A neural code for sensory reliability in the sound localization system of the barn owl

Abstract

Imprecision of sensory information is inherent in the physical nature of stimuli and sensory organs. To handle the uncertainty associated with sensory stimuli that are noisy and ambiguous, our brain must evaluate the level of reliability of sensory information and adapt behaviors accordingly. Yet to date, how the brain takes into account the reliability of the sensory input is still elusive. Therefore, the overall goal of my project was to define coding principles for how the brain deals with unreliable information. This thesis specifically addresses three questions: (1) Does sensory reliability drive neuronal response properties? (2) How is the level of sensory reliability represented in the brain? (3) How is sensory information decoded to adaptively guide behavior when reliability varies? The sound localization behavior of the barn owl was used as a model system, as it offers advantages towards understanding the neural coding supporting a specific and well characterized behavioral output, namely the head-orienting response. In particular, we have focused on the reliability of the interaural time difference (ITD), a primary cue used by owls and other animals to localize sound in horizontal space. This thesis first demonstrates that sensory neurons in the map of auditory space of the barn owl are tuned to the information that can be most trusted. Then, by recording neural activity of neurons in response to sound stimuli with quantitatively manipulated cue reliability, we showed that the degree to which a cue can be trusted is mainly represented in the spatial selectivity of sensory neurons. In light of these results, we investigated how the activity of neurons in the map is read-out to support the owl's orienting behavior as the reliability of the sensory input changes. We found that the response properties of premotor neurons downstream the map could be explained by convergence from sensory neurons in the map. These responses approximated the estimation of a probabilistic model and predicted the error owls made while localizing sounds. Therefore we uncovered a biological mechanism guiding the owl's orienting behavior when sensory reliability changes.