Behavioral and computational modeling studies of the precedence effect in humans

A sound generated in a room follows multiple paths, and the sound that arrives at a listener's ears includes multiple versions of the original sound, depending on the reflections and other interactions with walls and objects in the room. Remarkably, the ability of a listener to localize a sound source in a room is largely unaffected by the presence of these reflections. Since sound energy directly from the source reaches a listener's ears earlier than the reflected energy, it has been hypothesized that our ability to localize sounds in rooms benefits from specialized auditory processing that emphasizes location information in earlier arriving sound components relative to later components. This attribute of sound localization has come to be known as the “precedence effect” and has been studied for several decades in limited situations. Related physiological investigations of neural structures known to be involved in sound localization have suggested that inhibition processes within and between these structures may be useful in describing localization consistent with the precedence effect. Still, the functional processing that results in the precedence effect is not well understood. Evaluation of models of performance is made difficult by the narrow range of stimuli with which precedence has been examined.; This thesis includes both new behavioral experiments and computational modeling. The experiments were performed to measure the precedence effect over a wider range of source stimuli than used previously. The computational model can be used to predict results from a number of experiments and includes mechanisms modeled after the physiological inhibition processes.; The results of the psychophysical experiments indicate that the ability of subjects to suppress the influence of reflections on sound localization is robust to temporal properties of the sound source, yet dependent on the frequency content of the sound source. The computational model of localization is consistent with much of the behavioral data as well as physiological observations. Since this performance was directly attributable to the inhibition processing, we suggest that the physiological inhibition is of fundamental importance in enabling accurate localization of sounds in the presence of reflections.