Forward masking by maskers of uncertain frequency content

Abstract

Large amounts of simultaneous masking can be produced when the frequency content of multicomponent maskers is changed with each presentation. This study examined whether such masker uncertainty also influences forward masking. The forward maskers were a 1000-Hz sinusoid, a broadband noise, and tonal complexes composed of 2-100 sinusoids drawn at random from a 300 to 3000-Hz range. A 10-ms signal at 1000 Hz followed the offset of 200-ms, 60 to 80-dB maskers with delays of 4-32 ms. For comparison, simultaneous-masked thresholds were measured for a 200-ms signal at 1000 Hz for a subset of conditions or taken from an earlier study. Unlike simultaneous masking, forward masking showed little effect of masker uncertainty. As confirmed by fitting a three-parameter descriptive model of forward masking, the pattern of results was similar for the sinusoidal, noise, and multicomponent forward maskers and there was no reduction in masking with decreased masker uncertainty. Performance in simultaneous masking improved to equal that in forward masking when the signal was shortened from 200 to 10 ms. Thus temporal disparities between masker and signal can offset the effects of spectral uncertainty and help separate the relative contribution of peripheral versus more central processes.