Editorial: Dissociations between neural activity and conscious state: a key to understanding consciousness

Joel Frohlich, Julia Crone, Pedro A. M. Mediano, Daniel Toker, Daniel BorFrontiers in Human Neuroscience 2023 · DOI · OpenAlex

Abstract

EDITORIAL article Frontiers in Human Neuroscience, 02 August 2023 Sec. Brain Health and Clinical Neuroscience Volume 17 - 2023 | https://doi.org/10.3389/fnhum.2023.1256168

Introduction

Diagnosing Disorders of Consciousness (DoC), such as unresponsive wakefulness syndrome (formerly the "vegetative state"), currently relies on observing volitional, goal-directed behavior. However, this behavioral approach is notoriously error-prone, with misdiagnosis rates reaching up to 40%. A significant portion of these patients may be "covertly conscious"—internally aware but physically unable to manifest that awareness through movement.

To address this clinical gap, researchers are developing objective, physiological biomarkers, primarily using EEG and other neuroimaging techniques. The core challenge in this field is generalization: biomarkers derived from healthy brains (normal cortical dynamics) often fail or behave unpredictably when applied to severely brain-injured patients (abnormal cortical dynamics).

A New Conceptual Framework for Consciousness Research

In this editorial, the authors introduce a novel 2x2 conceptual framework designed to evaluate study populations and improve the training of consciousness classifiers. This matrix categorizes populations based on two critical axes: the normality of their cortical dynamics and the certainty of their conscious state labels.

The 2x2 conceptual framework categorizing study populations by cortical dynamics and the certainty of conscious state labels.

By mapping different groups—such as healthy sleepers, anesthetized patients, and those with rare genetic disorders—onto this matrix, researchers can better identify the risks of "overfitting" machine learning models to features of a healthy brain rather than consciousness itself. The framework highlights the importance of studying "paradoxical dissociations" (the top-right quadrant), where individuals have abnormal brain activity but known conscious states.

Key Insights: The Overfitting Trap

Standard research often focuses on healthy individuals during temporary unconsciousness (e.g., deep sleep or general anesthesia). Training models on these populations risks capturing signatures of a "normal" brain rather than the essential neural correlates of consciousness. When these models are applied to DoC patients with highly abnormal brain dynamics, they frequently produce false negatives.

The authors argue that the most robust biomarkers will come from studying populations that exhibit dissociations. For example, children with rare 15q genetic disorders or healthy adults given specific drugs like tiagabine exhibit highly abnormal EEGs while remaining behaviorally responsive. Training classifiers on these groups forces models to find universal signatures of consciousness that transcend specific brain states.

Summary of Research Findings

This Research Topic synthesizes several key quantitative and qualitative findings:

  • Sleep and Complexity: Aamodt et al. demonstrated that the complexity of spontaneous EEG signals—a common metric for consciousness—did not necessarily correspond to dream content during NREM2 sleep, suggesting that complexity alone may not be a perfect proxy for internal experience.
  • Recovery Metrics: Wu et al. found that whole-brain theta band functional EEG connectivity correlates with functional disability in patients emerging from a minimally conscious state, providing a potential tool for tracking clinical recovery.
  • Generalization Failures: Maschke et al. showed that common EEG measures of consciousness behaved unexpectedly in unresponsive wakefulness syndrome (UWS) patients when administered propofol, highlighting why models trained on healthy anesthesia subjects often fail in clinical settings.
  • Dementia and Consciousness: Huntley et al. raised profound questions by showing that patients with severe Alzheimer’s disease yielded Perturbational Complexity Index (PCI) values near or below the standard threshold for consciousness.

Conclusion and Future Directions

While the proposed 2x2 matrix is a heuristic and some conditions (like late-stage Alzheimer's) remain ambiguous edge cases, it provides a vital roadmap for the field. The authors advocate for a shift toward studying paradoxical dissociations between neural activity and behavior. By focusing on these outliers, researchers can build more robust, generalizable models to finally solve the problem of detecting covert consciousness in brain-injured patients.

Summary updated

Abstract

EDITORIAL article Front. Hum. Neurosci., 02 August 2023Sec. Brain Health and Clinical Neuroscience Volume 17 - 2023 | https://doi.org/10.3389/fnhum.2023.1256168
BibTeX
@inproceedings{frohlich2023editorial,
  author = {Joel Frohlich and Julia Crone and Pedro A. M. Mediano and Daniel Toker and Daniel Bor},
  title = {{Editorial: Dissociations between neural activity and conscious state: a key to understanding consciousness}},
  booktitle = {Frontiers in Human Neuroscience},
  year = {2023},
  doi = {10.3389/fnhum.2023.1256168},
}
Figure