Federal Project Grant Award Summary

Rochester Institute of Technology received a $562,247 Project Grant from the National Science Foundation's Office of Multidisciplinary Activities (CFDA 47.075: Social, Behavioral, and Economic Sciences program) awarded August 15, 2026, with a completion date of July 31, 2029. The project, "Advancing NeuroAI Through Biologically-Plausible Self-Supervised Learning," delivers research and development of neuroscience-informed artificial intelligence systems that integrate principles from brain science with machine learning. Specifically, the award funds the creation of self-supervised representation learning (SSRL) models that incorporate recurrent feedback connections—mimicking biological neural architecture—to develop AI systems that are more efficient, robust, and capable than current feedforward-only approaches.

The project's primary deliverables include: (1) advanced NeuroAI models for SSRL that integrate feedback loops to improve computational efficiency and performance; (2) mechanistic scientific theory explaining how the brain's sensory systems learn to process raw sensory input into useful internal representations; and (3) empirical validation demonstrating that the developed models exhibit phenomena known to depend on feedback mechanisms in biological brains. The research addresses a critical gap between artificial and natural intelligence, aiming to produce AI systems with improved performance on unexpected situations while reducing the computational resources required for training and deployment. Work will be performed at RIT's Rochester, New York campus.