Project Grant 2617523

The National Science Foundation Division of Computing and Communication Foundations awarded the University of Oregon $897,184 on September 1, 2026, under the Computer and Information Science and Engineering program (CFDA 47.070) to develop factorial state-space models that disentangle multiscale neural dynamics in brain-wide recordings.

The research addresses the challenge of separating overlapping brain processes occurring simultaneously across timescales ranging from milliseconds to minutes in large-scale neural recordings. The project develops mathematical models representing observed neural activity as the sum of multiple latent components evolving in parallel with distinct spatial footprints, timescales, and dynamical rules. The work includes scalable fitting methods, principled model selection, and uncertainty quantification to make these models practical for brain-wide recordings while preserving interpretability. Validation occurs on controlled simulations and benchmarking against state-of-the-art approaches, followed by application to cortex-wide calcium-imaging datasets from behaving mice. The analyses test three central questions: how optimal performance during decision-making relies on stronger routing of information from sensory to motor regions; and how fast and slow neural processes combine to support perception, decision-making, and natural behavior.

Performance occurs in Eugene, Oregon, with a period of performance from September 1, 2026, through August 31, 2030. The award is classified as a Project Grant.