Neural system identification model of human sound localization

This paper examines the role of biological constraints in the human auditory localization process. A psychophysical and neural system modeling approach was undertaken in which performance comparisons between competing models and a human subject explore the relevant biologically plausible "realism constraints." The directional acoustical cues, upon which sound localization is based, were derived from the human subject's head-related transfer functions (HRTFs). Sound stimuli were generated by convolving bandpass noise with the HRTFs and were presented to both the subject and the model. The input stimuli to the model were processed using the Auditory Image Model of cochlear processing. The cochlear data were then analyzed by a time-delay neural network which integrated temporal and spectral information to determine the spatial location of the sound source. The combined cochlear model and neural network provided a system model of the sound localization process. Aspects of humanlike localization performance were qualitatively achieved for broadband and bandpass stimuli when the model architecture incorporated frequency division (i.e., the progressive integration of information across the different frequency channels) and was trained using variable bandwidth and center-frequency sounds. Results indicate that both issues are relevant to human sound localization performance.