Back to All Events

A minimal network mechanism accounting for the statistics of babbling vocalizations

By David Hansel, CNRS and Paris Descartes University

Explorative behavior is a key concept in learning theory and is widespread during early stages of development. However how such behavior is actively generated by neuronal mechanisms is poorly understood. We show that babbling¬‐like vocalizations in human and non-human juvenile vocal learners share some common temporal features, which is in sharp contrast with the individual-specific learned vocalizations of the adults. These findings point toward the existence of a general neuronal mechanism for actively generating explorative behaviors. I will show how neuronal variability can intrinsically emerge as a network property and then transferred to the effectors to generate explorative behaviors. I will demonstrate how spatio-temporal correlations are naturally emerging in a  network operating in the balanced excitation-inhibition regime receiving structured feed‐forward inputs. I will then show how these activity patterns can be utilized by the effectors to generate explorative behavior with similar statistical properties as during babbling-like vocalizations. Finally, I will give several testable predictions regarding the structure of spatiotemporal correlations in cortical-homologues areas responsible for babbling‐like vocalizations in juvenile songbirds.