Spectral integration in binaural signal detection

For both monaural and binaural masking, the spectral content of the masker and of the signal to be detected are important stimulus properties influencing the detection process. It is generally accepted that the auditory system separates the incoming signals in several frequency bands. It is not clear a priori, however, how the auditory system combines these separate band-limited signals. In this paper, a spectral integration model is presented that weights the binaural information of various auditory filters adaptively to minimize the errors due to internal noise. The model is tested for several binaural.signal detection conditions. It can account for several phenomena concerning binaural signal detection, including the level-dependent wider binaural critical bandwidth in band-widening experiments and the effect of transitions of the interaural masker correlation at a certain frequency. These results suggest that the auditory system combines information across auditory filters under conditions where such processing enhances the detection performance