Topological Analysis of the Differential Effects of Ketamine and Propofol on Brain Dynamics Thomas Varley , Dr. Olaf Sporns, Dr. Alice Patania (Psychological and Brain Sciences, Indiana University, Bloomington, IN ) C5
Recent research into the neural correlates of consciousness has suggested that the level, and quality, of consciousness is related to the "complexity" of brain dynamics: loss of consciousness during anesthesia or following brain injury is associated with less complex patterns of activity and a reduced repertoire of available states, while psychedelic states are associated with increased complexity and a wider range of states. Here, we extend this line of research by constructing discrete, state-transition networks using electrophysiological data recorded from primates anesthetized with either propofol or ketamine. These state-transition networks approximate a dynamical attractor and allow us to explore the topology of the brain's evolution through state-space. Information about the dynamics of the system becomes encoded in the structure of the network, which is amenable to analyses using established techniques from network science. Being a discrete manifold, the state-transition graph is also amenable to information-theoretic analysis in ways that continuous signals and attractors are often not. By comparing data from primates anesthetized with ketamine or propofol and the respective awake controls, we attempt to determine which dynamical and information-theoretic properties are key to the maintenance of consciousness. It is well known that, while both being anesthetics, propofol and ketamine trigger markedly different states of consciousness: propofol anesthesia is associated with amnesia, loss of internal and external awareness, and at high doses, respiratory depression. In contrast, ketamine induces an exotic state known as "dissociative anesthesia", where external and bodily awareness are lost along with responsiveness to stimuli, but phenomenological awareness can continue in a dream-like state. We find several significant differences between brain dynamics in both states. As hypothesized, we found that, despite the behavioral similarities between the propofol and ketamine states, the propofol condition had a dramatically restricted state-space compared to ketamine, which was the same as the awake condition. Furthermore, we found that, in the propofol condition, the brain tended to return a small subset of all visited states far more than the awake condition, while ketamine had a much lower range in the number of times each state was visited when compared to the awake condition. Finally, we found that propofol significantly decreased the determinism and increased the degeneracy relative the awake condition, while ketamine increased the determinism and decreased the degeneracy, again compared to the awake condition. This indicates that propofol overall encodes less information in its causal structure compared the ketamine and the awake condition. The higher-order structure, as explored using topological data analysis and persistence homology reflects significant dynamical differences as well: the normalized persistence entropy shows a significant decrease in the propofol condition, which tracks the decrease in the Lyapunov exponent of the signal and the collapse of emergent structures following loss of consciousness. This work provides a suite of novel tools for work into the dynamical and information-theoretic characterization of altered and exotic states of consciousness. We anticipate that the further development and refinement of these analyses will help develop more comprehensive portraits of different conscious states.