An optimum value for detection probability

In many psychoacoustic experiments, an experimenter wishes to determine the effects of one or more treatments Ito the subjects, or to the stimuli, or to both) with as little experimental labor as possible. The experimenter is often able to control the similarity of the stimuli and is thus free to choose a base level of performance before the treatment is applied. On the basis of the model presented in this paper, the base level of performance that best enables any change in performance due to the treatment to be detected is predicted. A relatively high base-level performance of 95% correct discrimination results from the model, although practical constraints may dictate a somewhat lower value. An experiment (involving pitch discrimination) that provides data supporting the model is also reported. The model is applicable only to the two-alternative/forced-choice experimental paradigm.