Application of Wald sequential test in staircase up-down adaptive procedures

Abstract

An advantage of adaptive, staircase, up-down procedures in sensory threshold measurements is that such procedures are based on simple rules for signal level setting and are relatively robust to fluctuations of the subject’s attention during the measurement. A disadvantage of those procedures is that they allow to estimate the threshold level only for a few points on the psychometric function. In this study an adaptive, up-down staircase procedure is proposed which uses a decision rule based on the Wald sequential statistical test, similar to that applied in the PEST adaptive method. The Wald test allows to determine the signal threshold level for any point on the psychometric function and is simple to implement. Numerical simulations and results of an experiment performed on real subjects have shown that the staircase procedure combined with a Wald test for signal level setting reproduces the tracks of standard up-down procedures and may estimate the sensory threshold with similar accuracy. The modified procedure is thus plausible for various kind of threshold measurements in psychoacoustic studies.