Open data · CC BY 4.0 what each one claims, and what replicated no ads · no tracking
Bias Atlascognitive biases

Optimism bias

also known as — Unrealistic optimism · Comparative optimism · Optimistic belief updating · Pessimism bias · Unrealistic pessimism
Asked to compare themselves with their peers, people rate their own chances as above average for good events and below average for bad ones.
Does it replicate? Mixed The data pattern reproduces reliably. Whether it shows optimism, or shows what an unbiased reasoner would produce in the same task, is still disputed in print.
VerdictMixed
First published1980
FieldBelief and probability
Sources14
Last checked2026-08-15
What it claims

What does Optimism bias mean?

Two separate literatures travel under the name. The older one asks for a comparative judgement: how do your chances of divorce, or of owning a home, compare with the average student's? Since everyone cannot be below average on the same risk, a group whose answers skew one way is making a collective error, and that is what makes the bias testable at all.

The newer literature asks how people revise a personal risk estimate after being told the population base rate. Its claim is about asymmetry: good news moves the estimate more than bad news does. Both are claims about a mechanism, and both rest on the assumption that an unbiased person would produce a symmetric pattern in the measure being used.

That assumption is what the last fifteen years have been arguing about. Pessimism bias is the same claim inverted, and it has no paradigm of its own. It is what these two instruments report when a group's scores fall on the other side of the same benchmark, and most of the groups that do so are clinical.

Verdict

Repeats disagree, or the effect holds in some conditions and not others. The other 248.

The field
72 of the 192 entries in belief and probability share this verdict.
Does it replicate?

Has Optimism bias been retested?

Mixed The data pattern reproduces reliably. Whether it shows optimism, or shows what an unbiased reasoner would produce in the same task, is still disputed in print.

Harris and Hahn argued in Psychological Review that the comparative method cannot separate the two. Bounded response scales, sampling constraints and the rarity of genuinely bad events mean unbiased agents generate the apparent optimism for statistical reasons. Harris, de Molière, Soh and Hahn then went looking for a motivational signal and did not find one. In a rerun of Weinstein's task with 102 female undergraduates, comparative judgements tracked how common an event was, F(1, 101) = 146.50, p < .001, with no effect of whether it was good or bad, F(1, 101) = 1.32, p = .25.

Rare positive events drew pessimism. A design without the problematic scale, run on 389 Mechanical Turk participants, found no self versus other difference at all. The updating paradigm is where the two camps meet head on. Shah, Harris, Bird, Catmur and Hahn added positive events to Sharot's task and the asymmetry flipped: more updating on desirable news for negative events, t(12) = 4.73, and more on undesirable news for positive events, t(12) = 2.78, interaction F(1, 12) = 29.39, p < .001.

Garrett and Sharot replied that the positive events carried unvalidated base rates whose skew produces that reversal by itself, and reported optimistic updating for both event types in 142 participants, t(141) = -4.88 and t(141) = -2.88. Burton, Harris, Shah and Hahn then obtained the asymmetry using valence-neutral events across three preregistered experiments, combined N = 300. Garrett and Sharot failed to replicate that in a sample of 100, F(1, 81.94) = 0.11, p = 0.74, and reported that a fourth experiment of 200 participants sat unreported in the analysis code. Outside the dispute, Kube, Huhn and Menzel found no updating asymmetry about climate change in 109 people and a small one in 604, d = 0.16.

Read off Shah, P., Harris, A. J. L., Bird, G., Catmur, C., & Hahn, U. (2016). A pessimistic view of optimistic belief updating. Cognitive Psychology, 90, 71-127.

Examples

What are some examples of Optimism bias?

The show of hands on divorceillustration

A room of newly engaged couples puts its own chance of divorcing well below the rate it attributes to couples in general. Everyone cannot be below average on the same risk, so the group is collectively wrong whatever any single answer is worth.

The smoker’s arithmeticillustration

A smoker accepts the published risk of lung disease and places their own risk below the average smoker’s, on the strength of exercise and good genes. The bias lives in the comparison, not in the estimate of the population rate.

The new café on the high streetillustration

Someone opening a café agrees that most close within a few years and puts their own chance of lasting above the rest. The claim is not that people expect good things, but that they expect better than their own circumstances warrant.

Cases marked as illustrations describe nobody in particular. They are there to make the pattern recognisable, not to report that a particular thing happened.

The experiments

What experiments is Optimism bias based on?

The design

Weinstein asked 258 students at Cook College how their own chances of future life events compared with those of classmates of the same sex, on a fifteen-point scale from 100% less to 400% more. The 42 events were split at random across two rating forms, so each event was judged by 123 to 130 students rather than by all of them.

Means ran in the optimistic direction for 37 of the 42 events, p < .001 by the binomial test. Averaged over the 18 positive events the mean comparative judgement was 15.4%, t(255) = 6.8; over the 24 negative events it was -20.4%, t(255) = 13.9, both p < .001. The extremes were liking your postgraduation job at 50.2% and having a drinking problem at -58.3%.

The updating version comes from Sharot, Korn and Dolan, who scanned 19 participants estimating their risk of 80 adverse events, then gave them the average probability and asked again. Mean absolute update was 11.2 when the news was better than expected and 7.7 when it was worse, t(18) = 4.4, p < 0.001, a pattern present in 79% of participants.

5 primary research8 replication and reanalysis1 commentary and critique

Origin

Who first described Optimism bias, and when?

Year1980
WhoNeil D. Weinstein
WhereJournal of Personality and Social Psychology

Weinstein's paper named the effect and set the method every later study copied, though he cites earlier surveys about car accidents, crime and disease pointing the same way.

The claim, and the replication this entry reads its figures from.

Where it runs out

When does Optimism bias not apply?

Nobody has run a multi-site registered replication of either paradigm, so the core dispute is a series of small studies by two research groups who disagree about method. Sample sizes on both sides are modest: 19 in the original scanning study, 13 in the experiment that reversed it, 142 in the reply. The critique is narrower than it is often taken to be.

Harris and Hahn attack the inference from comparative and updating measures, not the proposition that people hold rosy beliefs, and they say the magnitude and mechanism are unknown rather than that the bias is absent. Where optimism is measured some other way, by comparing a stated expectation against a later outcome, none of this argument applies. Strunk, Lopez and DeRubeis did that.

They had 153 people recruited through internet announcements estimate the probability of 40 events over the next 30 days, then report which had happened. Low scorers on depressive symptoms showed an optimistic bias that did not reach significance, and high scorers showed a significant pessimistic bias. That paper is paywalled and the figures behind those two statements were not obtained.

Kube, Huhn and Menzel's small effect in 604 participants suggests any residual asymmetry is far below what the early papers implied. The optimistic pattern is also not general to everyone. Korn, Sharot, Walter, Heekeren and Dolan ran the updating task on 18 patients with major depressive disorder and 19 controls, across 70 adverse events. Controls updated more on desirable news, t(18) = 3.0, p = 0.008, and the patients showed no asymmetry, t(17) = -0.11, p > 0.9.

Within the patient group, Beck Depression Inventory scores correlated with the update bias at r = -0.50, p = 0.036, so the more symptomatic patients updated in the pessimistic direction. Patients also started higher, with a mean first estimate of 39.9 against 31.5 for controls, t(35) = 3.3, p = 0.002. Hobbs, Vozarova, Sabharwal, Shah and Button repeated this as a registered report, with 54 depressed and 56 healthy participants screened on the PHQ-9 and recruited through Prolific.

The group difference held for negative events, beta = 0.71, 95% CI 0.24 to 1.18, and the same test on positive events was inconclusive, beta = -0.51, 95% CI -1.16 to 0.15. Zabag, Vander Zwalmen, Kube, Hoorelbeke, Koster and Joormann pooled 29 papers and 3,130 people across several updating paradigms. Depressive symptoms tracked reduced updating to positive information, r = -0.11, 95% CI -0.16 to -0.07, over 33 effect sizes, and tracked nothing for updating to negative information, r = -0.05, 95% CI -0.12 to 0.02, over 17. All of these studies use the paradigm Shah and colleagues object to.

Commonly misread as

How is Optimism bias commonly misread?

The famous claim is not that people expect good things. It is that they expect better things than their own circumstances warrant, which requires a benchmark, and the benchmark is where the trouble lives. The figure of 80% of people being optimistically biased belongs to no study cited here. Weinstein's own data are not uniform: two of his 24 negative events, being a burglary victim and being injured in a car accident, came out in the pessimistic direction.

The reversal Shah and colleagues found for positive events is also not a demonstration of pessimism about life. It is a demonstration that the measure produces whatever the base rates make it produce. Neither is the debate settled by the neuroimaging. Sharot, Korn and Dolan located prefrontal tracking of estimation errors, and a neural correlate of a behavioural measure inherits every question hanging over that measure.

Pessimism bias is not a separate finding with a literature behind it. It is what these measures give when the scores fall the other way, and the bulk of that evidence comes from depressed samples rather than from a general human tendency to expect the worst. The clinical strands also differ. MacLeod and Byrne's anxious participants anticipated more future negative experiences than controls and did not differ on positive ones, while their mixed anxious-depressed participants did both.

The judgement-bias task run on animals travels under optimism and pessimism as well, and it scores how an animal responds to an ambiguous cue rather than what it expects of its own future.

Sources

Sources for Optimism bias

  1. 1Weinstein, N. D. (1980). Unrealistic optimism about future life events. Journal of Personality and Social Psychology, 39(5), 806-820.primary
  2. 2Sharot, T., Korn, C. W., & Dolan, R. J. (2011). How unrealistic optimism is maintained in the face of reality. Nature Neuroscience, 14(11), 1475-1479.primary
  3. 3Harris, A. J. L., & Hahn, U. (2011). Unrealistic optimism about future life events: A cautionary note. Psychological Review, 118(1), 135-154.commentary
  4. 4Shah, P., Harris, A. J. L., Bird, G., Catmur, C., & Hahn, U. (2016). A pessimistic view of optimistic belief updating. Cognitive Psychology, 90, 71-127.replication
  5. 5Garrett, N., & Sharot, T. (2017). Optimistic update bias holds firm: Three tests of robustness following Shah et al. Consciousness and Cognition, 50, 12-22.replication
  6. 6Harris, A. J. L., de Molière, L., Soh, M., & Hahn, U. (2017). Unrealistic comparative optimism: An unsuccessful search for evidence of a genuinely motivational bias. PLOS ONE, 12(3), e0173136.replication
  7. 7Burton, J. W., Harris, A. J. L., Shah, P., & Hahn, U. (2022). Optimism where there is none: Asymmetric belief updating observed with valence-neutral life events. Cognition, 218, 104939.replication
  8. 8Garrett, N., & Sharot, T. (2023). There is no belief update bias for neutral events: failure to replicate Burton et al. (2022). Journal of Cognitive Psychology, 35(8), 876-886.replication
  9. 9Kube, T., Huhn, J., & Menzel, C. (2025). Optimistic bias in updating beliefs about climate change longitudinally predicts low pro-environmental behaviour. British Journal of Social Psychology, 64(3).replication
  10. 10Strunk, D. R., Lopez, H., & DeRubeis, R. J. (2006). Depressive symptoms are associated with unrealistic negative predictions of future life events. Behaviour Research and Therapy, 44(6), 861-882.primary
  11. 11Korn, C. W., Sharot, T., Walter, H., Heekeren, H. R., & Dolan, R. J. (2014). Depression is related to an absence of optimistically biased belief updating about future life events. Psychological Medicine, 44(3), 579-592.primary
  12. 12Hobbs, C., Vozarova, P., Sabharwal, A., Shah, P., & Button, K. S. (2022). Is depression associated with reduced optimistic belief updating? Royal Society Open Science, 9(2), 190814.replication
  13. 13Zabag, R., Vander Zwalmen, Y., Kube, T., Hoorelbeke, K., Koster, E. H. W., & Joormann, J. (2025). Belief updating deficits in depression: A systematic review and meta-analysis. Clinical Psychology Review, 121, 102649.meta-analysis
  14. 14MacLeod, A. K., & Byrne, A. (1996). Anxiety, depression, and the anticipation of future positive and negative experiences. Journal of Abnormal Psychology, 105(2), 286-289.primary

Every claim on this page was held against these sources on 2026-08-15. Nothing here is written from memory. Written and checked by Krishna Chagti; how entries are written and corrected. Found a mistake? Report it, with the sentence and a source.

About Optimism bias

Is Optimism bias real?

The data pattern reproduces reliably. Whether it shows optimism, or shows what an unbiased reasoner would produce in the same task, is still disputed in print. The effect sizes from the original study and from the replication are printed side by side above, so the comparison is visible rather than asserted.