The process wherein high frequency noise bands or tone complexes above 75 to 80 dB SPL mask low frequency signals that lie more than one octave below the lower cutoff frequency of the noise is known as remote masking (RM). Its mechanism is not yet understood. It has been suggested that RM be used as a test to assess cochlear conduction. Prior to recommending its use as a diagnostic test, it is important to understand the underlying mechanisms of RM. The present study was, therefore, undertaken to investigate the mechanisms of RM.
Two versions of a cochlear saturating nonlinearity model namely the masker envelope detection model (MEDM) and the quadratic distortion products model (QDPM) were proposed in this study to predict RM. Both versions assume that the cochlea introduces low frequency distortions to its input due to its saturating nonlinearity. The resulting distortions mask low frequency signals and produce RM. The only difference between these models is that the MDEM assumes that the nonlinearity is realized by extracting the masker envelope whereas the QDPM assumes that the nonlinearity is realized by a quadratic process.
Two experiments were designed to evaluate these models. Remote masking was measured at 250, 350, 500, and 700 Hz probe frequencies in five normal hearing young adults (mean age = 22.3 years) using equal-power NBN and tone-complex maskers. The masker presentation levels were varied from 80 to 95 dB SPL in 5 dB steps for each of the four different maskers.
The findings of the study indicated that the models predicted RM for a variety of masking conditions reasonably well. However, both the MEDM and QDPM could not predict the effects of masker bandwidth completely. The observed RM at 700 Hz for 1 ERB wide TC2 maskers was also not consistent with the QDPM. A revision of the model is required. One possible revision might assume that distortions due to the cochlear saturating nonlinearity include higher order distortion products. Other possible revisions might include the complex role of envelope fluctuation cues during signal detection. These assumptions have not been tested and warrant further investigation.