BINAURAL SOUND LOCALIZATION

Binaural Sound Localization

Binaural Sound Localization

Introduction

The auditory system of humans and many other animals is able to localize sound sources with amazing precision. This ability is partially possible with only one ear (monoaural hearing), yet for localization in the horizontal plane two ears are necessary (binaural hearing). Sound source localization can be enhanced when the source and the receiver move relative to each other (Phillips and Brugge 1985). In this paper, however, we limit the task of sound localization to static sources using binaural cues at low frequencies. Binaural cues determine the azimuth defined as follows. The vector from a listener to a sound source is projected perpendicularly onto the horizontal plane. The angle between the projected vector and a reference vector, forming the intersection of the horizontal plane with the plane of head symmetry, oriented to the front, is called the azimuth.

One of the parameters influencing binaural sound source localization is the fundamental sound frequency. In mammals, for low fundamental frequencies (below 1,500 Hz) or for broadband sounds, the interaural time difference (ITD) is the dominant sound localization cue. For high fundamental frequencies, the interaural intensity difference (IID) is used. While this paper deals with lower frequency bands, it is possible that higher frequencies are processed with the use of similar neuronal algorithms, as we proposed in earlier studies by Marsalek and Kofranek (2004, 2005). This paper presents a theory of how binaural sound localization for low frequencies might be realized in mammals and particularly in humans. The theory of Jeffress (1948) is one of the first well-known attempts to explain how neuronal circuitry achieves this. His prescient work is still frequently cited (Joris et al. 1998). Jeffress' visionary hypothesis asserted that the ITD is converted to a binary signal in a higher order neuron through an array of delay lines of fibers in lower order neurons from both sides. Pioneering experiments by Carr and Konishi (1988) showed that Jeffress was correct in case of birds. As far as we know, the existence of an analogous delay line in mammals remains an open question (Grothe 2003, Joris and Yin 2007, McAlpine and Grothe 2003). What other neural circuit mechanism might be responsible for calculating the azimuth from the ITD? In this paper we propose an alternative to the delay line array model based on recent physiological evidence.

This alternative is a stochastic delay of a very small number of broadly tuned channels (McAlpine and Grothe 2003). The amazing time precision (Joris et al. 1998) in the range of tens of microsecond points towards another statement of Jeffress that the neurons of the circuit should be located among the lower order neurons of the auditory pathway. The lowest order suitable neuron is the first binaural neuron. The information about the sound source location contained in the ITD is implicitly encoded by spike trains of lower order neurons. The first binaural neurons function as encoders of the ITD. The circuit has to make the information accessible, in other words ...