M. G. Heinz
M. H. Goldstein, Jr.
Dept. of Elec. and Comput. Eng., Johns Hopkins Univ., Baltimore, MD 21218
Johns Hopkins Univ. School of Medicine, Baltimore, MD 21287
T. G. Forrest
Cornell Univ., Ithaca, NY 14853
Computer simulations were implemented for estimating detection thresholds for silent temporal gaps in sinusoidal markers that differed in frequency. The stimuli were presented in blocks of 50. 2AFC trials in the same adaptive paradigm used to measure human temporal gap detection (TGD) thresholds [Formby et al., companion paper at this meeting]. The stimuli were preprocessed by a multi-channel, multi-resolution cochlear model. The digital simulation of the cochlear model was implemented by a 30-channel basilar membrane filter bank, followed by 30 hair cell/synapse sections [Liu, Ph.D. dissertation, Johns Hopkins Univ. (1992)]. The output for each channel of the cochlear model was lowpass filtered, and a max--min ratio [Forrest and Green, J. Acoust. Soc. Am. 82, 1933--1943 (1987)] was estimated for each of the 30 channels. A majority vote across the 30 max--min ratio values on each trial was used to select the observation interval containing the silent gap. Results from this majority voting scheme were qualitatively similar to, but consistently poorer than, human psychoacoustic TGD thresholds [Formby et al., companion paper at this meeting]. Other decision rules are being investigated and will be presented for simulating human thresholds. [Research supported by NIH.]