Lucid dreaming—a rare state where the dreamer knows they are dreaming—activates the brain in ways that are distinct from both normal dreaming and wakefulness. A new study published in The Journal of Neuroscience has mapped the neural activity underlying lucid dreams with unprecedented precision. The researchers found that lucid dreaming produces a unique pattern of brain activity that includes altered communication between brain regions, increased gamma activity, and signatures linked to self-awareness and cognitive control.
Lucid dreaming occurs when a person becomes aware that they are dreaming, sometimes even gaining control over the dream’s content. Though vivid and immersive, this state is neurologically complex and still poorly understood. Previous studies had proposed possible brain markers of lucid dreaming, but results were often inconsistent. Many used small samples and lacked standardized methods to clean up confounding signals such as those produced by eye movements during rapid eye movement (REM) sleep. The team led by Çağatay Demirel at Radboud University Medical Center sought to overcome these limitations.
“I am a PhD candidate nearing completion, and this project represents the largest chapter of my dissertation,” Demirel explained. “Lucid dreaming felt like a strange crack in reality—a moment where you could witness your mind from within and perhaps even take control, even when nothing else felt truly graspable. That paradox—being awake inside a dream—captivated me.”
“Over time, that existential curiosity evolved into a scientific pursuit. Moreover, the inconclusive results regarding electrophysiological correlates in several studies using distinct, small-sample EEG datasets highlighted the need for refinement in the field—specifically, through an EEG mega-analysis by combining various datasets.”
The researchers assembled a large dataset from multiple laboratories in the Netherlands, Germany, Brazil, and the United States, bringing together a final sample of 26 lucid dreamers who contributed a total of 43 usable sleep recordings. These included both low-density and high-density EEG recordings, with up to 128 electrodes monitoring electrical activity across the scalp. Participants were instructed to perform a distinct sequence of eye movements (left-right-left-right) once they realized they were dreaming. This standard signal allowed the researchers to pinpoint the moment lucidity began.
A major innovation in the study was the development of a multi-stage preprocessing pipeline to clean the EEG data. This was essential because both spontaneous and voluntary eye movements during REM sleep can introduce artifacts that mimic brain activity, especially in the gamma frequency band (30–45 Hz). The team implemented techniques to identify and remove these saccadic artifacts using signal processing methods that worked even on low-density EEG setups. This preprocessing ensured that the signals analyzed truly reflected neural activity and not muscle movements or other noise.
The researchers then compared brain activity during lucid REM sleep to that during regular REM sleep and relaxed wakefulness. These comparisons used both broad frequency band analyses and more advanced techniques that measured brain signal complexity and functional connectivity. They found that lucid dreaming had a distinctive neural profile. While some features overlapped with normal REM sleep, such as lower alpha power and higher delta activity compared to waking, other features set lucid dreams apart.
One key finding was a reduction in theta and beta power in certain brain regions during lucid dreams, particularly in the posterior and right temporoparietal regions of the brain. These areas are involved in attention and self-awareness, suggesting that lucid dreaming might engage neural circuits similar to those used during reflective or metacognitive thinking. At the same time, the researchers observed increased gamma activity—especially in the 30–36 Hz range—around the moment the lucid dreamer became aware. This activity was most pronounced in the precuneus and prefrontal cortex, areas linked to consciousness and internal monitoring.
Functional connectivity analyses revealed that lucid dreaming was associated with greater long-range communication between brain regions, particularly in the alpha and gamma bands. These patterns involved areas of the brain known to support sensory integration, internal attention, and memory—functions likely involved in recognizing and maintaining lucidity within the dream. Notably, alpha connectivity during lucid dreams formed a network that included the superior temporal and superior frontal gyri, suggesting coordination between auditory, sensory, and executive systems.
Signal complexity analyses also distinguished lucid dreams from other sleep states. Measures like Lempel-Ziv complexity and entropy, which quantify the unpredictability or richness of brain signals, were higher in lucid dreaming than in regular REM sleep. However, these values were still lower than in the waking state. This suggests that lucid dreaming represents an intermediate state of consciousness—one that is more organized and self-aware than typical dreaming, but still distinct from being awake.
“We didn’t approach with any specific expectations (there is no null hypothesis in this project), as this is an almost entirely exploratory study due to the merging of large datasets from different labs,” Demirel told PsyPost. “However, the findings that captivated us most were in the source-level analyses (cortical estimation), which differ from the more traditional sensor-level EEG analyses we also applied. Sensor-level EEG patterns during lucid dreaming regarding power spectral density (PSD) analyses resemble REM sleep in statistical sense. However, source-level findings revealed heightened alpha connectivity during lucid dreaming that lies between REM sleep and wakefulness (there are evidences in the literature regarding alpha connectivity changes related to psychedelics).”
The researchers also studied how brain activity changed in the moments before and after lucid dreamers signaled their awareness with eye movements. In the seconds surrounding this signal, gamma activity spiked, and large-scale increases in connectivity were seen across the cortex. These changes began just before the eye movement signal, suggesting that the brain prepares for lucidity even before the dreamer communicates it. These moments may capture the emergence of self-awareness from a non-conscious dream state.
“The gamma activation in the precuneus around the onset of lucidity eye signaling was quite a surprising finding,” Demirel said. “Considering that this activation occurs in comparison to a baseline temporally close to the onset itself, it provides potentially conclusive patterns suggesting that the brain may simulate its own reality, reflecting self-referential awareness and possibly motor awareness. This could be interpreted as a potential sensory awakening into a simulated reality.”
This study sheds light on the neural mechanisms behind lucid dreaming by addressing previous methodological challenges and using both sensor-level and source-level analyses. The findings show that lucid dreaming is not simply a hybrid of dreaming and waking, but a distinct state of consciousness with its own brain dynamics. By tracking both spectral activity and functional connectivity, the researchers provide a more complete picture of how the brain supports self-awareness within a dream.
There are still several open questions. Lucid dreaming remains difficult to induce in experimental settings, which means that researchers often rely on naturally occurring instances. Dream content also varies widely between participants and sessions, making it harder to isolate what aspects of the experience are most responsible for the observed brain patterns. The study also relied on EEG, which has limited spatial resolution and cannot definitively rule out the influence of residual artifacts—especially in high-frequency bands where eye movements are most disruptive. Future studies using methods like fMRI or intracranial recordings could help resolve these issues.
“The rarity of very high-density EEG combined with fMRI data—or the lack of magnetoencephalography (MEG) data—on lucid dreaming poses a major limitation for performing volumetric estimations of deep brain structures,” Demirel said. “While interesting patterns can be captured at the source level, we had to restrict our analyses to cortical estimations. Although we can detect the onset of lucid dreaming through eye signaling via electrooculogram (EOG), the actual experience likely begins earlier, and we still don’t know exactly when lucidity truly emerges.”
“The limitations are shaping the goals. I am working on the development of more in-depth mathematical models to decode EEG patterns and improve sensitivity to phase shifts in non-stationary state decoding. Considering that lucid dreaming is seemingly a transient brain state, methodological reconsideration is essential to distinguish it from stationary signals. This could enable a more precise segmentation of the lucid state. I also view lucid dreaming as a tool for developing methods that could ultimately help redefine the dynamics of sleep and wakefulness, which will also indirectly support research into disorders of consciousness.”
“I’m just happy that the study has finally been published after years of very draining work that, at times, felt like it would never end,” Demirel added. “I’m really excited to finally share these findings with the community.”
The study, “Electrophysiological correlates of lucid dreaming: sensor and source level signatures,” was authored by Çağatay Demirel, Jarrod Gott, Kristoffer Appel, Katharina Lüth, Christian Fischer, Cecilia Raffaelli, Britta Westner, Xinlin Wang, Zsófia Zavecz, Axel Steiger, Daniel Erlacher, Stephen LaBerge, Sérgio Mota-Rolim, Sidarta Ribeiro, Marcel Zeising, Nico Adelhöfer, and Martin Dresler.
