Function-based modeling of binaural processing: interaural phase

The intent of function-based modeling is to predict from theoretical considerations the input/output characteristics of a neuron if it were to perform a specific signal processing task within a sensory system. In this study, the sensory task is localization within the horizontal plane of a sustained low-frequency tone using interaural phase cues alone. The stimulus is assumed to be represented by primary-like discharge patterns from each ear. These patterns serve as the input to a system whose task is to extract the information needed to determine azimuthal angle. The optimal system resembles the classic Jeffress model, but differs from it in important details. In particular, the Jeffress model is shown to be a sub-optimal approximation in that it ignores the characteristic frequency of the inputs. The physical realization of either model by a structure resembling currently known MSO anatomy and innervation is discussed.