Reliable discrimination thresholds in 17 trials

Abstract

This study explores potential weaknesses of the Zippy Estimation by Sequential Testing (ZEST) psychophysical procedure as a means of acquiring rapid psychoacoustic measurements. Although ZEST is promising for this purpose, previous work used a priori knowledge to make optimal choices for initial assumptions, a best case scenario for ZEST’s performance. Before the ZEST procedure can be employed as a clinical tool, an understanding of how it performs in the absence of a priori information is needed. Specifically, an investigator must choose: (1) the model psychometric function, (2) the starting level, and (3) the number of trials. The present study explores sensitivity to these choices when ZEST is employed in 2AFC, frequency-discrimination tasks. Data for six normal listeners were obtained for a wide range of initial conditions and compared with simulations. These data indicate that even when an inappropriate psychometric function is used, reliable thresholds can be obtained with only 17 trials when the starting level is within a factor of four times the listener’s “true” threshold. These results suggest that ZEST combined with a 2AFC paradigm is a promising candidate for rapid and reliable assessment of listeners' discrimination thresholds. [Work supported by NIH/NIDCD 1R03DC009071.]