Comparison of three adaptive psychophysical procedures

Abstract

Auditory thresholds were experimentally determined using three adaptive procedures. The procedures are a simple staircase procedure, a maximum-likelihood technique and PEST. The threshold of a sinusoid in noise (simultaneous task) and the threshold of a sinusoid in a forward masking experiment (successive task) were investigated. Number of trials per block, for the staircase and maximum-likelihood procedures, and stopping rule, for PEST, were varied in four different conditions. Nearly equal numbers of trials were obtained for each procedure. The observed variability in successive determinations of threshold was used to estimate standard errors of measurement. All three procedures produced standard errors of measurement that were nearly equal. Some slight differences among the procedures may be of interest for particular applications.

An observer-based psychoacoustic procedure for use with young infants.

“…On the other hand, a variation of the parameter estimation by sequential testing (PEST) rules for determining the frequency difference (Hall, 1981;Taylor &Creelman, 1967) gave thresholds for 6-month-olds in the same range as those we obtained in the conditioned head-turn procedure for 6-month-olds. This procedure uses more information than the last two trials to determine stimulus values on a given trial and tends to be more resistant to lapses of attention (Hall, 1981;Shelton, Picardi, & Green, 1982) on the part of the listener/observer.…”

Efficient estimation of sensory thresholds

“…In the same manner, the 4AFC paradigmis more efficientthan the other two because chanceperformance of 0.25 is easiest of the three to be identified. Shelton, Picardi, and Green (1982) comparedthe three methods in actual psychophysical experiments and reported that all three gave unbiased estimates of threshold. They noted that the maximum-likelihood methoddid not reach its theoretical efficiencyadvantage over the other two becauseof a tendencyafter a few correct trials (particularly those that are correct by chance to low stimuli)to jump to the lowest stimuluslevel possible and to remainthere for a numberof subsequent trials.…”

The step method: A new adaptive psychophysical procedure

“…If this experiment can be regarded as typical, the differences in bias and variability between the methods as found in the simulations are hard to detect when using real subjects. Other experiments that have compared psychophysical methods also found the methods to be more-or-Iess equivalent (Hesse, 1986;Madigan & Williams, 1987;Shelton et al, 1982). In this experiment, the range of presented stimuli was probably small (as it likely is in all experiments).…”

“…In other studies that compared different psychophysical methods (Hesse, 1986;Madigan & Williams, 1987;Shelton, Picardi, & Green, 1982), the threshold for only one experimental condition was obtained. Typically, however, we are interested in determining a threshold characteristic-the threshold as a function of some parameter.…”