Theory Comparison for Generalized Quantifiers

Premises and conclusions in classical syllogistic reasoning are formed using one of four quantifiers (All, Some, Some not, None). In everyday communication and reasoning, however, statements such as “most” and “few” are formed as well. So far only Chater and Oaksford’s (1999) Probability Heuristics Model (PHM) makes predictions for these so-called generalized quantifiers. In this article we (i) extend existing and develop new theories, (ii) develop multinomial processing tree (MPT) models for these theories, and (iii) conduct an experiment to test the models. The models are evaluated with G2, Akaike’s (AIC) and Bayesian Information Criteria (BIC), and Fisher’s Information Approximation (FIA). Mental modelbased accounts and PHM provide an equal account to the data.