Principles of an adaptive method for measuring the slope of the psychometric function

Abstract

Recent developments in the efficient estimation of threshold are here extended to the problem of how best to estimate the slope of the psychometric function. An adaptive method is described for selecting stimulus intensities that are optimal for slope estimation. A two-dimensional array of probabilities of different thresholds and slopes is used to calculate the stimulus intensity for the next trial; this array is updated after the trial, using Bayes' theorem to incorporate information from the subject's response. The practical implementation and efficiency of the method are demonstrated and discussed.