CRCNS Research Proposal: Neural and computational mechanisms of motor generalization in birdsong

This project addresses the question of how learned motor skills are transferred from one sequence of motor tasks to another, similar sequence of motor tasks. Understanding the processes involved in the generalization of motor skills across similar motor sequences has implications for management of recovery after brain injury as well as the development of more versatile systems of artificial intelligence (AI). Using the learned production of song in Bengalese finches as the motor skill in which to study this question, the investigators examine the role of similarity in sensorimotor context, i.e., the similarity in the acoustics and the sequence of song syllables, and in the underlying neural activity, in the generalization process, and the behavioral and neural mechanisms through which similarity in motor context plays this role. The investigators combine mathematical modeling with behavioral and neural recordings to develop a theory of motor generalization, for application beyond motor skills involved in birdsong. In addition to the broader impacts on motor rehabilitation and AI development, the project includes activities at the Atlanta Science Festival and the Science Gallery Atlanta to increase public science education. How motor skills generalize across motor contexts is a long-standing question, tackled here using song production in Bengalese finches. Evoking adaptations of the bird’s song in response to small acoustic perturbations in one part of the song, i.e., one context, the investigators examine the dynamics that underlie transfer of these adaptations to other parts of the song or contexts. Building on their prior model that focused on sensory feedback as the driver of adaptation and cross-adaptation (generalization to other parts of song), the investigators introduce an additional error signal, encoding discordances between motor actions produced in similar contexts, as a driver of generalization. The work aims to elucidate the dynamics of and the variables that critically shape motor skill generalization, and shed light on how the brain computes similarity and how that information, in turn, contributes to generalization. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.