Latent scope bias in categorization

• We investigate how people categorize exemplars given incomplete information.
• We define scope as the number of distinct features category membership implies.
• Results show bias for grouping exemplars in categories with narrower latent scope.
• Preferences extend to verbal and visual categorization tasks.

Categories often have unobservable diagnostic features. For example, if a person is a lawyer, one might expect him to be both well dressed and knowledgeable about the law. However, without observing the person in a courtroom, one cannot tell whether or not he is knowledgeable about the law. How might we categorize the well-dressed person before we know whether or not he possesses a particular category feature? Two studies showed that, all else equal, individuals prefer to group exemplars into categories that specify fewer unobserved and unobservable features — i.e., those that have a narrower latent scope — to those with a broader latent scope. In Experiment 1, participants were more likely to classify novel exemplars as part of a social category that had a narrower latent scope in a verbal task. Experiment 2 demonstrated that the scope bias generalizes to contexts in which category structure is never explicitly specified.