[Seminar] Colloquia 3/8/2011

[Amy Barnes]

Dept. of Physics & Astronomy
Lewis Hall Rm 101
Tuesday, Mar 8th @ 4:00 p.m.
Refreshments 3:45 p.m.

William M. Hartmann
Department of Physics and Astronomy
Michigan State University

Peripheral Problems and Central Solutions in Human Sound Localization

The localization of a sound in space begins with an interaction between the incident sound wave and features of the listener's anatomy. This interaction causes variations in the waves in the listener's ear canals that systematically encode information about the location of the source, but there are important spatial and spectral limits on the reliability and usefulness of the information. As a result, the human central nervous system has developed, and continues to develop, strategies to cope with ambiguities and contradictions in the spatial information. These strategies can be modelled as dynamic weightings of localization cues.

Even in a free field, steady-state interaural cues require reweighting because of frequency-dependent complexities with increasing source azimuths. These include the azimuth dependence of interaural coherence and the consequences of the acoustical bright spot. In a room, with early reflections and reverberation, the complexities are increased, and so are the demands on the reweighting of steady-state and transient cues. Experimental evidence on cue reweighting in the azimuthal plane reveals similar behavior for different listeners. By contrast, the weighting attached to the spectral cues in sagittal plane localization reveals large individual differences. This is especially true for the distinction between directly in front and directly in back, an area in which new experiments with unprecedented accuracy show that some listeners can make the distinction based only on information below 6 kHz, while other listeners require broad bands.

To model the cue weighting observed experimentally, there are signal processing circuits that make plausible contact with the known physiology of brain stem and midbrain nuclei. Some sound localization experiments can be successfully approached in terms of these models. Other experiments, particularly those with long time constants, seem to require a great deal of high-level processing. An appealing combined model postulates a centrally directed sound localization system that off-loads some of the less complicated tasks to peripheral neural sites.

ALL GRADUATE STUDENTS ARE REQUIRED TO ATTEND


Amy Barnes
Administrative Secretary
Dept. of Physics & Astronomy
108 Lewis Hall
University, MS 38677

Office: 662-915-7046
Fax: 662-915-5045