Lea-Maria Schmitt

The human brain is well equipped to turn arbitrary sensory signals into meaningful percepts. But how exactly is the interpretation of such incoming signals shaped by prior knowledge? What is a viable neural architecture within a cortical column for integrating bottom-up input with top-down predictions? And what constitutes the representational space of predictions along cortical hierarchies? In my research, I use laminar fMRI and deep learning models to answer these questions.

Before joining the Predictive Brain Lab, I completed my PhD at the intersection of Neuroscience, Linguistics and Machine Learning in the lab of Jonas Obleser. As a trained philologist and psychologist, I investigated how hierarchical event memory enables predictive processing in naturalistic speech comprehension, and soon started using artificial neural networks to better understand the underlying gating mechanisms in temporo-parietal cortex.