Interactions of masker bandwidth with signal duration and delay in forward masking

Forward-masked thresholds for 1-kHz sinusoidal signals were measured as a function of the bandwidth of a noise masker centered at 1 kHz, using a two-interval forced-choice technique. The noise spectrum level was 40 dB SPL/Hz, and noise bandwidth was varied from 50 to 1600 Hz. In experiment I signal duration was varied, with a fixed offset-onset time of 5 ms between masker and signal. For the shortest signal (5 ms) threshold at first increased with increasing bandwidth and then decreased. As signal duration increased, the bandwidth at which maximum masking occurred (the rollover bandwidth) decreased, and for the longest signal (45 ms) maximum masking occurred for the narrowest bandwith tested. In experiment II the silent interval between masker and signal was varied, for a signal of 5-ms duration. Again threshold at first increased with noise bandwidth, and then decreased. However, the rollover bandwidth decreased with increasing silent interval, and threshold varied less with bandwidth. In experiment III signal duration was varied for a fixed offset-offset time of 50 ms between masker and signal. Changes in threshold with noise bandwidth were small, except for the longest signal (45 ms) for which threshold increased markedly at the narrowest bandwidth used. For the wider nose bandwidths, threshold decreased gradually with increasing signal duration, whereas for the narrowest noise bandwidth, threshold increased with increasing duration. It is argued that interactions of excitation and suppression within the internal representation of the masker influence the results, but at least two other factors, detection of energy splatter in the spectrum of the signal and the presence or absence of quality differences between masker and signal, have a powerful influence.