A hemispheric two-channel code accounts for binaural unmasking in humans

The ability to localize sound sources relies on differences between the signals at the two ears. These differences are also the basis for binaural unmasking, an improvement in detecting or understanding a sound masked by sources from other locations. The neurocomputational operation that underlies binaural unmasking is still a matter of debate. Current models rely on the cross-correlation function of the signals at the two ears, the neuronal substrate of which has been observed in the barn owl but not in mammals. This disagreement lead to the formulation of an alternative coding mechanism where interaural differences are encoded using the neuronal activity within two hemispheric channels. This mechanism agrees with mammalian physiology but has not yet been shown to account for binaural unmasking in humans. This study introduces a new mathematical formulation for the two-channel model, which is then used to explain the outcome of an extensive library of psychoacoustic experiments.