Hedonic Adaptation

​

What is Hedonic Adaptation?

Hedonic Adaptation is the lessening, over time, of one’s emotional response to a circumstance that continues or repeats.

According to the hedonic treadmill model, good and bad events temporarily affect happiness, but people quickly adapt back to hedonic neutrality.

That sentence opens Diener, Lucas, and Scollon (2006). It is their statement of the strong model, not their conclusion. Hedonic neutrality means a midpoint, neither pleasant nor unpleasant. The treadmill picture is a walk that leaves you where you started, however long the belt runs. The 2006 paper says the picture needs five revisions. People do often move back toward a familiar level. That level is usually pleasant, it is not the same person to person, and some events never finish the trip. A useful split is a new coat and a lost job. The coat is still in the closet next winter, and the pleasure of buying it is mostly gone. The job can still be gone next winter, and the life-satisfaction score can still be low. Both are about time. Only one is a completed return. Impact Bias is a forecast made before the feeling: people often give a future emotion too much strength or too much time. Hedonic Adaptation is what the later report does. A forecast can miss because adaptation happens. It can also miss for other reasons. The two names are not one score. The Easterlin Paradox page is about income and reported happiness across people, years, and countries. This page is about the path after a named event in one person’s row. The Paradox of Hedonism is a different claim again: aiming straight at pleasure can spoil it. Adaptation can sit inside that story. It is not the definition of it.

​

Hedonic Adaptation in 3 Depths

• Beginner: When a purchase, a slight, or a piece of news feels as if it will set the mood for years, ask whether you are still inside the circumstance. A feeling that fades while the fact remains is a different object from a feeling that fades because the fact ended.
• Practitioner: Before a raise, a move, a wedding, or a job loss, write two lines. One is life satisfaction on an ordinary day in the first year. The other is the same question four years later, if the new state is still true. Clark, Diener, Georgellis, and Lucas (2008) show that those two years are not interchangeable across events.
• Advanced: “Full adaptation” in that paper’s Table 4 is a label for five events, not a law that every later year matches the years before the event. Unemployment is the exception they print. Women’s long marriage cell, and the long widowhood cells, are not printed zeros. A rule that says every shock dies is already wider than the table.

​

Origin

The treadmill has a chapter and a later correction. Diener, Lucas, and Scollon (2006) say Brickman and Campbell (1971) described a hedonic treadmill, on the model of sensory adaptation. A smell fades as the nose stays in the room. They read the 1971 proposal as this: people react briefly to good and bad events, then return to neutrality, and they keep chasing the next goal because they do not see that the chase returns them to the same point. The 1971 chapter was not opened, so its wording is not quoted. The 2006 paper traces the habituation model to Harry Helson (1948, 1964). Those works were not opened either.

We cannot reject the hypothesis of complete adaptation to marriage, divorce, widowhood, birth of child and layoff. However, there is little evidence of adaptation to unemployment for men.

That pair of sentences is the abstract of Clark, Diener, Georgellis, and Lucas (2008). The PDF was read. It is the counted case below. It is also the cleanest public correction of the idea that every event has the same ending. The famous cross-section is earlier. Brickman, Dan Coates, and Ronnie Janoff-Bulman, “Lottery winners and accident victims: Is happiness relative?”, Journal of Personality and Social Psychology 36 (1978), pages 917–927. The journal PDF was not opened. Diener, Lucas, and Scollon say the studies were taken to show that lottery winners were not happier than nonwinners and that people with paraplegia were not substantially less happy than people who can walk. They also say that reading is too strong. From Brickman and colleagues’ data they calculate a difference of about 0.75 standard deviations in general happiness between the spinal-cord-injured group and the controls, and they call that large. This page does not print the 1978 means, because the table was not in the file that was read. Diener, Lucas, and Scollon then cite Richard E. Lucas (2005a): two nationally representative panels, people followed on average seven years before and seven years after the onset of disability. Those certified 100 percent disabled reported life satisfaction 1.20 standard deviations below their own nondisabled baseline, with very little adaptation over time. That 2005 PDF was not opened. The sentence here is the 2006 paper’s summary, not a second reading of the panels. They also say the strong model had real support of a narrower kind. Silver (1982), a dissertation they cite and this page did not open, is their example of strong negative emotions one week after a spinal-cord injury, and of happiness as the strongest emotion two months later. Suh, Diener, and Fujita (1996) are their example that only recent events predicted happiness in one study. The review they point to is Frederick and Loewenstein (1999). The 2006 reference list spells the first surname Fredrick. That chapter was not opened.

​

Key Points

Four distinctions keep Hedonic Adaptation from turning into “nothing lasts” or into “circumstances never matter.”

​

Applications

These four moves use the questions the 2006 and 2008 papers actually separate. They are not a claim that a worksheet changes a household’s mood by a measured amount. Intervention papers cited by Diener, Lucas, and Scollon, including Fordyce (1977, 1983), Sheldon and Lyubomirsky (2004), Seligman, Steen, Park, and Peterson (2005), and Emmons and McCullough (2003), were not opened. Their designs are not copied here.

​

Case Study

The counted case is Clark, Diener, Georgellis, and Lucas (2008): twenty waves of the German Socio-Economic Panel, asking whether life satisfaction returns to a within-person baseline after unemployment, marriage, divorce, widowhood, birth of a child, and layoff. The sample they describe has more than 130,000 person-year observations. For ages 16 to 60 they print 65,658 person-years for men and 65,447 for women. Not everyone is present for all twenty waves. Some enter after 1984, and some leave before 2003. Life satisfaction is one item from 0, completely dissatisfied, to 10, completely satisfied. In the 16-to-60 table the modal answer is 8. The regressions are fixed-effects, or “within,” regressions, so the comparison follows the same person. The authors treat the 0-to-10 answers as cardinal for that reason, and they say an ordinal check did not change the qualitative pattern. Controls in the unemployment table include education, children, age bands, household income, health, marital status, German nationality, region, and year. They use only the first event they see a person enter during the panel. A person who was already unemployed, or already married, at the first interview is left out of that event’s adaptation analysis. The spells they count, men then women, are 1,087 and 1,019 for unemployment, 1,076 and 1,031 for marriage, 327 and 352 for divorce, 123 and 377 for widowhood, 1,497 and 1,592 for a birth, and 935 and 669 for a layoff. Births are ages 16 to 40. Widowhood is ages 16 to 80. A layoff, in their definition, is a job left after the previous interview because of a plant closing or a dismissal. The survey did not fully separate those two reasons until 2001, so the paper keeps them together. That event is not the same variable as being unemployed. The unemployment lags in Table 3, and again in Appendix Table A2, are the change in life satisfaction while the person is still unemployed, relative to the reference of being out of the labour force. For men the coefficients are −0.763 (SE 0.048) in the first year, −0.781 (0.080) at one to two years, −0.586 (0.112) at two to three, −0.667 (0.160) at three to four, −0.846 (0.224) at four to five, and −0.520 (0.192) at five or more. Each of those six is marked significant at 1 percent. For women they are −0.527 (0.048), −0.335 (0.082), −0.342 (0.126), −0.493 (0.173), −0.344 (0.281), and −0.147 (0.274). The last two are not marked significant. The authors say the long cells for women should be read with care. Appendix Table A1 prints the observation counts behind them: for men, 39 observations at four to five years and 70 at five or more; for women, 27 and 33. Those are observations, not a headcount of people. The same lag regression includes an employment dummy: 0.231 (SE 0.025) for men, marked significant, and 0.020 (0.018) for women, not marked significant. The authors say a person who was unemployed and is employed again has zeros on the unemployment-duration dummies, and the return to work is this other coefficient. In the male column of that table, major health problems are −0.817 (0.039). A first year of unemployment, −0.763, sits near that control. It is not a raw gap between two groups, and it is not “the same as” a major health problem outside this regression. Leads matter. Men who will enter unemployment within the next year are at −0.271 (SE 0.054). The male lead three to four years ahead is still −0.153 (0.059), marked significant. Women are at −0.233 (0.067) within the next year. Their earlier leads are not marked significant. Table 4 summarizes anticipation as four years for men and one year for women, and adaptation as “Little” for men and “Some, after 4 years” for women. A check that kept only continuous spells, combining the two longest male and female cells, still showed no adaptation for men and adaptation for women after four years. The contrast is the other five events, in Appendix Table A2 and Table 4. Marriage in the first year is 0.305 (SE 0.048) for men and 0.387 (0.052) for women. At five or more years, men are −0.075 (0.051), not significant, and women are −0.159 (0.059), marked significant at 1 percent. Table 4 still labels marriage adaptation “Full” for both. Widowhood in the first year is −1.021 (0.135) for men and −1.016 (0.084) for women. At five or more years those coefficients are 0.306 (0.144) and 0.317 (0.075), both marked significant, and Table 4 again says Full. A birth is 0.059 (0.045), not significant, for men in the first year, and 0.183 (0.046) for women. The authors say that four years later both sexes are equally unhappy. The three-to-four-year coefficients are −0.194 (0.062) for men and −0.170 (0.061) for women. The five-or-more coefficients are not significant. A layoff in the first year is −0.250 (0.052) for men and −0.144 (0.061) for women. At five or more years the coefficients are 0.057 (0.053) and 0.001 (0.057), not significant. Table 4 labels layoff adaptation Full. The lesson is the split inside one design. Staying unemployed does not fade on this scale for men, and the long female cells are small. A marriage, a child, a widowhood, and a layoff do not share that flat low path, and “Full” in Table 4 is not a printed zero in every long cell. These coefficients belong to that panel, those years, and those selections. They are not a happiness rate for another country, another decade, a hiring meeting, or a household. Lucas, Clark, Georgellis, and Diener (2004) is the companion paper about whether the drop remains after people are reemployed. The journal PDF was not opened. The published abstract, which was read, says a 15-year German panel of more than 24,000 people showed a strong reaction and a shift back that, on average, did not reach the old level even after reemployment. This page does not print coefficients from that paper. A February 2002 draft was read and was not used to fill the gap, because it was not checked against the journal.

​

Boundaries and Failure Modes

Hedonic Adaptation fails as a slogan in three places. First, it is not every fact about happiness, and it is not a forecast or a memory weight. Impact Bias is an overestimate of a feeling that has not happened yet. Duration Neglect is how little the length of a finished episode counts when someone rates it later. Focusing Illusion is attention standing in for a whole life or a whole choice. The Easterlin Paradox is a comparison of income and happiness, not the German duration dummies. Using this name for a misforecast, a remembered trip, or a national income chart goes past the reports these papers measured. A score above the midpoint after a bad event is also not, on the 2006 argument, proof that adaptation finished. Diener, Lucas, and Scollon say that pattern tells you more about where the baseline sits, and sometimes more about Impact Bias, than about a completed return. The test they want is the same people, before and after. Second, the German coefficients will not travel as a rate. The long unemployment cells are 39, 70, 27, and 33 observations. People already in the state at the first interview are excluded. Only the first observed entry is kept. Income, health, and marital status are in the regression, so −0.763 is not a raw mean and not the paycheck by itself. The 0-to-10 item is general life satisfaction, not satisfaction with the job. Women’s nulls after four years of unemployment are the cells the authors flag as thin. The positive long-run widowhood coefficients, and the women’s negative long-run marriage coefficient, sit in the same appendix as the word Full. A department that treats Full as “widowhood does not hurt,” or −0.763 as the cost of any job loss in any year, has left the table. The 0.75 and 1.20 disability figures are Diener, Lucas, and Scollon’s summaries of studies this page did not open. They are not a clinic’s rate. Third, the usual misuse is to hear adaptation as permission or as futility. The 2008 paper does not say a first year of widowhood feels like the years before it. Those first-year coefficients are about −1.02. It does not say a new child is a lasting lift. The middle years in that appendix are negative, and the authors say both sexes are unhappy four years on. It also does not say effort cannot move the baseline. Diener, Lucas, and Scollon say the strong form, in which no circumstance ever leaves a lasting change, is untenable, and they say some programs have raised reports for months. This page does not carry those programs as a protocol. A national contrast in their Table 2 is the other caution against “circumstances never show up”: life satisfaction in 1999–2001 is 8.14 in Mexico and 4.65 in Russia, on the 1-to-10 scores they take from the World Values Survey. Those are the paper’s printed means for those waves. They are not a ranking for a later year.

​

Common Misconceptions

Three shortcuts turn a time path into a rule about every feeling.

​

Related Concepts

These six pages sit next to Hedonic Adaptation. Each one answers a different question.

​

One-Line Takeaway