Publications

A key validity concern for EEG-based consciousness measures is whether they confound conscious level with general cognitive engagement. This study tests this directly by comparing proposed measures during passive sensory attendance versus an active working memory task. Several measures showed no reliable modulation by cognitive load, supporting the case for their specificity to consciousness level rather than to arousal or task engagement more broadly.

Extending human consciousness-measure methodology to rodents, this study compares perturbation-based and spontaneous EEG metrics under three anesthetics (propofol, sevoflurane, ketamine) in rats. While signal diversity measures behaved as predicted—suppressed by propofol and sevoflurane—integrated information metrics paradoxically increased under these same agents, suggesting that different measure families may be indexing distinct and potentially dissociable aspects of neural organization during anesthesia.

This perspective article argues that the standard criteria for declaring a subject unconscious—behavioral unresponsiveness and lack of memory recall—are insufficient to establish absence of experience. Drawing on empirical evidence showing frequent dream-like experience even during deep sleep and general anesthesia, the authors advocate for moving beyond binary conscious/unconscious distinctions and propose alternative frameworks for characterizing altered states.

Sub-anaesthetic ketamine — at doses sufficient to produce noticeable subjective effects — does not impair inhibitory control performance, contrary to the intuitive expectation that intoxication impairs behaviour.

This study benchmarks computational approximations and heuristic proxies for integrated information (Φ), the central quantity in Integrated Information Theory (IIT). Testing 14 candidate measures across over 2,000 small synthetic networks, the authors show that while some approximations closely track true Φ at limited computational savings, popular heuristic measures correlate only with state-independent aspects of Φ—raising doubts about their validity for empirically testing IIT.

Tsunami survivors show greater right temporal cortical thickness 7–8 years post-trauma compared to controls, suggesting structural brain correlates of resilience to PTSD.

Within 3 weeks of a road traffic accident, survivors show heightened activation and amygdala–somatosensory connectivity in response to trauma-specific stimuli, consistent with an attentional sensory processing bias toward trauma-related material.

The DREAM database: 20 standardised EEG-sleep datasets from 505 participants and 2,643 awakenings, enabling large-scale study of the neural correlates of dreaming.

EEG complexity measures distinguish sedation from wakefulness but do not reliably separate dreaming from non-dreaming periods within propofol sedation.

A pre-registered replication attempt testing whether Lempel-Ziv complexity (LZC) of high-density EEG tracks both sleep depth and the richness of dream experience within NREM sleep. LZC tracked sleep stage as expected, but failed to distinguish dreaming from non-dreaming awakenings within NREM2, and a previously reported correlation with perceptual dream quality did not replicate—raising questions about the specificity of LZC as a marker of phenomenal richness.

This study asks whether EEG signal diversity—proposed by several theories as a correlate of conscious richness—tracks both sleep depth and the vividness of dream experience. While diversity reliably decreased with sleep stage depth, it did not differentiate dreaming from non-dreaming awakenings within NREM2; a modest correlation emerged between posterior cortex complexity and the perceptual (versus thought-like) character of dream reports.

Using EEG from patients undergoing the Wada procedure—a clinical protocol that selectively anesthetizes one cerebral hemisphere—this study tests whether proposed consciousness measures detect the expected lateralized changes. Despite clear behavioral effects, signal diversity metrics showed no reliable hemisphere-specific differences, prompting discussion of cross-hemispheric confounds that may complicate consciousness measurement in neurological patients.

Sub-anaesthetic ketamine increases spontaneous EEG signal diversity (LZc, ACE, SCE) in parallel with subjective altered-state reports, but leaves TMS-evoked complexity (PCI) unchanged — suggesting distinct neural mechanisms for conscious capacity versus conscious content.

24 hours of sleep deprivation selectively impairs top-down cognitive control (sustained attention, P300, Pe) while leaving more automatic inhibitory processes (N200, ERN, stop-signal reaction time) largely intact.

Adolescents with CFS show reduced salience network connectivity to the right posterior insula — correlated with fatigue severity — revealing disrupted interoceptive and salience processing in paediatric CFS.

A popular science article published in MedisinBloggen (2021).

Systematic comparison of virtually all time-series EEG measures of consciousness proposed in the literature, showing that most measures behave similarly across states — suggesting diminishing returns from developing new metrics.

Three popular science articles published in Speilvendt, the student journal of psychology at the University of Oslo, between 2012 and 2014.

This doctoral thesis critically examines the conceptual and empirical foundations of consciousness measurement. It investigates whether standard criteria for declaring a system unconscious are valid, whether proposed EEG-based measures are specific to conscious level rather than to confounding factors, and how results generalize across anesthesia, sleep, cognitive load, and animal models—situating the findings within broader theoretical debates in consciousness science.

This master's thesis investigates the causal role of the right inferior frontal gyrus (rIFG) in voluntary motor inhibition using a real-time fMRI neurofeedback paradigm. Participants were trained to self-regulate rIFG activity, and the study examined whether this regulation influenced stop-signal task performance—contributing to ongoing debates about whether rIFG is causally necessary for response inhibition or merely co-activated alongside it.