Is Sleep Paralysis a Dream? What's Actually Happening When You Can't Move

I've had sleep paralysis three times. Twice it was only the paralysis: awake, aware, lying in my own room, unable to move. That on its own is unpleasant enough. The third time there was someone in the room.

I was on my back. I surfaced, and my bedroom was exactly my bedroom, the real one, down to the light. Someone I'd lost years earlier was over me with their hands at my throat. I couldn't move and I couldn't get a breath in properly, and at no point did any of it register as a dream. It registered as something happening in the room I was lying in.

That's the thing people get wrong about sleep paralysis from both directions. It isn't an attack, and it also isn't "just a bad dream," and being told the second one is almost as useless as believing the first.

The paralysis is real. The figure is not. Both of those facts have a fairly precise explanation, and knowing them is the difference between an episode you can name and an episode that follows you around for a week.

Your body has a nightly off switch, and it overstayed

Every night during REM sleep, your brain actively paralyzes your body. This isn't drowsiness. It's a deliberate signal, and researchers have traced the chemistry: GABA and glycine released together onto the motor neurons that would otherwise drive your muscles, shutting them down for the duration. The reason is obvious once you think about it. Your motor cortex is busy issuing instructions during a dream, and without that block you would carry them out.

Sleep paralysis is what happens when the block doesn't lift on schedule. Your cortex comes back online, awareness returns, your eyes work, you can see the actual room. The motor inhibition is still running. Sleep researchers call this a REM-wake dissociation, a hybrid state where you get the consciousness of waking and the muscle state of REM at the same time.

So when someone says they were awake and couldn't move, they're describing the mechanism accurately. They were awake. They genuinely could not move.

The same circuit that makes a sleeping dog paddle, running backwards

There's a neat symmetry here with something I wrote about a few weeks ago in the post on animal dreaming. In 1965 Michel Jouvet lesioned the brainstem region that produces this paralysis in cats, and the cats got up mid-REM and stalked things that weren't there. Same circuit, failing in the opposite direction: the paralysis switched off while the dream kept running, so the body acted the dream out.

Sleep paralysis is that fault run backwards. The dream state is ending, consciousness has returned, and the paralysis is the thing that won't switch off. One circuit, two failure modes, and between them they explain both a twitching dog and a demon on your chest.

It is much more common than people admit to each other

Because it's terrifying and sounds unhinged to describe out loud, most people assume they're unusual for having had it. They aren't.

A 2011 systematic review by Brian Sharpless and Jacques Barber pooled 35 studies covering 36,533 people and found that 7.6% of the general population had experienced at least one episode. Among students it was 28.3%, and among psychiatric patients 31.9%. For people with panic disorder it ran near 35%.

That student number is worth sitting with. Sleep paralysis tracks sleep debt and chaotic schedules, and an undergraduate population is essentially a natural experiment in both.

The hallucinations come in three shapes, over and over

This is the part I find genuinely strange. If these were ordinary dreams, their content should be as varied as dreams are. It isn't. J. Allan Cheyne's group at Waterloo ran factor analyses on large samples of sleep paralysis reports and kept recovering the same three clusters.

The Intruder is the sensed presence: someone in the room, often visual, often with footsteps or breathing attached. Cheyne's work on where these figures are placed in space found it reliably sits close by, usually within about a meter of the bed. The Incubus is the pressure, the weight on the chest, the suffocation, sometimes pain. The third, Vestibular-Motor, is the odd one out: floating, flying, leaving your body. That cluster is the one people describe as blissful rather than frightening, which is a strange thing for the same condition to produce.

Mine was the first two at once, which is the most commonly reported pairing there is: a figure directly over me, and the sensation of being throttled. At the time it felt like the most specific and personal thing that had ever happened to me. It turns out to be close to the median report.

The stability of this is old. Henry Fuseli painted The Nightmare in 1781: a sleeping woman with a squat demon crouched on her chest. Two centuries before anyone factor-analyzed anything, that's the Incubus cluster, painted from life or from report, hitting the same note.

Why it's that particular face

The clusters are predictable. Who shows up in them is not, and this is the part the research is thinnest on.

Something has to supply the figure, and it clearly isn't drawing at random. The bereavement literature is a useful neighbor here: sensing the presence of someone who has died is extremely common and not a sign of anything wrong, reported by a large share of grieving people, with sensed presence the most frequent form and visual encounters close behind. Researchers have only recently started looking at where that intersects with sleep paralysis specifically. A 2024 study by Cybele Blood and Joanne Cacciatore on traumatic grief and sleep paralysis found participants whose episodes began after a death, which is the first serious work I've seen connecting the two.

So I can tell you with confidence what the mechanism was doing to my body. I can't tell you why it reached for that particular person, out of everyone it could have picked. The honest answer is that nobody can yet, and I'd be suspicious of anyone who says otherwise.

The monster has a local accent

What varies isn't the shape of the experience. It's what people think it was.

In Newfoundland it's the Old Hag, and the folklorist David Hufford found the belief so embedded that locals used it as a verb: I was hagged last night. In Japan it's kanashibari, literally bound in metal. Baland Jalal and colleagues compared Egypt and Denmark and found 44% of Egyptians had experienced sleep paralysis against 25% of Danes, with 48% of the Egyptian sample attributing it to jinn.

Then the finding that actually matters. Half the Egyptian experiencers feared they might die from an episode, against 17% in Denmark. And within the Egyptian sample, believing the cause was supernatural was associated with both more fear and longer episodes of immobility.

Read that carefully, because it's correlational and I don't want to oversell it. But the direction is worth knowing: what you believe is happening to you appears to be bound up with how long it lasts and how badly it goes.

What you can actually do about it

The reliable levers are unglamorous. Episodes are strongly associated with sleeping on your back, with people reporting the supine position three to four times more often during an episode than at normal sleep onset. Sleep deprivation and irregular schedules are the other two consistent associations.

I was on my back for mine, which I only registered as meaningful years later. I don't remember whether I'd slept badly that particular night, but I slept badly most nights in that stretch of my life, so the honest version is that I was carrying both risk factors and had no idea either was a risk factor. None of this is a cure. It's three variables you control, which is three more than it feels like you have at the time.

The other lever is knowing the mechanism while it's happening. An episode that you can name as delayed REM atonia is a different experience from one you can only name as an attack, and it ends the same way regardless: it breaks on its own, usually within a couple of minutes.

If you keep a dream journal, this is worth logging properly rather than skipping because it wasn't technically a dream. Note the position you woke in, roughly how much sleep you'd had, what time it hit, and which of the three shapes it took. A handful of entries will usually show you your own pattern, and most people find it's the same two or three conditions every time.

There's something worth noticing in the fact that thousands of years of people, across cultures with no contact with each other, produced nearly identical monsters. That consistency isn't evidence of anything supernatural. It's evidence that the same piece of human hardware is running the same glitch, in every bedroom it has ever happened in.