Across-frequency integration in spatial release from masking

Spatial separation of a target (T) stimulus from a masker (M) often improves detectability of the target, a phenomenon known as the spatial release from masking (SRM). When the masker is a noise, two main factors contribute to SRM: changes in the target-to-masker ratio dominate the performance at high frequencies, while binaural processing dominates at low frequencies. Previous neurophysiological studies (e.g., Lane et al., ISH 2003) suggested that, at the level of inferior colliculus, the SRM of broadband stimuli is determined by a single unit – the one that is most sensitive in the given T/M spatial configuration. Based on this observation, Lane et al. proposed a simple model that used the assumption that the channel with the most favorable signal-to-noise ratio also determines behavioral performance. The current study evaluated this model psychophysically. First, several T/M spatial configurations were selected based on the criterion that they must have a narrowband spectral region with very favorable SNR (re. other spectral regions). The stimuli were then filtered so that they would activate mainly the peripheral channel with the most favorable SNR. Detection thresholds were then measured for the filtered and the unfiltered stimuli, both binaurally and monaurally. Large differences (up to 10 dB) in performance were observed, with binaural thresholds generally better than the corresponding monaural thresholds, which, in turn, were better than the single-channel thresholds. These results support the singlechannel model only partially. However, they do not prove that across-channel integration plays a role in spatial release from masking.