Sensory Awareness and Anesthesia: Predictive Coding, Oddballs and Reanimation

By Christian S. Guay, MD, from the IARS, AUA and SOCCA 2019 Annual Meetings*

During the panel, Anesthesia and Sensory Awareness: How Mechanism May Inform Clinical Practice, predictive coding provided attendees with a framework to understand the effects of anesthesia on visual and auditory stimuli, held on Monday, May 20 at the IARS 2019 Annual Meeting. A new thalamic stimulation method for reanimation from propofol anesthesia was also presented.  

The speakers focused in on sensory awareness during waking and anesthetic states, with special attention to predictive coding, cortico-thalamic circuits, visual and auditory evoked potentials.

They included Alexander Proekt, MD, PhD, from the Perelman School of Medicine, University of Pennsylvania, Mathew Banks, PhD, Aeyal Raz, MD, PhD, and Robert Sanders, MBBS, PhD, FRCA, all three from the University of Wisconsin School of Medicine and Public Health. Max B Kelz, MD, PhD, Perelman School of Medicine, University of Pennsylvania, served as the moderator of the session.

After reviewing the concepts of consciousness, sensory connection and behavioral responsiveness, previously presented at the symposium on anesthesia and consciousness, Dr. Robert Sanders began this panel discussion with an introduction to predictive coding. In the framework of predictive coding, the brain is constantly generating and updating sensory models of the world within its cortical hierarchies. High order predictions of the model are sent “top-down” to be compared with ascending sensory information. Errors generated in this way are then used to update the model in a perpetual feed-forward/feed-back loop. Paying attention to specific sensory stimuli serves to increase the weight applied to their prediction errors and therefore strengthens those particular aspects of the model. Within this framework, sensory disconnection during anesthesia can be conceptualized as a mismatch between prediction and sensory input, similar to dream states during natural sleep. In this way, predictive coding allows investigators to generate testable hypotheses on sensory awareness and anesthesia. In fact, propofol has been observed to selectively impair feedforward information flow through ascending visual processing networks.

Dr. Matthew Banks built on Dr. Sanders’ presentation to discuss auditory processing across awareness. The auditory cortical hierarchy has been well established, with bottom-up and top-down pathways identified. Interestingly, Dr. Banks reported that feedback auditory pathways are more sensitive to isoflurane than feedforward pathways, in contrast to the visual stimulus processing presented by Dr. Sanders. He also presented intracranial data on patients with epilepsy undergoing epileptic focus mapping. After implantation of electrodes directly on their (often temporal) cortex, patients were presented with an oddball paradigm. This classic experimental paradigm presents participants with a series of consistent sensory stimuli (in this case auditory syllables), which are eventually interrupted with an unexpected stimulus (i.e., the oddball). In the context of predictive coding, the oddball represents a strong violation of the descending predictions, leading to a well-described electrophysiological marker. Interestingly, Dr. Banks found that local but not global deviance responses within the auditory cortex are resistant to propofol sedation and anesthesia.

Shifting gears from predictive coding to spatiotemporal waves of cortical activity, Dr. Alexander Proekt started with a discussion of slow waves. EEG slow waves represent the synchronized oscillations of large groups of cortical neurons between depolarized and hyperpolarized states.  Both propofol and isoflurane induce slow oscillations in mouse cortex. Dr. Proekt’s team investigated the effects of these two hypnotics on visual-evoked responses. They found that visual stimuli could induce gamma-range EEG activity during each anesthetic, but that these evoked responses were temporally consistent during isoflurane anesthesia, leading to a more powerful signal when averaged across trials. Dr. Proekt also discussed analyses of neural connectivity and how current methods may confound theories of anesthesia. Specifically, he presented spurious correlations in nonlinear dynamical systems, and discussed the challenges of capturing global brain dynamics.

Dr. Aeyal Raz concluded the panel with a discussion of his work on thalamo-cortical circuits. After presenting the anatomy of the thalamus, he focused in on the central thalamic nucleus. This higher order nucleus serves to connect and synchronize different cortical areas. Working up from the principle that high frequency corticothalamic rhythms are integral to the processes of consciousness, Dr. Raz stimulated central lateral nuclei of non-human primates during propofol anesthesia at a rate of 50 Hz. This electrical stimulation consistently induced arousal from anesthesia and was associated with a decrease in the power and coherence of EEG slow waves. Stimulation also restored EEG markers of prediction-stimulus mismatch during an oddball paradigm. These findings suggest a critical role for the thalamic central lateral nucleus in consciousness and anesthesia.

*Coverage from the Panel: Anesthesia and Sensory Awareness: How Mechanism May Inform Clinical Practice, presented by Mathew Banks, PhD, Alexander Proekt, MD, PhD, Aeyal Raz, MD, PhD, and Robert Sanders, MBBS, PhD, during the IARS 2019 Annual Meeting