Auditory processing of complex sounds is critical for perception and communication in many species, including humans, but surprisingly little is known about how high level brain areas accomplish this task. We propose to tackle this question in the auditory forebrain of songbirds, using complex natural sounds, including bird songs. We hypothesize that the systematic use of this rich class of auditory stimuli, of ethological relevance to the animal, will enable us to reveal complex and nonlinear aspects of high level auditory processing, and will demonstrate that auditory forebrain neurons are optimized to efficiently represent these stimuli. In our first aim, we will therefore characterize the natural statistics of ensembles of bird song, and develop new covariance-based methods for quantifying neural responses to these sounds, including nonlinear aspects of the responses. In our second and third aims, we will use the methods developed in Aim 1 to describe receptive fields of the auditory forebrain areas that lie between the auditory thalamus and the song system. Comparison of neural responses in different areas should shed light on encoding of complex sounds within each area and on the functional stages of processing between areas. In addition, the use of tetrodes and multiple electrodes, combined with information theoretic analyses of results, should reveal the role of local circuitry in these neural responses, and test the further hypothesis that some sounds are encoded by populations of neurons rather than by single, highly selective cells. The neural coding principles revealed by this study are likely to be of general relevance to an understanding of auditory perception and its disorders.
