Predicting how we will feel

Choosing a job, a home or a departure date involves a forecast that is rarely put into words: what we will feel once the thing is obtained, or missed. These forecasts guide a large share of ordinary decisions, and they are almost never checked afterwards.

Daniel Gilbert and Timothy Wilson made them an object of study under the name affective forecasting. The method is simple: ask people what they will feel after an expected event, question them again once it has happened, and compare.

The most consistent result comes down to an asymmetry. The direction of the forecast is generally correct: we know that a failure will displease us and that a success will give pleasure. The intensity and above all the duration of the effect, on the other hand, are regularly overestimated, in both directions.

The authors called this mismatch the impact bias. It has been found in relation to break-ups, election results and long-awaited professional decisions. The gap concerned the moment itself less than the weeks and months that followed.

The first is focalism. When forecasting, we picture the event in isolation, without the rest of life that will carry on around it: work, family and friends, unrelated annoyances. Once the event has passed, these elements take back their place and dilute its effect.

The second lies in what the authors describe as a failure to recognise our own adjustment processes, which they call immune neglect. Faced with a setback, we reinterpret, put things in perspective, find ourselves reasons, almost always without noticing. Because this work remains invisible, it does not enter the forecast.

This mismatch is the preserve neither of worriers nor of enthusiasts. It affects happy forecasts as much as unhappy ones, and no way of anticipating comes out ahead: overestimating a joy sustains the drive towards what we desire, overestimating a sorrow keeps us alert to what we risk.

Nor does it mean that events matter little. Some lasting changes, notably those that alter daily life day after day, leave a trace that adjustment does not entirely erase. The bias bears on the estimate of the effect, not on its existence.

The same researchers compared personal forecasting with another source: the state reported by someone who had just been through the same situation. In some of their experiments, this account from another person predicted the future state better than personal simulation, something the participants themselves did not believe.

This result does not say what one ought to do. It merely recalls that predicting one's own states is an operation like any other, carried out with the means at hand, and that it errs in a regular rather than a random way.

What follows is a descriptive question, not a method: when we have long awaited or dreaded something, did the state we felt afterwards resemble what we imagined? The answer, whatever it is, is a piece of information that no questionnaire supplies.