Title: "Neural Coding of Pitch Cues in the Auditory Midbrain of Unanesthetized Rabbits"
Bertrand Delgutte, PhD – Harvard Medical School, Dept. of Otolaryngology (Advisor)
Kamal Sen, PhD – BU BME (Co-Advisor)
H. Steve Colburn, PhD – BU BME (Chair)
Oded Ghitza, PhD – BU BME
Laurel H. Carney, PhD – University of Rochester Medical Center, Dept. of Biomedical Engineering, Dept. of Neurology and Anatomy
Pitch is an important attribute of auditory perception that conveys key features in music, speech, and helps listeners extract useful information from complex auditory environments. Although the psychophysics of pitch perception has been extensively studied for over a century, the underlying neural mechanisms are still poorly understood. This thesis examines pitch cues in the inferior colliculus (IC), which is the core processing center in the mammalian auditory midbrain that relays and transforms convergent inputs from peripheral brainstem nuclei to the auditory cortex. Previous studies have shown that IC can encode low-frequency fluctuations in stimulus envelope that are related to pitch, but most experiments were conducted in anesthetized animals using stimuli that only evoked weak pitch sensations and only investigated a limited frequency range. Here, we used single-neuron recordings from the IC in normal hearing, unanesthetized rabbits in response to a comprehensive set of complex auditory stimuli to explore the role of IC in the neural processing of pitch. We characterized three neural codes for pitch cues: a temporal code for the stimulus envelope repetition rate (ERR) below 900 Hz, a “non-tonotopic” rate code for ERR between 60 and 1600 Hz, and a rate-place code for frequency components individually resolved by the cochlea that is mainly available below 800 Hz. While the temporal code and the rate-place code are inherited from the auditory periphery, the “non-tonotopic” rate code has not been currently characterized in processing stages prior to the IC. To help interpret our experimental findings, we used computational models to show that the IC rate code for ERR likely arises via temporal interaction of multiple synaptic inputs, and thus the IC performs a temporal-to-rate code transformation from peripheral to cortical representations of pitch cues. We also show that the IC rate-place code is robust across a 40 dB range of sound levels, and is likely strengthened by inhibitory synaptic inputs. Together, these three codes could provide neural substrates for pitch of stimuli with various temporal and spectral compositions over the entire frequency range.