Proceedings of the Annual Meeting of the Cognitive Science Society

Syllogistic reasoning is one of the oldest domains of reasoning research and has made great advances in understanding and modeling human reasoning processes. However, the field was mostly focused on a traditional set of quantifiers originating in first-order logic, thereby neglecting the large variety of quantifiers humans use when engaging in reasoning in their everyday life. The present work makes three main contributions: (I) we conducted a study yielding a dataset covering all traditional syllogisms and tasks containing generalized quantifiers ``most'' and ``most not'', providing a starting point for existing theories and models to transition to generalized quantifiers. (II) based on the dataset, we analyze the impact that the additional quantifiers have on the reasoning behavior. (III) We investigated the reasoning behavior with respect to the difference between traditional and generalized quantifiers, gaining insights into some of the peculiarities of the domain of generalized syllogisms.