ASA 124th Meeting New Orleans 1992 October

2pSP2. Evaluation and modeling of channel independence and uncertainty processing in cochlear implant subjects.

Leslie M. Collins

Gregory H. Wakefield

Dept. of Elec. Eng. and Comput. Sci., 1301 Beal Ave., Univ. of Michigan, Ann Arbor, MI 48109-2122

Results from previous experiments [Collins and Wakefield, Midwinter Research Meeting, Association for Research in Otolaryngology (February 1992)] have shown that performance in temporal pattern discrimination tasks is an efficient and valued psychophysical predictor of speech recognition. In the present paper, these measures are refined by extending the discrimination task to include temporal patterns across two electrode ``channels.'' Data from implanted subjects are presented and related to both speech and discrimination performance in the single channel case. A model that accounts for the discrimination data is proposed and used to evaluate channel independence. This model utilizes an alternative optimal statistic to that suggested by Sorkin [Sorkin, J. Acoust. Soc. Am. 87, 1695--1701 (1990)] for determining the performance of an ideal receiver. While this alternative model provides a better fit to the single channel data from Sorkins's experiment, both the alternative model and Sorkins's model fail to account for the worst performance of the subjects. Additional modifications to the model that reflect the impaired ability of some subjects to handle signal uncertainty are described and evaluated. [Research supported by NIH NIDCD NS21440.]