People update their beliefs accurately only when things turn out to be better than expected, not when they're worse than expected reports a study published online this week in Nature Neuroscience. Correspondingly, neural activity faithfully encodes feedback about things being better than previously thought, but the encoding of unexpectedly negative information is much weaker.
Tali Sharot and colleagues showed people a list of adverse life-events — such as getting Alzheimer Disease, or being robbed — and asked them to estimate how likely it was that these events would happen to them in the future. After each rating, people were provided with the actual probability of each of these events. Finally, these subjects rated the likelihood of the same events again, in order to measure how feedback about actual probabilities changes people's estimates of the likelihood of adverse life events.
The authors found that people were are far more likely to change their estimations later when the feedback indicated that, in reality, they were much less likely to suffer an adverse event than they had originally estimated. In contrast, when people were told that an adverse event was much more likely that they had originally estimated, they still tended to give the original, incorrect estimate.
The authors' team also tracked brain activity during this task, and, corresponding to the behaviour, they find that activity in frontal areas of the brain faithfully tracks estimation errors when things are better than expected, but its tracking of estimation errors when things are worse than people originally thought was much weaker.
These results highlight a seemingly built-in optimism bias in the human brain, which is resistant to accurate information about the world.