Link: https://doi.org/10.1525/collabra.147499
Abstract: Hindsight bias is the phenomenon that after learning facts about previously judged objects people tend to recall their previous judgments of the objects as closer to the facts than they actually were. Groß, Kreis, Blank, and Pachur (2023) found that hindsight bias emerged not only when people learned actual values for previously judged objects, but also when they learned the values of other objects in the same domain. Moreover, hindsight bias co-occurred with improved estimates for new objects. These findings challenge the traditional view that hindsight bias reflects a cognitive error and instead suggest that it results from adaptive knowledge updating. Groß et al. (2023) obtained their findings in the domain of country populations, a domain where people tend to be unfamiliar with the content and the numerical range (up to several million); this lack of familiarity may affect the link between knowledge updating and hindsight bias. In a high-powered conceptual replication (N = 300), we tested whether the findings generalize to the sugar content of food items—a domain where people are more familiar with both content and numerical range (up to 50 grams). Participants provided original judgments for items, learned numerical information, then recalled their original judgments, and lastly provided judgments for a new set of items. Our results replicate the key results of Groß et al. (2023), showing a close link between hindsight bias and knowledge updating in a more familiar domain. We discuss implications for theories of hindsight bias and propose directions for future research.