Biyu J. He, PhD, & Eric K. Oermann, MD, Are Mapping Brain Activity During Learning to Improve Computer Algorithms & Make AI More Human-Like
Collaboration and innovation are at the heart of research endeavors at NYU Langone. Biyu J. He, PhD, and Eric K. Oermann, MD, are exemplifying these qualities, merging their expertise to build an artificial intelligence (AI) model that imitates the human brain.
The Big Idea
Richard Feynman, a Nobel Prize–winning theoretical physicist, once said, “What I can’t make, I don’t understand.” The quote is one that aptly captures the spirit of the collaborative effort by Dr. He and Dr. Oermann to build an AI model that more closely mimics the human brain. They hope to use that computer algorithm as a more nimble and practical proxy for exploring the brain and plumbing the depths of its mysteries. “Building a computational model can help us understand what’s going on in the brain in a more quantitative, detailed way than traditional neuroscience allows us to do,” says Dr. He.
To fund their research, Dr. He and Dr. Oermann have been awarded $1.2 million from the W.M. Keck Foundation, a nonprofit that supports pioneering discoveries in science, engineering, and medical research.
The researchers will start by mapping the brains of volunteers as they complete a very specific, simple task—a process that neuroscientists call “one-shot learning.” For example, imagine seeing an abstract black-and-white drawing, and then seeing a photo of a recognizable object—say, a helicopter—that loosely resembles it. Once your brain recognizes the helicopter in the photo, it will forever see it in the abstract image as well. “Once you have that photograph in your head, it’s imprinted in your mind and forever alters the way you process the abstract image,” explains Dr. He.
The pair will use neuroimaging and electrodes to map the brain activity involved in one-shot learning, and then leverage advanced AI techniques to create a computer model. “One-shot learning is something the human brain does well, but algorithms do not,” explains Dr. Oermann. “We plan to use our analysis of the process to unpack the differences and make AI models more brain-like.”
Dr. He and Dr. Oermann are uniquely well suited to this venture. She is a neuroscientist who researches how the brain creates conscious awareness, while he is a neurosurgeon and machine learning expert. The seeds of their collaboration began in 2019 when Dr. He received an email from John G. Golfinos, MD, chair of the Department of Neurosurgery, announcing that Dr. Oermann was a candidate for a faculty role. In addition to his training in neurosurgery, Dr. Oermann had done a postdoctoral research fellowship in machine learning at the life sciences arm of Alphabet, Google’s parent company. “I looked at his website and thought, ‘It would be amazing if we recruited him,’” recalls Dr. He.
Likewise, Dr. Oermann was aware of Dr. He’s research before his arrival at NYU Langone. “Biyu asks some of the biggest questions about how our brains make us human,” he says. “As an engineer and a neurosurgeon, I was used to focusing on very specific problems. But understanding how the brain works is what drew me to AI.”
Once here, Dr. Oermann wasted no time in reaching out to Dr. He. “We immediately sensed there was a long-term research agenda that could benefit from combining our expertise,” he says. “We see this as the first step in a really ambitious collaborative process.”