A Investigation of the Contribution of Auditory - to the Comodulation Masking Release Effect

The comodulation masking release (CMR) effect involves improved detection of a pure-tone signal in an amplitude-modulated (AM) masker by an addition of AM noise spectrally distant from the signal (i.e, outside the critical -band of the signal). The underpinning of CMR has been elusive, and despite many previous investigations of this phenomenon it remains unexplained. A plausible explanation of CMR involves perceptual, auditory-grouping of acoustic signals. The purpose of this project was to investigate an hypothesis based on auditory-grouping to explain the CMR effect. A series of three experiments was completed towards this purpose. In the first, signal detection thresholds were obtained for a 2000-Hz pure-tone centered in narrow-band, AM maskers. Narrow -band, AM "flankers" centered at 1700-Hz were presented concurrently with the masker-band. The AM of the flanker -band was either identical to (correlated condition), or independent of (uncorrelated condition) the AM of the masker -band. Thresholds for the tone were about 3 dB better (lower) in the correlated condition, overall--a CMR effect of about 3 dB. In comparisons of specific correlated/uncorrelated noise-band pairs the magnitude of the CMR effect was found to vary substantially. CMR magnitude appeared to be related to the degree of envelope correlation of the "uncorrelated" noise-band pairs. In experiment two, strength of vertical -fusion of correlated and uncorrelated noiseband pairs was inferred from the interstimulus-interval (ISI) necessary to "capture" the flanker-band into a horizontal-stream. Shorter ISIs were needed to capture the flanker-band in correlated noise-band pairs than in uncorrelated noise-band pairs. This suggested that the strength of vertical-fusion of noise -band pairs was greater in correlated conditions than in uncorrelated conditions. In addition, the ISIs needed to capture the flanker-band were shorter for the "uncorrelated" noise-band pairs where the absolute value of the correlation coefficient was largest. In the third experiment, thresholds for the pure-tone were obtained in correlated noisebands. Two conditions were created, and denoted "weakly-fused" condition and "strongly-fused" condition. According to the auditory -grouping hypothesis of CMR, thresholds should have been better in the strongly-fused condition than in the weakly -fused condition. A significant difference for threshold was not found, however, between the two conditions. This finding suggests that auditory-grouping does not play a dominant role in CMR. Alternative explanations of why a difference for threshold was not found between the two conditions are offered.