You Wake, but the Body Is Gone
It usually begins like this. Four in the morning. Consciousness is clearly back, but the eyelids are heavy and not a finger will move. Something — or someone — sits on your chest. A dark figure stands in the corner, approaching. You try to scream; the vocal cords are locked. The clock seems frozen, though only seconds or minutes pass.
Koreans call it gawinullim, Japanese kanashibari (金縛り, 'metal binding'), Newfoundlanders Old Hag, Mexicans 'se me subió el muerto' ('the dead one climbed on me'), Caribbean kokma. The words differ; the descriptions are uncannily consistent. The Dutch physician Isbrand van Diemerbroeck recorded an identical case in 1664, attributing it to a demon on his patient's chest.
Modern sleep medicine gives the phenomenon the colorless name sleep paralysis. In their 2015 Oxford volume Sleep Paralysis: Historical, Psychological, and Medical Perspectives, Brian Sharpless and Karl Doghramji of Penn State College pulled together 800 years of clinical and cultural records.
7.6%, 28.3%, 31.9%
Sharpless and Barber's 2011 meta-analysis in Sleep Medicine Reviews (36 studies, 36,533 participants) demonstrated that this is hardly rare: lifetime prevalence 7.6% general population, 28.3% students, 31.9% psychiatric patients. Roughly one in thirteen adults and one in four students will have it at least once.
Incidence climbs sharply where sleep is irregular and stress is high. PTSD, panic disorder, and mood disorders are linked. Interestingly, belief in the supernatural correlates only weakly with frequency — the causality runs the other direction: people don't experience paralysis because they believe in ghosts; they invoke ghosts because they have experienced paralysis.
What Happens in the Brainstem: A REM-Atonia Leak
During REM sleep, every 90–120 minutes we drift into our most vivid dreams. If our muscles obeyed the dreaming brain, we would tumble from bed or strike our partners. The brainstem prevents this with an elegant safety lock.
Neurons around the locus coeruleus in the pons and in the ventromedial medulla bombard spinal motor neurons with glycine and GABA throughout REM. The result is near-total skeletal-muscle paralysis — REM atonia. Only the diaphragm and ocular muscles remain free, enough for breathing and the rapid eye movements that name the stage.
Sleep paralysis is a decoupling. The frontoparietal network responsible for awareness has flipped to 'awake,' while the brainstem still has the atonia switch flipped to REM. When this happens at sleep onset it is hypnagogic; on waking, hypnopompic.
Meanwhile the amygdala and parietal cortex remain hyperactive, painting REM's vivid hallucinations on top of the real bedroom. The sufferer does not feel they are dreaming; they feel they are awake in bed seeing something real. That is the crux — paralysis is not a poorly remembered dream but REM leaking into waking consciousness.
Cheyne's Three Hallucinatory Clusters
J. Allan Cheyne at Waterloo statistically analyzed thousands of paralysis reports and showed that the associated hallucinations cluster into three coherent types (Consciousness and Cognition, 1999–2003). The same pattern emerged across cultures and languages, suggesting the core phenomenology is universal.
| Cluster | Dominant sensations | Neural conjecture |
|---|---|---|
| Intruder | A presence in the room, footsteps, breathing, shadow at edge of vision | Amygdala hyperactivation + false-positive threat detection |
| Incubus | Crushing chest pressure, suffocation, choking, physical contact | Irregular REM respiration + failed voluntary breath attempts |
| Vestibular-Motor | Floating, falling, spinning, out-of-body, slipping out of bed | Vestibular input vs. motor command mismatch (cerebellum, parietal) |
Intruder and Incubus frequently co-occur. The classic Korean and Japanese 'a ghost sat on my chest' description maps precisely onto this combination. The Vestibular-Motor cluster overlaps with reports of out-of-body and near-death experiences and has drawn interest from consciousness researchers.
Triggers
Epidemiology has converged on a reliable risk list.
- Sleep deprivation and irregularity: fewer than four hours, post-shift work, exam weeks — elevated REM pressure increases leak probability.
- Supine posture: in several samples 50–60% of sufferers were on their backs. Partial airway compromise may amplify the Incubus cluster.
- Jet lag and shift work: melatonin–arousal desynchrony.
- Chronic stress, PTSD, panic disorder: elevated noradrenergic tone fragments REM.
- Caffeine, alcohol, abrupt withdrawal of REM-suppressing drugs (SSRIs, etc.).
- Genetics: monozygotic-twin heritability about 53% (Denis 2015).
Ghosts as Explanatory Models
Korean folk tradition has long read sleep paralysis as the work of ghosts, ancestral spirits, or wandering jabgwi. The word gawi may come from a verb meaning 'to press as with shears,' and traditional shamanic prescriptions sometimes included rituals or talismans.
This is not worth ridiculing. As Cheyne stressed, paralysis is a strikingly specific phenomenon — threatening presence + immobility + chest pressure within waking awareness. In the absence of brainstem neuroscience, the most parsimonious model on offer is 'a supernatural being is here.' Korean gwisin, Japanese makuragaeshi, Newfoundland's Old Hag, and the Islamic jinn are all culturally local explanatory models for the same neural event.
The interesting variation is in content. In Egyptian samples with strong belief in jinn, sufferers fear lasting supernatural harm and show more PTSD-like sequelae (Jalal & Hinton 2013). American student samples normalize the event faster and have fewer aftereffects. Culture doesn't manufacture the phenomenon, but it governs the aftermath.
Treatment: Sleep, Breathe, Submit — Jalal's MR Therapy
For mild cases, sleep hygiene alone substantially reduces frequency: seven to eight hours of regular sleep, less caffeine and alcohol, side-sleeping, dimmer screens before bed. Simply knowing that exam weeks and long-haul flights raise the odds reframes events from 'ghost' to 'brainstem clocked in late.'
Cambridge's Baland Jalal proposed Meditation-Relaxation Therapy (MR Therapy) in Frontiers in Psychology (2016): a four-step protocol applied during the episode. (1) Reappraisal: remind yourself, 'This is sleep paralysis; it is not dangerous.' (2) Psychological-emotional distancing: treat the hallucinations as external events, not threats to act on. (3) Inward attention meditation: focus on the breath or a single body region such as the toes. (4) Muscle relaxation: stop resisting and deliberately surrender the body. A 2020 pilot in Italian narcolepsy patients showed roughly 50% reduction in episode frequency after eight weeks (Jalal 2020).
For frequent or disabling cases — more than once weekly, or with PTSD, panic, or suspected narcolepsy — see a sleep specialist. CBT-I reduces REM fragmentation. Polysomnography and the Multiple Sleep Latency Test rule out narcolepsy. Low-dose SSRIs or tricyclics that suppress REM are occasionally prescribed.
Conclusion: Rename the Fear, and It Shrinks
Sleep paralysis is the narrowest meeting place of the supernatural and the neuroscientific. Consciousness is awake, the body is asleep, and through the small gap between them REM's shadows leak. Gawinullim, kanashibari, and the Old Hag are all cultural first-aid kits humanity has stitched together over 800 years for the same brainstem mismatch.
Next time, at four in the morning, you feel a weight on your chest, remember two things. One: you are not in danger. Two: it will end. Focus on the breath, stop trying to move a toe, and let go. The drowsy neurons of the locus coeruleus are about to clock in.
