Propofol anesthesia destabilizes neural dynamics across cortex

Adam Eisen, Leo Kozachkov, André M. Bastos, Jacob Donoghue, Meredith Mahnke, Scott L. Brincat, Sarthak Chandra, Emery N. Brown, Ila Fiete, Earl K. Miller

Abstract

Every day, hundreds of thousands of people undergo general anesthesia. One hypothesis is that anesthesia disrupts dynamic stability, the ability of the brain to balance excitability with the need to be stable and thus controllable. We tested this hypothesis using a new method for quantifying population-level dynamic stability in complex systems, Delayed Linear Analysis for Stability Estimation (DeLASE). Propofol was used to transition animals between the awake state and anesthetized unconsciousness. DeLASE was applied to macaque cortex local field potentials (LFPs). We found that neural dynamics were more unstable in unconsciousness compared to the awake state. Cortical trajectories mirrored predictions from destabilized linear systems. We mimicked the effect of propofol in simulated neural networks by increasing inhibitory tone. Paradoxically, increased inhibition also destabilized the networks. Our results suggest that anesthesia disrupts dynamical stability that is required for consciousness.

Introduction

General anesthesia is a cornerstone of modern medicine, yet the fundamental mechanisms by which it induces unconsciousness remain elusive. A prominent theory suggests that anesthesia disrupts the brain's "dynamic stability"—its capacity to maintain a balance between excitability and control. In a stable system, perturbations are quickly dampened, allowing the system to return to its baseline state.

Conceptual illustration of stability in dynamical systems, where trajectories from different initial conditions converge toward a common path, analogous to a damped pendulum.

Quantifying this stability in the brain is challenging due to its high-dimensional, nonlinear nature and the "partial observation" problem, where researchers can only monitor a fraction of total neural activity. Previous research has yielded conflicting results, with some studies suggesting anesthesia leads to excessive stability ("freezing") and others suggesting instability. This study introduces a robust mathematical framework to resolve these contradictions.

Methods: DeLASE

The researchers developed DeLASE (Delayed Linear Analysis for Stability Estimation) to quantify stability from empirical neural data. DeLASE integrates three key concepts:

  1. Takens’ Delay Embedding Theorem: Reconstructs the full state space of a system from limited observations by utilizing signal history.
  2. Koopman Operator Theory (via HAVOK): Approximates complex nonlinear dynamics as a high-dimensional linear system, which is more amenable to stability analysis.
  3. Delay Differential Equations (DDEs): Fits the data to DDEs to extract characteristic roots. The real parts of these roots serve as a direct metric of instability, indicating the timescale at which the system recovers from or diverges after a disturbance.

The method was validated using simulated Recurrent Neural Networks (RNNs) and applied to Local Field Potential (LFP) recordings from four cortical areas (vlPFC, FEF, PPC, and STG) in rhesus macaques during transitions between wakefulness and propofol-induced anesthesia.

Results

The application of DeLASE revealed that neural dynamics are significantly more unstable during unconsciousness compared to the awake state ($p < 1e-15$). Instability metrics nearly doubled during the maintenance phase of anesthesia.

Comparison of Local Field Potentials (LFP) and spiking activity across four cortical areas in the awake versus unconscious state. The unconscious state shows characteristic bursty, irregular patterns.

The Inhibition Paradox

Propofol acts as a GABA$_A$ agonist, increasing inhibitory tone. While intuitively this might seem to stabilize the brain, simulations and empirical data showed that increased inhibition can paradoxically push the neural network into a chaotic, unstable regime. This instability manifests as a "slower" brain; the return to baseline after sensory stimuli was nearly twice as slow under anesthesia.

State-Space Divergence

During wakefulness, neural trajectories in state space quickly returned to a stable baseline after perturbation. Under anesthesia, these trajectories exhibited prolonged, slow oscillations, failing to recover before subsequent stimuli arrived. This suggests that dynamical stability is a prerequisite for the rapid, widespread neural communication ("ignition") associated with conscious awareness.

Discussion

Our findings suggest that propofol-induced unconsciousness is characterized by a loss of dynamical stability. This metric tracked the depth of anesthesia monotonically, positioning it as a potential marker for consciousness.

Future work should investigate whether this destabilization is a universal feature of all anesthetics, including those with different pharmacological profiles like ketamine. Furthermore, the DeLASE framework could offer insights into psychiatric conditions. If consciousness requires a "sweet spot" of stability, disorders like schizophrenia or depression might be characterized by dynamics that are either pathologically stable or excessively chaotic.

Summary updated

Abstract

Abstract Every day, hundreds of thousands of people undergo general anesthesia. One hypothesis is that anesthesia disrupts dynamic stability, the ability of the brain to balance excitability with the need to be stable and thus controllable. We tested this hypothesis using a new method for quantifying population-level dynamic stability in complex systems, De layed L inear A nalysis for S tability E stimation ( DeLASE ). Propofol was used to transition animals between the awake state and anesthetized unconsciousness. DeLASE was applied to macaque cortex local field potentials (LFPs). We found that neural dynamics were more unstable in unconsciousness compared to the awake state. Cortical trajectories mirrored predictions from destabilized linear systems. We mimicked the effect of propofol in simulated neural networks by increasing inhibitory tone. Paradoxically, increased inhibition also destabilized the networks. Our results suggest that anesthesia disrupts dynamical stability that is required for consciousness.
BibTeX
@inproceedings{eisen2024propofol,
  author = {Adam Eisen and Leo Kozachkov and André M. Bastos and Jacob Donoghue and Meredith Mahnke and Scott L. Brincat and Sarthak Chandra and Emery N. Brown and Ila Fiete and Earl K. Miller},
  title = {{Propofol anesthesia destabilizes neural dynamics across cortex}},
  booktitle = {Neuron},
  year = {2024},
  doi = {10.1016/j.neuron.2024.06.011},
}
Figure