Proceedings of the Annual Meeting of the Cognitive Science Society

Inductive generalizations about the properties of kinds arebased on evidence. But evidence can come either from ourobservations, or from the testimony of knowledgeableinformants. The current study explores how we combineinformation from these two sources to make inductiveinferences. Participants learned about a novel object category,and observed the property occur with some frequency in asample of category members. Different groups of participantsalso heard an informant making either Generic, Quantified, orSpecific claims about the prevalence of the property.Participants who heard generic claims were more resistant toa straightforward use of statistical evidence in theirgeneralizations. Moreover, participants who rated theinformant as more knowledgeable (across conditions) gavehigher prevalence estimates. The results suggest twopathways through which testimony translates into evidencefor category learning, and raise questions on how to bestcombine evidence from these different sources into acommon representational form.