HNDS-I: A Unified National Infrastructure for Human EEG

The human brain produces electrical signals that can be measured from the scalp, a method called electroencephalography, or EEG. Scientists record these signals while people engage in cognitive science experiments, which enables the study of how the brain supports learning, cognition, decision making, and other human behaviors. Researchers around the world have collected thousands of such studies, but the data sit in many separate online collections that do not work well together, and much of it is hosted outside the United States. As a result, scientists cannot easily search across studies or combine them to ask bigger questions. This project builds a single, well-organized U.S.-based resource that brings these studies together, organizes them in a common language and labeling system, and connects them to the software tools. Bringing many studies together in one place also makes it possible to train modern artificial intelligence on brain data, which can reveal patterns that are harder to find one study at a time. By making years of brain and cognition data easier to find, share, and reuse, the project speeds scientific discovery, trains the next generation of scientists through community events and competitions, and protects the public’s investment in research by keeping these valuable datasets openly available for the long term. The project addresses the fragmentation and inconsistent annotation of human EEG data. More than 2,000 datasets formatted with the Brain Imaging Data Structure (BIDS) standard are spread across international repositories, yet fewer than 15 percent carry the semantic labels needed to search by experimental design, stimulus type, or cognitive process. The research team pursues two integrated aims. The first aim builds an annotated infrastructure that joins EEG recordings with accompanying cognitive behavioral measures, stimulus events, response times, and eye tracking. The project includes application of Hierarchical Event Descriptors to enable queries across studies, implements automated quality-assessment pipelines, consolidates datasets from major repositories with provenance linked to their original sources, and provides software bridges and programming interfaces (including EEGLAB, MNE-Python, FieldTrip, and Braindecode). This unified, well-labeled resource is designed to support training artificial-intelligence models on large collections of brain recordings. The second aim establishes community ownership, ensuring that the infrastructure persists, through a distributed steering committee drawn from major tool ecosystems, training hackathons that develop long-term maintainers, open competitions that drive adoption, and institutional hosting partnerships that provide independent, durable support. Together these activities create a sustainable national resource that enables cross-study analyses linking neural activity to behavior at a scale not previously possible. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.