A physiological model for comodulation masking release based on cross-frequency coincidence detection

Abstract

Detection thresholds for tones in noise can be significantly improved when the masker is amplitude modulated, or when its various components are comodulated [e.g., Hall et al., J. Acoust. Soc. Am. 76, 50–56 (1984)]. Several stimulus-based signal-processing models have been proposed to explain this phenomenon, referred to as comodulation masking release (CMR); however, no physiologically based models have successfully described the psychophysical data. These models have tested hypotheses that CMR involves either cross-frequency processing strategies or within channel envelope-based processing strategies. Cross-frequency coincidence detection is a physiologically realistic, nonlinear processing mechanism that is consistent with suggestions that information from different frequency channels is combined to explain CMR. Furthermore, cross-frequency coincidence detectors are sensitive to envelope cues, thus this mechanism is also consistent with within-channel envelope-based models for CMR. The sensitivity of coincidence detectors to envelope cues is due to the nonlinear timing of low-frequency auditory-nerve (AN) fibers—changes in amplitude of the stimulus result in systematic changes in the relative times of phase-locked AN fibers tuned to different frequencies [Heinz et al., J. Acoust. Soc. Am. 110, 2065–2084 (2001)]. A cross-frequency coincidence-detection model can explain the psychophysical results for the basic CMR paradigm described in Hall et al. (1984). [Work supported by NIH-NIDCD.]