Hypnagogia: what does science say about the drowsy state?
Hypnagogia, or the hypnagogic state, is the drowsy passage from waking into sleep, when images and sounds can arrive on their own, and sleep researchers have been measuring it with electrodes for decades. They can show which brain rhythms fade and which rise, and one 2021 study found that a minute of it helped people spot a hidden rule in a number task. No study tests what Neville Goddard claimed about it.
Schematic trace drawn by us to show the shapes researchers score. It is not a recording.
Frequency
The five bands, in cycles per second
EEG reads tiny voltage swings at the scalp, and a band is a range of how fast they swing. Theta at 6 Hz means six rises and falls a second. Labs do not agree on the edges, so the chart shows both the usual range and everything used across 184 studies reviewed in 2019. Pick a band.
The band ranges as a table
| Band | Usual range | Lowest start used | Highest end used |
|---|
Theta is the band everyone quotes for this state, usually as 4 to 8 Hz, and the slope of the change is well measured. A 2014 paper by Michael Prerau and colleagues put it this way: “The most obvious changes to the EEG during the SOP are a progressive decrease in alpha (8-12 Hz) power, as well as a progressive increase in slow (<1 Hz), delta (.5-5 Hz), and theta (5-8 Hz) power.” SOP is the sleep onset period. Note that their theta starts at 5, and the 2019 review warns of “a great deal of variability and confusion” in where any band begins.
This site runs on the idea that everything has a frequency, so it is worth being exact about what these numbers are. They count swings of electrical voltage at the scalp, summed over huge numbers of neurons, and they sit in the single and double digits per second. Atkinson’s 1906 claim in Thought Vibration put thought far above light: “Light and heat are manifested by vibrations of a far lower intensity than those of Thought.” EEG cannot settle that either way, because it reads a voltage rhythm and says nothing about what a thought is made of.
Sleep onset
Nine steps down into sleep
Standard scoring cuts the night into 30-second blocks and calls the drowsy part stage 1, or N1. In 1994 Tadao Hori’s team split the same stretch into nine EEG stages, “leaving the line between wake and sleep blurry,” as the Dormio authors put it. Step down the stairs.
What people experience here has been sampled by waking them. The review by Romain Ghibellini and Beat Meier gathers forty publications and lands on one difference from dreams that matters for anyone reading Neville. In dreams you act. In hypnagogia, they write, we would expect people to experience it “much more so as a passive spectator of internal events.” As sleep nears, reports of thinking fall away and reports of movement imagery grow. And content follows the day. In Robert Stickgold’s 2000 Tetris study, as later summarized, 17 of 27 players reported the game at sleep onset, including amnesic patients who could not remember playing it.
Science Advances, December 2021
The study that made N1 famous
Célia Lacaux and colleagues at the Paris Brain Institute gave 103 people a number task with a hidden shortcut. The 16 who spotted it early were dropped. Everyone else got a 20-minute break in a dark room with EEG running, eyes closed, holding a bottle in one hand, then went back to the task.
The people who touched N1 and went no deeper were 2.7 times likelier to find the rule than those who stayed awake. Those who slid on into stage 2 lost the gain entirely. The team’s summary: “spending at least 15 s in N1 during a resting period tripled the chance to discover the hidden rule.”
| Group | Found it |
|---|---|
| Stayed awake | 15 of 49, 30.6% |
| Reached N1 only | 20 of 24, 83.3% |
| Reached N2 | 2 of 14, 14.3% |
What the paper does not say
- The people were picked for it. Researchers selected those who fall asleep easily and asked them to sleep about 30 percent less the night before.
- The N1 group was 24 people on one kind of problem. The authors write that “more research is needed to determine whether our findings are generalizable to other types of creative tasks.”
- The insight was late. It came about 94 trials after the break, roughly half an hour on.
- The dream imagery itself did nothing measurable. “neither hypnagogic experiences in general (χ2 = 0.14, P = 0.71) nor task-related hypnagogia (χ2 = 0.05, P = 0.83) had an effect on insight.”
- Theta was not the marker. The frequencies that predicted insight were “the delta (3.2 to 4.4 Hz) and alpha (9 to 9.8 Hz) bands,” with moderate alpha best and high delta worst.
Two old tricks
Edison’s spheres and Dalí’s key
The bottle in the Paris study is a tribute. The authors say the procedure was “inspired by the famous inventor Thomas Edison, who allegedly napped while holding spheres in his hands.” Allegedly is their word, and they cite nothing for it. The 2023 Dormio paper files both men under “Anecdotal reports.” We looked for a document from Edison himself and did not find one, so treat the story as a story.
Dalí is better attested because he wrote his method down. In 50 Secrets of Magic Craftsmanship (1948) he seats you in a bony armchair with a heavy key held between thumb and forefinger over “a plate upside down on the floor,” aiming at what he calls “sleeping without sleeping.” The clatter wakes you. He says the slumber lasts “less than a quarter of a second long,” a figure nobody has checked. We read his passage as transcribed by the sleep researcher Michelle Carr, and did not see the book.
The old trick did get a real test. Across 63 drops in the Paris experiment, 41 percent came after scored N1, and 78 percent once very short microsleeps were counted. Delta power climbed for about 24 seconds before the bottle fell, and people woke with hypnagogic reports far more often after a drop. The authors call it a good marker of sleep onset, and say a run without the bottle would be needed to know whether it fosters insight.
MIT Media Lab
Dormio: putting a word into the drowsy state
Adam Haar Horowitz and colleagues built a hand-worn sensor that watches finger tone, heart rate and skin conductance. It spots sleep onset, wakes you one to five minutes later, asks what was going through your mind and sends you back with a voice prompt: “Remember to think of a tree.” Serial reports from one sleeper ran from “Trees, many different kinds” to “Trees splitting into infinite pieces.”
| Condition | Reports | With the tree |
|---|---|---|
| Sleep onset, prompted | 67 | 45, 67% |
| Sleep onset, no prompt | 69 | 2, 3% |
| Awake, prompted | 56 | 29, 52% |
| Awake, no prompt | 48 | 0, 0% |
Prompted sleepers and prompted wakers did not differ significantly (p = 0.129), so the prompt worked in both. The 2023 review adds a caution that “it remains questionable whether reports from subjects can be trusted due to influence from demand characteristics.” People told to think of a tree may say tree.
A follow-up in Scientific Reports in 2023 asked whether it matters. Forty-nine people did a 45-minute session in one of four groups, then took three creativity tasks. The group prompted at sleep onset beat all three others on a combined creativity index (0.79 against −0.07, −0.32 and −0.47). The authors call it, to their knowledge, “the first controlled experiment investigating a direct role of incubating dream content in the enhancement of creative performance.” Each group had about a dozen people, the only theme tested was tree, the tasks were about trees, and the device stood in for EEG, “due to the lack of polysomnographic confirmation of sleep staging within the study.”
Where the evidence stops
What is measured, and what is not
| Statement | What the record shows | Status |
|---|---|---|
| Alpha fades and slower rhythms rise as people fall asleep | Tracked in the lab with EEG, behavior and muscle tone (Prerau 2014 and many before it). | Measured |
| A short spell of N1 helps insight | Found once on a number task with selected easy sleepers. A smaller MIT study points the same way on creativity tasks. | Measured, narrow |
| A prompt at sleep onset can steer what people report | Yes in Dormio, with a warning about demand effects, and a prompt did nearly as well awake. | Measured, narrow |
| Hypnagogia is a theta state | Theta rises overall, but alpha and delta predicted insight in Lacaux, and alpha and theta power fell just before visual reports in two studies. | Too simple |
| Edison napped holding steel balls | Called alleged and anecdotal by the scientists who cite it. | Anecdote |
| A scene imagined in this state changes later events in the world | This is Neville’s claim. No study tests it. A PubMed search on his name returns 23 records, every one a clash of author names in medical papers. | Not tested |
The labs and Neville are also looking at different things. The research is mostly about imagery that shows up uninvited, in people who have been told to rest, and the review finds them watching it as spectators. Neville asks for the reverse: a scene you choose, held on purpose, with you acting inside it. His state akin to sleep sounds like the same drowsy stretch, and we found no study that put a practitioner on an EEG to check. Dormio’s repeated prompt is the nearest thing to his method, and what it measures is words in a report. Whether the law of assumption does what he says is outside anything measured here.
Questions
Hypnagogia
What is the hypnagogic state?
It is the transition from waking into sleep, and the images, sounds and bodily feelings that can arise on their own during it. The word was coined by Alfred Maury in 1848 from Greek words for sleep and leader. Researchers place it in the first stage of sleep.
What is the difference between hypnagogic and hypnopompic?
Timing. Hypnagogic refers to the way into sleep and hypnopompic to the way out of it, on waking. Frederic Myers coined the second word. The Ghibellini and Meier review uses hypnagogic states as the term for both.
Are hypnagogic hallucinations normal?
The 2023 review says these experiences fall in the normal range of perception and notes that the DSM-5 mentions them as normal. Its authors keep the word hallucination for cases where they cause a complaint or discomfort. This page reports the research and is not medical advice.
Is the hypnagogic state a theta state?
That is too simple. Theta power rises as people fall asleep, and one study scored microsleeps by theta with alpha loss. But in Lacaux 2021 the frequencies that predicted insight were in the delta and alpha bands, and two studies saw alpha and theta power fall just before visual imagery was reported.
How common is hypnagogia?
Estimates swing widely because studies define it differently. A UK survey of 4,972 people found 37 percent reporting it at least twice a week. A survey of 13,057 people in the UK, Germany and Italy found about 25 percent. Older reviews put it at 72 to 77 percent.
Can you stay in hypnagogia on purpose?
Researchers have tried hypnosis, sensory deprivation by ganzfeld, and the Dormio device. The 2023 review calls induction not sufficiently tested. Lacaux 2021 found that holding an object, which drops as you doze, kept most people from going deeper before the first drop.
Sources
Papers read in full
- Lacaux C, Andrillon T, Bastoul C, Idir Y, Fonteix-Galet A, Arnulf I, Oudiette D. Sleep onset is a creative sweet spot. Science Advances, 2021, 7(50).
- Haar Horowitz A, Cunningham TJ, Maes P, Stickgold R. Dormio: A targeted dream incubation device. Consciousness and Cognition, 2020.
- Horowitz AH, Esfahany K, Gálvez TV, Maes P, Stickgold R. Targeted dream incubation at sleep onset increases post-sleep creative performance. Scientific Reports, 2023, 13.
- Ghibellini R, Meier B. The hypnagogic state: A brief update. Journal of Sleep Research, 2023, 32(1). Our source for Maury, Myers, the Hori stage table, prevalence figures and the studies by Hayashi, Germain and Nielsen, Noreika and Speth, which we know through this review.
- Prerau MJ and colleagues. Tracking the sleep onset process. PLoS Computational Biology, 2014.
- Newson JJ, Thiagarajan TC. EEG Frequency Bands in Psychiatric Disorders. Frontiers in Human Neuroscience, 2019. Used only for its table of band definitions across 184 studies.
- Second hand, marked as such: Stickgold’s 2000 Tetris study as summarized by Haar Horowitz 2020, and Dalí’s 1948 text as quoted in Michelle Carr, How to Dream Like Salvador Dali, Psychology Today, 2015.
- William Walker Atkinson, Thought Vibration, 1906. Nothing here is medical advice or a health claim.