About 39% of Americans say they believe in ghosts, according to Gallup’s 2025 survey. That number has barely moved in two decades. A reader asked us to look at what science actually has to say about this, and it turns out the answer is more interesting than either side of the debate usually admits. The science of ghosts is not really about ghosts at all. It is about the ways human perception breaks down under specific, measurable conditions, and why the results feel so convincingly supernatural.
This is not a debunking piece in the sneering sense. The experiences people report are real. The question is what causes them.
The Frequency You Cannot Hear but Your Body Can
In 1998, a researcher named Vic Tandy published a paper in the Journal of the Society for Psychical Research with an unusually honest title: “The Ghost in the Machine.” Tandy worked in a medical equipment laboratory that had a reputation among staff for being haunted. People reported anxiety, cold chills, and peripheral visual disturbances. Tandy himself saw a grey figure appear at the edge of his vision one night, only for it to vanish when he turned to look directly at it.
The next day, Tandy brought a fencing foil to work for repairs. He clamped it in a vice, and the blade began vibrating on its own. This was unusual. He investigated and discovered that a recently installed extractor fan was generating infrasoundSound waves at frequencies below human hearing range (typically below 20 Hz) that can trigger physiological responses like visual disturbances and feelings of dread.: sound waves at approximately 18.98 Hz, just below the threshold of human hearing.
Infrasound at this frequency is significant because 18-19 Hz is close to the resonant frequency of the human eyeball. Vibrations at this range can cause the vitreous humor (the gel inside the eye) to distort slightly, producing peripheral visual artifacts. The same frequency range can trigger feelings of unease, pressure on the chest, and the distinct sensation that someone is watching you. Your ears cannot hear it. Your body reacts to it anyway.
Tandy went on to investigate other allegedly haunted locations, including the cellar beneath a medieval tourist attraction near Coventry. He found elevated infrasound levels at each site, often generated by wind patterns interacting with the architecture, nearby traffic, or mechanical systems. The science of ghosts, in these cases, was the science of acoustics.
It is worth understanding why this matters beyond ghost stories. Infrasound is generated by wind turbines, industrial ventilation, highway traffic, and certain weather patterns. The physiological effects are real and documented. If you have ever felt inexplicably uneasy in a building and could not explain why, a standing wave you cannot hear is a surprisingly common explanation.
The Furnace That Haunted a Family
In 1921, ophthalmologist William Wilmer published a case study in the American Journal of Ophthalmology that reads like a ghost story but ends with a gas bill. A patient he identified only as “Mrs. H” had moved into an old house with her family. They heard footsteps in empty rooms. They saw figures, including a woman dressed in black who vanished when approached. They felt pressure on their bodies while lying in bed. The children were pale and listless. The houseplants kept dying.
The family was being poisoned. Their furnace was defective, venting carbon monoxide directly into the living spaces instead of channeling it up the chimney. Carbon monoxide at sub-lethal concentrations produces a specific cluster of symptoms: headaches, dizziness, confusion, auditory and visual hallucinations, and a persistent feeling of dread. The combination is, from the inside, indistinguishable from a haunting.
Toxicologist Albert Donnay has proposed what he calls “Haunted House Syndrome,” linking the historical prevalence of ghost sightings to the widespread use of gas lighting and coal heating in the Victorian and Edwardian eras. Gas lamps and coal furnaces were prolific sources of carbon monoxide. The golden age of ghost stories coincides almost exactly with the period when indoor carbon monoxide exposure was highest and least understood.
This is not a niche concern. Carbon monoxide is colourless and odourless. The CDC estimates that around 50,000 Americans visit emergency departments each year for accidental CO exposure. In 2015, journalist and skeptic Carrie Poppy described feeling a “dark presence” in her apartment, chest pressure, and auditory disturbances. A gas company inspection found unsafe levels of carbon monoxide. The ghost was a faulty appliance.
Your Brain Paralyses Your Body Every Night
Sleep paralysisA state in which consciousness returns before the motor paralysis of REM sleep (REM atonia) releases, leaving the person awake but unable to move. is a state in which the brain wakes up before the body does. During REM sleep, your motor system is largely inhibited to prevent you from acting out dreams (a mechanism called REM atonia). Occasionally, consciousness returns before this paralysis releases. The result: you are awake, aware of your surroundings, and completely unable to move.
That alone would be unsettling. But sleep paralysis frequently comes with hypnopompic hallucinationsVivid sensory experiences generated by the brain during transitions between sleep and wakefulness, often including a sense of presence, figures, or sounds.: vivid sensory experiences generated by a brain that is still partly in dream mode while partly processing real sensory input. The most common report is a “sensed presence,” the overwhelming feeling that someone or something is in the room. Many people report seeing shadowy figures, feeling weight on their chests, or hearing breathing or footsteps.
Prevalence estimates vary, but a meta-analysis published in Sleep Medicine Reviews in 2011 found that roughly 7.6% of the general population has experienced at least one episode of sleep paralysis, with higher rates among students (28.3%) and psychiatric patients (31.9%). The experience is so consistent across cultures that it has generated its own folklore: the “Old Hag” in Newfoundland, the kanashibari in Japan, the Pisadeira in Brazil. Different cultures, identical neurological event, culturally specific interpretation.
The science of ghosts intersects here with the science of sleep. The brain, caught between states, generates experiences that are subjectively indistinguishable from a genuine encounter with something in the room. The person is not lying. They are not confused. Their brain is producing a coherent hallucination while their body refuses to respond, which is exactly the combination most likely to be interpreted as supernatural.
Pattern Recognition Gone Rogue
PareidoliaThe tendency of the brain to perceive meaningful patterns like faces or figures in random or ambiguous visual stimuli, such as clouds or shadows. is the tendency to perceive meaningful patterns in random or ambiguous stimuli. You see a face in a cloud, a figure in a curtain, a ghostly shape in a long-exposure photograph. This is not a malfunction. It is a feature of a visual processing system that prioritises speed over accuracy.
The science of ghosts overlaps here with evolutionary psychology. From an evolutionary perspective, the cost of seeing a face where there is none (false positive) is trivial. The cost of failing to see a face where there is one (false negative, perhaps a predator or an enemy) could be fatal. Natural selection favoured the paranoid observer. Your visual cortex is tuned to find faces, figures, and intentional agents in ambiguous data because, for most of human evolutionary history, the penalty for being wrong in one direction was much higher than the penalty for being wrong in the other.
Research published in Cortex has shown that individuals who report paranormal experiences score significantly higher on pareidolia tasks: they are more likely to see faces in noise, patterns in randomness, and figures in shadow. This is not a sign of lower intelligence. It is a measurable difference in perceptual sensitivity. The dial is turned slightly higher, and the world fills with shapes that are not quite there.
Combine pareidolia with confirmation biasThe tendency to search for, interpret, and recall information in ways that confirm your existing beliefs, while ignoring evidence that contradicts them. (paying more attention to evidence that supports what you already believe), and the mechanism becomes self-reinforcing. This is closely related to anti-motivated reasoningThe tendency to seek out, interpret, and remember information in a way that confirms what you already want to believe.Reasoning away from a conclusion you find unwelcome by actively searching for flaws in the evidence, rather than evaluating it impartially. The direction is chosen before the analysis begins., where we reject evidence that contradicts what we want to be true. If you believe a house is haunted, your brain’s pattern-matching system will work overtime to find evidence, and it will find it. It always does. That is what it is designed to do.
The God Helmet and Its Instructive Failure
In the 1980s and 1990s, neuroscientist Michael Persinger at Laurentian University in Canada developed what journalists called the “God Helmet”: a modified snowmobile helmet fitted with solenoids that generated weak, complex magnetic fields over the temporal lobes. Persinger reported that approximately 80% of participants experienced a “sensed presence,” often described as a ghostly or spiritual entity in the room.
The claim was dramatic. If true, it would mean that the subjective experience of ghosts, spirits, and even God could be reliably induced by tickling a specific part of the brain with magnetic fields. The implications for the science of ghosts would be profound.
In 2004, a Swedish team led by Pehr Granqvist at Uppsala University attempted to replicate Persinger’s results under double-blind conditions (meaning neither the participants nor the experimenters knew who was receiving real stimulation versus a sham). The result: the magnetic fields had no effect. The “sensed presence” experiences were predicted entirely by the participants’ suggestibility and personality traits, not by whether the helmet was actually on.
This is worth sitting with. Persinger’s original experiments were not blinded. Participants knew they were in a study about unusual experiences, were wearing a device on their heads, and were in a darkened room. The expectation alone was sufficient to produce the experience. The magnetic fields were, at best, a placebo trigger.
The God Helmet failure is actually more interesting for the science of ghosts than the God Helmet success would have been. It demonstrates that the human brain can generate vivid sensory experiences of invisible presences without any external stimulus at all. You do not need infrasound, carbon monoxide, or electromagnetic fields. Under the right conditions (darkness, suggestion, expectation), the brain does it on its own.
Grief Makes You See Them Again
Perhaps the most human explanation in the science of ghosts has nothing to do with environmental contaminants or acoustic anomalies. It has to do with loss.
A 2019 systematic review and meta-analysis published in the Journal of Affective Disorders by Kamp and Due examined 21 studies on bereavement hallucinationsSensory experiences of seeing, hearing, or sensing a deceased loved one during the grieving process, considered a normal part of bereavement by clinicians.. Their finding: 56.6% of bereaved people experience some form of hallucination involving the person who died. These include seeing the deceased, hearing their voice, feeling their touch, or sensing their presence in a room.
These are not signs of pathology. Bereavement hallucinations are considered a normal part of the grieving process by most clinicians. A study published in BJPsych Open found that 73.4% of those who experienced them reported comfort, and 68.4% considered the experience important for their bereavement process. The brain, confronted with the absence of someone deeply familiar, fills in the gap. It has spent years building a predictive model of that person (where they stand, how they sound, when they enter a room), and it does not delete the model the moment the person dies.
This means that a significant proportion of ghost sightings, particularly the emotionally significant ones (seeing a deceased spouse, hearing a parent’s voice, feeling a child’s hand), are the brain’s predictive machinery running on outdated data. The experience is genuine. The interpretation is where the science diverges from the supernatural.
The Science of Ghosts and Why 39% Is Not a Surprising Number
Given everything above, the question is not why so many people believe in ghosts. It is why the number is not higher.
Human perception is a constructive process. Your brain does not passively record the world; it builds a model and updates it with sensory data. When the data is ambiguous (dim lighting, unfamiliar sounds, emotional distress, sleep disruption, chemical exposure), the model fills in gaps using expectations, fears, and cultural templates. If your culture has a concept of ghosts, the model reaches for it.
None of this means that every ghost sighting has been explained. Science does not work by explaining every individual case. It works by identifying mechanisms that account for the patterns. The mechanisms here are well-documented: infrasound, carbon monoxide, sleep paralysis, pareidolia, suggestion, and bereavement hallucinations. Each one is independently verified. Together, they account for the vast majority of reported experiences.
The honest answer to “should you believe in ghosts?” is that the experiences people describe are real, the explanations are natural, and the evidence for anything supernatural remains at zero. That last point has not changed in over a century of investigation. The mechanisms behind contradictory expert claims apply here too: when the evidence consistently points one direction but belief persists, the interesting question is about the belief, not the evidence.
If you hear footsteps in your attic, check the furnace before you call a priest. It might save your life.
This article discusses carbon monoxide poisoning, which can be fatal. If you suspect a CO leak, evacuate immediately and call emergency services. Install carbon monoxide detectors in your home. This article is for informational purposes and is not a substitute for professional medical or safety advice.
nn
Sources
- Gallup: Paranormal Phenomena Met With Skepticism in U.S. (May 2025)
- Vic Tandy, “Something in the Cellar,” Journal of the Society for Psychical Research (2000)
- Vic Tandy and Tony R. Lawrence, “The Ghost in the Machine,” Journal of the Society for Psychical Research (1998)
- Kamp and Due, “How many bereaved people hallucinate about their loved one?” Journal of Affective Disorders (2019)
- Simmonds-Moore et al., “Exceptional Experiences Following Exposure to a Sham ‘God Helmet,'” Imagination, Cognition and Personality (2019)
- IFLScience: “An Awful Lot Of Hauntings Can Be Attributed To Carbon Monoxide Poisoning”
- Smithsonian Magazine: “Five Scientific Explanations for Spooky Sensations”
