Conference Agenda

Studies aimed at isolating neural signals linked with conscious perception overwhelmingly utilize a binary contrastive approach, manipulating awareness to compare brain responses between conscious and unconscious conditions. This method, while informative, often yields numerous candidate signals – which can be further confounded by signals related to reporting. Seeking more precise isolation, Cohen et al. (2024) used a no-report visual masking paradigm in which stimulus visibility was parametrically manipulated such that perception was either below, at, or above perceptual threshold. Neural responses in the no-report condition were compared with the psychometric function of visibility obtained in the report condition. One event-related potential (ERP) in the no-report condition showed bifurcation dynamics closely matching the pattern of behavioral responses in the report condition: a fronto-central negativity (fcN2) from ~250-300ms. Earlier signals, such as the VAN, however, were difficult to measure in this paradigm due to ERP latency shifts caused by the different mask latencies.

Here, we present a variation of this experiment, with simpler stimuli and no masks, in which visibility was manipulated by varying stimulus contrast across five levels (2-below, 1-at, 2-above threshold). VAN amplitudes (250-350ms) as well as a late bilateral posterior positive wave (450-650ms) showed bifurcation dynamics during no-report that matched the behavioral pattern in the report condition. In addition, temporal generalization of decoding showed significant spreading from 200-400ms for the visible but not invisible stimuli. These results confirm the close link between VAN and perceptual awareness and warrant further investigation into the late posterior positivity and meta-stable decoding.

Slow oscillations, characterized by alternating UP and DOWN states, are fundamental components of cortical dynamics and represent the default mode of activity in the brain. Occurring in the 0.1-4 Hz range, these oscillations are implicated in both normal and pathological conditions, including sleep, anaesthesia, and disorders of consciousness. Although the cyclic transitions between UP and DOWN states have been extensively documented in in-vivo and in-vitro animal models as well as in clinical settings, theoretical work that precisely identifies these periods and explains the underlying recurrent network processes is relatively sparse. This study introduces a novel algorithm for detecting UP and DOWN states in brain signals using the framework of Recurrent Quantification Analysis. By applying this methodology, we uncover recurring patterns in network activations that underpin oscillatory dynamics. Furthermore, our approach offers innovative network‐based tools for probing the dynamical mechanisms of slow wave propagation and to explore how these processes are modulated under varying brain conditions.

The extent to which conscious intentions contribute to the causal chain leading to action has been debated for decades. A key reason for that is that attempts to decode upcoming movement from preparatory brain activity have so far been largely unsuccessful. Using microelectrode arrays implanted in the motor and somatosensory cortices of four tetraplegic participants, we recorded spiking neural activity during two experimental paradigms: a self-paced intertemporal choice game ("Flip That Bucket", where players strategically time their flips of a self-filling slime bucket to soak their opponent before being slimed themselves), and a version of Grey Walter’s anticipatory-projector experiment (where participants scroll through visual content at their own pace). Using machine learning, we decoded the neural signals both in real-time and offline. Our real-time decoding predicted action onset with ~80% accuracy, ~250 ms before movement initiation and ~100 ms before muscle activation; offline models achieved nearly 90% accuracy 315 ms in advance of movement. Moreover, participants reported conscious decisions to move at the moment the real-time system initiated bucket flips on ~87% of the trials. Importantly, a decoder trained on the game transferred its predictive power seamlessly to the anticipatory-projector task. We describe the details of the machine-learning method we developed and discuss the results we obtained as a step in understanding the causal role of consciousness in decision making and action formation.

This theoretical model explains how sensory input from the sensors transforms into behavioral output on actuators, through four interrelated information processing units: body, self, mind, and ego. Rooted in cognitive science and philosophy of mind, it explores the processes underlying perception, cognition, and action.

Body serves as the primary interface with the external world, receiving sensory stimuli and transmitting raw data to higher-order cognitive systems. Embodied cognition suggests cognition is deeply grounded in bodily interactions with the environment (Clark, 1997). Mind functions as the domain of higher-order cognition, encompassing memory, reasoning, planning, and problem-solving. It constructs mental representations based on sensory data, integrating them with prior knowledge and conceptual frameworks, aligning with predictive processing theories (Friston, 2010). Self integrates sensory experiences into a coherent sense of identity, forming the foundation of conscious experience (Gallagher & Zahavi, 2008). While classical psychoanalysis (Freud, 1923) views ego as a mediator of unconscious drives, metacognitive models emphasize its role in self-regulation and reflective awareness (Carver & Scheier, 2012).

Reflective self awareness and self-referential cognition are closely related to attention which acts as a controller unit between the four processor blocks contribute each processor’s output to the decision/action. Consciousness is a product of the communication and information sharing among body-self-mind and ego blocks

The model provides a holistic framework for understanding awareness, decision-making, and behavior. Its implications extend to artificial intelligence, computational cognition, and human-computer interaction, bridging empirical cognitive science and philosophy in studying selfhood, consciousness, and behavior across disciplines.

Brain complexity measures, capturing the critical balance between integration and segregation in neuronal circuits, are increasingly recognized as promising markers of consciousness. Various metrics have been proposed to estimate brain complexity, from entropy-based measures to network-theoretic approaches. These metrics can be applied in two ways: observationally, by analysing spontaneous activity patterns, or through perturbation, by examining the responses evoked by direct brain stimulation. Despite their widespread use, the relationships between these metrics and the specific aspects of brain structure they capture remain poorly understood. Here, we use computational models to investigate how different complexity metrics reflect different structural arrangements in neuronal networks.

We employed a mean-field model of excitatory and inhibitory neural populations to simulate activity across networks with diverse architectures. By systematically rewiring network connectivity, we achieved varying degrees of integration and segregation. We then applied multiple complexity metrics, both observational (i.e., Lempel-Ziv Complexity, Functional Complexity, Neural Complexity) and perturbational (i.e., Perturbational Complexity Index), to assess their sensitivity to these structural modifications.

We found that different complexity metrics differently respond to network architecture. The observational metrics exhibited biases toward either integration or segregation, or were insensitive to network modifications. However, the Perturbational Complexity Index identified a peak in complexity when integration and segregation were optimally balanced, aligning more closely with the theoretical notion of brain complexity.

These findings emphasize the need for a deeper understanding of how structural and functional properties influence complexity estimates. They also highlight the importance of selecting appropriate metrics when investigating brain dynamics and consciousness.

Consciousness is shaped by the quality of how incoming stimuli are perceived and processed within the brain. Fluctuations in perception and cognition are critically supported by broadly projecting neuromodulatory systems, like the cholinergic system. Cholinergic activity overall has a stabilizing effect on attentional performance, while short term activity bouts enhance performance on attention shifts. The cholinergic basal forebrain provides the primary source of cholinergic innervation. We hypothesized that phasic events of fMRI activity within the basal forebrain would lead to increased activity and regional heterogeneity (ReHo) in the cholinergic projection field.

We used the Human Connectome Project (HCP) 7 Tesla resting-state fMRI data. Functional connectivity to maps of vesicular acetylcholine transporters was calculated using the REACT toolbox (Receptor-Enriched Analysis of Functional Connectivity by Targets; Dipasquale et al., Sci Rep 2023), yielding neurotransmitter-informed subject-specific maps (Ni-FC). Phasic firing events in the basal forebrain were determined from fMRI time courses (Munn et al., Nat Comm 2021). fMRI activity and ReHo in the Ni-FC maps in response to these events were analyzed.

Our results show a significant hemodynamic response in fMRI activity (p<0.001) as well as ReHo (p<0.001) in the Ni-FC maps after phasic events in cholinergic nuclei.

We demonstrated with a novel analytical approach that phasic firing from the cholinergic basal forebrain leads to changes in activities in brain areas projected to by these nuclei. This approach could serve to quantify cholinergic activity in different states of consciousness and to understand behavioral relevance of phasic cholinergic modulation.

Recent studies using pharmacological interventions have evaluated the effectiveness of empirical brain complexity measures in tracking changes in consciousness. Specifically, both observational and perturbational indices—Lempel-Ziv complexity (LZc) and the Perturbational Complexity Index (PCI)—applied to EEG have been assessed under various drug-induced states, including psilocybin and low-dose ketamine (Ort et al., 2023; Farnes et al., 2020).

These studies reveal a notable dissociation between the two measures: PCI remains consistently high across conditions, reflecting preserved consciousness capacity, while LZc increases following serotonergic and glutamatergic drug administration. The LZc increase is thought to reflect the enriched phenomenology of hallucinatory states but may also result from changes in EEG alpha power, a regular oscillation suppressed by both psilocybin and ketamine.

To investigate the role of alpha power, we systematically assessed LZc and PCI in a minimal contrast between eyes-closed and eyes-open conditions in healthy participants (n=20). PCI was derived from spatiotemporal EEG patterns evoked by Transcranial Magnetic Stimulation (TMS) of the occipital (BA 19) and premotor (BA 6) cortices, while LZc was calculated from spontaneous EEG signals.

Our findings show that LZc increases when regular alpha EEG oscillations are suppressed during eye opening. Conversely PCI shows high and stable values across conditions. These results highlight the sensitivity of observational measures to alpha power fluctuations and underscore the need for caution when linking dynamical EEG complexity measures to phenomenological richness.

Background/Aims: Organophosphate (OP) pesticides, widely used in Latin America, act as cholinesterase inhibitors, leading to excessive acetylcholine accumulation and dysregulation of the cholinergic system, which is critical for wakefulness, attention, and cognitive function. Chronic OP exposure has been linked to fatigue, excessive daytime sleepiness, and altered sleep patterns. However, no studies have assessed these effects using resting-state electroencephalography (EEG). This study examines the dose-response relationship between OP exposure, measured via dialkyl phosphate (DAP) metabolites in urine, and alterations in alertness and drowsiness in an exposed population from the Maule region, Chile.

Methods: Resting-state EEG was recorded in 68 participants (10% women) at two time points over two years. Urine samples were collected at four time points to quantify DAP metabolites as biomarkers of OP exposure. EEG-based alertness and drowsiness were assessed using the micro-measures algorithm of alertness. A dose-response approach was applied to examine associations between exposure and neurophysiological alterations.

Results: Preliminary analyses suggest a potential dose-dependent relationship between OP exposure and EEG markers of wakefulness with neurophysiological effects linked to chronic exposure. Further analysis is needed to confirm these associations.

Conclusions: This is the first study to use resting-state EEG to assess altered alertness and drowsiness in an OP-exposed population. Findings highlight the need for stricter pesticide regulations and improved occupational health monitoring to mitigate potential neurocognitive risks.

Unveiling the neural correlates of conscious perception is a central focus of consciousness research. A recent study by Sergent et al. suggests that stimuli at an individual’s detection threshold can result in two modes of processing during wakefulness. The first mode involves unconscious processing, where early brain responses do not persist, and stimuli cannot be consciously perceived or reported. The same stimuli can sometimes be processed consciously, with early brain responses that extend over time, leading to late activations. These late activations recruit widespread brain regions, including sensory, motor, and prefrontal areas, forming a "Global Workspace" as described by the Global Neuronal Workspace Theory. When no task is required, conscious perception correlates with a subset of these regions, excluding motor areas, termed the "Global Playground" by Sergent et al.

In our study, we aim to investigate how this bifurcation between conscious and unconscious processing during wakefulness is modulated during sleep. Specifically, we will explore whether a shift from the Global Workspace to the Global Playground occurs as the brain transitions through sleep stages. We will present auditory stimuli to sleeping participants and assess brain and behavioral responses via EEG and EMG recordings. We hypothesize that the sleeping brain retains the capacity for conscious auditory processing, particularly during early sleep stages. We predict that reduced behavioral responses during N1 and N2 sleep will correspond to a shift to the Global Playground, with conscious processing ceasing during N3 sleep, marked by the absence of responses and late activations in the Global Playground.

Phi-ID is a powerful recent extension to information theory (Mediano, 2021), which decomposes functional interactions in the brain into synergistic and redundant components. To date, Phi-ID has largely been applied to resting state, with results demonstrating connections between synergy and conscious state - greater synergy observed in awake compared to drowsy participants (Rognone, in prep) and indicating a link between synergy and complex cognition (Luppi, 2022). However, this method has not yet been applied to task-based fMRI to investigate the link between synergy and cognition directly. Here we do this by leveraging a large-scale fMRI dataset of ~600 participants measured across multiple tasks (Shafto, 2014).

We analysed a subset of tasks with strong cognitive components and identified significant increases in synergy relative to resting state in each task. Additionally, in those tasks with a performance measure, we demonstrated a positive correlation between synergy and performance (Fluid Intelligence r=0.20, p=0.0023, Stop-Signal/Go-NoGo  r=0.22, p=0.0355). Initial results also indicate that the brain networks involved in synergistic interactions change in relation to task type. For example, in a simple sensorimotor response task significant increases in synergy were limited to Visual, Somatomotor, and Dorsal Attention networks, while in a semantically-laden movie-watching task we found a marked increase in synergistic interactions involving temporal lobe regions associated with semantic processing.

Consciousness has been proposed to relate to synergistic ‘integrated information’ in the brain (Luppi, 2024). By investigating these synergistic interactions in the context of cognition, this work aims to characterise a functional role for consciousness in higher-order cognition.

In Alzheimer’s disease (AD), a mismatch between neurological damage and cognitive functioning is often attributed to individual differences in cognitive reserve. Understanding the neural mechanism of cognitive reserve could help assessing the therapeutic effectiveness of interventions in AD. To address this, here, 38 elderly participants performed a sustained attention task during high-density EEG while awake and during drowsiness. Operationally, the degree to which performance was impaired under drowsiness signalled the extent of cognitive reserve, with less impairment indicating a higher level of cognitive reserve. Investigating performance variations during the active management of neural challenges offers a novel approach to studying cognitive reserve, capturing dynamics that mirror everyday cognitive demand. We related cognitive reserve to various measures, including informational complexity using the Lempel-Ziv (LZSUM) algorithm. We found a significant interaction effect between arousal and performance, where LZSUM values increased in high performers when drowsy but decreased in low performers. This effect was most pronounced in the frontal and central areas. Our findings suggest LZSUM to be indicative of a compensatory mechanism and thus show potential for LZSUM as a neural marker in assessing cognitive reserve. However, we found no consistent relationship between performance and structural brain measures, and proxies of cognitive reserve. Critically, our findings present a counterexample to the prevailing view that informational complexity purely reflects conscious level. Further research, such as a study with the same paradigm in patients with mild cognitive impairment (MCI) and AD, may lead to additional insights of whether we are truly measuring cognitive reserve.

Introduction:

Serotonergic compounds can induce profound changes in subjective experience – not only acutely but also long-term. While much work has focused on the acute effects (e.g., Tolle et al., 2023), here we investigate pre-treatment neuroimaging markers of long-term mood changes. Using advanced, interpretable machine learning, we achieve high-accuracy predictions of post-treatment depressive symptoms and identify key neurobiological predictors of hedonic experience over time.

Methods:

We analyze two independent datasets of depressed patients undergoing treatment with psilocybin or escitalopram (n=42, n=16; Daws et al., 2022). Pre-treatment fMRI and clinical assessments serve as model inputs. A variational graph autoencoder (VGAE) learns a latent representation of each patient’s brain network, incorporating functional connectivity and brain-regional couplings with three serotonin systems (5HT1A, 5HT2A, 5HTT; Believeau et al., 2017). A multilayer perceptron then predicts post-treatment depression scores (QIDS) from these latent representations. We also introduce an analytical method that directly maps predictions to interpretable neurobiological patterns.

Results:

We achieve a true-vs-predicted correlation of r=0.75 (p<1.0e-8). Notably, our model generalizes to other brain parcellations without additional training, and to an independent dataset with only minimal finetuning (r=0.69, p<2.9e-3). Interpretability analysis reveals three distinct subtypes of neurobiological patterns, with shared markers in serotonin, dopamine, and attention systems, and subtype-specific markers in sensory-attention integration and the noradrenaline system.

Conclusions:

Overall, our work reveals robust, interpretable neurobiological predictors of long-term mood changes following serotonergic interventions, highlighting the need for stronger psychiatry-consciousness science cross-talk. Additionally, we present a powerful framework for directly linking macroscale brain dynamics to phenomenological reports.

Adaptive systems must navigate the tension between stability and exploration: minimizing free energy to maintain homeostasis while maximizing entropy to discover globally favorable states. This study presents primary research into this trade-off using a computational agent-based simulation grounded in a Markov blanket framework.

In a 100×100 grid-world, each cell is assigned a random “surprise” value. The agent interacts with its environment through noisy sensory observations, which are processed via Bayesian inference to update its internal beliefs. Actions are taken based on two behavioral modes: (1) Pure Minimize, which prioritizes local surprise reduction, and (2) Entropy Injection, which periodically increases belief uncertainty or randomizes actions with a fixed probability (20%) to encourage broader exploration. Over exactly 5,000 steps, the entropy injection mode achieved significantly greater coverage (810 vs. 159 distinct cells), reduced average observed surprise (0.456 vs. 0.503), and maintained higher belief entropy (6.684 vs. 4.013), reflecting broader exploration and adaptability.

This research highlights how strategic bursts of entropy allow adaptive agents to momentarily tolerate increased uncertainty, enabling them to escape local minima where they might otherwise remain trapped, discover globally low-surprise regions that offer better alignment with the environment, and effectively balance exploration with stability. These findings have implications for reinforcement learning, autonomous systems, and multi-agent dynamics, extending theoretical frameworks like predictive processing and free energy minimization. Future work will investigate hierarchical Markov blankets and multi-agent systems, offering further insights into how adaptive systems thrive in complex, partially observable environments.

Conscious access, according to the Global Workspace Model, is an all-or-none phenomenon : an internal or external representation is either consciously accessed or not. Despite empirical evidence in favor of this view, there is still a debate on that matter (Windey and Cleeremans, 2015).

Recently, Sergent and collaborators, 2021, explored the following prediction using electroencephalography : if access consciousness is an all-or-none phenomenon, the neural correlates of processing stimuli around the perceptual threshold should display a bifurcation pattern, with the bifurcation variable being the intensity of the sensory stimulation.

Here, we deepen and formalize this hypothesis within the dynamical systems framework. Leveraging simulation-based machine learning methods, we optimize stochastic differential equation models to reproduce Pr. Sergent’s EEG data and further analyze the qualitative properties of their dynamics (especially, non-linear and asymptotic properties).

Importantly, we show that a simple model, similar to mean-field potential models, adequately reproduces the dynamics of cortical encoding patterns over time. Our results indicate that neural representations, at a whole-brain level, evolve in a nonlinear, potentially bifurcative way.

Windey, B., & Cleeremans, A. Consciousness as a graded and all-or-none Phenomenon : a Conceptual Analysis. Consciousness and Cognition. (2015)

Sergent, C., Corazzol, M., Labouret, G., Stockart, F., Wexler, M., King, J.R., Meyniel, F, Pressnitzer, D. Bifurcation in brain dynamics reveals a signature of conscious processing independent of report. Nature Communications. (2021)

The search for the Neural Correlates of Consciousness (NCCs) has generated multiple electrophysiological candidates, including the Visual Awareness Negativity (VAN) and the P3b. However, no-report paradigms suggested that the P3b is associated with report rather than with conscious perception. Yet in these paradigms, awareness and/or reports were always manipulated across blocks. Here we overcome these challenges using a novel dual-task paradigm designed as part of the Cogitate adversarial collaboration: participants played an engaging video game as the primary task, while irrelevant faces and objects were occasionally presented in the background. In 1/3 of the trials (report trials), a secondary task was presented: the game was paused and participants were asked to report if they just saw a face or object. Here we analyzed EEG data, focusing first on report trials. We then trained a support vector machine to differentiate seen and unseen trials, and used it to label the remaining 2/3 no-report trials. In report trials, we found the VAN (~220-260ms) and a novel signal characterized by a fronto-central negativity (fcN2) and bilateral posterior positivity at ~350-550ms. The P3b was elicited only after the report probes. In the classifier-labeled no-report trials, we found evidence for the VAN and fcN2 and no evidence for the P3b. These results demonstrate the potency of this novel paradigm and the classification approach for detecting NCCs while minimizing task-related effects. They also highlight the importance of the fcN2, suggesting that there might be a second necessary stage of processing for visual awareness.

Since Block's seminal article in 1995, phenomenal consciousness and access consciousness are regarded by many as two types of perceptual consciousness, reflecting the qualitative nature of subjective experience and it being available for cognitive processes, respectively. Although this was a conceptual distinction aimed at clarifying the vocabulary used in the field, its adoption by the neuroscientific community has led to confusion concerning the theories and neural correlates of perceptual consciousness. As suggested by others before, rather than two types of consciousness, phenomenal aspects and access may better be conceived as two necessary conditions for perceptual consciousness. In this view, a percept is considered conscious if and only if its content is (a) encoded with an appropriate, phenomenal format, and (b) it is accessed. I will describe the implications of this shift in perspective and its power to reunite the so-called perceptual and cognitive theories of consciousness and their neural implementations. Moreover, I will present a leaky evidence accumulation model describing how those two conditions for consciousness are met in time, shedding light on three distinct thresholds between unconscious perception, conscious access, and conscious report. Finally, I will illustrate how the model can account for temporal aspects of conscious perception that are often neglected, including its subjective duration.

Dreaming during anesthesia presents a unique opportunity to study consciousness and its neural correlates while also offering potential clinical applications. However, there are currently no established protocols for the intentional induction of anesthesia dreaming, limiting systematic research into its mechanisms, clinical outcomes, and broader applications. Here, we describe a protocol designed to induce a state conducive to dreaming during anesthetic emergence and investigate its electrophysiological (EEG) correlates in an experimental setting. In an ongoing study with healthy volunteers (N = 15), we administer target-controlled infusions of propofol while continuously monitoring EEG and vital signs, followed by structured interviews assessing dream content. We are testing multiple anesthetic protocols (4-6 sessions per participant) to reliably induce pre-emergent dream vs. no-dream states and test their corresponding EEG signatures. Preliminary analyses suggest that dream experiences during pre-emergence anesthesia are associated with increased beta/alpha and gamma/alpha ratios, increased spectral slope, and decreased beta/alpha phase-amplitude coupling in frontal and posterior regions. These experimentally induced anesthesia dreams are positively valenced and share phenomenological similarities with psychedelic experiences. This study establishes a novel experimental framework and EEG-guided anesthetic protocols to facilitate future research on the neural correlates of consciousness and the therapeutic potential of anesthesia-induced dreaming.

Dissociation includes a constellation of symptoms and perceptual states characterised by discontinuities in experience, such as depersonalisation and derealisation. Although dissociation is hypothesised to play an integral role in a range of perceptual phenomena (e.g., hallucinations), experimental research on depersonalisation and derealisation is limited in part due to a lack of consensus regarding an experimental model of dissociation. Nitrous oxide (N2O), a dissociative anaesthetic that functions as an NMDAR antagonist, represents a potentially viable laboratory model of dissociation because of its rapid onset/offset effects. Here I will describe three (pre-registered) controlled experiments examining the impact of N2O inhalation or placebo (medical air) on dissociation and different perceptual effects. In Experiment 1, we found that inhalation of N2O was associated with both elevated dissociation and more false alarms in an auditory signal detection task than the placebo condition. The response pattern closely paralleled that of clinical and non-clinical hallucinators, thereby further strengthening the hypothesised link between dissociation and hallucinations. In Experiment 2, inhalation of N2O was not associated with increased responsiveness to direct verbal suggestions in contrast with a hypothesised coupling of dissociation and suggestibility. Finally, in Experiment 3, we found evidence that a placebo identified as N2O could reproduce moderate dissociative and psychedelic effects. These results highlight N2O as an efficacious method for reliably inducing dissociation, and studying its impact on perception, in a controlled manner with implications for the experimental study of dissociation and the perceptual effects of psychedelics.

Background: Ketamine is commonly used in emergency settings, where NDEs are prone to occur. However, the relationship between ketamine uses in a medical context and the emergence of NDE remains scarcely discussed.

Aims: This case study of a 73-year-old woman, who underwent ketamine treatment in emergency care, addresses the evolution of NDE memories over time and its lasting impact.

Methods: Two semi-structured interviews were conducted, at 2 weeks and 2 months post-event. These included free recall, the NDE Content Scale (NDE-C), and questions about the impact of the experience.

Results: The patient presented a coma (Glasgow Coma Scale: 3/15) secondary to severe hypercapnia, hypoxia, and respiratory acidosis requiring intubation during which she received 500 mg of ketamine. She met the NDE-C criteria (total score≥27/80) for an NDE at the first (34/80) and second (32/80) interviews. While vividness remained unchanged, some NDE features were inconsistent. Notably, the patient described an out-of-body experience (i.e., seeing herself above her coffin) during the first interview but not during the second interview. The patient identified her NDE as her most meaningful life experience, reporting lasting positive impacts including increased empathy and appreciation for love.

Conclusion: This case study suggests that ketamine may trigger or enhance the phenomenology of NDEs in clinical settings. Moreover, our findings challenge the reliability of NDE memories – as suggested by anecdotal reports–instead highlighting their possible dynamic evolution. This underscores the need for longitudinal studies on the mechanisms shaping NDE memories and their relationship with the subsequent lasting impact.

The genesis of different states of consciousness, such as wake, sleep or anesthesia, requires to take into account actions on multiple synaptic receptors in central neurons. However, how to link this receptor scale to the emergence of global activity states in the brain is presently unsolved. Here, we show a modeling approach that takes into account actions at synaptic receptors and evaluates its effect at the whole-brain level. We use biophysically-grounded mean-field models that integrate membrane conductances and synaptic receptors, to generate population-level models, that in turn can be used at the basis of whole-brain models. Using the example of general anesthesia, we show that anesthetics targeting GABA-A or NMDA receptors can switch brain activity to generalized slow-wave patterns, as observed experimentally in deep anesthesia. To validate our models, we use several measures that were previously used to compare awake and conscious states, to anesthetized and unconscious states. The first measure is the responsiveness to external stimuli using the Perturbational Complexity Index (PCI). The PCI calculated from the model is high in asynchronous, awake-like states, and drops to lower values in simulated anesthetized states, as shown experimentally. Second, we investigated how the functional connectivity (FC) differs from the structural connectivity (SC) in the model. In simulated anesthestized states, FC remained close to SC, while in awake-like states, FC differed from SC, as also found experimentally across species. In conclusion, mean-field models that incorporate molecular realism provide a robust framework to understand how molecular-level drug actions impact whole-brain dynamics.

The resting primate brain is traversed by spontaneous functional connectivity patterns that show striking differences between conscious and unconscious states (Barttfeld et al., PNAS 2015; Uhrig et al., Anesthesiology 2018; Demertzi et al., Sci. Adv. 2019; Castro et al., Commun Biol 2023). Transcranial direct current stimulation, a non-invasive neuromodulation technique, can improve signs of consciousness in disorders of consciousness (Thibaut et al., Neurology 2014; Angelakis et al., Arch. Phys. Med. Rehabil. 2014; Hermann et al., Sci. Rep. 2020), but can it influence conscious and unconscious dynamic functional connectivity? We investigated the modulatory effect of prefrontal cortex (PFC) transcranial direct current stimulation (tDCS) on brain dynamics in awake and anesthetized non-human primates using functional MRI. In awake macaques receiving either anodal or cathodal tDCS, we found that cathodal stimulation robustly disrupted the repertoire of functional connectivity patterns, increased structure-function correlation, decreased Shannon entropy, and favored transitions towards anatomically-based patterns. Under deep sedation, anodal tDCS significantly altered brain pattern distribution and reduced structure-function correlation. The prefrontal stimulation also modified dynamic connectivity arrangements typically associated with consciousness and unconsciousness. Our findings offer compelling evidence that PFC tDCS induces striking modifications in the fMRI-based dynamic organization of the brain across different states of consciousness. This study contributes to an enhanced understanding of tDCS neuromodulation mechanisms and has important clinical implications for disorders of consciousness.

Drug-based fMRI studies are increasingly used to understand drug actions, with many reports on drug-related functional connectivity (FC) changes. Typically, a fixed hemodynamic response function (HRF) is assumed, though HRF varies across the brain. Using a canonical HRF likely overestimates FC, whereas deconvolving the BOLD signal with a data-derived HRF provides more reliable metrics. Additionally, low-frequency oscillations (LFO) in the BOLD signal can artifactually inflate FC estimates. The extent to which drug effects on the brain align with receptor density distributions of drug sensitive receptors remains unclear.

We examined HRF modulation using fMRI data from healthy controls under different levels of LSD (n=10), Ketamine (n=8), and Propofol (n=10). Data was preprocessed using FSL Melodic, HRF was estimated via rsHRF, with and without LFO removal (RAPiDTiDE). Voxelwise group comparisons of the HRF’s shape was performed using 3dMSS implemented in afni. Finally, REACT was used to estimate the estimate target-enriched functional connectivity for each drug. Results show that HRF is locally modulated by all three drugs, suggesting prior studies may have overestimated drug effects on neural activity. Without HRF deconvolution, significant variance in BOLD signal is attributed to neurotransmitter densities—an effect lost after deconvolution. Interestingly, even in drug-free conditions, significant correlations persist, with no difference in strength between drug and non-drug states. These observations provide an interesting starting point for a more detailed investigation of drug effects which eventually might explain their phenomenology better.

The identification of features that are uniquely linked to conscious experience lies at the forefront of consciousness research, with important implications for the diagnosis and prognosis of disorders of consciousness patients. In the auditory domain, the local global paradigm has been proposed as a test for assessing consciousness level. This paradigm allows for the investigation of auditory regularity encoding at the local level, based on the repetition of single sounds, and at the global level, relying on a memory trace over the repetition of groups of sounds. In a pre-registered study (1), we investigated the specificity of this test by assessing global regularity encoding in comatose patients, in the absence of consciousness. We administered the local-global paradigm in a cohort of comatose patients (N=30) on the first days of coma after cardiac arrest and a cohort of healthy volunteers (N=15), who served as controls. The outcome of comatose patients was assessed at three months using the cerebral performance category (CPC) and classified as favourable (CPC=1-2) or unfavourable (CPC=3-5). We recorded 63-channel electroencephalography while patients and volunteers passively listened to auditory sequences, and we utilized multivariate decoding analyses to classify the neural response to standard and deviant sounds in local and global regularities. We found that both local and global regularity encoding was observable in comatose patients, irrespective of their outcome, and in healthy volunteers. These findings suggest a preservation of complex auditory regularity processing in the absence of consciousness in acute coma.

(1) Pelentritou et al.(2025), Brain Communications, 7(1): fcae466

Disorders of consciousness (DoC) are clinically heterogeneous conditions with significant prognostic challenges. Diffusion MRI (dMRI)-based structural connectivity (SC) captures static architecture, building the scaffolding of the dynamics that the brain can generate. We hypothesize that the critical dynamics expressed by the structure of these matrices can relate to the conscious capacities of recovery in DoC.

We obtained dMRI data from 28 DoC patients and constructed personalized SC matrices. These matrices were used as the coupling substrate for an Ising spin system. Spins served as proxies for local neural activity, and global coupling was systematically varied to identify critical points. Metrics derived from the Ising model such as heat capacity, critical temperature, and susceptibility were used along with renormalization exponents were inputted to a Leave-one-out decision tree to predict positive prognosis

The classifier showed low performance, indicating that the chosen features were not useful to classify recovering patients. We compared each marker’s distance to a null distribution generated by randomizing connections. Deviance from this null distribution did not differ between conditions. Standard connectivity measures were not significantly different when comparing diagnosis and prognosis groups.

We tested whether dynamical markers obtained by informing an Ising model with personalized SC matrices could predict recovery in DoC patients. Our classifier could not distinguish between classes due to the similarity of the SC matrices. We hypothesize that the low signal to noise ratio from these patients might be incompatible with the current tractography algorithm. Next steps include expanding the dataset and exploring alternative pipelines.

Traditional behavioral assessments for patients with Disorders of Consciousness (DoC) often result in high misdiagnosis rates. Advances in fMRI, however, have revealed significant cognitive abilities in many DoC patients (Xie et al., 2017). A strong link has been established between olfactory responses and consciousness. Unlike other sensory modalities, the olfactory system lacks an obligatory thalamic relay that may provide direct conditions for inducing consciousness. Its unique neuroanatomy may help distinguish conscious from unconscious states (Mori et al., 2013).

While olfactory task-related EEG has been conducted to assess consciousness, fMRI research on consciousness and olfactory stimuli remains limited (Merrick et al., 2014). This study will examine the neural correlates of consciousness in DoC and the diagnostic value of olfactory stimuli responses using fMRI by enrolling 25 participants with DoC and 25 healthy controls. They will be exposed to olfactory stimuli (vanillin, decanoic acid) and a baseline odorless condition, to isolate brain regions involved in olfactory processing (Wu et al., 2023). fMRI will record task-related brain activity and whole-brain analysis of relative power and functional connectivity will be performed. A block design with four stimulus blocks will be used, alternating 5 seconds of stimulation and 30 seconds of rest to avoid habituation. Nasal airflow will serve as a biomarker and behavior will be assessed using the Coma Recovery Scale Revised (CRS-R) before the experiment and after three months.

We hypothesize that olfactory stimuli generate quantifiable cerebral activity in individuals with DoC and may serve as neural indicators of consciousness, improving diagnostic accuracy.

Background

With very few treatments available, post-comatose disorders of consciousness (DoC) represent one of the greatest challenges in modern neurology. Following promising clinical trial results in psychiatry and a growing understanding of their brain mechanisms, psychedelics have been proposed as potential therapeutic agents for DoC patients due to their ability to increase the entropy and complexity of spontaneous brain activity in healthy individuals. However, no studies have yet investigated the effects of typical psychedelics (i.e., 5-HT2A receptor agonists) in DoC patients.

Methods

In this case report, we describe the first-ever administration of psilocybin to a 41-year-old patient in a minimally conscious state (one year post-traumatic brain injury). We assessed behavioural responses, electroencephalographic (EEG) changes, and autonomic parameters, including blood pressure and heart rate.

Results

We observed no improvement in overt behaviour as assessed with validated scales (i.e., SECONDs: unresponsive). However, novel spontaneous behaviours not previously seen, such as leg movements, were detected. EEG analyses revealed a decrease in relative power of slower frequencies (delta and theta), an increase in higher-frequency activity (beta and gamma), and an increase in brain complexity as measured by Lempel-Ziv complexity (LZC). No serious adverse effects were reported.

Conclusions

The increase in LZC during a state of unresponsiveness may suggest a state of disconnected consciousness, a hypothesis that should be further investigated in future studies. This report contributes to our understanding of the potential role of psychedelics in DoC, their broader applications in medicine, and the relationship between brain complexity and consciousness.

Breathing is a vital physiological process essential to human survival, and as expected from such an important function, it occurs automatically without the requirement of conscious control. Nonetheless, breathing automatism can be consciously modified and is regularly modulated by our conscious experience and consciousness state. For example, respiratory patterns are altered by a melancholy sigh, a scream of fear, or a yawn. In addition, breathing has an erratic structure during rapid eye movement sleep and a more regular form in deep non-rapid eye movement sleep. Yet, it is unclear precisely how consciousness shapes breathing. Disorders of consciousness (DoC) provide a unique model to investigate the interaction between consciousness and breathing and offer a rare opportunity to track breathing dynamics during recovery of consciousness. To characterize the interplay between breathing dynamics and consciousness state, we recorded nasal respiration repeatedly over time in brain-injured patients with DoC. We found that respiratory patterns can differentiate between minimally conscious state (MCS) and vegetative state/unresponsive wakefulness syndrome (VS/UWS) patients as well as between VS/UWS patients who remained unresponsive and VS/UWS patients who recovered consciousness and transitioned to MCS. These results uncover how consciousness shapes breathing and may provide an accessible bedside tool that signals consciousness and recovery in brain-injured patients.

INTRODUCTION: Zolpidem is a non-benzodiazepine sedative approved for the treatment of insomnia. However, it exerts paradoxical awakening effects in a fraction of the general population (~16%). To characterize these sedative and awakening effects, this study investigated neuropsychological, electrophysiological, and phenomenological effects of zolpidem in neurotypical individuals.

METHODS: This pilot cross-over double-blind placebo-controlled randomized clinical trial (RCT) was performed in two sessions during daytime with a one-week washout period. During each session, baseline electroencephalography (EEG) and electrocardiography (ECG) were recorded. Then, zolpidem (10 mg) or placebo (mannitol) was administered in a randomized order with EEG and ECG being recorded continuously for 45 minutes. One hour after placebo and zolpidem intake, post-intervention neuropsychological assessment was conducted, and participants reported their phenomenological experiences.

RESULTS: Preliminary results in two neurotypical individuals (25 and 28yo; female) responding non-paradoxically to zolpidem showed impaired executive functions with lower number of correct responses, more changes in decisions, as well as longer thinking and execution time under zolpidem effect, compared to placebo. Phenomenological reports indicated that zolpidem altered conscious perception as evidenced by visual distortion of reality in one participant. Preliminary EEG results were indicative of variations in power spectrum and connectivity measures after zolpidem vs. placebo intake. ECG measures illustrated decreases in time-domain heart rate variability after zolpidem intake.

CONCLUSION: This pilot study provides preliminary evidence on how zolpidem affects cognitive functions and neurophysiological activity of the brain. It also confirms the feasibility of a large-scale RCT in paradoxical and non-paradoxical responders to zolpidem.

Loss of consciousness (LOC) is a hallmark of some epileptic seizures. We recently demonstrated distinct ictal LOC mechanisms, with increased high-gamma power in focal to bilateral tonic-clonic seizures (FBTCS) and cortical sleep-like activity in focal impaired awareness seizures (FIAS). Microelectrode recordings suggest sustained firing increases in FBTCS, but this remains untested for FIAS and focal aware seizures (FAS).

Here we analyzed 22 seizures (6 FBTCS, 8 FIAS, 8 FAS), classified based on behavioral responsiveness, amnesia, and tonic-clonic activity, in six epileptic patients implanted with Behnke-Fried macro-micro electrodes recorded with Blackrock Microsystems. Multi-unit activity (MUA) was extracted from filtered signals (300–3000 Hz), sorted using UltraMegaSort2000, and normalized by 10-min baseline. Mean firing rates were compared between seizure halves (or pre-/post-generalization) in seizure onset zone (SOZ) vs. non-SOZ areas.

MUA analysis revealed a 1.8-fold increase in cortical firing rate at FBTCS onset, 4.6-fold increase post-generalization (p<0.05) in non-SOZ areas, whereas SOZ electrode showed a 28-fold increase at onset, persisting through generalization. FIAS exhibited a sustained 2.6-fold increase at onset (p<0.001), decreasing in the second half (p<0.05). FAS showed no firing change at onset (0.06-fold, ns) but increased slightly when seizure progress (0.5-fold, p<0.05).

These findings support the feasibility of characterizing MUA during ictal period and suggest distinct firing rate patterns for seizures with impaired vs. preserved consciousness. We confirmed increased firing in non-SOZ areas during generalization. FIAS showed less increase with a declining trend, while FAS exhibited a minor but progressive rise. Future work will examine neural firing in a larger sample.

Disorders of consciousness (DoC) are chronic states where conscious awareness is lost or limited. They represent a condition in which humans have preserved physiological functions but no consciousness. Prominent theories of consciousness highlight the importance of information integration. At the same time, recent work has pointed to brain organisational principles that follow a gradient from unimodal sensory regions to associative cortex. We might therefore suppose that loss of consciousness corresponds with brain changes that follow this gradient.

To investigate this, we obtained glucose metabolism measures (FDG-PET) from DoC patients (n = 69) and conscious controls (n = 19). A group difference (controls-DoC) in metabolism was calculated for 180 cortical regions. Gradient values from previously reported canonical functional gradients were obtained for the same regions. Finally, metabolism difference values were correlated with gradient values.

Differences in glucose metabolism were positively correlated with the first functional gradient, going from unimodal to associative gradient ends. A negative correlation between glucose metabolism difference and gradient position was seen for the third and fifth gradients.

These results suggest that, regardless of retained activity in sensory regions, consciousness also requires activity in information integration brain regions. The third and fifth gradients have, respectively, been associated with salience processing and the olfactory system. In the latter case, the correlation may relate to work showing that olfactory responses are associated with recovery of consciousness. Together, these results provide insights into the neuro-metabolic changes seen in DoC patients, highlighting how these correspond to core organisational principles of the brain.

Introduction

Predicting outcomes in pediatric disorders of consciousness (DoC) is challenging due to limited vali-dated clinical scales and developmental brain plasticity. Coherently, international guidelines provide limited recommendations for evaluating and managing pediatric DoC. To address these challenges, we developed a neurophysiological tool using pediatric-specific quantitative EEG (qEEG) features to im-prove prognostic accuracy.

Methods

A support vector machine algorithm was developed using qEEG indicators: relative powers in delta, theta, alpha and beta bands, spectral exponent, Higuchi Fractal Dimension (HFD). These metrics dis-criminate wakefulness from sleep across ages, capturing complementary properties. Relative powers and spectral exponent reflect EEG slowing in unconscious states, while HFD quantifies signal com-plexity. Based on adult studies, we hypothesized that EEG from pediatric unresponsive wakefulness syndrome (UWS) patients resembles sleep patterns, whereas minimally conscious state (MCS) patients exhibit wake-like activity. The algorithm was trained on wakefulness and sleep EEG data from 89 healthy children and tested on 8 pediatric DoC patients diagnosed as UWS.

Results

Among the 8 patients, 2 had unfavorable outcomes, while 6 improved to at least MCS. The algorithm accurately predicted unfavorable outcomes in both patients, showing no transitions toward wake-like activity in serial recordings. It also correctly predicted favorable outcomes in 4 of the 6 cases, identify-ing EEG transitions toward wakefulness that preceded clinical recovery. Predictions failed in 2 cases, likely due to the lack of serial recordings.

Conclusion

This study highlights the potential of qEEG features to predict outcomes in pediatric DoC, offering promising tools to enhance clinical decision-making and improve patient management.

Out-of-hospital cardiac arrest (OHCA) is a significant global health concern, reaching numbers of approximately 55 treated per 100,000 person years among adults, and frequently leads to hypoxic-ischemic brain injury (HIBI) and subsequent disorders of consciousness (DoC). Predicting long-term outcomes for this patient group, ranging from recovery of consciousness to persistent vegetative states or death, remains a significant clinical challenge. Current predictors, including age, anoxic episode duration, motor responses, and electroencephalography, provide limited precision, particularly for evaluating levels of consciousness. Magnetic resonance imaging (MRI) is underutilized, as the diffuse nature of HIBI damage complicates interpretation using conventional methods.

In this study, we seek to enhance outcome prediction for DoC by leveraging machine learning models trained on classical predictors in combination with advanced diffusion-weighted imaging (DWI)-derived features. Our approach aims to provide a more granular understanding of the relationship between structural brain changes and consciousness recovery after HIBI.

We present an experimental framework and initial findings from a cohort of patients scanned 3–14 days post-cardiac arrest. The presentation aims, in part, to solicit collegial feedback on our methods, including the integration of advanced MRI metrics and machine learning, while results remain preliminary.

Our goal is to refine this framework further in collaboration with the cross-sectional consciousness research community. By combining advanced neuroimaging techniques and machine learning models, we aim to contribute to the development of approaches that could ultimately improve outcome prediction and clinical decision-making in DoC following HIBI.

Transcranial direct current stimulation (tDCS) over the dorsolateral prefrontal cortex (DLPFC) can improve behavioural responsiveness in minimally conscious state (MCS) patients, as assessed by the Coma Recovery Scale-Revised (CRS-R) scale. Previous research has been limited by an arbitrary timing of stimulation, while MCS patients are known to present vigilance and responsiveness fluctuations. This first pilot sham-controlled randomized crossover trial aimed at evaluating the effects of tDCS applied over the DLPFC in a brain state-dependent manner. We used a customized 20-channel EEG and tDCS software (Neuroelectrics) to compute a spectral entropy index, previously shown to correlate with vigilance (ultradian cycles 70min) in MCS patients. Fluctuations of this index were used to trigger the tDCS application in three different conditions: high, low, and random vigilance. Our primary outcome measure was the change in CRS-R score while the secondary outcomes focused on EEG power and connectivity changes between the three conditions. We included 12 patients: 11 in MCS and one emergent from MCS (5 females, 5 traumatic aetiologies, 50.3 ± 17.4 years old and 75.0 ± 127.6 months since injury). The group median CRS-R score increased in the high condition (pre post: 9.5-11), remained identical in the low one (9.5-9.5) and decreased in the random one (9.5-8.5). However, these differences were not statistically significant (Kruskal-Wallis p=0.24). The secondary EEG analyses are still undergoing. This primary research shows potential beneficial effects of brain-state dependent application of tDCS in MCS patients but further trials on larger groups are warranted.

Consciousness involves the flow of neuronal information strongly linked to oscillatory synchronization in the brain. The alignment between environmental rhythms and brain oscillations is known as neural entrainment, and it might play a role in states of consciousness. When it occurs in response to auditory stimuli, it is termed auditory neural entrainment - brain waves syncing with sound waves.

Building on previous research about the effects of music on people with disorders of consciousness, our study aimed to characterize and modulate states of consciousness (focused attention and mind wandering) through matching and mismatching of auditory neural entrainment. Furthermore, we aimed to investigate the effect of musical training on entrainment.

This study used electroencephalography (EEG) and auditory stimulation (binaural beats) to investigate auditory synchronization in 5 binaural beat (BB) conditions, matched in frequency with the brain waves delta, theta, alpha, beta and gamma.

We hypothesise that neural synchronization caused by low frequency BB will increase the power of delta, theta and alpha brain waves, and high frequency BB will entrain beta and gamma brain waves, correlating with behavioral measures of focused attention and mind wandering.

Understanding neural synchronization with auditory stimulation in diverse states of consciousness holds potential for future research focused on improving the rehabilitation of disorders of consciousness, as well as deepening our knowledge about the effects of sound on the brain.

Background

Two influential theories of consciousness imply different necessary brain features. Global neuronal workspace theory (GNW) centres consciousness in prefrontal regions. Notably, work on GNW has highlighted the inferior frontal gyrus (IFG) as a core region for awareness. Information integration theory (IIT) implies that consciousness be more centred on posterio-medial brain regions, such as the posterior cingulate cortex (PCC). To date, there have not been experimental results that allow these theories to be discriminated. Lesion studies provide one tool for investigating this question by presenting situations where theoretically implied brain regions are removed.

Methods

We investigated a patient who had most of their prefrontal cortex surgically removed but who remained conscious and capable of complex thought. Delineating the IFG and PCC in this patient, we measured regional neural function through arterial spin labelling (ASL). Patient ASL values for each region were compared to normative values from participants with normal brain function.

Results

Patient’s PCC activity was within normal range bilaterally. The left IFG was within normal range but the right IFG showed ASL values indicative of no neural function.

Conclusions

This study highlights how consciousness can be preserved even with extensive prefrontal cortex loss, which is partly in tension with the assumptions of GNW. However, the preservation of IFG function in one hemisphere may support a more specific version of this theory. Normal activity in the PCC lends some support to IIT. This case adds context to the discussion of neural correlates of consciousness in relation to clinical conditions.

Following severe brain injury, patients may present with a disorder of consciousness (DoC), whose diagnosis remains challenging. Recent advancements comparing behavioral scales with cerebral recordings have identified a novel DoC phenotype: cognitive-motor dissociation, characterized by preserved cerebral functions without overt behavioral expression.

These findings highlight the absence of a single, unimodal signature for consciousness recovery and emphasize the need for multimodal and continuous evaluation of DoC patients. However, no such integrated tool currently exists.

We developed a new methodology leveraging computer vision to quantify fine-grained and comprehensive behavioral metrics (e.g., gaze, movements, facial expressions) synchronized with EEG (e.g., spectral power, dwPLI connectivity) and ECG (e.g., R-R interval) recordings. Results in healthy individuals revealed unique multimodal signatures capable of effectively discriminating experimental conditions (varying by the presence of a sound or an individual, and/or by emotional intensity). For instance, the presence of a person increased gaze and body movements while reducing the R-R interval and alpha spectral power. Temporal dynamics revealed evolving patterns across modalities, with sequences of coupling and decoupling.

We are now applying this methodology to DoC patients. Preliminary findings indicate that multimodal signatures elicited by similar environments differ significantly from those of conscious individuals, both globally and temporally. These insights into brain-body-environment interactions provide a promising framework for exploring the mechanisms underlying consciousness and improving DoC assessment.

While the diagnosis of disorders of consciousness (DOC) can be reliably achieved with batteries of behavioral tests, the outcomes of such patients are much harder to predict.

Here, we present a proof of concept for a method which could allow for both a higher-throughput diagnosis, as well as prediction of outcomes of patients with DOC. EEG recordings were collected (Engemann et al. 2018) both during resting state and during a task meant to elicit surprise responses on short and long timescales (Bekinschtein et al. 2009). Our method couples a novel type of autoregressive (AR) modelling on EEG data with machine-learning classification. Instead of modelling the signal as a weighted sum of the past N values, like in traditional AR (Box et al. 2009), we also explore lags spaced according to an exponential factor. This approach enables us to generate multiple AR models per dataset by varying the exponential spacing factor, which determines how far back in the signal's history each model incorporates. We then use multi-layer perceptrons to assess how predictive the AR coefficients are for determining patient diagnosis and outcomes.

We first show that the quality of the fit of these models correlates to how much information they contain about the diagnosis of the participant, but that the factor which generated the models doesn't. Furthermore, we show that it is possible, using the best models per dataset, to diagnose patients and to at least distinguish between favorable and unfavorable outcomes.

Severe brain lesions causing disorders of consciousness [DOC] can arise from various origins (traumatic brain injury [TBI], stroke, or anoxia) and heterogeneous mechanisms. Detecting and characterizing lesions (number, volume, location) is crucial for clinical and research purposes in both diagnosis and prognosis dimensions. Most studies focus on single etiology based on manual segmentation which are time consuming, non-reproducible and operator-dependent.

We aimed at developing an automatic method for lesion segmentation that can be applied across these different etiologies and injury mechanisms. We collected MRI data from 71 DOC patients within the first month post-injury, 67 of them had a follow-up longitudinal MRI after 4 months.

Our approach relied on lesion maps obtained from T1 or FLAIR MRI with three different segmentation models (Pixyl®): 1) multiple-sclerosis, 2) TBI/stroke and 3) glioblastoma. (1) and (2) provided probabilistic maps that we thresholded at an optimal individual threshold without any tissular classification. (3) gave complementary labels of tissular classification: neo-volumes intra- and extra-cerebral (intracerebral versus subarachnoid hematoma) and surrounding edema. Incomplete neo-volumes were manually corrected in the minimal number of cases.

We combined these maps to obtain the following integrated features: maximal brain lesion load (all labels except extra-cerebral neo-volume), parenchymal lesion load (all labels except neo-volume), extrinsic and internal deforming brain lesions load (neo-volume labels).

Overall, this semi-automatic procedure simplifies lesion segmentation and makes it more objective compared to manual delineation. Our method addresses most challenges of lesion heterogeneity that clinicians are facing, ensuring more consistent and reliable segmentation across different DOC etiologies.

INTRODUCTION: Consciousness is thought to be supported by the brain through interactions between the thalamus, basal ganglia, and cortex. Within the mesocircuit hypothesis, the EEG power spectrum’s shape, visually classified in “A”, “B”, “C”, and “D”-type patterns, is thought to reflect the extent of thalamocortical deafferentation in patients with Disorders of Consciousness (DoC) after severe brain injury, ranging from most to least impaired. This ABCD model has shown diagnostic value in the acute (<28 days) stages of DoC.

METHODS: Here, the ABCD classification is validated in a large sample (140 prolonged DoC, 29 controls), and associated to the cerebral glucose metabolism FDG-PET through a region-based analysis (RBA) in a subsample (n=108). Three raters performed the ABCD rating twice, at least a month apart, blinded to behavioural diagnosis, previous ratings and each other’s ratings. Consensus was reached in a separate session.

RESULTS: There was substantial inter-rater (κ=0.64) and intra-rater (κ=0.70) agreement. Increased consciousness was associated with more favourable patterns (χ2(16, n=169)=85.22, p<.001). RBA including controls highlighted that the glucose metabolism associated with “D”-type patterns was significantly higher across the brain compared to all other patterns. Without controls, differential glucose metabolism was observed between “A” and “B”-types, and more locally between “C” and “D”-types.

CONCLUSION: These results support the notion that simple, visual EEG spectral power inspection in prolonged DoC can be clinically informative. The RBA will promote development of objective guidelines for pattern identification to enhance rater reliability, support clinical translation, and pinpoint critical brain regions necessary for sustaining thalamocortical electrogenesis.

Background and Methods

To inform the debate about including fMRI in assessment and prognostication of people in prolonged disorders of consciousness (PDOC), we report the long term follow up (mean[stDEV] 90[22]) months post imaging (MPI) of a research cohort who had had fMRI at 20 [median 11 months SD 4] months MPI. Initial admission included detailed neurological examination, recording of Coma Recovery Scale-Revised (CRS-R), and fMRI. Follow-up involved CRS-R recording in person, or video/telephone, with relatives/professionals.

Results;

Of the total cohort (n=72) 22% had emerged and 47% had died.

13/53 who were in VS or MCS- at initial inclusion, had demonstrated positive fMRI BOLD responses to a command following paradigm (imagining playing tennis or navigation); 2/13 emerged at very long term follow-up. However, 4/40 in VS/MCS- at first assessment, who had not demonstrated positive command following responses in fMRI, had also emerged at follow-up.

6/19 in MCS+ at first admission demonstrated command following in fMRI. Only 3 of these 6 had emerged at follow-up;

Of the 13 initially in MCS+, who did not demonstrate fMRI command following, 7 had emerged.

Conclusions;

The group is small and heterogenous with wide variations in times since injury, however these findings illustrate that

1.a significant number of people in PDOC may emerge late after injury, and

2. fMRI paradigm responses alone, performed many months after injury, may not be predictive of long term emergence.

In addition, negative findings on command following fMRI paradigms do not exclude possible later emergence.

Experimental literature suggests that information integration and awareness are closely related. Some even claim that cerebral markers of perceptual integration (like the P3 wave) are also markers of awareness in non-communicative patients following a coma. In contrast, we have recently shown that ⅓ of these patients could exhibit a discriminative brain response to their own name, whatever their diagnosis (coma, vegetative state, minimally conscious state), and that delirium patients show no such response despite being conscious, evidencing a cognition-consciousness dissociation. We hypothesize that there are specific markers of information integration and awareness that can be disentangled through the study of neuronal synchronies on a local and/or large scale.

We recorded high-density EEG in 129 patients (coma=40, vegetative state=20, minimally conscious state=17, delirium =17, conscious=17) at rest and during exposure to two auditory protocols (neutral and emotional sounds). Then we performed power spectral analysis and phase synchrony analysis (dwPLI) as measures for local and long-distance neural synchronies, respectively.

Preliminary results show that conscious and unconscious patients can be distinguished by the delta-to-alpha power ratio and the strength of long-distant connectivity in these two frequency bands. They also show that the different levels of information integration (rest, neutral sounds, emotional sounds) are associated with variation in power and connectivity in high-frequency bands (notably in the alpha and beta bands). These results could allow caregivers to better assess non-communicative patients by dissociating their cognitive abilities and awareness.

Methods for assessing disorders of consciousness (DoC), including neurobehavioral scales, EEG, and fMRI, have limitations in acute care due to reliance on overt motor responses, cost, and feasibility constraints. Functional near-infrared spectroscopy (fNIRS) detects changes in oxygenated (HbO) and deoxygenated hemoglobin (HbR), providing an objective, real-time method for evaluating cerebral hemodynamics. Grounded in neurovascular coupling (NVC), fNIRS captures cerebral blood flow changes linked to neuronal activity. In auditory oddball paradigms, EEG captures event-related potentials such as mismatch negativity and P300, while fNIRS measures associated hemodynamic changes during auditory processing.

We evaluated fNIRS’s ability to detect neurovascular responses to auditory stimuli in healthy individuals, establishing baseline activation patterns for future comparisons in DoC. A 30-second auditory oddball paradigm was implemented, consisting of standard 500 Hz and deviant 2000 Hz tones. Hemodynamic responses were recorded using 18-optode continuous-wave fNIRS. Data preprocessing included motion artifact correction and finite impulse response filtering. Analyses examined HbO and HbR changes across task periods to assess responses to standard and deviant tones.

Thirteen healthy participants (mean age: 37 ± 15yo; 69% female) were recruited. Results demonstrated significant HbO increases following deviant tones, suggesting cortical engagement during auditory processing. Variability across optodes and participants reflected neurovascular dynamics, aligning with NVC and novelty detection. Establishing these response patterns provides a reference for assessing brain function in clinical populations, including DoC patients. fNIRS enables real-time, noninvasive cortical monitoring and offers a practical alternative where conventional neuroimaging is impractical due to cost, accessibility, or patient instability.

Transcranial ultrasound stimulation (TUS) is an emerging non-invasive brain stimulation technique that has shown promise as a potential treatment for Disorders of Consciousness (DOCs). However, existing studies have only investigated the effects of TUS on the central thalamus, and the precision of sonication of the intended target in TUS interventions for DOC requires further validation. Additionally, existing interventions have not addressed the distinct roles of thalamic regions beyond the central thalamus in DOC pathologies.

This study therefore aims to advance the development of TUS as a reliable, adaptable and mechanistically informed treatment for DOCs. To this end, we will validate a novel TUS protocol in healthy participants by demonstrating differential modulation of two thalamic nuclei implicated in two distinct DOC clinical phenotypes: patients who show no behavioural or neuroimaging evidence of awareness, and those with cognitive motor dissociations (CMD). Specifically, we will present data on the behavioural effects of offline sonication of the mediodorsal and ventrolateral thalamic nuclei on performance on a backward masking task and a motor task involving graded force production. By establishing task- and region-specific effects, this study seeks to lay the groundwork for future applications of TUS as a personalised and evidence-based treatment for DOCs. Moreover, this work may extend our understanding of the causal roles of specific thalamic nuclei in mediating awareness versus responsiveness, offering insight beyond our intervention’s therapeutic application.

Introduction:

Individuals with hypersomnia, including narcolepsy type 1 (NT1) and idiopathic hypersomnia (IHS), can experience altered states of consciousness when tired, including intense mind wandering (MW), mind blanking (MB), and hallucinations. In healthy individuals, MW and MB are associated with sleep-like slow waves (SW). This study explores altered consciousness in NT1 and IHS patients and their relationship with SW dynamics.

Methods:

23 healthy participants, 16 NT1, and 8 IHS performed a Sustained Attention-to-Response Task (SART) for one hour each in the morning and afternoon. Periodic task interruptions captured changes in the content of subjective experiences (including hallucinations), volitional control over the stream of thoughts, and sleepiness. We computed accuracy and reaction times as measures of behavioral performance and recorded high-density EEG to detect SW.

Results:

Compared to controls, NT1 and IHS patients exhibited slower reaction times and more misses. NT1 patients showed more false alarms and hallucinations, while IHS patients experienced more frequent MW. Increased frontal SW density in NT1 and steeper occipital SW slopes in IHS were observed.

Conclusion:

Distinct SW patterns in NT1 and IHS highlight disruptions in wakeful consciousness. In NT1, increased frontal SW density may contribute to perceptual distortions and hallucinations. In IHS, steeper occipital SW slopes are suggested to impair sensory integration, promoting inwardly directed cognition and MW. These findings extend prior research in healthy controls, situating hallucinations, MW, and MB on a continuum between wakefulness and sleep. Our results suggest that fluctuations in consciousness may stem from local imbalances in wake and sleep dynamics.

Dreaming is a phenomenon that is of great interest to consciousness research. It is very common and easily attainable; however, it is also very elusive and difficult to study effectively. Many theories and hypotheses have been created over the decades, such as Revonsuo’s threat simulation theory (2009) and Domhoff’s neurocognitive theory of dreaming (2019). Many of which are very specific to the research context in which dreaming was investigated. This led to many, very context dependent views and even definitions of dreaming. In recent years, the need for a more unified understanding of dreaming has become more and more obvious – to facilitate sharing of information and the conversation around this phenomenon.

The Multilevel Framework is an analytical framework to investigate multiple aspects of biological phenomena. It is based on Tinnbergen’s four questions on the proximate and ultimate causes. The idea to apply this framework to dreaming was first introduced by Valli (2011).

Ideally, this classification of dream theories will contribute to the conversation by simplifying the comparison of existing theories as well as identifying interesting overlaps, contradictions and blind spots in them which will hopefully lead to more effective study questions and paradigms in the future. An effective comparison of these theories will also help with sharing and will be a great expansion on multi-center studies and shared databases.

Dreams have been a source of inspiration for millennia. Investigations of the contributions of dreams to creativity, on the other hand, have been limited by the difficulty of experimentally manipulating REM-sleep dreams. A few studies showed that memory reactivation during non-REM sleep can promote creativity. Here, we combined new strategies for influencing dreams with a procedure in which people went to sleep after failing to solve four different puzzles, and then attempted to solve them the next morning and then again after a longer delay. We recruited individuals for overnight sessions who claimed to be frequent lucid dreamers, aware that they are dreaming during their dreams. Before sleep, they attempted to solve various puzzles, each associated with a unique sound. Half of the sounds linked with unsolved puzzles were presented again during REM sleep and when possible, during a lucid dream signaled by eye movements. Participants were told to attempt to solve a puzzle if they heard its sound in their sleep and indicate that they were working on a puzzle via sniffing signals. We found that sound cues increased lucid and non-lucid dreams of associated puzzles, and that dreaming of specific puzzles was associated with increased solving both the next morning. Whereas it is well-known that sounds can be incorporated into dreams, here we induced dreams for the specific goal of reactivating memories of pre-sleep experiences. Our findings support the notion that REM sleep can boost creativity in relation to the specific content of experienced dreams.

Sleep is a reversible condition of reduced awareness and responsiveness to the external environment. Nevertheless, even during sleep, organisms must regularly sample the environment, create predictions, and detects their violation. Indeed, compelling evidence indicates that the sleeping brain can detect simple sensory deviation. However, only a few studies investigated more complex predictions, and it remains unclear how sleep modulates the formation of predictions and surprise responses. To answer this question, we recorded high-density EEG from healthy participants in sleep and wakefulness while they passively heard an auditory oddball-omission paradigm. The paradigm included expected and unexpected omitted sounds with intermediate complexity rules, which enabled to disentangle between the neural response to the “pure” prediction error and the neural response to the stimulus’s physical properties. ERP analysis showed a significantly increased negativity at 100-300ms following omission onset in the unexpected omission condition compared to the expected omission in wakefulness, however, was not evident in NREM and, REM sleep. This result implies that the sleeping brain's ability to create predictions more complex than a mere sensory deviation is compromised.

In cognitive psychology, perceiving a stimulus can be defined in two key ways: Conscious Perception, when a stimulus breaks into your conscious experience, and Attentional Capture, when a stimulus automatically grabs attention. Both processes prioritize cognitive resources, making it reasonable to assume they are at least somewhat aligned. This study examines the link between non-conscious prioritization speed (NPS), which reflects how quickly stimuli are detected under masking, Attentional Capture (AC), and Inattentional Blindness (IB) across two experiments.

Experiment 1 measured NPS using repeated breaking masking suppression (bRMS). An additional singleton task measured AC. IB was measured by sustained IB. Experiment 2 measured NPS similarly to experiment 1. IB was measured using a traditional single-trial sustained IB and a modified version with multiple trials, which contains noisier distracting tasks and variance on target stimuli location.

Surprisingly, we found no significant effects, suggesting that our tasks may not measure what we intended. One possibility is that the assumed link between NPS, AC, and IB does not hold due to limitations in task design. Alternatively, visual noise or overload may have interfered with participants' ability to process stimuli as expected, masking potential effects. It is also possible that there is no inherent correlation between NPS, AC, and IB, indicating that the cognitive processes may be more distinct than initially hypothesized. These findings raise the need to reassess our paradigms and consider adjustments to better isolate the involved cognitive mechanisms better.

Music’s temporal and hierarchical structure offers a powerful tool for probing the brain’s predictive mechanisms. Sleep, in turn, reveals how arousal and consciousness modulate these processes. We propose that the brain’s capacity to track statistical regularities in music during NREM sleep depends on both stimulus features and neural state, with second-order prediction errors, like surprise shaped by both.

Participants (N=23) monitored with high-density EEG slept to a three-hour long monophonic musical composition with gradual transitions across four melodic & rhythmic conditions. To estimate predictability, we computed note-level Information Content (IC) of pitch and note-onset timing using IDyOM, a predictive model with varying N-gram scales. Longer time-scales (>n=4) exhibited statistically higher mean ICs during transitions between conditions, indicating increased unexpected events. We synchronised EEG data with acoustic events (note onset) and IC time-series. Using multivariate temporal response functions (mTRF), we assessed how acoustic vs melodic expectation components contributed to distinct cortical responses during wake and slow-wave sleep.

Preliminary results suggest that the brain's response in both wakefulness and NREM sleep is driven by the note-onset strength envelope, showing preserved auditory processing across consciousness levels. We expect these responses to be more pronounced during transitions between conditions, and in high IC notes, indicating that the sleeping brain continues to partially process the anticipation of auditory inputs in NREM sleep.

Our findings contribute to the broader exploration of cognition in altered conscious states, by exploring the temporal integration of acoustic objects during sleep, whilst leveraging implicit learning from naturalistic stimuli.

Lucid dreaming (LD) is the phenomenon during which dreamers are aware they are dreaming and may be able to influence dream content. Dream manipulation, or dream control, plays a critical role in the applications of LD across both clinical and non-clinical domains and is a key determinant of therapeutic success. However, not all lucid dreamers are able to effectively influence their dreams, and achieving dream lucidity does not automatically translate into meaningful control. The unpredictable nature of dreams means that dream manipulation does not always work as planned. Challenges with dream control can hinder the LD experience as well as its therapeutic effects. Despite the growing interest in LD and its clinical and non-clinical applications, research on the mechanisms, reliability, and effectiveness of dream manipulation remains limited.

This presentation critically reviews the current state of research on dream manipulation within LD, identifies gaps in knowledge, and proposes future directions for addressing challenges in reliably controlling dreams. It begins by exploring the elements of dreams that can be influenced through dream manipulation (such as the dream environment, body, and narrative) and discusses strategies, challenges, and individual differences in dream control. Additionally, it explores the possible clinical and non-clinical applications of manipulating lucid dreams. Lastly, it calls for future research that focuses on developing standardised methods to assess dream control skills, empirical studies evaluating the effectiveness of dream manipulation strategies, and a deeper understanding of the cognitive, emotional, and neurophysiological factors that contribute to successful dream manipulation.

Introduction: Despite important progress, the neural mechanisms associated with dreams, and our ability to remember them, remain poorly understood (Stickgold et al. 2001; Nir & Tononi, 2010; Ruby et al. 2011,2013; Eichenlaub et al. 2014; Vallat et al. 2017; Siclari et al. 2017). The goal of this study was to leverage machine learning tools to harness novel insights through data-driven comparisons of brain activity recorded during sleep in individuals with high dream recall (HR) versus those with low dream recall (LR).

Methods: We used sleep polysomnography alongside state-of-the-art machine learning techniques, including Riemannian space classifications, linear discriminant analysis and deep learning, to analyze sleep EEG from 36 participants. Our investigation focused on differences in EEG across sleep stages and examined a range of features including covariance and cross-spectral matrices, as well as spectral power.

Results: We found that sigma band (11-16 Hz) power spectra and cross-spectra during stage S2 sleep most effectively predicted dream recall levels. Additionally, during REM sleep, EEG covariance matrices and delta band (1-4 Hz) cross-spectra were most informative. A region-specific analysis highlighted the significance of slow oscillations over prefrontal areas in distinguishing between HR and LR groups. Importantly, running a feature-agnostic convolutional neural network classifier on the raw EEG data yielded results largely confirming those obtained through hand-crafted features.

Conclusion: These findings confirm and extend previous research on the role of neural oscillations in dream recall. Crucially, our cross-validated ML insights demonstrate the transformative potential of data-driven approaches in elucidating the neural mechanisms of dream recall.

Sustained cognitive effort, such as prolonged attention to a task, is cognitively challenging. Subjectively, fatigue manifests through diverse experiences. These include tiredness or drowsiness, which are often conceptualized as distinct mental states linked to different physiological processes. Tiredness is typically seen as a use-dependent process, whereas sleepiness is seen as a time-dependent process. Regardless of its origin, mental fatigue leads to a deterioration of performance. Yet its effects are complex: individuals may exhibit sluggishness at times and impulsivity at others.

I argue that the variability in fatigue’s behavioral and experiential consequences does not preclude a common neural origin. I will draw on recent studies employing sustained attention paradigms to induce mental fatigue, assessing its impact on behavior and subjective experience. Neurophysiological data were collected using EEG or EEG-fMRI, with participant groups including neurotypical individuals, as well as those with attention deficits (ADHD) or sleep disorders (hypersomnia).

I propose that sleep-like intrusions underlie local neuronal off periods, disrupting cortical processing and altering brain connectivity. These brief episodes of neuronal silencing would account for the paradoxical effects of fatigue. Furthermore, the differences in the behavioural and experiential consequences of these sleep-like intrusions would arise from the variability in the location of these local changes, not in the physiological nature of the changes themselves.

A framework centered on sleep-like intrusions offers a unified perspective on mental fatigue by linking together fluctuations in attention and arousal. More generally, this model suggests that differences in subjective experiences do not necessarily reflect distinct underlying neural mechanisms.

Affective experiences are integral to both wakefulness and dreaming, yet their neural underpinnings in dreams remain largely unexplored. Frontal alpha asymmetry (FAA)—the relative difference in alpha-band power (8–13 Hz) between the right and left frontal cortical regions—is a well-established trait-like marker of affective processing and affect regulation in wakefulness. A prior sleep laboratory study found that greater FAA (reflecting lower right frontal activity or higher left frontal activity) during rapid eye movement (REM) sleep and evening resting-state predicted increased anger in dreams, suggesting a shared neural basis for affect regulation across wakefulness and dreaming. The present study aims to conceptually replicate and extend these findings in a larger sample, using more ecologically valid home dream reports. Eighty-six participants (18-70 y, M=27.28, 74 female) recorded their dreams and rated their dream affect each morning for 14 consecutive days. On day seven, waking FAA was assessed during an 8-minute resting-state EEG session conducted in the laboratory. Multi-level regression models will be performed to test the relationship between FAA (F4–F3) and self-reported dream affect, particularly anger and interest. These findings will contribute to a deeper understanding of how trait-like neural markers of affect regulation in wakefulness relate to affective dynamics in dreams, offering insights into the continuity of emotional processing across different states of consciousness.

Are we self-conscious when we dream of ‘being someone else’? Self-consciousness in dreams raises several important philosophical issues. When we dream, self-consciousness, the awareness one has of oneself as oneself, can be highly altered, reduced or even absent. Yet individuals report ‘vicarious dreams’, dreams in which the protagonist appears to be self-conscious but believe themselves to be a different person than the dreamer. Under certain interpretations of the psychological continuity theory of personal identity, the dream protagonist appears to be a different person than the dreamer. Thus, when self-consciousness arises in vicarious dreams, the dreamer themselves may not be conscious of anything. Drawing on philosophical and scientific literature about personal identity and dreams, I argue that dream protagonists can be psychologically isolated from the waking self in important ways. Taking into account features that are often considered to be important for psychological continuity such as sense of self, personality, values, and, in particular, memory, the protagonist can appear to be psychologically discontinuous, believe they are a different person, and be unaware of any connection with the sleeping individual. Autobiographical memory, memory of one’s own experience, is often seen as the key to psychological continuity. However semantic and procedural memory, memories of facts and skills respectively, have unduly received less attention, and are also important for self-consciousness. All three types of memory can be highly disrupted in dreams, supporting the hypothesis that a dream protagonist can attain self-consciousness of a self who is not the dreamer of the dream.

Parasomnias (e.g., sleepwalking/talking and night terrors) are intriguing examples of wake-like behaviours occurring in the deepest sleep stages with no conscious awareness or memory of these behaviours having occurred. Rather than being fully awake or asleep, parasomnias show that we often experience a mixture of characteristics from both waking and sleeping states. But what affects the blurring between wakefulness and sleep and are some people more vulnerable to experiencing sleep-state dissociation? We addressed these questions in a cross-sectional study (N = 133) examining the role of individual differences in sensory sensitivity (Adult Sensory Profile), arousal predisposition (Arousal Predisposition Scale), and cognitive/somatic pre-sleep arousal (Pre-sleep Arousal Scale) on parasomnia frequency (Munich Parasomnia Screening). Across the majority of parasomnia types, those who experienced parasomnia had significantly higher levels of (1) sensory sensitivity, (2) pre-sleep somatic arousal, and to a lesser extent (3) pre-sleep cognitive arousal. A parallel moderated mediation analysis further demonstrated that sensory sensitivity directly predicted parasomnia occurrence and was mediated by somatic but not cognitive pre-sleep arousal. Importantly, there was a conditional indirect effect, where arousal predisposition moderated the relationship between sensory sensitivity and parasomnias through somatic pre-sleep arousal. At average and high levels of arousal predisposition, sensory sensitivity led to greater somatic pre-sleep arousal, which increased parasomnia occurrence. Findings reveal how individual differences in sensory sensitivity and somatic arousal may contribute to an increased vulnerability to parasomnias and emphasise the importance of both trait-like vulnerabilities and state-like arousal in sleep disruptions characterised by increased permeability between conscious and unconscious states.

Sleep is integral to consciousness, and its disruption, particularly in Insomnia Disorder (ID), offers a unique window into altered cognitive and affective states. ID is marked by hyperarousal and impaired emotion regulation, with neural alterations linked to negative affect and reduced positive mood (Van Someren, 2021). Recent findings suggest distinct ID subtypes based on subjectively characterized overall distress and arousal levels (Blanken et al., 2019), potentially reflecting divergent neural configurations.

With this study protocol, we will investigate subtype-specific differences in dynamic functional connectivity states and arousal fluctuations in ID. Forty participants aged 20-50 years (20 by group; slightly and highly distressed) will undergo ultra-high-field (7T) MRI. First, by employing a Dynamic Functional Connectivity (DFC) brain-state approach on resting-state fMRI, each temporally resolved connectivity pattern will be assigned to an integrated, segregated, or small-world network state, characterizing the fluctuations of these recurring and transient global brain states. Next, we will explore the causal effect of arousal fluctuations on whole-brain state changes. Arousal levels will be estimated from the BOLD signal of the Locus Coeruleus (LC), a key noradrenergic center, localized through a validated segmentation process and LC-specific MRI sequence (Koshmanova et al., 2023).

We hypothesize that subtype-specific arousal differences will align with different DFC organizational properties, shedding light on ID’s neural mechanisms. Beyond the study of insomnia, this work will potentially advance fMRI-based arousal tracking with implications for clinical, cognitive, and consciousness research.

As one falls asleep, a rich repertoire of conscious experiences arise, ranging from fleeting thoughts to immersive dream-like experiences. Tracking these fluctuations may help pinpoint the neural mechanisms underlying spontaneous conscious experiences. However, their conceptual and neurophysiological definitions remain unclear, limiting our understanding of mental states during the transition to sleep.

Here, we aimed to classify the various conscious experiences that emerge during the sleep onset period without presupposition. We regularly interrupted 103 healthy participants during two 20-minute nap periods and asked them to evaluate their mental content across several dimensions, including bizarreness, fluidity, spontaneity, and subjective wakefulness. Brain activity was monitored using a 64-channel EEG to track sleep onset.

We characterized mental experiences in a data-driven way by applying a k-means clustering approach, grouping mental reports (N=375) based on their dimension scores. We found 4 clusters, each reflecting distinct phenomenological properties: 1) fleeting past memories, 2) environmental perceptions, 3) dream-like imagery, and 4) prospective thinking. These clusters appeared across Wake, N1 and N2 stages, stressing the limitations of associating specific types of conscious experiences with coarsely defined brain states. Independently of the sleep stage, dream-like imagery was associated with reduced cortical activation (i.e., lower total power and aperiodic offset) and decreased frontal-occipital connectivity (wSMI), while the other clusters exhibited differences in complexity (Kolmogorov, Permutation entropy).

By linking distinct mental states to specific EEG signatures, our approach offers a powerful framework for probing the neural basis of spontaneous conscious experiences across sleep and wakefulness.

The auditory steady-state response (ASSR), particularly at 40 Hz, is a sensitive marker of changes in arousal levels.While it was found to reduce its amplitude during deep sleep, the changes in cortical connectivity were not studied before. In this study, we examined how wakefulness, NREM (N1, N2, N3) and REM sleep affect the propagation of 40 Hz ASSR. We analyzed alterations in the direction and extent of signal flow between bilateral frontal regions (anterior and dorsolateral prefrontal cortex), auditory areas (primary auditory cortex, supratmeporal gyrus), and posterior associative areas (temporoparietal junction, posterior intraparietal lobule).

We hypothesized that (i) long-range connectivity weakens with increasing NREM sleep depth, (ii) front-parietal, fronto-auditory, and auditory-parietal connectivity patterns will be mostly affected by the arousal level, (iii) strong fronto-auditory connectivity is most prominent during wakefulness. We assessed effective connectivity using the Directed Transfer Function (DTF) in the delta (1-4 Hz) and low-gamma (37-43 Hz) bands during periodic 40 Hz auditory stimulation.

Main analysis in the gamma band showed sleep-dependent connectivity changes for 40 Hz ASSRs. In the N1 stage, fronto-auditory weakened but recovered in REM sleep. During N2 stage, connectivity between auditory areas and temporoparietal junction was notably reduced. In deep sleep, N2 and N3 stages, fronto-parietal connections were significantly diminished. These findings highlight the dynamic nature of effective connectivity across arousal states, offering new insights into auditory (40 Hz ASSR) processing alterations during sleep, and confirm the sensitivity of 40 Hz ASSRs to state changes in consciousness.

Research supported by the NSC no. 2018/31/B/HS6/03920.

Perceptual experience fluctuates significantly throughout the sleep-wake cycle, varying according to the level of consciousness. While awake, we are acutely attuned to our surroundings and can report our conscious percepts. During sleep, however, it is unclear whether conscious percepts disappear or persist but go unreported due to the inability to produce conscious reports. Our study investigates whether islands of consciousness emerge during sleep, reflecting conscious experience of external stimuli that remain unreported.

This study was conducted in patients with drug-resistant epilepsy undergoing stereoelectroencephalography with intracerebral depth electrodes for presurgical evaluation, during naps or full-night sleep. Babbling sounds of varying intensities around the detection threshold were presented, and high-gamma activity, as a proxy of neuronal activity, was measured in predefined regions of interest, including the superior temporal gyrus, inferior frontal cortex, insula, and posterior parietal cortex. We focused primarily on NREM sleep due to its prevalence and permeability to external stimuli.

Neural correlates of consciousness during wakefulness were defined as the difference in high-gamma activity between detected and undetected sounds. Taking advantage of the opportunity to record stereoelectroencephalography in the same individuals while awake and asleep, we are currently exploring the transferability of neural correlates of consciousness from wakefulness to sleep. Namely, we test such transfer from wakefulness to N2 to assess the presence of unreported conscious percepts other than dreaming during sleep.

Dreams are a core feature of our subjective experiences: we spend almost half of our lives asleep. Depersonalisation Disorder is a condition that makes people feel detached from themselves (depersonalisation) and the world around them (derealisation). People commonly report feeling like they are “living in a dream” and experience vague boundaries between awake and dream states. But for people living in a dream on a daily basis, how and what do they dream about? To address this question we have examined the relationship between sleep quality, dream experiences and depersonalisation-derealisation, depression and anxiety symptoms in 246 participants aged 18 to 64. We found that sleep quality was highly negatively correlated with depression and anxiety, but also with depersonalisation-derealisation. Sleep quality accounted for 31% and 21% of the variance in depression and anxiety scores, respectively, as well as for 14% of the variance in depersonalisation-derealisation scores. Multiple linear regressions controlling for sleep quality revealed that depersonalisation-derealisation symptoms were the strongest independent predictor of paramnesias (dream-reality confusions), while anxiety symptoms were the strongest independent predictor of the frequencies of nightmares that wake the sleeper. Depression was the strongest independent (negative) predictor of the pleasantness of dreams. These findings indicate that depersonalisation-derealisation, anxiety and depression each present with a unique phenomenological dream signature, indicative of distinct types of ‘self-continuity’ between waking and dream states. Our work has profound implications for therapy, as dreams are key indicators of the quality of one’s inner life.

Introduction:

While most people dream, we do not always remember these dreams or their content. This study examines how state factors predict the likelihood of reporting a dream experience and its content upon awakening.

Methods:

Thirty-six participants (M=21.8±4.1 years; 24 female; ≥3 recalls/week) completed daily dream diaries and subjective measures over six consecutive weeks, capturing sleep quality, frequency of nocturnal awakenings, presleep affect, and alcohol/caffeine intake. Using these factors as predictors, two binomial generalized linear mixed models were fit, with random intercepts for participants and day-in-study as a covariate: Model 1 compared no dream recall vs. any dream recall (with or without content), while Model 2 explored dream recall without content vs. with content.

Results:

In Model 1 (n=424), better sleep quality (OR=1.47, p=0.023) and more frequent awakenings (OR=2.54, p=0.002) increased the odds of any dream recall, while alcohol intake (OR=0.48, p=0.032), caffeine intake (OR=0.47, p=0.030), and negative affect (OR=0.85, p=0.031) lowered them. In Model 2 (n=306), caffeine intake significantly reduced the odds of recalling dream content (OR=0.53, p=0.031). Random intercepts accounted for substantial variance across participants in both models.

Conclusion:

Frequent nocturnal awakenings may facilitate dream recall by increasing opportunities for memory encoding. However, it remains unclear how the remaining factors interact with sleep and memory processes to shape dream recall. Our next step is to refine our understanding of the underlying mechanisms by incorporating objective sleep measures and mediation analysis.

Introduction:

Altered states of consciousness including coma and sleep are finely regulated by brainstem nuclei. Due to the lack of an atlas of brainstem nuclei in living humans, extrapolations from postmortem atlases are performed to localize these nuclei in in-vivo human neuroimages with reduced precision and limited understanding of arousal/sleep/allostatic-interoceptive circuits in humans.

Purpose:

To develop the Brainstem Navigator toolkit (i.e. an atlas/connectome of neuromodulatory brainstem nuclei in living humans, and a tutorial of coregistration routines to map this atlas to conventional/advanced in-vivo human MRI) and to show its proof-of-concept application to investigate arousal/sleep/allostatic-interoceptive mechanisms in humans.

Methods:

Brainstem Navigator toolkit development: Brainstem nuclei atlas: The Brainstem Imaging Lab performed semi-automatic segmentation of 68 brainstem/10 diencephalic nuclei of 7 Tesla multi-contrast (T2-weighted/diffusion) MRI (n=12,28±1years), and computed probabilistic atlas labels after coregistration to stereotactic space. Tutorial of coregistration routines/connectomes generation: We developed scripts/documentation for precise brainstem coregistration of structural/diffusion/functional MRI to stereotactic atlas space and developed functional/structural connectomes of neuromodulatory brainstem nuclei using 7 Tesla MRI (n=20,29.5±1.1years).

Results:

We generated and publicly released the Brainstem Navigator toolkit (https://www.nitrc.org/projects/brainstemnavig/, 2000+ downloads). Neuromodulatory brainstem nuclei showed strong connectivity with subcortex/frontal cortex and weak connectivity with visual cortex. Application of the Brainstem Navigator to traumatic coma (collaborator: Brian Edlow, MGH), REM-sleep-behavior-disorder (Aleksandar Videnovic, MGH) and to investigate the allostatic-interoceptive network (Lisa Barrett, Northeastern) showed alterations in specific brainstem nuclei microstructure/connectivity in line with animal studies.

Conclusions:

We developed a toolkit able to expand investigation of altered states of consciousness in living humans.