II. ADAPTATION

The problem of adaptation has been introduced in the context of welfare measurement by Amartya Sen. Sen has stressed the importance of adaptation in considering the ‘utilitarian approach’,Footnote ¹⁸ or ‘happiness approach’ repeatedly in his work.Footnote ¹⁹ In On Ethics and Economics he formulates it as follows:

In more recent formulations, Sen focuses more specifically on the happiness scales used in SWB research: ‘the use of the happiness scale can be quite misleading if it leads to ignoring the significance of other deprivations that may not be at all well-judged in the scale of happiness’.Footnote ²¹

That these examples pose a serious threat to the SWB-WB view should be clear. If a person clearly has a low well-being, but nevertheless has a high reported happiness or satisfaction with life, reported happiness or satisfaction with life are bad measures of her well-being. However, Sen does not specify exactly why he believes this to be the case. On an axiological reading, the problem is that happiness and well-being do not correlate well even if we are perfectly able to judge our own happiness. On this reading happiness and well-being are simply not the same:

The Axiological Argument

1. (1) The happy deprived are genuinely happy.

2. (2) The happy deprived do not have a correspondingly high well-being.

3. (3) Hence, genuine happiness and well-being are distinct.

Those who endorse the Epistemic Argument do not accept the first premise of the Axiological Argument – the deprived in Sen's examples do not appear to be genuinely happy. While they agree that the people in Sen's example do not have a high degree of well-being, the reason for this is not (only) that happiness does not constitute well-being, but that under these conditions people are bad evaluators of their own happiness.

The Epistemic Argument

1. (1) The deprived are not happy.

2. (2) The deprived do report to be happy.

3. (3) Hence, reported happiness is a bad indicator of genuine happiness.

The Axiological Argument is essentially a philosophical argument about the nature of well-being. On this interpretation, the problem of adaptation is a problem for philosophical theories that identify well-being with happiness. In particular, both accounts of happiness we have discussed (pleasure or satisfaction with life) conceive of happiness in terms of mental states.Footnote ²² While mental-state accounts of well-being were prominent in the nineteenth century, the view lost popularity due to a number of prominent counter-examples that were formulated against it.Footnote ²³ Most notable among them is Robert Nozick's experience machine objection. Imagine that there would be a machine that would create a dream world for you if you plug in, in which you can experience the most incredible things.Footnote ²⁴ Your life would be highly pleasurable and you would feel deeply satisfied. Nevertheless, the objection goes, a life lived in an experience machine would not be considered a fully good life, and thus, while this life would be optimal in terms of happiness, it would not be so in terms of well-being.Footnote ²⁵

The view that adaptation is mostly a problem for the conceptual identification of well-being and happiness – just like Nozick's experience machine example – is not rare in the literature. Erik Angner, for instance, puts the two problems on a par: ‘Philosophers have traditionally made the point [that happiness is not well-being] by reference to thought experiments like Robert Nozick's experience machine,. . ., or evocative vignettes like Amartya Sen's descriptions of destitute beggars, landless labourers, overworked servants, and subjugated housewives.’Footnote ²⁶ Similar interpretations are given by many philosophers writing on the topic.Footnote ²⁷

While the Epistemic Argument is less widespread, there is some recognition that the Axiological Argument is not the only way of analysing what problem(s) adaptation poses to the SWB-WB view. In particular, Marc Chekola, in a wider discussion about happiness, observes that there are two different ways of looking at the problem:

While Chekola prefers the Axiological Argument, he acknowledges that it is not the only way to see the problem. There are also philosophers writing on the topic who emphasize the epistemic challenge that adaptation poses. In an article called ‘Do we know how happy we are?’ Daniel Haybron writes:

The conclusions of the two separate arguments can both be held at the same time – happiness may not identify with well-being, while at the same time, happiness and people's perceptions of happiness may diverge as well. However, it is also perfectly possible to accept one argument without accepting the other.Footnote ³⁰ The two problems for the SWB-WB view are thus analytically independent but compatible.

III. ADAPTATION AS AN AXIOLOGICAL PROBLEM

Is the problem of adaptation viewed from the perspective of the Axiological Argument convincing? While many philosophers commenting on happiness research take this perspective, I will argue that it is not. The Argument suffers from a number of weaknesses, but the most serious weakness is that the counter-example it is based on leaves much room for interpretations compatible with happiness-accounts of well-being; this makes the argument vacuous. Rather than showing that such views fail to be plausible in cases of adaptation, it presupposes it.

First, regardless of whether those defending a happiness-account of well-being should be convinced by the argument, it is important to point out that few happiness researchers have been affected much by the force of the Axiological Argument. Many researchers take measures of SWB to be a direct measure of general human well-being even if they acknowledge Sen's adaptation argument.Footnote ³¹ Perhaps this is due to a substantive disagreement with the argument, but that is not often spelled out. Rather, Sen's argument is often rejected on the basis of simply being mistaken empirically.Footnote ³² The poor do not regard themselves as happy, neither do the unemployed. This empirical argument avoids the more crucial conceptual point that philosophers have in mind.

There are at least two prima facie problems for the Axiological Argument. At least when it comes to the hedonic interpretation of happiness, it is not at all clear whether the first premise – that the happy deprived are genuinely happy – is plausible.Footnote ³³ The compatibility of deprivation and pleasure that is assumed is not at all obvious. First, we must observe that the deprived are highly likely to experience many disadvantages in their life which negatively affect their happiness: becoming ill, being overworked, the loss of loved ones who die too soon, etc. Even if they do adapt to their circumstances must they feel the strain of these disadvantages. The claim that the deprived can be happy only seems plausible if the standards of what it means to be happy are not set very high, in which case the conclusion of the argument loses support. A second prima facie problem with the first premise is that even if the deprived happy are genuinely happy when they report it, their lives as a whole may be less happy in the probable case that deprivation leads to a reduction in the length of their life. In short, being deprived – i.e. lacking necessities of life – does not go without consequences, even on happiness-accounts of well-being.

The point that it is conceptually possible to be deprived and still be genuinely happy is not undermined if these observations are correct. But scrutiny reveals that such conceptual possibilities may be more hypothetical than Sen makes it seem.

The second premise – that the happy deprived do not have a correspondingly high well-being – is also contentious, as it is based on intuitions that are very unlikely to be shared by someone who also has intuitions supporting a happiness account of well-being. Like the Experience Machine counter-example, the Axiological Argument uses the adaptation phenomenon as an intuition-pump. Just as the experience machine objection shows that, even from the perspective of the mental-state theorists, there is something counter-intuitive about mental-statism in certain cases, the Axiological Argument counts on the reader intuitively agreeing with the second premise. Even a fierce defender of mental-state accounts has to admit that there is something unsatisfactory about saying that a person in the experience machine lives a fully prudentially good life.Footnote ³⁴

But does the same apply to adaptation? If we start from the intuition that well-being consists in happiness, it is unlikely that we also have the intuition that a genuinely happy person living in deprivation is not leading a good life. In fact, the intuition that an adapted poor person does not lead a happy life may seem quite paternalistic to someone with intuitions supporting happiness-based accounts on well-being. For instance, Marc Fleurbaey and Didier Blanchet write:

The Axiological Argument does not demonstrate that the judgement that the happy deprived are leading lives with a high degree of well-being is counter-intuitive; it simply assumes that it is. In this sense, it is question begging. The intuition is not obvious and not independent of the view one holds on the nature of well-being. Someone defending a happiness-account of well-being may agree that the experience machine example is counter-intuitive for her view, but is unlikely to share the intuition that the happy deprived are not leading good lives.

If the deprived really do have the same quality of experience (and live equally long lives) it is not obvious that their lives are still worse in virtue of having lowered aspirations. There has been a wide array of schools of thought that have emphasized that the key to a good life is living a simple life, satisfying only a minimal amount of ambitions and desires, such as Buddhist, Epicurean and Stoic thought; but the idea arguably also captures a key Christian insight.Footnote ³⁶ It is not at all implausible that a reduction of desires and ambitions may genuinely improve a person's life by making her happier. In short, neither of the premises supporting the Axiological Argument is obvious, and taking them for granted begs the question.

There is a final, more pragmatic reason to resist the Axiological Argument in the context of well-being measurement, which does not concern the specifics of the Axiological Argument. The well-being debate has known a number of instances where intuitions of philosophers have gone in different directions. Despite Nozick's experience machine, there are still mental-statists,Footnote ³⁷ and despite well-known counter-examples to desire-satisfactionism (such as Rawl's grass-counterFootnote ³⁸ or Derek Parfit's stranger on the bus)Footnote ³⁹ there are still defenders who are willing to bite the bullet.Footnote ⁴⁰ Because intuitions clash, the debate is not easily resolved. Adaptation, as an Axiological Argument, suffers from the same fate. There is something unfruitful about the usage of intuition-pumps in this context: because intuitions clash, they are unlikely to resolve debates. Thus, regardless of whether the Axiological Argument is question begging or not, an argument that relies on contentious intuitions does not stand on firm ground.

IV. ADAPTATION AS AN EPISTEMIC PROBLEM

The epistemic version of the adaptation argument, as I will argue in this section, succeeds in providing a convincing argument that does not rely on contentious intuitions about well-being.

The idea that we are imperfectly able to evaluate our own happiness is actually a widespread idea.Footnote ⁴¹ There are a variety of ways in which the Epistemic Argument can be made precise. An important question that needs to be addressed is: how could one possibly be incorrect about one's own happiness? In fact, there are a number of ways in which adaptation may lead to a misjudgement of a psychological state. Below I briefly discuss these issues and then provide a thought experiment – borrowed from Marc Fleurbaey and Didier BlanchetFootnote ⁴² – to argue that adaptation poses a severe epistemic challenge to the SWB-WB view.

To appreciate the epistemic argument one must first consider how different happiness-accounts of well-being vary. Hedonism about happiness comes in various forms, each with different epistemic challenges. While some hedonists maintain that pleasure is a specific experience,Footnote ⁴³ othersFootnote ⁴⁴ take pleasure to be heterogeneous, with some types of pleasurable experience being of a higher quality than others.Footnote ⁴⁵ Life-satisfaction views are epistemically significantly different from these views because they take happiness to be a person's self-assessment of life, which seems more accessible from a researcher's perspective than pleasurability of experience.Footnote ⁴⁶ A different notion of happiness is Haybron's recently developed account on which happiness is an emotional state, which may or may not directly enter experience.Footnote ⁴⁷

What is required for the Epistemic Argument to be sound is a conceptual distinction between the judgement of one's own happiness and happiness itself. For Haybron's emotional state view this is most evidently the case. On Haybron's view stress, anxiety, and even depression, may (to some extent) go unnoticed by people even if these emotional states do constitute unhappiness.Footnote ⁴⁸ A person may judge herself to be happy, while she is under more stress than she realizes. For qualitative hedonists, like Mill, there is also quite clearly such a distinction.Footnote ⁴⁹ Mill famously argued that people who have only experienced low-quality pleasures believe that they are happy, while people who have experienced high-quality pleasures may be less satisfied (by comparison) but nevertheless are happier. Even monist hedonists draw the same distinction. Torbjörn Tännsjö, in an article defending monist hedonism, writes:

However, on the life-satisfaction view, the distinction between judgement of one's own happiness and happiness itself is not as obvious. Could someone judge her own satisfaction with life wrongly? In order to understand this question it is important to see the difference between content errors of judgements and scale errors of judgements. Footnote ⁵¹ Making a content error of judgement involves a failure to take into account a relevant part of the concept in a judgement of happiness. When someone misidentifies her own happiness, or misses out on an important part of happiness, she commits a content error. Making a scale error, on the other hand, means that the degree of happiness is not proportionately identified in relation to an intended scale. The concept of scale error is illustrated in cases of making a judgement about length. Because the concept of length is quite clear, it is not easily misidentified. However, in judging whether something is ‘tall’ or not, different people may have different views. A Filipino would most likely judge me – roughly 1.80 metres in height – to be tall, while for Dutch standards, I am actually quite small.Footnote ⁵² Different people may use different standards to make judgements of degree.

Content errors of judgement may be definitionally ruled out for the life-satisfaction conception of happiness, because it may be conceptually impossible to misidentify one's own satisfaction. Scale errors, however, are not. There is at least one way in which two people who report equal degrees of satisfaction may not in fact be equally happy, even on a life-satisfaction view of happiness. Just as in the case of length, people may hold different standards about the meaning of ‘very satisfied’, in particular about the meaning of ‘very’. A person who has experienced a large amount of satisfaction in life and is back at a normal level will most likely judge her satisfaction to be lower than someone who comes from a very dissatisfied state. So, even for life-satisfaction views – for which the possibility of content errors may be excluded – there is a conceptual distinction between actual happiness and happiness judgement, which is due to the possibility of scaling errors. Two people who equally judge themselves to be ‘very satisfied’ with their life may not be equally satisfied, if their interpretation of degree (in this case ‘very’) is not the same. By focusing on scaling errors, the Epistemic Argument can now be precisely formulated so that it applies to the variety of happiness conceptions that I discussed above:

1. (1) When someone undergoes adaptation, she changes the standards by which she evaluates her happiness.

2. (2) For this reason, people who have undergone adaptation apply different standards of evaluation to their happiness than people who have not undergone the same (degree of) adaptation.

3. (3) Reported happiness does not represent the degree of happiness well in comparisons, when two reports have resulted from different degrees of adaptation.

So far, I have argued that the Epistemic Argument is valid. But is it also sound? One example which illustrates its soundness is the debate about the Easterlin paradox. The paradox is based on findings which suggest that while the rich are reportedly happier than the poor within countries, and rich countries are happier average than poor countries, the increase in prosperity of the past four or five decades has not led to significant increases in happiness scores over time in western societies.Footnote ⁵³ Betsey Stevenson and Justin Wolfers for instance report the noted problem that SWB may not be informative of quality of life over time because it captures people's judgements of their ‘relative well-being’: ‘If individuals assess their life relative to contemporary standards, then as countries and the world grow richer, reported satisfaction may not change’.Footnote ⁵⁴ Marc Fleurbaey and Didier Blanchet provide us with the following thought experiment to illustrate this point:

There are two aspects of Fleurbaey and Blanchet's illustration that are particularly powerful. First, it not only shows that it is plausible that people do adapt their standards of evaluation, but it also illustrates how absurd it would be if people would not. It is almost unthinkable that Europeans would judge their own happiness by the same standards in the scenario where the United States maintains the standards of living of today, as they would in the scenario that Fleurbaey and Blanchet describe. In other words, it is almost unthinkable that people's benchmarks would not adapt, even though their lives do not change substantively. Second, Fleurbaey and Blanchet's suggestion that our standards adapt explains – or is compatible with – many of the behaviours of the SWB data. It explains why the poorest countries in the world also report the lowest levels of happiness – the standards used are influenced by their disadvantaged position. It also explains why the richest people report highest levels of happiness within countries – their standards will be affected by their advantaged position. At the same time, it explains why countries do not appear to get happier over time if they do get richer: people apply contemporary standards to their self-evaluation. In short, it provides a plausible story that is compatible with the evidence, while their thought experiment shows that its alternative – the idea that happiness has truly been left unaffected by societal changes over time – is implausible.

If applied to Sen's examples, the Epistemic Argument would illustrate that the lives of the deprived who report to be happy are not in fact happy. Their standards of judgement have adjusted, but their actual levels of happiness have not. Deprivation not only makes a person resilient in the face of bad external circumstances, but it reduces the standards that they use to evaluate their happiness. Thus, it is plausible that people who are genuinely unhappy in perpetually deprived circumstances report less unhappiness than those who have not endured it for the same length of time. The Epistemic Argument does not take the process of adaptation to genuinely improve happiness, but only to improve the perception thereof. Not only is this a plausible interpretation of Sen's examples but, if the arguments in the previous section are convincing, it is a better argument for considering why there is a gap between reported happiness and well-being than the Axiological Argument.

V. CAN THE EPISTEMIC ARGUMENT BE TACKLED EMPIRICALLY?

The Epistemic Argument turns the adaptation problem into an empirical challenge: can we really capture how happy people are? If correct, the Epistemic Argument provides reasons to be sceptical about this challenge. Nevertheless, even if correct, it is still an open question how bad the problem is for happiness research. It may very well be that while adaptation would distort happiness measures, the effect is small. An empirical challenge may have to be tackled by empirical technology. In fact, a potential objection to the efficacy of the Epistemic Argument is that SWB measurements have been under extensive scrutiny and that none of these tests indicate severe measurement problems with the SWB measurements.

The most common way measures are empirically tested is through construct validation. Construct validation – as it is used in psychometric practice – implies an assessment of the statistical evidence of the measures given our expectations of how these measures should behave if they were good measures of what they are aimed to measure. In the present discussion this is the concept of happiness.Footnote ⁵⁶ Examples of construct validation results include high correlations of SWB measures with unemployment,Footnote ⁵⁷ facial expressionFootnote ⁵⁸ or neurological activity.Footnote ⁵⁹ Andrew Oswald and Stephan Wu construct an objective well-being measure on the basis of real-estate prices, and find it correlates well to aggregate happiness measures in the areas.Footnote ⁶⁰ John Helliwell discusses the predictive value of SWB measures for suicide probability.Footnote ⁶¹ Lucas et al. show that there is a correlation between optimism and reported happiness, but the correlation is not high enough to suspect that they measure the same concept.Footnote ⁶²

SWB measures have done reasonably well for the standards of construct validation. If SWB were ‘just noise’,Footnote ⁶³ as some sceptics in fact have suggested,Footnote ⁶⁴ the measures would not pass a construct validation test. This should provide some confidence that these measures indeed capture something consistently. However, the adaptation problem does not imply that SWB is ‘just noise’. The Epistemic Argument drives a wedge between perceived degrees of happiness and actual degrees of happiness in certain contexts. In particular, it does so when people have suffered prolonged deprivation. If the Epistemic Argument succeeds, measures of SWB capture perceived degrees of happiness. Similar to actual degrees of happiness, we would expect perceived degrees of happiness to correlate to unemployment, facial expressions, suicide probability, etc. In fact, it is likely to be correlated to actual happiness and in many cases very closely.

Unfortunately, an implication of the Epistemic Argument is that the wedge between perceived and actual degrees of happiness implies that there is a limit to the empirical testing of measurement problems. As measures of SWB cannot provide us with direct information on actual degrees of happiness, we cannot test how closely related actual happiness and measures of SWB truly are. What we would need are variables that are correlated with actual, not just perceived, happiness. It is difficult to think of variables that can fulfil this role. This implies that there is a limit to how much confidence construct validation can provide for measures of SWB.

As the extent of the problem the Epistemic Argument poses cannot be assessed empirically, its assessment must remain suggestive. On the one hand, we can worry that the problem is ubiquitous and severe. After all, there is no reason to assume that everyone applies the same standards of evaluation. Nevertheless, given that the study of SWB, and happiness economics in particular, concentrates on relationships at the group level, general scepticism may not be warranted. In lack of a specific worry that two groups being compared have adapted differently, it is likely that differences in measures due to different applied standards of assessment ‘average out’ on a group level.

The real worry arises when comparisons are made between groups that have presumably adapted differently and for that reason apply different standards to assess their happiness. The problem is that in these cases, the degrees of perceived happiness do not correspond to degrees of actual happiness. To provide an example, consider the finding that Latin American countries reach high levels of reported happiness despite low levels of economic prosperity.Footnote ⁶⁵ Without questioning the authors’ conclusions (Latin American countries may of course indeed be very happy), it follows from the Epistemic Argument that there is good reason to doubt that the observed measures of SWB can be interpreted as degrees of happiness that can be compared among countries. Growing up in Latin American countries would probably result in different degrees of adaptation than growing up in, for example, European countries. This would effectively make comparisons in actual happiness on the basis of SWB difficult.

Another crucial example pertains to the empirical results discussed in section III that have been used to argue that adaptation does not actually occur. For example, the fact that the unemployed do report low levels of SWB has been taken as evidence that adaptation does not occur. However, if the Epistemic Argument is correct, we still have good reasons to be sceptical about interpreting the degrees of perceived happiness in SWB measures as degrees of actual happiness. It may be that actual levels of happiness are more sensitive to unemployment than measures of SWB. As such, even in light of corroborating empirical findings, whether adaptation affects empirical results in such cases remains an open question.