Dr Sandra Martin | Executive Networks in Language Prediction

Language prediction plays a pivotal role in understanding and facilitating everyday communication. Thus far, it remains open how much linguistic predictions rely on language-specific and domain-general neural networks and how these interactions are modulated by age. This project aims to answer these questions by implementing a self-paced reading task with varying cognitive load in an fMRI experiment. Specifically, we will explore how an increasing cognitive load affects language prediction on different time scales, ranging from words to paragraphs, and how this might be modulated by age. Using univariate and multivariate analysis techniques, we will explore the effects of surprisal and age as well as their interaction on network connectivity and deploy language models to explain observed neural responses.