Bayesian Learning in the Infant Brain

ended

Infants come into the world equipped with amazingly powerful learning competences. The past decades have witnessed an ever-growing series of discoveries about young infants’ early cognitive achievements, illustrating these fast and unique adaptive abilities. However, the brain bases of learning mechanisms in the developing brain remain poorly understood. Meanwhile, in the past decade, the Bayesian brain has been put forward as a promising computational model accounting learning processes in the mature adult brain. In the present project, I propose to bring the two communities together and inspect the validity of the influential Bayesian brain hypothesis when learning processes are most crucially shaping the brain. Two assumptions derived from the Bayesian theory will challenged with studies conducted in young infants, using non-invasive behavioral and brain imaging techniques. In a first study, I will inspect whether the infant brain actively propagates predictions about upcoming events, as hypothesized in the Bayesian framework. In a second study, I propose an integrated application of computational modeling, neural recordings and behavioral measurements to test a second assumption of the Bayesian brain which postulates that the brain continuously tracks and adjusts to the progressive discovery of regular patterns in the input. I will establish an international research network and rely on two hosts with distinct expertise to provide me with new skills. By bridging multiple levels of description, this research program will open up translational perspectives for the understanding of developmental processes and ultimately for the diagnosis of atypical development.