Brain Prediction

Brain prediction research aims to understand how brain activity reflects language processing by using machine learning models to predict neural responses to speech and text. Current research focuses on leveraging large language models (like transformers) and self-supervised speech models, analyzing their internal representations to map onto fMRI data and exploring the relationship between model architecture (e.g., convolutional and recurrent networks) and brain hierarchies. These studies reveal strong correlations between model predictions and brain activity, suggesting that these models capture aspects of human language processing, potentially advancing our understanding of the neural basis of language and enabling improved brain-computer interfaces.