Combining whole-brain models with deep learning, researchers mapped the low-dimensional space of patients with disorders of consciousness and simulated pharmacological interventions by altering neuromodulatory levels. Serotonergic and opioid receptor activation shifted the models toward brain dynamics seen in healthier states, with improvements correlating with the mean density of activated receptors across the brain. This approach provides a way to explore therapeutic potential of psychedelic drugs within ethical and methodological constraints, marking progress toward treatments for disorders of consciousness and other brain diseases.
A multivariate pattern classifier can decode single-trial EEG responses to auditory novelty, distinguishing local (automatic) from global (conscious) novelty detection. In 38 high-density EEG, MEG, and intracranial EEG recordings, the method overcame individual variability and multiple-comparison issues. Local responses were robust to distraction, while global responses depended on attention. Among 104 patients in vegetative state (VS), minimally conscious state (MCS), and conscious state (CS), local response decoding was significant in about 60% of recordings regardless of consciousness state. For global responses, significant decoding occurred in 14% of VS patients, 31% of MCS patients, and 52% of CS patients, indicating that global novelty decoding tracks consciousness level.