In naturalistic settings, it is sometimes ambiguous whether or not a particular exemplar exhibits various properties indicative of category membership. We show that individuals prefer to group exemplars into categories that have a narrower latent scope – categories that specify fewer unobserved features – to those with a broader scope. In Experiment 1, participants classified suspects with fewer unobserved features as more likely to have committed a crime. In Experiment 2, participants were more likely to categorize novel creatures as part of a group that had a narrower latent scope. Experiment 3 showed that participants were more willing to make inductions about unknown features based on a narrower latent scope categorization.