THE QUESTION LIBRARY EXTRAORDINARY CLAIMS

Are ghosts real? What does the evidence say?

Ghostly experiences are common and deeply felt; here is what careful studies show about where they come from, and what is still open.

Not supportedGod Plays Dice Editorial · AI-assisted and independently fact-checked · corrections welcomeChecked Sep 26, 202610 min read

Henry Fuseli’s 1781 painting The Nightmare shows a creature crouched on a sleeper. In sleep paralysis, people can sense an intruder nearby and feel pressure on the chest.Henry Fuseli, The Nightmare (1781), Detroit Institute of Arts, via Wikimedia Commons · Public domainImage source ↗

THE SHORT ANSWER

Not as spirits of the dead, as far as the evidence shows. The experiences, though, are real and common: in a 2009 Pew survey, 18% of U.S. adults said they had seen or been in the presence of a ghost. Studies link such moments to sleep paralysis, suggestion, features of a place such as its lighting and, in at least one well-documented case, carbon monoxide. The studies so far have not found a spirit at work.

In a 2009 Pew survey, 18% of U.S. adults said they had seen or been in the presence of a ghost, up from 9% in 1996.
About 7.6% of people have had sleep paralysis, which can bring a sense of an intruder and a weight on the chest.
At two famous “haunted” sites, odd feelings clustered in some spots even for visitors who didn’t know their stories.
In a 1921 medical report, a family’s “haunted” house had a broken furnace leaking carbon monoxide.
Whether sound too low to hear, called infrasound, adds to eerie feelings is still being tested.

WHERE THE EVIDENCE STANDS

How sure are we?

EstablishedMany people have experiences they describe as ghosts or as a visit from someone who has died.Pew: 18% of U.S. adults in 2009 said they had been in the presence of a ghost; 53% in 2023 said a dead relative had visited them in a dream or other form.
EstablishedOrdinary causes such as sleep paralysis, suggestion, settings and carbon monoxide can produce ghost-like experiences.Sleep paralysis surveys (Cheyne 1999; Sharpless and Barber 2011), studies at Hampton Court and Edinburgh (2003), the Haunt room (2009), a lab-made presence (2014) and the 1921 carbon monoxide case.
Open questionInfrasound, sound too low to hear, helps cause eerie feelings in “haunted” places.A 2026 study of 36 people linked it to irritability and higher cortisol; the 2009 Haunt study found no link to odd sensations; the famous 1998 case was never directly measured.
Not supportedGhost sightings show that spirits of the dead are present.Controlled studies at haunted sites and in a lab-built room point to settings, bodies and expectations; none has shown a spirit at work.
Beyond scienceSome part of a person continues after death.Central to many faiths; science has no accepted way to test it. See What Happens When We Die?

What the labels mean

WHAT WE DON’T KNOW YET

Nobody fully knows why some people have these experiences often and others never do. Whether infrasound plays a real part is still being tested, and small studies point different ways. A haunting that nobody measured at the time can’t be checked now.

WHAT WOULD CHANGE THIS ANSWER

A haunting that separate teams could record with instruments, again and again, with no ordinary cause left, would change this answer. So would a “ghost” giving true facts that no one present could have known, in a test that rules out leaks and lucky guesses.

Maybe it happened in an old house: footsteps overhead when no one was upstairs. Maybe you woke up unable to move and felt someone standing by the bed. Maybe, weeks after a loss, you felt a loved one close by. Moments like these are common. They can stay with a person for life. So it makes sense to ask: are ghosts real?

Here is the core picture. The experiences are real. People truly do hear, see and feel these things. In a 2009 Pew survey, 18% of U.S. adults said they had seen or been in the presence of a ghost. In 2023, 53% told Pew that a dead relative had visited them in a dream or some other form. Having such a moment does not make anyone foolish or dishonest.

What the evidence does not support is the usual explanation: that spirits of the dead are present. When researchers study these moments, they keep finding ordinary causes. Sleep paralysis is a short spell between sleep and waking when you can’t move. It can bring a sense of an intruder and a weight on the chest. Suggestion, meaning what we are told or expect, shapes what we notice and remember. Some spots in old buildings draw more odd reports than others, even from visitors who don’t know their stories. And in a 1921 medical report, a family’s “haunted” house turned out to have a broken furnace. It was leaking carbon monoxide, a poison gas you can’t see or smell.

Healthy brains can produce these moments too. Our brains are quick to see faces, even in plain objects. In one lab, a robot made healthy people feel that someone was near them.

So the fair answer has two halves. The experiences are real and worth taking seriously. The claim that they show spirits of the dead is not supported by the evidence so far. Whether any part of us goes on after we die is a separate question. Science has no accepted way to test it.

THE LONG ANSWER

What do people actually observe?

Researchers who study hauntings list apparitions, sudden changes in temperature and a strong sense that someone is present.

In an August 2009 Pew survey of 4,013 U.S. adults, 18% said they had seen or been in the presence of a ghost. Pew noted that this share had doubled since 1996, when it was 9%. In a 2023 Pew survey of 5,079 adults, 34% said they had “felt the presence” of a dead relative in the past 12 months.

One of the most detailed accounts sits in a medical journal. In 1921, the eye doctor W. H. Wilmer published a mother’s written record of the winter her family spent in a rented house, starting in November 1912. She heard footsteps in empty rooms. Her husband felt someone watching him from behind and woke to doorbells that had not rung. She saw a dark-haired woman in black walk toward her, then realized it was her own reflection in a mirror. The children’s nurse told her, “This house is haunted.” The nurse had once woken to see a man and a woman at the foot of her bed while she was unable to move.

What do people infer from it?

The step from “I experienced this” to “a spirit caused it” is an inference, a conclusion drawn from the experience. It is a natural one. Tales of ghosts have been reported in virtually all cultures, as one neuroscience team noted.

Nor is it always reached lightly. In Wilmer’s account, the mother first laughed at the idea. The nurse answered that she would have said the same thing three months earlier, but now she was convinced. The family then learned that the previous occupants had reported the same kinds of experiences.

The Society for Psychical Research formed in London in 1882 to study such claims scientifically. Today it says it holds no official view on whether they are real.

Does the evidence support that inference?

Researchers have tested ordinary causes one by one.

Sleep paralysis

Sleep paralysis is a brief spell, while falling asleep or waking up, when the mind is awake but the body can’t move. A 2011 review of 35 studies, covering 36,533 people, found that about 7.6% of the general population has had at least one episode. Vivid, dream-like images often come with it, and an intruder in the room is a common theme. A University of Waterloo team found these experiences can include a sensed presence, fear, sounds or shapes, and chest pressure, and that such accounts are strikingly similar across time and cultures. The nurse in Wilmer’s report described something close to this.

Suggestion

What we expect shapes what we notice. In a 2003 study, volunteers sat through a fake séance in which an actor said a table was rising. It never moved. Afterward, about a third said it had. Believers in the paranormal were more likely to say so, and a second test found this happened only when the suggestion fit their belief.

A London team then tried to build a haunted room. In the 2009 “Haunt” study, 79 people each spent 50 minutes in a chamber with infrasound (sound usually too low to hear), complex electromagnetic fields, both, or neither. They had been told they might feel odd sensations. Many did, but the number did not depend on which condition they got. The simplest explanation, the researchers said, was suggestibility.

The setting

Nearly 900 members of the public visited Hampton Court Palace in England or the South Bridge Vaults in Edinburgh, Scotland, and noted anything unusual. They reported more odd experiences in areas already known as haunted, whether or not they knew the reputation. Those spots also differed in ordinary ways, such as lighting, room size and shifts in the local magnetic field. In the vaults, the visitors themselves rated only 5 of 150 experiences as definitely or probably caused by a ghost. The authors concluded that people were responding, perhaps without noticing, to normal features of their surroundings.

A presence made in the lab

In 2014, researchers in Switzerland created the feeling that someone was nearby in healthy volunteers, using a robot that scrambled the link between their movements and what they felt. The team concluded that the feeling comes from the brain misreading signals from the person’s own body.

Faces in things

In a 2022 NIH study of more than 3,800 adults, people readily saw an emotion, an age and a gender in faces that appear in everyday objects. Seeing faces in objects is called pareidolia. Keep it in mind when a face appears in a shadow or a photo.

Carbon monoxide

The house in Wilmer’s report had a hidden cause. An expert sent to inspect it found the old furnace burning badly and pouring carbon monoxide into the rooms instead of up the chimney. The family had headaches and tiredness, and the children grew pale and listless. The CDC lists headache, dizziness, weakness and confusion among the most common symptoms of carbon monoxide poisoning. It says more than 400 Americans die each year from accidental poisoning not linked to fires.

Infrasound

Infrasound is sound below about 20 hertz, usually too low for people to hear. In 1998, V. Tandy and T. R. Lawrence suggested that infrasound near 19 hertz from a faulty extractor fan was behind odd experiences in one case. According to a Society for Psychical Research summary, that frequency was estimated, not directly measured. In 2000, Tandy measured infrasound close to 19 hertz in a reputedly haunted cellar in Coventry, England. Since then the evidence has been mixed. The 2009 Haunt study found no link between infrasound and odd sensations. A small 2026 study of 36 people found that infrasound near 18 hertz, which they could not detect, was linked to more irritability and higher levels of the stress hormone cortisol. It measured mood, not ghostly experiences.

What is the strongest objection, and what do people get wrong?

The strongest objection: these experiences feel real, and they come from large numbers of “seemingly trustworthy witnesses,” as the Hampton Court and Edinburgh researchers put it. Sometimes several people in one house report similar things.

Honesty is not in question. The studies above take the reports seriously; that is how they found where experiences cluster. Shared experiences can also have shared causes. In Wilmer’s house, the same bad air reached everyone.

Grief deserves special care. In a 1971 study in mid-Wales, almost half of 293 widowed people reported hallucinations or illusions of their late husband or wife, often for many years. Few had told anyone, even close family. The researcher, W. Dewi Rees, considered these experiences normal and helpful. If you have sensed someone you lost, you are far from alone. An ordinary explanation does not make that moment less meaningful.

Two mistakes show up on both sides. One is treating “unexplained” as “spirit.” A report with no measurements often can’t be explained at all, and that is different from being explained by a ghost. The other is treating an ordinary cause as an insult. Sleep paralysis alone has touched about 1 in 13 people.

What is still unresolved?

Nobody fully knows why some people have these experiences often and others never do. The Haunt study found one clue: people who reported more odd sensations also scored higher on a questionnaire about unusual mental experiences. Whether infrasound plays a real part is still being tested. And a haunting that nobody measured at the time can’t be checked now.

Whether any part of a person continues after death is a separate question. It is central to many faiths, and science has no accepted way to test it. The verdict on this page is narrower: ghost experiences, as studied so far, do not show that the dead are present. People of faith and skeptics can agree on a lot here. The experiences are real, they deserve respect, and carbon monoxide should be ruled out first. For more, see What Happens When We Die? and What Do Near-Death Experiences Tell Us?

How can you check it yourself?

Write it down right away. Note the time, the place, whether you had just woken up, and who else was there.
Look for patterns. Does it happen only at night, on waking, in one room, or when a furnace or fan is running?
Check the air. The CDC advises carbon monoxide detectors near every sleeping area. If one sounds, the CDC says to leave your home right away and call 911. If you feel dizzy or sick and suspect carbon monoxide, get medical help quickly.
Ask a fresh visitor. Does someone who doesn’t know the story notice the same thing in the same spot?
Test a message. If a “ghost” seems to share information, write it down before anyone checks it. Could it have been known, guessed or looked up?

The door stays open: if a haunting ever passed careful, repeated tests, that would be worth knowing.

Deep dive (optional)

Optional: how would a fair ghost test work? The best studies of hauntings share a few design features. You can use them to judge any new claim.

A comparison condition. The Haunt study exposed people to infrasound, electromagnetic fields, both, or neither. Without the “neither” group, the many odd sensations could have been blamed on the equipment.
Checking what people already know. The Hampton Court and Edinburgh studies recorded what visitors knew about each area beforehand. That let the team separate a room’s reputation from the room itself.
Keeping people unaware. In the 2026 infrasound study, people could not tell above chance whether the infrasound was on, and the team reported no effect of expectation on the results.
Measuring the room, not just the person. The Edinburgh team measured light levels, room size, air movement, temperature and magnetic fields in each vault.
Deciding in advance. A strong test states beforehand what result would count as a hit, so a near miss can’t be counted later.
Independent repeats. One striking result is a starting point. The same result from separate teams is evidence.

THREE THINGS TO REMEMBER

Ghost experiences are common, and they are real to the people who have them.
Studies link them to sleep paralysis, suggestion, settings and, at times, carbon monoxide.
The evidence does not support spirits of the dead; whether anything of us survives death is beyond science.

WORDS WORTH KNOWING

Sleep paralysis: A brief spell, while falling asleep or waking, when the mind is awake but the body can’t move; it can bring a sense of presence, sounds, shapes or chest pressure.
Suggestion: The way hints, stories or expectations shape what people notice, feel and remember.
Pareidolia: Seeing a face or figure in something that has none, like a pattern in wood or a shadow.
Infrasound: Sound below about 20 hertz (vibrations per second), usually too low for people to hear.
Carbon monoxide: A colorless, odorless poison gas made by fuel-burning devices such as faulty furnaces; it can cause headache, dizziness and confusion.
Inference: A conclusion drawn from what was observed, as opposed to the observation itself.

Sources & further reading

Many Americans Mix Multiple Faiths (section: Ghosts, Fortunetellers and Communicating With the Dead) ↗Survey of 4,013 U.S. adults, Aug. 11-27, 2009: 18% say they have been in the presence of a ghost (9% in 1996); 29% have felt in touch with someone who died.
Interacting with dead relatives: What Americans say they have experienced ↗Survey of 5,079 U.S. adults, March 27-April 2, 2023: 53% ever visited by a dead relative in a dream or other form; 34% felt a dead relative's presence in the past year.
An investigation into alleged ‘hauntings’ ↗Hampton Court (678 participants) and Edinburgh's South Bridge Vaults (218): more reports in reputedly haunted areas, unrelated to prior knowledge; linked to magnetic-field variance, lighting and room size.
The “Haunt” project: an attempt to build a “haunted” room by manipulating complex electromagnetic fields and infrasound ↗79 participants, 50 minutes each, infrasound, complex electromagnetic fields, both or neither; anomalous sensations unrelated to condition but related to Temporal Lobe Signs scores; suggestibility the most parsimonious explanation.
Belief in the paranormal and suggestion in the seance room ↗In a fake séance, about one third wrongly reported that a stationary table moved; believers were more likely to report movement, and in a second experiment were more susceptible only when the suggestion fit their belief.
Lifetime prevalence rates of sleep paralysis: a systematic review ↗35 studies, 36,533 people: 7.6% of the general population, 28.3% of students and 31.9% of psychiatric patients report at least one episode; vivid hallucinations often accompany episodes, and intruders in the bedroom are a common theme.
Hypnagogic and hypnopompic hallucinations during sleep paralysis: neurological and cultural construction of the night-mare ↗Sensed presence, fear, sounds and sights, and chest pressure are common in sleep paralysis; accounts are consistent across time and cultures.
Effects of Carbon Monoxid upon the Eye ↗A family's winter from November 1912 in a house with a faulty furnace: footsteps, bells, figures, a sense of being watched; the furnace was pouring carbon monoxide into the rooms.
Infrasound exposure is linked to aversive responding, negative appraisal, and elevated salivary cortisol in humans ↗36 participants; infrasound near 18 Hz went undetected but raised cortisol and irritability; measured mood, not ghostly experiences.
The hallucinations of widowhood ↗293 widowed people in mid-Wales: almost half had hallucinations or illusions of their late spouse; rarely disclosed; judged normal and helpful.

KEEP ASKING

See a mistake? Report a problem. Corrections are made openly.