The reliability and precision of auditory nerve responses

Sensory neurons encode environmental stimuli by transmitting trains of nearly identical all-or-none action potentials (spikes) to the brain. It follows that the number and occurrence times of spikes are the only source of information conveyed to the brain about the outside world. These two features, count and timing, give rise to two neural coding strategies commonly referred to as rate coding and temporal coding, respectively. A stimulus that is repeated over multiple trials elicits responses that vary in the number and timing of spikes from trial to trial. This trial-to-trial variability of the neural response is noise that constrains the amount of information sent to the brain about the stimulus. All sensory systems exhibit some level of noise in their neural responses and the amount of noise is determined by many factors that are only partially understood at both the cellular and systems levels. The current dissertation examines neural mechanisms which determine the noise in rate and temporal coding. I have chosen to work on the peripheral auditory system of the chick (Gallus domesticus), which uses rate and temporal coding simultaneously to transmit information about the acoustic world via an ordered array of neurons tuned to specific acoustic frequencies. By presenting young anesthetized chicks with repeated pure tones and recording the occurrence times of action potentials in single fibers, I discovered that two basic neural properties, found in all peripheral sensory neurons, helped determine the level of noise inherent in rate and temporal coding. The responses of auditory nerve fibers to repeated tones revealed that these two properties, rate-adaptation and refractoriness, directly influenced the level of noise. Rate adaptation, a decrease in spike discharge rate during a constant stimulus, degraded spike-count reliability but did not affect spike-time precision. In contrast, refractoriness, a decrease in the probability of spike occurrence immediately following a preceding spike, enhanced both spike-count reliability and spike-time precision. The results suggest that rate-adaptation constrains the amount of information sent by a rate code, whereas refractoriness enhances the amount of information sent by both a rate and temporal code