Journal

A psychometric function is used to summarize classifi-cation performance (such as detection or discrimination) from a psychophysical experiment. An observer classifies events within a limited set of response categories over a series of trials. The psychometric function is a sigmoidal curve that describes the probability of a correct classifica-tion as a function of stimulus strength (Falmagne, 1982; Klein, 2001). If there are n choices, the lower asymptote should approach 1/n. Typically, the upper asymptote is ex-pected to approach 1. Given a sufficiently difficult task, however, the observer might not achieve perfect perfor-mance over the realizable range of stimulus values (see, e.g., Higgins, Arditi, & Knoblauch, 1996). The raw data consist of the numbers of trials on which the observer correctly and incorrectly classified the stimu