Machine Learning at the Flatiron Institute Seminar: Leila Wehbe

Title: Learning representations of complex meaning in the human brain

Abstract: It has become increasingly common to use representations extracted from modern AI models for language and vision to study these same processes in the human brain. This approach often achieves accurate prediction of brain activity, often accounting for almost all the variance in the recordings that is not attributable to noise. However, better prediction performance doesn’t always lead to better scientific interpretability. This talk presents some approaches for the difficult problem of making scientific inferences about how the brain represents high-level meaning. We also discuss how to go beyond aligning AI representations and brains. Instead, we directly learn the representations used in a brain region from its activity recordings. Using modern AI tools, data from naturalistic neuroimaging experiments and other large scale datasets, we reconstruct the representations and preferences of individual voxels and suggest new subdivisions that are more refined than existing regions of interest. This perspective draws a close connection between brains and AI models, reveals new aspects of brain function, and can serve as the basis for more powerful brain computer interfaces.