Abstract

Generic sentences such as “Birds lay eggs” are used frequently and effortlessly, but there is no simple quantitative rule that determines whether they are true or false. For instance, while “Birds lay eggs” is considered true, “Birds are female” is considered false, even though there are necessarily fewer birds that lay eggs than birds that are female. In this article, we adopt a cognitive perspective on genericity. Specifically, we draw on learning principles that predict asymmetries in the acquisition of category representations, which in turn might determine the acceptance of generic sentences. Our key hypotheses were that generics are more likely accepted when the attributes they refer to are distinctive (i.e., more prevalent in the category relative to comparison categories) and that this pattern is sensitive to the temporal order in which category information is acquired. We report three preregistered experiments to test these hypotheses. In all experiments, we employed a trait‐learning paradigm in which participants received information about exemplars of two fictitious kinds (human‐like sea creatures in Experiments 1–3, stones in Experiment 2) in sequential order. We manipulated the prevalence of attributes within kinds, as well as their status as being shared between kinds or distinctive for either the first‐ or second‐learned kind. As hypothesized, generic sentences were more likely accepted when referring to distinctive (vs. shared) attributes, but only for the second‐learned kind. We discuss implications for theories of generics as well as stereotype formation and representation.