A descriptive Bayesian account of optimism in belief revision

A number of findings suggest that people's expectations about the future are unrealistically optimistic (e.g. Sharot et al., 2011). This bias is thought to result from the “motivational modulation” of evidence, driven by the desire to feel positively about one's own future (Sharot, 2011). However, evaluating “bias” in belief revision requires careful comparison against a rational standard, and recent arguments and findings (Shah et al., 2016) give reason to doubt much of the evidence for optimism bias. Descriptive Bayesian models allow for a direct comparison of human belief updating against the Bayesian rational standard (Tauber et al., 2017). Here, these analyses indicate widespread “conservatism, ” or weaker-than-rational belief revision. However, in contrast to the widely-reported “optimism bias, ” participants more commonly displayed pessimism than optimism in their belief revision. Both effects were marked by significant heterogeneity, with a sizable fraction of participants engaging in largely rational updating.