons of experimental convenience) emotions are often identified with short episodes of intense feeling. But we have no reason to accept this restriction.

Happiness and unhappiness also involve dispositions. The sentence “John is happy” can, it is true, sometimes be used to make a statement about John’s current state of mind. (“John is happy. He took Ecstasy an hour ago.”) But unless it is qualified in some such way, it is more naturally understood in a dispositional sense, as a statement about John’s standing tendency to feel and act in certain ways. It tells us that John is often in a good mood, that he smiles readily, that he is likely to confront misfortune without despair, and so forth. Participants in happiness surveys are usually asked how happy they are “in general” or “taking their lives as a whole”, implying that what is at issue is dispositional, not occurrent happiness.

The dispositionality of happiness is one reason why it, like other emotions, is not reliably accessible to introspection. For it is a well-known fact that people are often very bad judges of their own dispositions. They tend to ascribe a false permanence to their current feelings, forgetting how often they have felt differently in the past and how readily they may feel differently in the future. This is particularly true of the young, who have yet to learn the fickleness of human passions. Pushkin’s novel in verse, Eugene Onegin, provides us with a nice example. Olga feels genuinely and acutely unhappy over the death of her betrothed, Lensky, but we cannot call her deeply unhappy because we know – Pushkin tells us so – that her native cheerfulness will soon reassert itself. Presented with a happiness questionnaire, Olga might well respond in a way that a perceptive onlooker would regard as unduly pessimistic.

Some have denied that happiness is dispositional. As mentioned above, Daniel Kahneman holds that happiness in an interval is simply the sum of happiness at moments within that interval.1 This seems to me implausible. Let us suppose that Olga has a sister who is also in love with Lensky. A week after Lensky’s death, the two women are equally unhappy, yet whereas Olga will recover quickly her less flexible sister will remain grief-stricken for years. We would want to call Olga superficially, her sister deeply unhappy. Yet by hypothesis, the two women’s lives contain, to date, an equal share of happy and unhappy feelings. Therefore happiness is not purely occurrant.2

And even if happiness is occurrent, it is still vulnerable to introspective error, though from a different direction. On an occurrant understanding of happiness, the question “how happy are you in general?” can only mean “how happy have you been, moment by moment, within some past interval?” And answers to this latter question are open to the doubts raised by Kahneman about the reliability of emotional memory. These doubts might be assuaged in the way that Kahneman suggests, by monitoring happiness at regular intervals and summing the results, but the difficulties of this procedure are such that it has never been implemented on any scale. Besides, the mere act of reporting regularly upon one’s happiness might well be expected to have a depressing effect upon it. “Ask yourself whether you are happy, and you cease to be so” wrote John Stuart Mill (Mill, 1924, p. 120).3

Emotions are intentional. Most recent literature on the emotions has emphasised their intentionality – their directedness towards an object.4 Generally speaking, we don’t just feel frightened or angry; we feel frightened of x or angry about y. Where this object is a complex state of affairs, emotion furthermore implies belief – belief, at a minimum, that the state of affairs obtains. If I am angry that the local hospital is in a mess, I presumably believe that the local hospital is in a mess.

The conceptual connection between emotion and belief gives us yet another reason to doubt the authority of emotional self-reports. It is a familiar if puzzling fact that people can believe they believe things that they do not really believe: this is the phenomenon of insincerity or bad faith. And if beliefs can be insincere or in bad faith, emotions based on those beliefs can also be insincere or in bad faith. If we are sceptical of condemnations of private education from people who send their sons to Eton, we can also be sceptical when such people express outrage over private education. We needn’t deny that the “outrage” feels real to those experiencing it, or that it is accompanied by its usual behavioural manifestations – shrill voices, knotted brows etc. What makes it insincere is the absence of the connections that should normally exist between it and the general course of life. The case is not unlike that of the skilled actor who works himself up into a frenzy of indignation over some purely fictional wrongdoing.

Happiness, too, has an essential connection to beliefs about the world. Generally speaking, one is not just happy, but happy that such-and-such is the case. This understanding of happiness as intentional faces two apparent counter-examples. First, one can be “in a happy mood” without being happy about anything in particular. However, as Peter Goldie has convincingly argued, “a mood involves feeling towards an object just as much as does an emotion, although … what the feeling is directed towards will be less specific in the case of a mood.” (Goldie, 2000, p. 143) When I’m in a happy mood I’m not happy about this or that but about many things or things in general. I warm to the dull old gentleman on the bus; I forgive the insult I received this morning; I may even, if I’m metaphysically inclined, start looking on the world as intrinsically just and beautiful. “The world of the happy is quite another than that of the unhappy” wrote Wittgenstein in the Tractatus (Wittgenstein, 1922, 6.43).

Second, one can be said to be simply happy, without further qualification. “How’s Jane these days?” “She’s very happy.” Such “all-in-all” happiness looks, on the face of it, non-intentional. But it cannot really be so. If Jane is not happy about anything, unhappy about many things, and has no tendency to be in a happy mood, it makes no sense to call her happy. All-in-all happiness is logically tied to happiness about specific things or things in general, which is not to say that it can be derived from them by means of some algorithm.

If happiness is essentially grounded in beliefs about the world, it can share in the insincerity of those beliefs. This kind of insincerity is, I suspect, quite common. Think of the man who, as part of a positive thinking course, is required to repeat the mantra “every day in every way I’m getting a little bit better.” After a while, he comes to affirm this thought quite spontaneously, though in sober moments he acknowledges that it is not really true. He is happy that his life is getting better. He scores himself 8 out of 10 on happiness questionnaires. But is he really happy, if his considered belief is that his life is not getting better at all? It is at least plausible to suggest that the answer is “no”.

Emotions are publically manifest. If “fear” were the name of a purely private sensation, with only a contingent relation to public circumstances and behaviour, it would make good sense to ascribe it to a man who neither displays nor has cause to display fear. But such an ascription makes no sense. It merely suggests a misunderstanding of what the word “fear” means. The point is familiar from Wittgenstein. Words referring to inner states stand in need of outward criteria of application. We manifest understanding of the word “fear” by using it correctly, on the basis of our knowledge of the situations, gestures and actions that typically accompany fear. These situations, gestures and actions constitute the frame of reference within which alone “fear” has meaning.

One implication of this is that it is possible for an individual to be mistaken as to whether or not he feels an emotion. John may think and say that he respects his colleague Sarah, but if his behaviour towards her is not of a respectful kind – if he continually ignores her and puts her down, say – we might reasonably be doubtful. Perhaps John has managed to conceal from himself his real feeling, which is one of contempt. Were emotions purely phenomenal states, such an ascription would be puzzling – like ascribing to John a pain that he does not feel. But if they are criterially connected to outward behaviour, it is perfectly intelligible. People are often blind to the emotional tenor of their actions. We might even agree with Proust that “it is only with the passions of others that we are ever really familiar, and what we come to discover about our own can only be learned from them.” (Proust, 1992, p. 181)

Happiness, too, is criterially connected to public actions and circumstances, which gives us yet another reason to doubt the authority of first-person happiness reports. A woman who says that she is happy but whose actions and circumstances suggest otherwise is not self-evidently credible. Perhaps she is “in denial”. Perhaps she is a Stoic philosopher with an unusual understanding of happiness. Of course, an individual’s assertion that he is happy may be among the grounds for ascribing happiness to him: this is, after all, one common way in which happiness manifests itself. But it is not the only way. Where verbal and non-verbal manifestations of happiness conflict, only the particularities of the individual case can tell us which to trust.5

It is, then, simply not true that “if people say they are happy then they are happy”. Individuals are not authoritative judges of their own happiness. They can be mistaken.

Are people in general reliable judges of their own happiness?

But perhaps advocates of the survey method needn’t insist on the infallibility of happiness self-reports. All they need say is that such reports are, on average, accurate. Sure, there will be errors of optimism and pessimism here and there. But given a large enough sample, these errors will “wash out”.6

Supporters of this hypothesis take comfort from studies showing correlations between self-reported happiness and other measures associated with happiness. These measures are of three broad kinds: physiological, behavioural and circumstantial. On the physiological side, it has been shown that people who declare themselves happy tend also to have good immune systems and high levels of electrical activity in the left forebrain (see Layard, 2005, pp. 17-20). These look like significant correlations, but they invite the question: how do we know that such physiological indices themselves track happiness? The answer clearly cannot be that they track self-reported happiness, since that is the very thing in question. Such studies may strengthen our conviction that happiness self-reports latch onto something real, but they do not establish what this something is. For all we know, it might simply be a propensity to answer happiness questionnaires optimistically.

Other studies show a correlation between self-reported happiness and the actions and circumstances associated with happiness. Andrew Oswald and Stephen Wu have established a correlation between quality of life in U.S. states, as measured by sunshine hours, commuting times, crime figures etc., and the self-reported happiness of their inhabitants. (Oswald and Wu, 2010. Adjusting for age and wealth, New York comes out bottom on both counts.) Other studies have shown that people who rate themselves happier also tend to smile more frequently (see Diener and Suh, 1999, p. 437).

If non-verbal actions and circumstances are, as I have claimed, among the criteria of happiness, then these studies must indeed increase our confidence in happiness self-reports. But they also raise an important epistemological puzzle. Happiness self-reports are not, I have said, self-evidently authoritative. They require external validation. But in seeking such validation, do we not presuppose a measure of happiness independent of self-reports? How then can happiness surveys tell us anything new? Either they tally with prior estimates of human happiness, in which case they seem to be redundant, or else they do not, in which case they seem to be flawed. Their function, it appears, is essentially ceremonial: it is to bestow the blessings of social science on the deliverances of common sense.

This is too strong a conclusion.7 In science, the fact that measure A must be validated against measure B does not mean that it cannot in turn refine, extend and, on occasion, correct B. In fact, this is a common occurrence. Hasok Chang has written a fascinating book on the history of modern thermometry, showing how each new measure of temperature has had to be validated with reference to earlier, less sophisticated measures. We confront, he writes, “the paradoxical situation in which the derivative standard corrects the prior standard in which it is grounded” (Chang, 2004, p. 44). How is this possible?

Chang identifies two principles governing the advance of temperature measurement. The first is an “imperative of progress” (Chang, 2004, p. 44). Progress here has a number of aspects. A new thermometer can be superior to existing instruments in accuracy (it registers smaller intervals), in range (it measures temperatures above and below what was previously possible) or in reliability (it is less prone to error). A mercury thermometer is superior to a human hand – the primordial thermometer – in all three respects. It can register finer gradations of temperature. It can measure temperatures above the point at which a hand burns and below the point at which it goes numb. And it is not prone to the illusion that the same object is cold (to a hot hand) and hot (to a cold hand).

But alongside the imperative of progress Chang also posits a “principle of respect” (Chang, 2004, p. 43). This states that a new measure must, by and large, agree with the measure it replaces. Otherwise, we have no grounds for saying that it is measuring the same thing, or indeed anything at all. The principle of respect limits the degree to which a new measure can correct an old one. A thermometer may on occasion override the evidence of our unaided senses, but if it does so too often, it is no longer clear that it is measuring temperature.

Chang’s work provides a useful framework for thinking about the achievements and limits of happiness surveys. Like thermometers, happiness surveys lend mathematical precision to the rough and ready verdicts of intuition. Common sense tells us that health, love and honour all contribute to happiness, but it does not tell us how much they contribute. Surveys inform us that the effect of unemployment on happiness is 20 per cent greater than that of divorce (Layard, 2005, p. 64).8 They tell us that men adapt better to divorce than do women (Layard, 2005, p. 66). And they tell us that no one adapts well to the irritations of background noise (Frederick and Loewenstein, 1999, p. 311).

These are not banal findings. They tell us something new. Notice, though, that they do not contradict our prior understanding of what makes people happy. They merely refine it, in the same way that a thermometer with 0.1°C intervals refines one with 1°C intervals. Might happiness surveys do more than this? Might they not just refine but substantially revise our common-sense understanding of the conditions of happiness? I suspect not, because our only ground for trusting happiness surveys is that they by and large agree with common sense. Were they to disagree significantly, we would not trust them. We would suspect some computational error, or question the sincerity, self-knowledge or linguistic competence of the participants. Here Chang’s “principle of respect” comes into play. A new measure can contradict its predecessor only on occasion, as in the case of the thermometer versus the hot and cold hand. If it contradicts it across the board, we have no reason to accept it.

The problem is not just hypothetical. Sometimes happiness surveys do yield strikingly counter-intuitive results. An example comes from the aforementioned paper by Oswald and Wu. “Although it is natural to be guided by formal survey data”, they write (2010, p. 578),

it might be thought unusual that Louisiana – a state affected by Hurricane Katrina – comes so high in the state life-satisfaction league table. Various checks were done. It was found that Louisiana showed up strongly before Katrina and in a mental-health ranking done by Mental Health America and the Office Applied Studies of the U.S. Substance Abuse and Mental Health Services Administration …. Nonetheless, it is likely that Katrina altered the composition of this state – namely, those who were left were not a random sample of the population – so some caution in interpretation is called for about this state’s ranked position, and that position may repay future statistical investigation.

This is a revealing admission. While recognising that they should, in all consistency, “be guided by formal survey data”, Oswald and Wu allow themselves to be swayed by what they intuitively know about the effects of hurricanes on happiness. When it comes to the crunch, they question the data; they do not revise their views on what makes people happy.

But what if “future statistical investigation” proved the Louisiana survey to be representative after all? Should we conclude that natural disasters do not, contrary to widespread belief, detract from happiness? Not necessarily. We might first of all ask whether the Louisiana respondents were being honest, whether they had correctly understood the scale presented to them, and whether they were using the word “satisfied” with the same meaning as respondents elsewhere. We might also wonder – to return to the issue raised in the first half of this article – whether or not they knew their own minds. Ruling out these and other possibilities would require further, more detailed interviews, and perhaps also behavioural studies. We might in the end conclude that Katrina had indeed failed to dampen the good spirits of the citizens of Louisiana: this would be a case of a survey overturning prior expectations. But we would accept this conclusion only with some resistance, and only if accompanied by some intuitively plausible explanation of the psychological mechanism involved. (Perhaps disaster had awakened a spirit of community, as is said to have happened in London during the Blitz.)

Let me summarise. The criteria of happiness are, I have said, threefold: verbal, behavioural and circumstantial. All three categories are on a par, epistemologically speaking. There is no automatic presumption in favour of the first. Hence while happiness self-reports might occasionally be allowed to overrule the testimony of behaviour and circumstances, there is no necessity for them to do so. Indeed, Chang’s principle of respect implies that it is only occasionally that self-reports can be allowed to overrule the testimony of behaviour and circumstances. Were they to do so too often, they would simply cease to be credible. In short, nothing that surveys might tell us can upset our common-sense conviction that health, love, freedom, security and respect all standardly contribute to happiness.

My conclusion, then, is a qualified endorsement of the survey method. Happiness surveys can add numerical precision to our prior understanding of the causes of human happiness. They can arrange those causes in order of importance, perhaps even assign cardinal values to them. These are significant achievements. But we cannot expect surveys to fundamentally revise our prior understanding of what makes people happy and unhappy. It is comforting to know that modern statistics confirm Solomon’s dictum that a dinner of herbs with love is better than a stalled ox with hatred.9 But if they failed to confirm it, we would quite rightly side with Solomon, and not with the statistics.

Acknowledgements: I am grateful to the audience at the conference “Happiness and Well-Being” (Uehiro Centre, Oxford, 20-21 June, 2013), where I read an earlier version of this paper. I am particularly grateful to my colleague at Exeter University, Sabina Leonelli, who directed me to the work of Hasok Chang. I am also grateful to my two anonymous referees, whose useful criticisms of an earlier draft I have striven to address.

REFERENCES

Argyle, M., The Psychology of Happiness (London: Routledge, 1987).

Chang, H., Inventing Temperature: Measurement and Scientific Progress (Oxford: Oxford University Press, 2004).

Diener, E. and Eunkook M. S., “National Differences in Subjective Well-Being”, in D. Kahneman, E. Diener and N. Schwarz, Well-Being: The Foundations of Hedonistic Psychology (New York: Russell Sage, 1999), pp. 434-450.

Feldman, F., What is This Thing Called Happiness? (Oxford: Oxford University Press, 2010).

Frederick, S. and George L., “Hedonic Adaptation”, in D. Kahneman, E. Diener and N. Schwarz, Well-Being: The Foundations of Hedonistic Psychology (New York: Russell Sage, 1999).

Goldie, P., The Emotions: A Philosophical Exploration (Oxford: Clarendon Press, 2000).

Haybron, D. M., “Do We Know How Happy We Are? On Some Limits of Affective Introspection and Recall”, NOÜS 41:3 (2007), pp. 394-428.

——— The Pursuit of Unhappiness: The Elusive Psychology of Well-Being (Oxford: Oxford University Press, 2008).

Kahneman, D., “Objective Happiness”, in Well-Being: The Foundations of Hedonic Psychology, eds. D. Kahneman, E. Diener and N. Schwarz (New York: Russell Sage Foundation, 1999), pp. 3-25.

Kenny, A. and C. Kenny, Life, Liberty and the Pursuit of Utility (St. Andrews Studies in Philosophy and Public Affairs, 2006).

Layard, R., Happiness: Lessons from a New Science (London: Penguin, 2005).

Mill, J. S., Autobiography (London: Oxford University Press, 1924).

Myers, D. G., “The Funds, Friends, and Faith of Happy People”, American Psychologist 55/1 (2000), pp. 56-67.

Nussbaum, M., Upheavals of Thought: The Intelligence of Emotions (Cambridge: Cambridge University Press, 2001).

Oswald, A. J. and S. Wu, “Objective Confirmation of Subjective Measures of Human Well-Being: Evidence from the U.S.A.,” Science 327 (29 Jan. 2010), pp. 576-79.

Proust, M., Swan’s Way, trans. C. K. Scott Moncrieff and Terence Kilmartin (New York: Random House, 1992), p. 181.

Schwarz, N. and Fritz S., “Reports of Subjective Well-Being: Judgemental Processes and Their Methodological Implications”, in Well-Being: The Foundations of Hedonic Psychology, eds. D. Kahneman, E. Diener and N. Schwarz (New York: Russell Sage Foundation, 1999), p. 61-84.

Skidelsky, R. and Skidelsky, E., How Much is Enough: The Love of Money, and the Case for the Good Life (London: Allen Lane, 2012).

Solomon, R., The Passions: The Myth and Nature of Human Emotions (New York: Doubleday, 1984).

Wierzbicka, A., “’Happiness’ in Cross-Linguistic and Cross-Cultural Perspective”, Daedalus 133/2 (2004), pp. 34-43.

James, W., “What is an Emotion”, Mind 9 (1884), pp. 188-205.

Wittgenstein, L., Tractatus Logico-Philosophicus (London: Routledge and Kegan Paul, 1922)