Do We Know What Will Make Us Happy?

In this world there are only two tragedies. One is not getting what one wants, and the other is getting it.

— Oscar Wilde, “Lady Windermere’s Fan”

Photography students were asked to pick one of two prints they had made themselves. They kept one; the researchers took the other. Some were told the decision was final. Others were given a few days to swap. The right to change your mind should only help: a bad choice can be fixed, a good one kept. The participants expected exactly that. But later the people with the final choice liked the print they kept more, while the option to swap kept satisfaction from settling in.

The experiment shows that preference does not just precede choice, it also shifts afterwards. A closed alternative stops entering the daily internal contest, and the chosen option gets more attention and more justification. The freedom to revise preserves the chance to correct a mistake — and the comparison itself. From this come two opposite traps: treating a present feeling as an accurate forecast of a future life, or deciding that adaptation will smooth out any outcome anyway. Both errors need checking.

A future lit by one spotlight

When a person imagines a future life, he lights up one object and then judges the whole scene by that patch. Residents of the Midwest and California both thought the California climate was a major reason for a possible difference in happiness, though the actual overall life satisfaction of students in the two regions barely differed. The weather was vivid and easy to picture; everyday life also consisted of classes, money, friends, sleep, health and a bad Tuesday.

In relationships the spotlight is especially strong. “Once we are married everything will finally be fine,” “after the divorce everything will collapse,” “a child will fill life with meaning,” “with this person I will never be lonely again.” Imagination builds the scene of the wedding, the empty apartment or the infant in your arms, and forgets that the future person will be sitting in traffic, answering work email, getting sick, laughing at someone else’s joke and getting used to what today looks like the main event. The event matters, but in the forecast it fills the whole screen.

The psychological immune system

Affective forecasting often overestimates the intensity and the duration of future emotions. People routinely underweight their own adaptation: after a bad outcome they change the explanation, the routine, the circle of support and the object of attention. These processes do not soften the loss itself, but they keep it from remaining forever the only event in consciousness.

The same system works after a good choice. New love, a wedding, a higher income or the apartment you wanted gradually become the environment rather than the celebration. This is not devaluation. Constant delight sits badly with ordinary life: the brain notices change more than a steady level. The promise “this will make me happy forever” is usually wrong about the trajectory, not about the sign. A good event can improve a life and stop being felt daily as an improvement.

Hedonic adaptation does not return every person to a single starting point after every event. A meta-analysis of 188 publications found different trajectories for marriage, divorce, loss of a partner, the birth of a child, unemployment and other events. After marriage the average rise in cognitive satisfaction faded in roughly two years; after unemployment, divorce and bereavement recovery was slower and often incomplete. The events differed, the people differed, and so did the measure of well-being.

The formulas “people get used to anything” and “after this I will never be the same” erase the spread equally well. The data describe a capacity to rebuild, with limits that depend on health, income, social ties, how often the stress repeats and whether the environment can be changed.

Three different measures of happiness

The word “happiness” again folds several quantities into one. There is experienced well-being — what a person feels over an ordinary day. There is life evaluation — how he answers the question whether he is satisfied with it on the whole. There is memory of an event and forecast of the future. They are related, but they can diverge. Parenthood can reduce pleasure in particular hours and raise the sense of meaning; divorce can bring acute stress and improve life evaluation after leaving a destructive union.

The method of measurement changes the picture. A survey right after the wedding catches the event and the expectations, a diary catches recurring emotions, longitudinal observation catches the trajectory before and after. But the people who marry, divorce and have children are not randomly assigned, and a meta-analysis inherits the differences between the questions asked in the original studies. So before drawing a conclusion you have to name the measure: an experienced state, an overall evaluation of life, memory or forecast.

The difference between measures changes the decision, not just the methodology section. A person can predict accurately that he will consider the choice correct and be wrong about how his days will go; he can fear brief acute pain and underestimate long financial damage. A forecast is useful only when it measures the consequence the decision is being made for.

The ability to come to love the photograph you chose is easily turned into advice to close doors. In relationships the same process can also lock in a bad union: a person recalls the years invested, explains a violation as an exception, lowers expectations and calls the absence of alternatives maturity. Synthetic satisfaction is real, but it describes adaptation to a choice, not the quality of the original choice.

Here two chapters of this book meet. Sunk costs make you continue because so much has already gone in. Affective adaptation can make continuing subjectively bearable. Together they produce a durability that can be mistaken for quality. The opposite error is possible too: constant comparison with open alternatives keeps satisfaction from arising even in a good union. Neither calm after a choice nor unease with the doors open is a sufficient diagnosis.

The choice is not the life after the choice

In sales, a signature on the contract does not yet prove the purchase was good: the test starts with use. In relationships the difference is harsher. A mutual like, sex, a shared address, a wedding or the birth of a child record that a decision has been made. None of these events by itself shows how well the resulting system will work.

People easily optimise the event of choosing: the number of dates, the speed of moving to exclusivity, the beauty of the ceremony, the fact that the marriage was preserved. Such a metric can look successful while an ordinary day produces loneliness, overload or dependence. The real result appears later — in the division of labour, the ability to repair conflict, the safety of saying no, and whether each participant’s life remains liveable for him.

So a decision should be compared not with the promise made at the moment of choice, but with the experience that follows and with a realistic alternative. The difficulty of obtaining something does not prove the value of what was obtained; the length of time something was kept does not prove the quality of what was kept.

An average that lives in no one’s house

Large surveys usually find higher life satisfaction among people who are married or in a stable partnership. An average difference does not yet say how much the union itself produced. Healthier, more sociable, better-off and initially more contented people find partners and keep relationships more easily. Longitudinal work sees a rise around the wedding and a partial return to the previous level; other analyses find a more durable advantage, especially when the spouse is also a close friend. The argument is not about a binary benefit of marriage, but about the shares that belong to selection, to the institution and to the quality of the particular relationship.

For predicting a particular life, relationship quality is more informative than marital status alone. Meta-analyses link marital quality to psychological and physical health, but most of this evidence is observational and does not fully separate causation in both directions. Support, closeness and predictability can buffer stress; health and initial resources influence relationship quality at the same time. A relationship under control, contempt or constant threat can add load to every day. Comparing “marriage” and “being single” as two uniform products is therefore close to meaningless. The spread inside each category is larger than a convenient average is willing to show: a friendly marriage and a hostile marriage share a legal name, and a chosen independent life and an agonising isolation share a line on the form.

Economics can handle such distinctions when it is not lazy. It can say that a union produces joint goods, insures risks, splits fixed costs and creates specialisation; then it is obliged to ask who gets the gains, who pays for the specialisation and whether a participant can leave. It can note that marriage is associated with higher satisfaction in a number of samples, but it has to check whether the association survives controls for income, health, personality, age, quality of the tie and selection into marriage. An average does not lie, but it answers the average question. No one lives in the average marriage, raises the average number of children or wakes up next to the average partner. Statistics are necessary when we know almost nothing about the particular case; once you have a history of your own, they must give way to more precise data.

Loneliness with no witness and loneliness for two

The word “loneliness” turns into a moral verdict especially easily. It mixes at least three states: the absence of a romantic partner, living in a separate household, and the subjective sense that there is less closeness than a person needs. These states are related but not identical. Psychologists define loneliness not as the physical absence of people but as the perceived gap between desired and actual connection; so a person can live alone and not feel lonely, and can have a spouse, children, colleagues and a full calendar and still experience isolation. Life-course research also shows that loneliness changes with age and circumstances, but does not form a simple line in which every year without a partner automatically makes things worse.

Separating these states matters not for the sake of a consoling line about being happy alone too. It changes the choice problem itself. The alternative to a particular relationship is not an abstract void but a specific life outside it: housing, friends, work, children, relatives, sexuality, health, money, the chance to meet someone new and the ability to sit with silence. For one person, leaving a bad union really does mean dangerous social isolation; for another it means returning to a network the relationship had gradually crowded out. Living alone can be desired autonomy, forced waiting, a transitional period, the result of a loss, or all four at once. The same outward form carries a different price and a different freedom.

Romantic culture compares asymmetrically. It sets a real relationship, with all its fatigue, domestic labour and conflict, next to a caricature of solitude — an empty apartment, a cold dinner and a person with no one to call. When freedom needs advertising, the comparison flips: the real difficulties of an independent life are set next to a caricature of the couple, where every evening is taken up by compromise and someone else’s socks. Both images are convenient and both get in the way of a decision. What should be compared are two plausible trajectories: this tie as it is actually developing, and the life outside it that the person is in fact able to build. A bad union does not become good because solitude carries risks; solitude does not become a victory merely because a union ended.

At this point economics offers a useful but limited reminder about opportunity cost. By staying, a person gives up not “all other partners” but a certain set of future possibilities; by leaving, he loses the problem and the real goods of the existing tie. The model, however, cannot assign the right weights to these things. It does not know what a familiar voice in the next room is worth to this person, or the right to close a door without explaining, or a child living in one house, or constant fear. Economics keeps these variables in view, but only the person who has to live with the result can weight them.

An option to correct the error

An imprecise forecast raises the value of reversibility, but constant revision has a price too. Irreversibility is justified where it creates a real gain — trust, investment, stability — and is dangerous when it serves only as cultural proof of seriousness. A decision should close enough doors for joint work and leave an exit for the error the present person cannot yet see.

It is economically useful to separate the decision from the right to revise it. You can choose a partner and end the daily contest while keeping your own income, your friends and a legal way out. You can commit to living together without immediately merging all assets and reproductive plans. You can close some alternatives while keeping insurance against a mistake the present person is unable to imagine.

This is where the trade-off between commitment and option appears. Too many options make investment impossible: everyone lives like a short-term tenant. Too few make an error catastrophic and therefore intensify the fear of choosing. Good architecture does not promise you will not be wrong. It lets a decision become real without turning a forecasting error into a life sentence.

It is more useful to assess the repeating day than to pass an advance verdict on happiness: what will happen every week, what will disappear, what costs the body, the time and other relationships will bear, what will stay the same, how fast the choice will become ordinary, and which consequences will remain after the feeling changes. Such a forecast is less vivid, but it is closer to actual daily life.

For decisions about relationships one more correction matters: the future result is produced by the person chosen and by a sequence of joint actions. You cannot predict the happiness of a marriage by looking at the wedding; you can assess the division of labour, the ability to repair conflict, the safety of saying no, financial dependence and the quality of ordinary time. You cannot know the whole pain of a divorce in advance; you can assess housing, support, children, income and which parts of life already exist outside the union. A forecast gets better when feeling is translated into mechanism, without pretending to be complete accounting for the future.

We are wrong about the future not because we do not know ourselves at all. We imagine too narrow a scene and forget that the future person will not live inside a single event. He will get used to part of the good, adapt to part of the bad, change the explanation, meet new circumstances and sometimes stay hurt worse than a comforting theory of adaptation promises. So it is more useful to predict not the exact size of future happiness but the conditions under which an ordinary day gets better or worse: division of labour, the safety of saying no, the quality of the tie, support, dependence and the possibility of revision. A sensible choice does not require guessing a future emotion correctly. It requires building a trajectory in which a forecasting error does not remove the ability to change the decision and does not shift the whole price onto someone else.

Main sources

Gilbert, D. T., & Ebert, J. E. J. (2002). Decisions and revisions: The affective forecasting of changeable outcomes. Journal of Personality and Social Psychology, 82(4), 503–514.

Wilson, T. D., & Gilbert, D. T. (2005). Affective forecasting: Knowing what to want. Current Directions in Psychological Science, 14(3), 131–134; Gilbert, D. T., Pinel, E. C., Wilson, T. D., Blumberg, S. J., & Wheatley, T. P. (1998). Immune neglect: A source of durability bias in affective forecasting. Journal of Personality and Social Psychology, 75(3), 617–638.

Schkade, D. A., & Kahneman, D. (1998). Does living in California make people happy? A focusing illusion in judgments of life satisfaction. Psychological Science, 9(5), 340–346.

Luhmann, M., Hofmann, W., Eid, M., & Lucas, R. E. (2012). Subjective well-being and adaptation to life events: A meta-analysis. Journal of Personality and Social Psychology, 102(3), 592–615.

Lucas, R. E., Clark, A. E., Georgellis, Y., & Diener, E. (2003). Reexamining adaptation and the set point model of happiness: Reactions to changes in marital status. Journal of Personality and Social Psychology, 84(3), 527–539; Grover, S., & Helliwell, J. F. (2019). How’s life at home? New evidence on marriage and the set point for happiness. Journal of Happiness Studies, 20, 373–390.

Robles, T. F., Slatcher, R. B., Trombello, J. M., & McGinn, M. M. (2014). Marital quality and health: A meta-analytic review. Psychological Bulletin, 140(1), 140–187.

Hawkley, L. C., & Cacioppo, J. T. (2010). Loneliness matters: A theoretical and empirical review of consequences and mechanisms. Annals of Behavioral Medicine, 40(2), 218–227.

Luhmann, M., & Hawkley, L. C. (2016). Age differences in loneliness from late adolescence to oldest old age. Developmental Psychology, 52(6), 943–959; Mund, M., Freuding, M. M., Möbius, K., Horn, N., & Neyer, F. J. (2020). The stability and change of loneliness across the life span: A meta-analysis of longitudinal studies. Personality and Social Psychology Review, 24(1), 24–52.

At the start of this book a dating app asked questions with the briskness of a mortgage officer and promised that setting the filters correctly would be enough. By the end of the book it is still a useful tool: distance exists, attitudes to children matter, smoking is entirely real. But now a more complicated object is visible behind the interface. Choice depends on the local pool, mutual attention, information, institutions, demography and a person’s ability to be wrong about his own future. Feeling remains part of the problem, but none of these variables lets us take it for a complete description.