web page banner

Jack Gallant

Prof. Jack GallantBig data approaches to understanding the human brain
Jack L. Gallant, Ph.D.
Professor, Department of Psychology
University of California-Berkeley

The human brain is the most sophisticated computer processing system known, capable of impressive feats of recognition, evaluation and prediction under challenging natural conditions. Reverse-engineering the human brain might enable us to design artificial systems with the same capabilities. Efforts to reverse-engineer the brain are inevitably limited by the technology available for recording brain activity and by the amount of data that can be recorded. Recent advances and ongoing development of human neuroimaging offer an opportunity to use a data-driven, big-data approach to the reverse engineering problem. In this talk I will summarize ongoing work in my laboratory that uses a data-driven system identification approach to tackle this reverse-engineering problem. Our approach consists of several broad stages.

  1. We use functional MRI to measure brain activity while people perform natural tasks, such as watching movies or listening to stories. We divide these data into two parts, one use to fit models and one for testing model predictions.
  2. We use a system identification framework based on multiple linearizing feature spaces to model activity measured at each point in the brain. We also use a generative modeling framework to create probabilistic maps that characterize functional selectivity across individuals.
  3. We use a variety of dimensionality reduction and visualization techniques to understand how the brain represents different aspects of information in naturalistic signals.
  4. Finally, we use the fit models to decode brain activity, reconstructing movies or stories directly from brain activity measurements.

As neuroimaging technology advances and more data can be recorded from the brain, the quality of the fit models and the quality of brain decoding will increase dramatically, and data-driven approaches to brain mapping will become increasingly powerful and important.