Affective forecasting
Predicting how future events will make us feel — usually badly.
What it means
Affective forecasting is the process of predicting one's own future emotional states, and a large body of research shows that people do it systematically poorly. The most pervasive error concerns not the direction of feelings, which people usually get right, but their intensity and especially their duration: people consistently overestimate how strong and how long-lasting their reactions to future events will be, a pattern known as the impact bias. Several mechanisms drive these errors — focalism, in which we fixate on the focal event and neglect everything else that will fill our lives; immune neglect, in which we underestimate our psychological resilience and capacity to rationalize and adapt; and projection from our current state onto the future. Because we rely on these flawed forecasts to choose, the errors steer major life and consumer decisions, leading people to over-invest in things expected to bring lasting happiness and to over-fear setbacks that will fade faster than expected. It matters for understanding well-being, decision-making, and policy, and it underlies the cautionary insight that we are surprisingly bad witnesses to our own future feelings.
Why the forecasts miss
Forecasting works by simulation. We run a mental preview of the event, read our feelings off the preview, and hand back the reading as a prediction, so the forecast is only as good as the preview is representative. It rarely is. A simulation is a construction rather than a recording: it is assembled now, out of what the present makes available, and it renders what is easy to render. That is why the error is systematic rather than random, and why knowing about it does not dissolve it. Worse, the preview is silent about the part that will matter most, because the mind's sense-making machinery works best unobserved: watch yourself concluding the job was never right for you and the consolation stops consoling. The machinery that will do the most to determine how you actually feel is precisely the machinery the preview cannot show you, and the preview's confidence is unaffected by the omission. Gilbert and colleagues found this immune neglect across denied tenure, electoral defeat, and employer rejection.
What the evidence shows
The direction of a forecast is rarely wrong; the dispute is over the magnitude, and it is live. Levine and colleagues argued in 2012 that much of the impact bias is a question-wording artifact: forecasters predict how they will feel about an event, then later report how they feel in general, and the mismatch gets scored as error. Clarify the question and the overestimation shrinks. Wilson and Gilbert replied that the reanalysis omitted contradicting studies and mishandled data; Levine's group answered that the bias is both dead and alive, depending which question was asked. Lench and colleagues mapped the seam with field data on exam grades and the 2016 US election: intensity forecasts were reasonably accurate, while forecasts of how often the feeling would recur, and how far it would tint general mood, were not.
Where the errors matter
Medicine is where they cost most. People asked to imagine dialysis, a colostomy or paraplegia predict a bleaker life than patients living those conditions report, and such forecasts shape advance directives, treatment refusals, and the quality-of-life weights health systems use to price interventions. Consumer choice runs the mild version: the feature that dominates the showroom is the one focalism inflates, so buyers pay for a difference they will stop noticing. Hiring errs in both directions, with candidates weighing title and salary while discounting the commute and the colleagues, and rejected applicants dreading a blow that fades in weeks. Any method that asks unaffected people to price an experience they have never had inherits the bias, which is why contingent-valuation and cost-utility exercises deserve caution.
Limits, and what actually helps
The bias is a big-event phenomenon more than a daily one. Moeck and colleagues sampled ordinary days and found people reliably knew when a day would be better or worse than usual, erring more on absolute levels than on direction; everyday forecasts are good enough to steer everyday choices. What does not help is imagining harder, since the simulation is the source of the error. Two things do. Defocusing works: prompt someone to lay out the ordinary hours around the event and the overprediction shrinks, the redirection manipulation Wilson and colleagues used to isolate focalism. Surrogation is the more counterintuitive candidate, and the evidence is thinner than its reputation. In a small 2009 study, 25 women forecasting a speed date, knowing how one previous woman had felt predicted their own reaction better than reading the man's profile did, and participants refused to believe it. Surrogation outperformed simulation in the study that tested it, but that is a striking result from a single experiment that has not been independently replicated at scale. Treat it as a promising heuristic rather than a settled finding: ask what happened to people who already did this, not what you picture.
Examples
People expect a breakup or a job loss to devastate them for far longer than it actually does.
A promotion feels like it will change everything. Three months on, the commute, the inbox and the same Tuesday evenings have crowded back in, and the glow has gone.
Buyers are sure the bigger kitchen will make them happy for years. By the second month they notice it about as often as they noticed the old one.
A patient signs an advance directive refusing dialysis, certain such a life would be unbearable. The asymmetry is the point: accept treatment and a mistaken forecast gets corrected by living through it, with the option to stop still open; refuse, and the forecast executes itself, leaving no adapted self around to disagree with it.
Someone weighing a job offer reads the careers page and the role description twice, and never calls the person who left the post last year. Offered both, most people want the materials: one account of how the job actually felt reads as a sample of one, while the brochure feeds the simulation they already trust.
First described in Wilson & Gilbert (2003).
Key references
- Moeck, E. K., Grewal, K. K., Mehta, A., Greenaway, K. H., Koval, P., & Kalokerinos, E. K. (2026). Affective forecasting accuracy in everyday life. Affective Science, 7(2), 306-318. doi.org/10.1007/s42761-026-00364-x
- Lench, H. C., Levine, L. J., Perez, K., Carpenter, Z. K., Carlson, S. J., Bench, S. W., & Wan, Y. (2019). When and why people misestimate future feelings: Identifying strengths and weaknesses in affective forecasting. Journal of Personality and Social Psychology, 116(5), 724-742. doi.org/10.1037/pspa0000143
- Wilson, T. D., & Gilbert, D. T. (2013). The impact bias is alive and well. Journal of Personality and Social Psychology, 105(5), 740-748. doi.org/10.1037/a0032662
- Levine, L. J., Lench, H. C., Kaplan, R. L., & Safer, M. A. (2013). Like Schrodinger's cat, the impact bias is both dead and alive: Reply to Wilson and Gilbert (2013). Journal of Personality and Social Psychology, 105(5), 749-756. doi.org/10.1037/a0034340
- Levine, L. J., Lench, H. C., Kaplan, R. L., & Safer, M. A. (2012). Accuracy and artifact: Reexamining the intensity bias in affective forecasting. Journal of Personality and Social Psychology, 103(4), 584-605. doi.org/10.1037/a0029544
- Wilson, T. D., Wheatley, T., Meyers, J. M., Gilbert, D. T., & Axsom, D. (2000). Focalism: A source of durability bias in affective forecasting. Journal of Personality and Social Psychology, 78(5), 821-836. doi.org/10.1037/0022-3514.78.5.821
- Gilbert, D. T., Killingsworth, M. A., Eyre, R. N., & Wilson, T. D. (2009). The surprising power of neighborly advice. Science, 323(5921), 1617-1619. doi.org/10.1126/science.1166632
- 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. doi.org/10.1037/0022-3514.75.3.617
- Riis, J., Loewenstein, G., Baron, J., Jepson, C., Fagerlin, A., & Ubel, P. A. (2005). Ignorance of hedonic adaptation to hemodialysis: A study using ecological momentary assessment. Journal of Experimental Psychology: General, 134(1), 3-9. doi.org/10.1037/0096-3445.134.1.3
- Ubel, P. A., Loewenstein, G., Schwarz, N., & Smith, D. (2005). Misimagining the unimaginable: The disability paradox and health care decision making. Health Psychology, 24(4, Suppl), S57-S62. doi.org/10.1037/0278-6133.24.4.S57