Harnessing the Second Power of Anesthesia for Disorders of Consciousness: 2025 T. H. Seldon Memorial Lecture

Anesthesia’s ability to ablate consciousness and eliminate the perception of pain revolutionized surgical practice; its discovery was named one of the most important medical developments in a thousand years by the editors of the New England Journal of Medicine.1 The state of reversible, controlled unconsciousness caused by anesthetics facilitates hundreds of millions of surgeries across the world each year.2 Although this common, international use suggests that its powers are widely understood and accepted, anesthesia harbors another power that has been largely untapped to date. In 1947, Harvard anesthesiologist Henry Knowles Beecher published a prescient article in Science entitled “Anesthesia’s Second Power: Probing the Mind.”3 In it, he wrote: “…we seem to have a tool for producing and holding at will, and at little risk, different levels of consciousness—a tool that promises to be of great help in studies of mental phenomena.” This “second power” of anesthesia can probe the neural substrates that undergird the mind, which has particular promise for individuals in disorders of consciousness (DOC). These disorders involve prolonged and reduced levels of wakefulness and awareness of self and environment after brain injury or progressive brain damage. In acute DOC, families and clinical care providers are faced with 2 sets of urgent questions. The first involves the patient’s current level of consciousness. Although this is commonly assessed at the bedside using behavioral measures such as the Glasgow Coma Scale (GCS), the misdiagnosis rate of patients with liminal consciousness (ie, minimally conscious state [MCS]) as unconscious (ie, unresponsive wakefulness syndrome [UWS]) is extraordinarily high.4 International clinical guidelines have begun to support the use of functional magnetic resonance imaging (fMRI) and electroencephalography (EEG) to detect consciousness through a patient’s brain activation in response to cognitive tasks.5,6 The resultant state, named cognitive motor dissociation (CMD), can be detected in approximately 25% of patients without an observable response to commands.7 The second set of questions faced by care providers involves the DOC patient’s capacity for consciousness: whether they will regain consciousness, the timeline for recovery, and what treatments and resources will be required to get there. Although the answers to these questions are critical determinants of the clinical goals of care, knowledge of post-acute care recovery trajectories has remained remarkably limited, and predictions of recovery are often imprecise. Although converging evidence suggests that late recovery and functional independence are possible for many patients with severe brain injuries, 70% of deaths during acute care for traumatic brain injury have been associated with withdrawal of life-sustaining treatments.8 How can anesthesia address these diagnostic and prognostic questions in DOC? In the field of consciousness science, it is a common practice to define consciousness by its absence: it is “what vanishes every night when we fall into a dreamless sleep.”9 Let us consider a variation of this definition: consciousness is “what vanishes when we are anesthetized.” Through this lens, the appearance or disappearance of neural markers of consciousness, when anesthesia is induced or withdrawn from a patient whose underlying awareness is unknown may indeed fulfill Beecher’s prediction of anesthesia’s second power to probe the mind. WHAT VANISHES WHEN WE ARE ANESTHETIZED The idea of using brain perturbation to assess the level of consciousness of an unresponsive patient has deep roots. Today’s most validated index of consciousness is the perturbational complexity index (PCI), which captures the complexity of the brain’s response to a direct and noninvasive cortical perturbation using transcranial magnetic stimulation (TMS) and EEG. The spatiotemporal complexity of the brain’s response to TMS perturbation successfully differentiates consciousness and unconsciousness induced by various anesthetic agents, and the different diagnostic categories in DOC.10 The exquisite accuracy of TMS-EEG assessment of consciousness is reflected in the European and American Academy of Neurology’s inclusion of the PCI in its guidelines for the diagnosis of coma and DOC patients. Despite this, the PCI faces large translational barriers to broad clinical implementation, including the low availability of the required equipment and techniques, challenges of integrating a TMS into the acute care environment, and the length of the testing time. Our team built on this idea using a different type of brain perturbation: anesthesia. We provided a brief and reversible perturbation using a propofol challenge that reliably disrupted neural activity and allowed its reconfiguration. By measuring the EEG changes to the anesthetic exposure in patients with DOC, we repurposed propofol as a stress test for the patient’s level and capacity for consciousness. The underlying hypothesis was straightforward: the EEG of patients with high levels or capacity for consciousness would significantly change during propofol exposure, while the EEG of patients with low levels or capacity for consciousness would remain the same as before anesthesia. This approach was highly compatible with the acute critical care environment where many DOC patients are admitted. For one, EEG and propofol can be administered at the bedside with limited patient distress or contraindications. Our acute patients were all exposed to propofol when intubated in the ICU without negative effects, suggesting a high likelihood for safe administration of propofol for this study; and all participants remained hemodynamically stable throughout the experimental protocol. Additionally, testing consciousness through neural responses to propofol exposure did not require patients to perform any sensory, motor, or cognitive tasks, and was thus independent of their capacity for or willingness to react to external stimuli or commands. The anesthetized state is hypothesized to be a network-level process, as it is not associated with gross suppression of neural processing in the primary sensory cortex.11 Accordingly, we first examined alterations on the brain network level that are known responses to propofol in the healthy brain. Functional connectivity networks undergo large-scale reconfiguration upon propofol exposure, with a reversal of posterior and anterior areas with rich functional connectivity (eg, network hubs), and a corresponding reversal in the dominant direction of functional connectivity.12,13 We calculated the reconfiguration of these 2 metrics, and combined them into an index called the Adaptive Reconfiguration Index (ARI). In a pilot case series with 12 adults with DOC after acquired brain injury, the ARI predicted with 100% accuracy the recovery of responsiveness at 3-month follow-up.14 All patients who recovered consciousness had a high adaptive reconfiguration of their brain network in response to the propofol challenge (Figure 1); all patients who did not have brain networks that remained static.Figure 1.: Reproduced with permission from Blain-Moraes et al.15 A case example of the EEG features of a patient diagnosed with unresponsive wakefulness syndrome who recovered responsiveness within 3 mo of the recording. The EEG was recorded from the participant during (A) baseline, propofol-exposure, and postexposure periods; using (B) spectrogram; (C) alpha topographic power map; (D) phase-amplitude coupling; (E) phase lag index; (F) directed phase lag index; and (G) network hub location. EEG indicates electroencephalogram.In contrast to the remarkable prognostic accuracy of the ARI, its diagnostic accuracy did not exceed chance levels. In other words, brain network reconfiguration in response to propofol perturbation predicted a patient’s capacity for consciousness, but not their current level of consciousness. Investigating possible explanations for this result, we turned our attention to the power spectral density (PSD) of our study participants. The PSD of the EEG represents the distribution of the signal power across different frequencies: oscillatory peaks in the PSD co-occur with broadband nonoscillatory (ie, aperiodic) activity. Traditionally, EEG analysis has been focused on the oscillatory patterns within specific frequency bands. We followed this conventional approach when calculating the ARI, focusing on functional connectivity networks constructed from the alpha band (ie, 8–14 Hz). However, the PSD of DOC patients often exhibits a total absence of theta and alpha spectral peaks. Indeed, when we examined a dataset of 43 DOC and coma patients, an oscillatory peak could only be identified in 13 of these participants.16 As anesthesia is known to alter the aperiodic EEG component during general anesthesia,17,18 we were concerned that changes in alpha network activity before and during propofol exposure might be underpinned entirely by alterations in the aperiodic component of the signal. Accordingly, we examined the change in the aperiodic slope of an expanded group of DOC participants (n = 16) at baseline and during exposure to propofol. As expected, anesthesia significantly steepened the spectral slope in the 30 to 45 Hz range, and the 1 to 45 Hz range (Figure 1A, 1C). Importantly, the absolute change in the spectral slope in the 30 to 45 Hz range predicted the participant’s current level of consciousness, as measured through the Coma Recovery Scale-Revised (CRS-R) score (Figure 1B, 1D). Patients with a higher CRS-R score demonstrated a larger steepening in the aperiodic slope in response to propofol. In other words, anesthesia-induced changes in the non-oscillatory component of EEG corresponded to a patient’s diagnosed level of consciousness. A CRITICAL MECHANISM What is this anesthetic-induced change in EEG revealing about the underlying brain dynamics of conscious awareness? A compelling explanation can be found in the concept of criticality, which comes from the world of complex systems science. Briefly, criticality describes the behavior of a system that is precisely poised between 2 dynamical regimes, such as chaos and stability. At this fine balance point, the system displays optimal computation capacity, information richness, and maximal sensitivity to perturbation.19 A system’s reaction to a perturbation will depend on its preperturbed distance from criticality, just as a stone thrown into water will generate a different reaction depending on whether the water is an icy pond, a calm lake or a stormy sea. Criticality is increasingly explored as a requirement for healthy brain function and the emergence of consciousness. Many different approaches have been taken to measuring criticality in the brain, including the edge of chaos—which describes the meeting point between the dynamical regimes of chaos and stability—and avalanche criticality—which describes the meeting point between activity amplification and dissipation. Additionally, many signal characteristics are known to be associated with criticality in general, including Lempel-Ziv complexity (LZC)—a measure of the number of unique patterns embedded within the dynamics of a neural signal. To investigate criticality as a potential mechanism underpinning our observations, we thus assessed the relationship between the anesthetic-induced change in these criticality metrics and the aperiodic slope pre-perturbation in our group of DOC participants.16 We discovered that the change of the spectral slope, information-richness, and criticality relied on the patient’s pre-anesthetic aperiodic EEG characteristics. These results suggested that the brains of DOC patients operate far from a critical point, resulting in weaker network susceptibility to global perturbations such as propofol anesthesia, and that the baseline aperiodic EEG slope could predict the magnitude of the brain’s reaction (Figure 2).Figure 2.: Alterations of the spectral slope and distance from criticality in disorders of consciousness during exposure to propofol anesthesia. (A) Power spectral density of Baseline (red) and Anesthesia (blue) state in a log-log scale, averaged across participants in a minimally conscious state (MCS; top), unresponsive wakefulness syndrome (UWS; middle), and coma (bottom). (B) Illustration of criticality hypothesis: a system at criticality is poised between 2 dynamical regimes, and the conscious brain is posited to be critical. Propofol anesthesia shifts the dynamics of the system to a subcritical state. Patients in MCS (purple), UWS (green), and coma (orange) are increasingly distant from a critical state, resulting in different degrees of change in system dynamics on exposure to propofol.While compelling, these results were based only on correlational analysis. To strengthen the evidence, we then conducted a secondary analysis of a dataset where EEG was collected from participants exposed to 3 different types of anesthesia (propofol, xenon, and ketamine) at doses that were roughly comparable vis-à-vis loss of consciousness in humans. Crucially, this dataset also contained a direct measure of the brain’s sensitivity to perturbation: the PCI.20 We hypothesized that states of unconsciousness during exposure to propofol and xenon would diverge from criticality, while wakefulness and exposure to ketamine (where participants were unresponsive but experienced vivid dreams) would maintain close-to-critical dynamics. We further hypothesized that the distance to criticality of the resting-state cortical activity could predict the complexity of the brain’s response to perturbation using TMS (i.e., PCImax). Our results confirmed our predictions: propofol and xenon, but not ketamine, induced a shift away from avalanche criticality and increased brain chaoticity.21 Most importantly, the criticality of the resting-state EEG reliably predicted the PCImax, in other words, the response to external perturbations (Figure 3). These results provide further evidence that critical neuronal dynamics are a necessary condition for the emergence of consciousness in the brain. Beyond the explanatory mechanism, the results may also have major practical implications for the clinical assessment of consciousness in DOC patients. We predicted individual subjects’ PCImax using a short resting-state 60-channel EEG recording obtained just before the TMS intervention with a mean average error below 7%, raising the intriguing possibility that assessment of consciousness could be accurately conducted without requiring a perturbation of TMS or anesthesia.Figure 3.: Prediction of the PCImax based on resting-state EEG dynamics. Individual points represent individual subjects during predrug wakefulness (light green: eyes closed [EC]; dark green: eyes open [EO]), propofol (blue), xenon (black), and ketamine (red) conditions. Conscious states (yellow) can be separated from unconscious states (blue) on an individual-subject level (based on a figure in Maschke et al21). PCI indicates perturbational complexity index.PROBING THE MIND Do these last results suggest that we have been chasing a red herring, that anesthetic perturbation does not reveal anything beyond resting-state EEG criticality of a patient’s level of consciousness? In short—no. First, the results predict a patient’s current level of consciousness, rather than their capacity for consciousness, and anesthetic perturbation remains a promising tool to support prognostication in DOC patients. Second, the EEG changes due to anesthetic perturbation contain meaningful clinical insights above and beyond the final PCI index that cannot be predicted based on resting-state criticality alone. The DOC patients in our study exhibited a heterogenous set of reactions in response to propofol anesthesia: while the majority demonstrated a steepening of the spectral slope during anesthesia, some displayed the opposite pattern—a flatter spectral slope, accompanied by a more complex signal and increased criticality.16 This same heterogeneity was observed in Toker et al’s study that included an assessment of the proximity of the EEG to the edge-of-chaos critical point in 4 DOC patients22: 1 patient unintuitively exhibited increased chaoticity after regaining consciousness. DOC are a heterogeneous set of conditions. Although 2 patients may be phenotypically indistinguishable (eg, similar lack of behavioral responsiveness), 1 patient could theoretically deviate from criticality towards a subcritical state (eg, too stable), while the other patient could deviate from criticality towards a supercritical state (eg, too chaotic). Endotyping patients has been named 1 of 3 major scientific gaps in the DOC field by the Curing Coma Campaign,23 and a patient’s response to anesthesia presents an intriguing possibility for identifying the underlying mechanism of unconsciousness and suggesting potential treatment. In a case study of 3 DOC patients, we demonstrated a paradoxical increase of consciousness-related markers (eg, brain complexity) on exposure to propofol.24 The results recapitulate the arousing effect that zolpidem (a GABAergic drug) has on approximately 10% of DOC patients.25 GABAergic drugs may shift subcritical patients further from criticality, but may shift supercritical patients closer to a critical point, supporting their return to consciousness. Other intriguing possibilities of probing the mind with anesthesia are scattered like breadcrumbs throughout the literature. For example, the LZC of EEG during anesthesia has been associated with impairment of response times in cognitive tasks after return of responsiveness,26 and the complexity and entropy of the EEG of brain-injured children in the PICU variously exposed to propofol, midazolam, or dexmedetomidine has been associated with the degree of their recovery.27 Such examples illustrate the emerging embrace of anesthesia’s of innate ability to probe the global reactivity and network resilience of the brain, and a powerful tool to reveal hidden capacities, and suggest potential mechanisms and treatments to re-establish the mind of patients with persistent unconsciousness. ACKNOWLEDGMENTS This article is based on the 2025 T. H. Seldon Memorial Lecture delivered for the International Anesthesia Research Society. DISCLOSURES Conflicts of Interest: None. Funding: S. Blain-Moraes is funded by a Canada Research Chair (Tier II) in Consciousness and Personhood Technologies; a Canadian Institutes for Health Research Project Grant (480995) and a Natural Sciences and Engineering Research Council Discovery Grant (RGPIN-2023-03619). This manuscript was handled by: Peter A. Goldstein, MD.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.