CRCNS: The role of sound statistics for neural discrimination and coding of sounds

The brains ability to accurately recognize and categorize sounds despite highly variable acoustic structure and in the presence of environmental noises is a remarkable trait of the normal functioning auditory system. How the brain achieves this seemingly simple task is poorly understood and new working models are necessary to understand how the brain deals with and exploits statistical structure for basic auditory recognition tasks. Since background noise and acoustic variability present significant challenges for individuals with hearing loss, this knowledge is essential to develop biologically informed auditory prosthetic that account for normal acoustic variability and can deemphasize unwanted background noise.

This study proposes a novel framework to understand how the normal auditory system encodes high-level statistical regularities in sounds that are essential for sound discrimination and recognition behavior. Computational approaches will be used to identify salient high-level statistical features in sounds that contribute to sound recognition. Neural recordings in two critical mammalian auditory structures (inferior colliculus, IC; auditory
cortex, AC) of the awake rabbit will be obtained to identify neural coding mechanisms that exploit statistical structure for sound recognition. The study will provide the groundwork for developing a general theory of how the brain encodes and discriminates sounds based on high-order statistical features and will enable future auditory prosthetic that complement normal hearing function.