Proceedings of the Annual Meeting of the Cognitive Science Society

Research in psychology on reasoning has often been restrictedto relatively inexpressive statements involving quantifiers.This is limited to situations that typically do not arise inpractical settings, such as ontology engineering. In orderto provide an analysis of inference, we focus on reasoningtasks presented in external graphic representations wherestatements correspond to those involving multiple quantifiersand unary and binary relations. Our experiment measuredparticipants’ performance when reasoning with two notations.The first used topology to convey information via node-linkdiagrams (i.e. graphs). The second used topological andspatial constraints to convey information (Euler diagrams withadditional graph-like syntax). We found that topological-spatial representations were more effective than topologicalrepresentations. Unlike topological-spatial representations,reasoning with topological representations was harder wheninvolving multiple quantifiers and binary relations than singlequantifiers and unary relations. These findings are comparedto those for sentential reasoning tasks.