Project Grant 2543127

Grant Award Summary

New York University received a $599,994 Project Grant from the National Science Foundation's Office of Multidisciplinary Activities under the Social, Behavioral, and Economic Sciences program (CFDA 47.075), awarded September 1, 2026, with completion targeted for August 31, 2031. The award funds research to reverse-engineer human deliberation through process-tracing and neural networks, investigating how the human brain learns to think efficiently by selecting which possibilities to consider, memories to recall, and reasoning steps to follow. The research team will train recurrent neural networks using meta-reinforcement learning to model both mental actions (what to think about) and external actions (what to do), then validate these models against rich human behavioral data including choices, response times, and eye movements.

The project deliverables include empirical research outputs explaining how humans decide which scenarios to simulate during planning and how memory retrieval strategies adapt to task demands, as well as significant educational and workforce development components. These include a new university course on human and artificial intelligence cognition, free online training materials reaching thousands of learners, and an outreach program introducing high school students to human and machine thinking. By understanding how the human brain achieves computational efficiency, the research aims to inform the development of artificial intelligence systems that are less costly, more energy-efficient, and better aligned with human cognition than current systems.