Stanford Interdisciplinary Graduate Fellowships (SIGFs)

Dr. Darian Hadjiabadi aims to identify higher-order features of neuronal circuits responsible for seizure initiation and propagation by quantifying mesoscale-network reorganization in genetic models of zebra sh that faithfully recapitulate seizure dynamics in humans.

Schwann cells (SCs) sort and myelinate peripheral axons, and impairments in either process can cause long-term disability. There are no therapeutic strategies for targeting SC dysfunction, underscoring the need to investigate mechanisms of sorting and myelination. Both processes require highly motile SC cytoplasmic protrusions, but the basis of this motility is unclear.

Understanding the relationship between structure and function in the human brain is a key interest in neuroscience. In recent years the focus is turning to understanding the role of the white matter in human cognition, brain function and neurological disorders.

Many neurological injuries and diseases such as brainstem stroke and Amyotrophic Lateral Sclerosis (ALS) result in severe speech impairment, drastically reducing quality of life. Recent progress in brain-computer interfaces (BCI) has allowed these individuals to communicate, but performance is still far lower than typical spoken conversation speeds.

Yi Lui's project aims to use deep brain microstimulation (DBMS), which causes even less brain damage and has higher spatial resolution than DBS, for memory recovery.

Behavioral state—such as alertness or exhaustion—dramatically impacts how our brains function. Yet, in spite of the key role that it plays in cognition, how behavioral state influences brain function remains a central mystery in neuroscience.

Many neurological diseases feature the death of neurons, but the mechanisms that mediate cell death in these disorders are unknown. Astrogliosis, the response of a cell-type called “astrocytes” to injury, is common to most diseases of the central nervous system (CNS), and recent studies in our lab suggest that some reactive astrocytes may release a protein that is potently toxic to neurons.

We propose a novel framework for efficient Bayesian cognition called Inference via Abstraction (IvA), which learns to approximate complex world models with simpler abstractions that capture main dependencies, but leverage structure in the prior distribution for efficient inference. We instantiate IvA with a combination of probabilistic graphical models and deep neural networks.