Dr. Pospisil will deliver a Mind in Vitro seminar on September 25, 2026 at 4:15pm in 4100 Sidney Lu Mechanical Engineering Building.
Abstract: Connectomes, which reconstruct the morphology of a nervous system, provide a detailed description of the physical substrate of neural computation, but understanding function requires understanding the dynamics that emerge from that structure. In this talk, I will outline work leveraging the recent whole-brain connectome of the fly to study how anatomy constrains neural dynamics. I will describe approaches for identifying the circuits and dynamical modes that are most consequential for neural activity, and I will introduce a complementary experimental and statistical framework that combines perturbation experiments with connectomic priors to estimate causal interactions between neurons. Finally, I will discuss ongoing work that leverages natural symmetries in the nervous system to constrain dynamical equations derived from morphological structure. Together, these approaches illustrate how static wiring diagrams can be used to generate hypotheses about neural dynamics and computation.
Biography: Dean A. Pospisil is an Assistant Professor of Psychology at the University of Illinois Urbana-Champaign. His research combines statistical inference, machine learning, and neuroscience to understand how neural circuits compute. He earned his Ph.D. in Neuroscience from the University of Washington and completed postdoctoral research at Princeton University. His work has examined the geometry of neural population responses, deep neural networks as models of visual representation, and causal and dynamical models of neural circuits based on connectomic data.