Relative informativeness of quantifiers used in syllogistic reasoning

Three experiments tested a possible resolution of the probability heuristics model (PHM) of syllogistic reasoning proposed by Chater and Oaksford (1999), with their experimental results apparently showing that the generalized quantifier few was not as informative as suggested theoretically. Modifying the interpretation of few to take into account the distinction between positive and negative quantifiers (Moxey & Sanford, 1993) indicated two orderings over the quantifiers all, most, few, some, none, and some... not that are more consistent with the results. Experiments 1-3 tested these orderings empirically by having participants rank whether a quantifier applied to a particular probabilistic state of affairs. Experiments 1 and 2 showed that participants agreed on when a quantifier applied and that the empirically derived informativeness orderings were consistent with the proposed modifications of the order. Experiment 3 showed that this finding was robust even when response competition was eliminated