Journal of Neuroscience
September 10, 2018
Francesca Siclari, Giulio Bernardi, Jacinthe Cataldi et al.
148 citations
Dreaming during non-rapid eye movement (NREM) sleep is linked to fewer, smaller, and shallower slow waves and faster spindles, especially in central and posterior brain regions. A minority of very steep, large slow waves in frontal areas, occurring against a background of reduced slow wave activity and accompanied by high-frequency power increases (local microarousals), preceded successful dream recall. The findings suggest that the brain's ability to generate experiences during sleep is reduced when neuronal off-states are present in posterior and central regions, and that dream recall may be aided by intermittent activation of arousal systems during NREM sleep.
Sleep
May 14, 2021
Laura Sophie Imperatori, Jacinthe Cataldi, Monica Betta et al.
30 citations
Functional connectivity metrics, which describe how brain regions interact, can reveal differences across stages of sleep and wakefulness that power-based analyses alone may miss. Analyzing overnight sleep and resting-state wakefulness recordings from 24 healthy adults, the study found that combining power features with two connectivity measures—weighted Phase Lag Index (wPLI) and weighted Symbolic Mutual Information (wSMI)—improved the accuracy of classifying four vigilance stages (wakefulness, NREM-N2, NREM-N3, and REM sleep) compared to using any single feature type. Delta-band connectivity (0.5–4 Hz) was most important across all classifications, suggesting slow waves play a role in consciousness and sensory disconnection.
Clocks & Sleep
September 19, 2021
Valentina Elce, Giacomo Handjaras, Giulio Bernardi
29 citations
Dream reports—oral or written accounts given upon awakening—are a key source of insight into dreaming. Traditionally, structural information (e.g., word or sentence counts) and semantic content (scored by human coders using predefined systems) are extracted from these reports. This review argues that linguistic analysis methods—graph analysis, dictionary-based content analysis, and distributional semantics—can complement or replace classical measures. These techniques allow direct, operator-independent extraction of quantitative information, enabling fully objective and reproducible analysis of conscious experiences during sleep. They can be partially or fully automated, making them suitable for large datasets.
May 16, 2023
William Wong, Kátia C. Andrade, Thomas Andrillon et al.
21 citations
preprint
A new open-access database, DREAM, combines sleep magneto/electroencephalography (M/EEG) recordings with standardized dream reports to enable large-scale neurocognitive research on dreaming. The initial release includes 20 datasets from 561 participants and 2649 awakenings, each with at least 20 seconds of high-frequency M/EEG data and a classification of the subject's experience. Analyses demonstrate that features extracted from EEG can predict whether a person reports having had a conscious experience during both REM and NREM sleep. The database aims to overcome the limitations of small sample sizes and methodological variability in dream research, allowing new questions to be addressed at a scale unattainable by individual labs.
Nature Communications
August 13, 2025
William Wong, Rubén Herzog, Kátia Cristine Andrade et al.
10 citations
A new open database, the DREAM database, combines standardized sleep magneto/electroencephalography (M/EEG) recordings with dream reports from 505 participants across 20 datasets, totaling 2,643 awakenings. Each awakening includes at least 20 seconds of high-resolution sleep EEG (≥100 Hz, ≥2 electrodes) and a classification of the sleeper's reported experience. Analyses showed that reports of conscious experiences during sleep can be predicted from objective EEG features in both REM and NREM sleep. The database aims to overcome limitations of small sample sizes and methodological variability in dream research, enabling larger-scale investigations of the neurocognitive basis of dreaming.
March 6, 2023
Claudia Picard-Deland, Giulio Bernardi, Lisa Genzel et al.
2 citations
preprint
Memory traces formed during waking hours are spontaneously reactivated during sleep, a phenomenon first observed in the 1990s that has been proposed as the neural basis for dreaming. Recent animal and human studies, along with advances in sleep and dream engineering, show both similarities and differences between memory reactivations and dream experiences. The authors argue that memory reactivations may influence subjective experiences across various states of consciousness but are unlikely to be directly experienced as dreams. They highlight shortcomings in current research methods and propose new approaches to empirically investigate the relationship between memory reactivation and dreaming.
bioRxiv (Cold Spring Harbor Laboratory)
May 21, 2025
Valentina Elce, Giorgia Bontempi, Serena Scarpelli et al.
1 citation
preprint
Dreams are shaped by both stable individual traits and major external events. A large adult cohort provided dream and wakefulness reports alongside demographic, psychometric, cognitive, and sleep measures. Natural language processing revealed semantic features that distinguish dream from wakefulness reports, with this distinction significantly modulated by individual-specific factors. Longitudinal and cross-sample analyses showed that the COVID-19 pandemic left lasting traces on dream content. The findings highlight a dynamic interplay between personal characteristics and external events in shaping dream experiences, offering insights into the cognitive and emotional architecture of dreaming.
arXiv (Cornell University)
February 28, 2023
Lorenzo Bertolini, Valentina Elce, Adriana Michalak et al.
1 citation
Dream research usually depends on human experts manually scoring dream reports, a time-consuming process. While natural language processing tools have been explored for automatic analysis, they could not reason over a report's full context, needed extensive preprocessing, and were rarely validated against manual scoring. This work used large language models, both off-the-shelf and bespoke, to replicate manual annotation of dream reports, focusing on emotions. The off-the-shelf method performed poorly, likely due to linguistic differences across individuals. In contrast, the bespoke text classification method achieved high performance and was robust against biases. This approach may enable analysis of large dream datasets and improve reproducibility and comparability across studies.
Communications Psychology
April 28, 2026
Valentina Elce, Giorgia Bontempi, Serena Scarpelli et al.
Dreams are universal yet highly personal. By analyzing 3,366 reports of dreams and waking experiences from 207 adults collected between 2020 and 2024, along with demographic, cognitive, psychometric, and sleep measures, researchers found that dreams differ from waking thoughts: they are less self-referential and thought-centered, and more perceptual, dominated by visuo-spatial details, multiple characters, and bizarre events. Stable traits like attitude toward dreaming, mind-wandering propensity, and subjective sleep quality selectively influenced dream content. During the first 2020 COVID-19 lockdown, dreams from an independent dataset of 80 participants showed increased references to limitations and heightened emotional intensity, which gradually normalized over subsequent years. Stable individual traits and incidental experiences jointly shape dream semantics.
bioRxiv (Cold Spring Harbor Laboratory)
September 8, 2025
Adriana Michalak, Davide Marzoli, Francesco Pietrogiacomi et al.
preprint
Perceived sleep depth, a key factor in subjective sleep quality, is traditionally linked to unconsciousness and reduced brain activity. Using high-density EEG and a serial awakening paradigm during NREM sleep, researchers found that deeper sleep corresponded to lower high-to-low frequency power ratio, indicating reduced cortical activation. However, this relationship weakened when dreaming occurred, suggesting immersive conscious experiences can counteract the effect of cortical activation on perceived depth. Perceived sleep depth was lowest during states with a mere sense of presence and highest during immersive dreaming or deep unconsciousness. As the night progressed, physiological sleep pressure and subjective sleepiness declined, but perceived sleep depth increased alongside rising dream immersiveness. These findings challenge the view that deep sleep stems solely from reduced brain activity.
PLoS Biology
March 1, 2026
Adriana Michalak, Davide Marzoli, Francesco Pietrogiacomi et al.
Perceived sleep depth, a key part of subjective sleep quality, is traditionally linked to unconsciousness and reduced brain activity. Combining high-density EEG with a serial awakening paradigm during N2 sleep in healthy participants, the study found that deeper sleep was associated with a lower high-to-low frequency power ratio, indicating reduced cortical activation. However, this relationship weakened when dreaming occurred, suggesting immersive conscious experiences can counteract the impact of cortical activation on perceived depth. Perceived sleep depth was lowest during minimal awareness (a mere sense of presence) and highest during immersive dreaming or deep unconsciousness.
Communications Psychology
February 18, 2025
Valentina Elce, Damiana Bergamo, Giorgia Bontempi et al.
Dream recall varies widely between people and across nights. In 217 healthy adults aged 18 to 70, attitude toward dreaming, tendency to mind-wander, and sleep patterns were associated with whether a person reported a dream upon waking. The likelihood of recalling dream content was predicted by age and vulnerability to interference. Night-to-night changes in sleep patterns and seasonal fluctuations also influenced dream recall. These findings help explain why some people remember dreams more often than others and why recall varies within the same person.
Workshop on Computational Linguistics and Clinical Psychology
January 1, 2024
Lorenzo Bertolini, V. Elce, Adriana Michalak et al.
Dream reports are typically analyzed by trained human annotators, a time-consuming process. While earlier natural language processing tools could automate some analysis, they could not reason over full report context, required extensive preprocessing, and were rarely validated against manual scoring. This study used large language models (LLMs) to replicate manual annotation of dream reports, focusing on references to emotions. An off-the-shelf LLM method performed poorly, likely due to linguistic differences between reports from different individuals. A bespoke text classification method achieved high performance and was robust against biases. The approach may enable analysis of large dream datasets and improve reproducibility across studies.
Scientific Reports
June 20, 2019
Laura Sophie Imperatori, Monica Betta, Luca Cecchetti et al.
Two methods for measuring brain connectivity, wPLI and wSMI, detect different types of neural interactions. Using simulated EEG data, wPLI is sensitive to couplings that mix linear and nonlinear dependencies, while only wSMI detects purely nonlinear interactions. In real EEG recordings from 12 healthy adults during wakefulness and deep sleep, both methods showed different sensitivity to changes in brain connectivity across these states. The findings suggest that using both methods together provides a more complete picture of the functional basis of consciousness in both healthy and altered states.
Valentina Elce, Damiana Bergamo, Giorgia Bontempi et al.
preprint
Dream recall varies greatly between people and across nights. In a study of 204 healthy adults aged 18 to 70, attitude toward dreaming, tendency to mind-wander, and sleep patterns were associated with how likely someone was to report a dream upon waking. The probability of recalling specific dream content was predicted by age and vulnerability to interference. Dream recall also fluctuated with nightly changes in sleep patterns and showed seasonal variation. These findings help explain why dream recall differs both between individuals and within the same person over time.