Neural circuits can exhibit remarkable stability (e.g., when supporting long-term memory) as well as flexibility (e.g., when supporting rapid learning). While neuroscientists have begun to individually characterize these two aspects of neural computation, developing more complete scientific theories will require new experiments and theoretical perspectives that simultaneously capture both short-term and long-term changes to neural circuit behavior.
Neural recording technologies now enable simultaneous activity measurements from several hundreds of neurons over long time periods (weeks or even months). However, characterizing the resulting datasets poses severe data analytic challenges, since we currently lack reliable and interpretable statistical methods that track changes to large-scale network interactions over time.
To overcome these challenges, I will develop statistical methods that extract functional ensembles, or sub-populations, of neurons that work together to encode animal behaviors and sensations. Importantly, these methods will not only identify these ensembles, but elucidate their emergence and stabilization over long-term recordings. To anchor these broad goals in tangible biological contexts, I will collaborate with experimental neuroscience labs at Stanford studying the neural basis of navigation (Dr. Lisa Giocomo) and movement production (Dr. Krishna Shenoy).
By developing novel statistical methodologies, this research will enable scientists and clinicians to monitor changes in neural circuit behavior over long time periods in laboratory experiments and eventually in human patients with chronic electrode implants. This will lead to a better fundamental understanding of how neurobiological systems maintain operation over long time periods, provide insights into long-term progression of neurodegenerative conditions, and improve the long-term effectiveness of neural prosthetics, which must adapt to changes in the underlying neural substrate.