Objective: This study was designed to evaluate threshold estimation procedures for cochlear implantees, both for clinical and research purposes. Precision and testing time of eight procedures were evaluated. These procedures included three variations of two adaptive tasks and two variations of an adjustment procedure.
Design: Electrical thresholds were obtained with eight different procedures. For each procedure, one of three tasks was combined with one of four decision strategies. With the adjustment task, current level was adjusted from "inaudible" to "just audible" or from "clearly audible" to "just audible." With the two adaptive tasks (i.e., "count the number of pulses" and "choose the interval with the pulse"), current level was adjusted during the sequence of trials by subjects' preceding responses according to three decision criteria (Levitt, 1971). Experimental data were collected on both an apical and a basal channel of six Laura cochlear implantees, and test-retest measures were obtained for each procedure, channel and subject. Moreover, total testing time was recorded in each test, to examine the trade-off with precision, and to investigate the feasibility of the procedures in a clinical setting.
Results: In general, the "count the pulses" procedures yield higher threshold values than the "choose the interval with the pulse" and adjustment procedures. Precision was not a statistically significant factor, although the test-retest variation was approximately a factor of 2 larger in the adjustment procedures than in the adaptive procedures. Test duration differences were large. Moreover, analyses of the adaptive procedures showed that a stable estimate requires at least six reversals and that the first four reversals should not be taken into account for threshold estimation.
Conclusions: In a clinical setting, where precision and time are important parameters, precise thresholds can be obtained with the adjustment procedure. However, for research purposes, it is advised to use the "choose the interval with the pulse" task together with one of the transformed up-down decision criteria.