AI News
All stories
Official

Stanford funds second round of AI and neuroscience grants for five research projects

Stanford's Wu Tsai Neurosciences Institute and Institute for Human-Centered AI have funded five projects that link brain research and AI.

Research · · 2 min read

Stanford's Wu Tsai Neurosciences Institute and the Institute for Human-Centered Artificial Intelligence (HAI) have awarded the second round of their AI and Neuroscience Grants to five projects that link brain research with machine learning. The projects range from depression subtypes to how infants learn. Kang Shen, director of the Wu Tsai Neurosciences Institute, called it "an incredible moment for both neuroscience and AI research."

The program began in 2022 with one-year awards of $125,000 each to four teams. Those teams studied stroke, Parkinson's disease, speech decoding, and the fine-scale structure of links between neurons. The second round covers a wider set of topics. Stanford's announcement doesn't state the size of the second-round awards, so readers can't tell whether the money per project has changed.

The depression project is led by Cammie Rolle of psychiatry and Ellen Kuhl of mechanical engineering. It starts from recent research showing that fMRI brain scans can reveal distinct subtypes of depression. If those subtypes are real, psychiatrists might need to treat each one differently. Rolle and Kuhl plan to use AI methods to connect fMRI results to electroencephalography, or EEG. They will then test whether EEG alone or a combination of the two can sharpen the subtypes and guide treatment.

The infant project is led by Cameron Ellis, Michael Frank and Dan Yamins. It begins with an observation about learning efficiency: human infants learn faster than AI systems, even though they see a small and biased slice of the data those systems are trained on. The team will record fMRI brain scans alongside video shot from a baby's point of view. They will then build a model trained on both the brain data and the video. The aim is to understand how babies learn so quickly and whether that understanding can help AI systems learn more efficiently.

Marius Wernig and Anshul Kundaje are studying why a developing human brain produces many different cell types from cells that are essentially genetically identical. They plan to use new genetics and AI methods to build a foundation model of neuronal cell types, aiming to find the molecular basis of how those cell types differentiate. A psychotherapy project led by Corey Keller and Manish Saggar will record intracranial electrical activity from patients undergoing epilepsy monitoring while they take part in therapy sessions. The team will use large language models to characterize therapeutic dialogue and topological methods to map how brain states change over time. The long-term goal is a neural atlas of psychotherapy.

The smell project is led by Todd Coleman and Emily Fox. Coleman and Fox argue that machine learning is good at pattern recognition but poor at capturing subjective, context-dependent experience. They say smell is an underappreciated part of everyday life that these systems don't reach. The pair will use electromagnetic recordings from the brain and large language models to connect brain activity to people's reported experiences of smell. The announcement says that linking these experiences to neural encodings could reveal new principles for AI that better understands human experience.

Why it matters

This is early-stage academic research with no product changes. Its most relevant idea for buyers is that the infant-learning project aims to make AI systems learn from less data more efficiently.

Original source

Where we saw it

1 source

Stanford HAI

Stanford HAIStanford UniversityWu Tsai Neurosciences Institute

AI wrote this summary from the sources listed, and nobody reviewed it individually before publication. Check the original source before you rely on it. How it works