Tuesday, 11 November 2014

Optimism Bias and Belief Updating

Anneli Jefferson
I'm Anneli Jefferson, Visiting Research Fellow at King's College London, and I'm interested in philosophy of psychology and ethics, particularly in issues at the intersection of these two fields.

People tend to systematically overestimate their own abilities and their likelihood of positive future outcomes and to underestimate the likelihood of negative outcomes. Most commonly, this optimism bias is measured at a group level. While it is clear that some people must be making overly optimistic judgments if, for example, 70% of people asked think that they are less likely than their peers to experience a car accident, it can be difficult to tell which individual person’s outlook is actually unrealistically optimistic.

In a recent study, Sharot et al. used an update paradigm to measure the optimism bias at the individual level. Participants were asked to rate their likelihood of experiencing various negative events and were subsequently confronted with the average likelihood of experiencing the respective events. All estimations were then repeated, so that participants had the opportunity to update their initial estimate. It turned out that on average, participants were more likely to update their estimate in response to information that was desirable than when they received information that was undesirable.

For example, if a participant rated her likelihood of suffering from cancer at 30% and found out that the population-wide likelihood was 20%, she was likely to revise her estimates towards the more favorable base rate. If, on the other hand, a participant estimated her likelihood of suffering from cancer at 10% and was subsequently confronted with an average likelihood of 20%, she was significantly less likely to adjust her estimate upwards and, even when she did, the update tended to be smaller.

We modified this elegant paradigm in order to find out whether we could reproduce the bias towards neglecting undesirable information. In addition, we tested whether this tendency to exhibit optimistically biased belief updating was stronger for self-related judgments than for other-related judgments. To this end, we let participants estimate the likelihood of experiencing negative events in the future, then presented them with the average likelihoods of the respective events, and subsequently gave them the opportunity to revise their initial estimate. In half the trials, estimations referred to oneself, in the other half to a similar other (i.e. someone of a similar age and with a similar socio-economic background).

It turned out that by and large, belief updates were again greater in response to desirable than undesirable information and this effect was more pronounced when making judgments regarding one’s own future than when making predictions for a similar other. What is more, we found that the tendency to disregard undesirable information to a greater extent in self-related than in other-related judgments correlated with optimism as a personality trait, which we tested for separately with the Life Orientation Test (LOT-R).

So, overall, the findings demonstrate the optimism bias in belief updating, because new information calling for unfavorable updates was selectively dismissed. This tendency to neglect undesirable information was significantly stronger in judgments referring to oneself than to others, but only in persons with high trait optimism. Thus, the generally positive future expectations in persons with high trait optimism may reflect the perception that one’s own outlook is better than that of others, rather than reflecting positive expectations about the future more generally.


Sharot, T., Korn, C. W., & Dolan, R. J. (2011). How unrealistic optimism is maintained in the face of reality. Nat Neurosci 14(11), 1475-1479. doi: nn.2949 [pii]

Kuzmanovic, B., Jefferson, A. & Vogeley, K. Self-specific optimism bias in belief updating is associated with high trait optimism. Forthcoming in Journal of Behavioral Decision Making.


  1. Hi Anneli,

    What exciting results! I’ve thought a bit about doxastic biases, and what adaptive role they play. Peter Railton in his ‘Truth, Reason, and the Regulation of Belief’ claims that ‘[i]t would appear to be part of the normal, healthy operation of one’s self image that one discount negative evidence and defy the odds’ (1994: 92). I’m sympathetic to that view, it seems like something is going *right* when we make these judgments. I was just wondering what you think of this kind of claim? Given how prolific optimism bias is, might it be something which is (biologically?) useful for us to have, and indeed, no accident that we do have?

    Also, I wonder quite how much ‘negative evidence’ the bias can stand. I noticed that in your experiment, half of the participants were asked to give self-related judgements, and half other-related judgements. Do you think that the optimism bias with respect to self-related judgements would be maintained if participants were asked to give *both* self-related and other-related judgements after being told the average likelihoods? In that case, the contrast between self and other is presented to the subject, and so it might be harder the bias to be influencing the subjects' judgements?

    1. Hi Ema,
      Thanks for your comment. We did indeed have every participant do estimates for self and for similar others, half the trials where for self, the other half for similar other. (Sorry for not getting that across clearly enough) Whether there would be a change if participants were asked to re-estimate for themselves and a similar other at the same time, I am not sure.It's an interesting thought.
      Regarding the question of the beneficial nature of the optimism bias, I think there is certainly something to that. It probably helps resilience etc. if we are somewhat to optimistic. But it may of course end up being harmful when we are smoking 40 a day but are still very optimistic about our chances of contracting lung cancer...
      Also, I think it is actually quite hard to say when people are actually making mistakes in their estimates. We can tell that their updating pattern is skewed, but they may still have good reasons to think of themselves as special in any given scenario.

  2. Hi Anneli

    Thank you for your post and reply to Ema. I was wondering whether one could summarise your response in this way: if we are excessively optimistic about the future, we may be risk-takers and not act in a way that keeps us safe, but when something bad actually happens we may be better able to react positively and "bounce back".

    I wonder whether there is literature supporting this latter thought (which I am intuitively very sympathetic to). In the relationship literature, what I found is that general optimism gives rise to better coping strategies when difficulties in the relationship emerge. However, optimistic beliefs specific to the relationship or the partner do not predict good relationship outcomes (somehow they are a bit like smoking 40 cigarettes a day and underestimating the risk of lung cancer).

    Does this make sense to you? Have you got any other examples that would help understand how optimism and coping work?

    1. The exact effect of the optimism bias on behaviour and overall resilience is frustratingly hard to assess. So there certainly seems to be an association between positive affect and bias, or at the very least one between negative affect and not being biased, as there are numerous studies showing that depressed people do not have the same kind of bias. However, whether being optimistically biased is a good thing because it allows us to undertake endeavours we would not otherwise, or to be more successful in the ones we do undertake because we are less likely to give up, is a different question. And it is frequently perceived as detrimental in getting people to look after their health because of the 'it won't happen to me' attitude. There is a good summary of work on effect on behaviour at the end of Shepperd et al 2013 'Taking Stock of Unrealistic Optimism' (Perspectives on Psychological Science) though I am afraid their take home message seems to be: we don't know enough yet.


Comments are moderated.