Proceedings of the Annual Meeting of the Cognitive Science Society

We present a natural logic for reasoning with quanti-fiers that can predict human performance in appro-priate reasoning tasks. The model is an extension ofthat in (Geurts, 2003) but allows for better fit withdata on syllogistic reasoning and is extended to ac-count for reasoning with iterated quantifiers. Weassign weights to inference rules and operationalizethe complexity of a reasoning pattern as weightedlength of proof in our logic – this results in a measureof complexity that outperforms other models in theirpredictive capacity and allows for the derivation ofempirically testable hypotheses.