Believing is seeing: Pre-existing beliefs influence information search and shape decisions in dynamic social environments

We investigated whether gender-based societal beliefs shape the way individuals gather information and draw conclusions—an effect that may support the persistence of gender stereotypes. Three pre-registered experiments (N = 210) used a dynamic sampling paradigm, in which participants estimated the gender distribution of the workforce of fictitious companies representing highly gendered professions (e.g., mechanic, hairdresser). Participants could learn about the workforce of each company by viewing up to ten pictures of its employees, in which the gender of the first two faces either confirmed, disconfirmed, or remained neutral with regard to societal expectations about the profession. When initial evidence matched (vs. mismatched) expectations, participants sampled fewer faces. And when participants did view additional faces in belief-confirming sequences, these additional faces also skewed towards the gendered expectations, suggesting that participants actively employed strategies leading to selective exposure. Furthermore, both gender-based beliefs and newly sampled information influenced estimates of the workforce gender distributions, with larger effects of newly sampled evidence as the participant viewed more faces. Thus, pre-existing gendered beliefs shaped information sampling and social judgements. Furthermore, early exposure to belief-disconfirming information elicited more extensive and less biased information search. More extensive information search, in return, led to less biased social judgements. This research shows that when social expectations are quickly confirmed, people tend to not look any further, whereas initial belief-challenging information elicits more extensive exploration and less biased conclusions.