Abstract:
Noise discrimination ability was measured with a cued 2IFC procedure in which a wideband noise carrier was modulated by bandlimited noise. Modulator bandwidth was constant for the three stimulus presentations on each trial with the signal-interval modulator a repetition of the cue modulator. The modulator bandwidth ranged from 10 to 5000 Hz. The carrier was either a random or frozen sample of wideband noise across the three presentations per trial. In these conditions, discrimination among noise samples is based solely on envelope and not fine-structure characteristics. Performance was better with the frozen rather than random carrier samples with discrimination ability inversely related to modulator bandwidth. Data trends were modeled with an envelope detector providing the input to a sequential equalization-cancellation mechanism. In a second set of conditions, rippled noise was used as the modulator for the cue and signal intervals. Ripple delay ranged from 1 to 64 ms with the number of iterations ranging from 1 to 16. The ability to discriminate between rippled and Gaussian modulators of equal bandwidth increased with the number of iterations and was also inversely related to delay, suggesting a type of envelope processing similar to that involved in pitch perception. [Work supported by NIH.]