Reproducible noise discrimination with concurrent narrowband noises

Abstract

Reproducible noise discrimination ability was measured with a cued 2IFC procedure. Stimuli were either one, two, or three concurrent noise bands with bandwidths of either 12.5, 50, or 200 Hz and center frequencies of either 500, 1250, or 3125 Hz. In the multiband conditions, the concurrent noise samples either shared a common bandwidth and a common pattern of envelope fluctuation, had a common bandwidth with asynchronous envelopes, or had different bandwidths. Performance in the single versus multiband conditions was used to estimate cross-spectral correlations in the integration of envelope information. Single-band d′’s were inversely related to both bandwidth and center frequency. Performance in the multiband conditions was generally better than with the component bands presented individually. For subjects whose single-band d′’s were widely separated, greatest integration was obtained in the synchronous multiband conditions. Correlation estimates were significantly higher in the synchronous versus the two asynchronous condition sets. In all conditions, estimated correlation was inversely related to bandwidth. Results are used to model internal noise levels in cross-spectral envelope processing. [Work supported by NIH.]