Perception and Neural Representation of Auditory Restoration in the Songbird

It is well established that the human auditory system is capable of filling in missing pieces of a speech stream through a process known as phonemic restoration, yet a neural understanding of this phenomenon has been limited by the lack of a suitable animal model. Songbirds, and especially zebra finches, have proven to be a fruitful model for speech perception and production, making them a promising model for studying restoration of complex vocalizations. In this work, I established through behavioral testing that zebra finches experience this illusion and then recorded single-unit neural data to characterize the neural representation of restored birdsong. Using a decoding method, I was able to identify neural responses consistent with restoration. These responses were widely distributed through the avian auditory cortex but especially strongly represented in the deep auditory pathway. I also found that the evidence for restoration grew stronger over the course of the missing syllable, suggesting the involvement of an internal predictive model that suppresses the incoming auditory signal.