Department of Brain and Cognitive Sciences (BCS)
Adam Eisen Thesis Defense: Stability and Control: Signatures of the Conscious Brain
June 8, 2026
12:00 pm
Location
Building 46,Singleton Auditorium 46-3002
Description
Title: Stability and Control: Signatures of the Conscious Brain
Abstract: The brain is composed of interconnected cortical areas that dynamically coordinate to support conscious functioning. Consciousness is reliably disrupted by general anesthetics, providing a means through which the signatures of consciousness can be identified by contrasting across states. We leveraged dynamical systems theory to explore changes in neural dynamics between consciousness and anesthetic-induced unconsciousness. We focused on two fundamental dynamical systems properties: dynamic stability, the robustness of dynamics to perturbation, and control, how inputs can be coordinated with system dynamics to steer behavior. To evaluate these phenomena, we developed and applied two complementary methodologies. First, Delayed Linear Analysis for Stability Estimation (DeLASE) infers dynamic stability from delay-embedded neural recordings. We applied DeLASE to local field potentials (LFPs) recorded from non-human primates during propofol infusion and found that propofol anesthesia destabilized cortical dynamics relative to the awake state. This destabilization generalized to recordings under anesthetics with distinct molecular mechanisms (ketamine and dexmedetomidine), suggesting a shared dynamical signature of anesthetic-induced unconsciousness. Second, we constructed JacobianODE, a deep learning framework that estimates the time-varying Jacobian of nonlinear dynamics from data, and used it to characterize how subsystems within a complex network control one another. We applied this control-theoretic analysis to the same multi-area cortical recordings during propofol anesthesia. We found a broad disruption of intracortical control, both in terms of how easily brain areas could drive each other toward novel states, and how well they could control each other to produce reliable dynamics. Together, these contributions offer rigorous, data-driven tools for probing stability and control in neural dynamics. More broadly, they frame anesthetic-induced unconsciousness as a joint breakdown of the stability and directional control required for consciousness.
Zoom link: https://mit.zoom.us/j/95355449682