Ghosts
in a Hurry

The Whole Thing in One Page

Ghosts arrive with a false choice attached. Either a dead person returned, or a frightened witness invented the whole thing. That choice is too crude for the evidence. People can have forceful, sincere experiences of a figure, voice, touch or presence when no ordinary person is there, and the fact that an experience has a natural cause does not make it unreal as an experience.

A ghost begins when the mind supplies a person more confidently than the world supplies a body. Human perception is built for that risk. Faces are found in stains, voices in noise and intention in movement because perception finds important patterns quickly, even at the cost of false alarms. In darkness, fatigue, expectation or grief, sparse signals carry more weight. A coat becomes a watcher. A settling house acquires footsteps. A familiar voice appears in the instant before waking thought corrects it.

Some encounters begin deeper inside the system. During sleep paralysis, awareness can return while normal REM muscle inhibition persists and dream imagery overlaps the bedroom. A threatening presence may feel close enough to touch because the body cannot move and the brain is trying to explain why. Sensed presence can also occur without a clear sight or sound when the systems that keep self, body and nearby space aligned become uncertain. Laboratory work can induce a version of this experience through delayed sensorimotor feedback. That does not explain every apparition. It proves that presence is something a brain can construct.

Grief supplies a different route. The dead remain active in memory, habit and prediction. A bereaved person may hear a spouse call, glimpse them in a crowd or feel them sitting nearby. Such experiences are often benign and can be comforting. Calling them proof of survival exceeds the evidence. Calling them madness mistakes a continuing relationship for a diagnosis.

Culture decides what the experience is allowed to become. A night visitor may be an ancestor, revenant, hungry ghost, purgatorial soul, demon, family presence or neurological event. The underlying experiences overlap; the biographies assigned to them do not. Places contribute too. A house with a story trains attention towards draughts, knocks and cold corners, while other people supply names, motives and remembered details. Once a site is haunted, every ambiguity arrives pre-labelled.

Technology has not removed this process. It has made new material for it. Spirit photographs, electronic voice phenomena, camera orbs, electromagnetic meters and thermal images convert large uncertain environments into small suggestive traces. A fragment can look objective while its production conditions disappear. The machine narrows the evidence, and the mind restores the person.

History therefore changes the costume without ending the encounter. Ancient ghosts demanded burial and food. Medieval ghosts returned with unfinished obligations. Victorian spirits tapped through tables and appeared on plates. Modern ones inhabit recordings, night-vision screens and message feeds. The forms endure because death leaves social models unfinished, perception must act before certainty and every new medium creates fresh gaps that look as though somebody is inside them.

The disciplined position is neither automatic belief nor automatic contempt. Preserve what happened, separate sensation from interpretation, test ordinary causes and ask what evidence would distinguish the rivals. A witness can be honest, the experience can be consequential, and the supernatural explanation can still be unproved.

That is the book.

Why You Should Care

In 2014, researchers placed healthy participants between two small robots. A person moved the front robot with one hand. The rear robot copied the motion and touched the person's back. When the touch arrived in time, the arrangement felt odd but coherent. When the researchers inserted a delay, some participants felt another presence behind them, even though they knew the room contained a machine and an experimenter.

Nothing in that result proves that every ghost is a sensorimotor error. It establishes a more unsettling fact. The feeling that somebody is there is not a passive reading from the room. It is an achievement of the brain, normally kept stable by agreement between movement, touch, vision and position. Disturb the agreement and a second person can appear in experience before one appears in space.

That matters because ghost stories expose the machinery of ordinary certainty. You rarely ask how you know another person is nearby. A face, a footstep, a shift in the mattress and a familiar rhythm of movement are combined too quickly for inspection. The result arrives as presence. Ghost encounters occur where that inference becomes visible: at the edge of sleep, in poor light, after loss, in an unfamiliar building or through a recording too thin to settle its own meaning.

They also show how badly we handle testimony. A witness says, “I saw my father at the end of the bed.” One listener hears proof of life after death. Another hears stupidity, dishonesty or illness. Both add more than the report contains. The useful questions come first. Was the person awake? Was there light? How long did it last? Did the figure move? Was the father expected? Was the account recorded before other people supplied details? A careful inquiry respects the witness by refusing to turn one experience into a verdict before understanding its shape.

Ghosts matter historically because the dead have never been socially simple. A body can be gone while obligations, property, guilt, rank, love and resentment remain. Societies have used returning dead to demand burial, expose murder, enforce promises, warn descendants and explain misfortune. The ghost gives unfinished business a person. It lets a relationship continue after one participant can no longer answer, and that makes it useful for religion, law, literature, family memory and fraud.

The modern story is not one of science driving ghosts away. Each technology has supplied a new form. Telegraphy made communication without visible presence ordinary just as Spiritualism expanded. Photography appeared to preserve a person after death and soon produced spirits on glass. Audio recording created voices that could be replayed, slowed and searched. Digital cameras filled dark rooms with dust, compression artefacts and infrared insects, then presented the results with a timestamp. Artificial intelligence can now imitate a dead person's voice or conversational style. The question is moving from whether a trace is authentic to what kind of authenticity anyone expected.

There is a practical reason to understand the subject. Some apparent hauntings begin with a loose window, plumbing, animals, neighbouring sound or the physiology of sleep. Some experiences belong to grief and need neither ridicule nor treatment. Some new or distressing perceptions deserve clinical attention. Some buildings contain hazards: carbon monoxide can cause headache, dizziness, weakness and confusion, and an alarm is not part of a paranormal experiment. Some investigators create the evidence they think they are measuring through suggestion, contaminated recordings or instruments used outside their purpose.

The deeper reward is intellectual. Ghosts teach a clean distinction between reality, experience and explanation. An experience can be real without its first explanation being right. An explanation can be possible without being supported. A mystery can remain after a bad theory is removed. Learning to hold those differences does not make the world less strange. It stops strangeness being settled by whichever story arrives first.

The Core Ideas

A Ghost Is an Experience Plus an Explanation

The word ghost looks like the name of a thing. In practice it joins two different claims. First, something happened: a figure appeared, a voice spoke, the bed dipped, a room felt occupied, an object moved or a recording contained an unexpected trace. Second, the event was attributed to a dead person or another disembodied agent. The first claim may be secure while the second remains open.

That separation prevents the standard insult. A natural explanation does not require a fabricated witness. Consider a person who wakes unable to move, sees a dark figure and feels pressure on the chest. The immobility, figure and terror are experiences. Sleep paralysis with dream imagery is a causal account. A dead intruder is another causal account. Rejecting the intruder does not erase the figure from the person's night. It changes what the figure was.

The same discipline applies outside sleep. A pale form crosses a corridor. The event may begin with another person, a reflection, an animal, a curtain, a visual afterimage or internally generated imagery. A deceased relative may then be recognised in it. Recognition feels like discovery because the completed figure reaches awareness before the uncertain fragments that produced it. The mind reports the result, not its assembly history.

Ghost reports also vary more than the familiar transparent figure suggests. Some are visual apparitions with clear faces. Some are peripheral shadows. Some are voices, footsteps, touches, smells or changes in temperature. A sensed presence can contain no definite sensory image at all. Bereaved people may encounter a familiar person briefly in a crowd, hear a habitual call or experience companionship without believing a body has returned. Poltergeist reports centre on movement and noise rather than a visible dead person. Photographic and audio ghosts may be discovered only after the event, when nobody present felt anything.

These differences matter because one explanation cannot be stretched across them. Pareidolia can explain why a face emerges from a pattern, but not why a fully awake person feels a companion without seeing one. Sleep paralysis can explain a particular class of bedroom intruders, but not a daylight photograph noticed weeks later. Grief can alter attention and expectation around a known person, but it cannot explain every anonymous apparition. Fraud can produce striking evidence, yet most private experiences offer no audience or profit.

A useful model therefore has layers. At the bottom is the event or stimulus, which may be external, internal or mixed. Above it is the experience constructed from that input. Above that is identification: mother, soldier, child, demon, stranger. Then comes the larger explanation: survival, haunting, memory, sleep, error, illness, hoax or something unresolved. Each layer can be more certain than the next.

This is not a device for explaining everything away. It is what makes a claim testable. If the figure was reflected light, changing the viewing angle should matter. If suggestion shapes reports, people given different histories of the same room should notice different events. If a device detects a spirit, it should distinguish that spirit from phones, wiring, radio interference and ordinary heat. If no possible observation could count against the preferred explanation, the account may still carry meaning, but it has stopped functioning as evidence.

The first rule of ghosts is therefore the first rule of serious inquiry: do not demand one verdict from several questions.

Perception Builds People from Fragments

A face can survive brutal compression. Two dark marks above one lighter mark may be enough. Turn a plug socket, car grille or slice of toast towards a human observer and the arrangement can acquire eyes and a mouth before conscious judgement arrives. Face pareidolia is not a childish mistake. It is the visible edge of a system that normally finds socially important patterns quickly and from incomplete input.

Ghosts exploit the same priority at a larger scale. People are unusually good at extracting bodies and intentions from movement. A few points of light attached to a walking person's joints can reveal gait, direction and sometimes identity. A rustle can sound like somebody shifting weight. Alternating knocks can become a reply. Once personhood is inferred, the next details organise around it: the shape is facing you, the footsteps are approaching, the pause means listening.

The system is not guessing at random. Perception combines current signals with learned expectations about what usually causes them. In daylight, with time to inspect, the incoming evidence can correct a poor hypothesis. At the end of a dark corridor, resolution is low, contrast is weak and the observer may have only a fraction of a second. Prior expectation then carries more of the load. A person who has been told that a woman appears on the stairs has been given a model with which to organise a pale vertical patch.

Low light changes more than confidence. Rod-dominated vision is sensitive but poor at fine colour and detail. Peripheral vision detects movement well while representing precise form badly. Small eye movements, adaptation, glare and shadows can make a stable scene appear to shift. Looking directly at a faint figure may make it vanish because the visual conditions have changed, which feels exactly like intentional evasion if a person has already been inferred.

Hearing is equally constructive. Speech perception must recover words from variable voices, reverberation and noise. Expecting a phrase makes that phrase easier to hear. Repeated listening can stabilise an interpretation until alternatives become difficult to recover. This is why an electronic voice clip becomes more persuasive after an investigator announces what it says. The listener is no longer approaching noise without a model.

None of this means human perception is defective. A system that refused to recognise a person until every feature was verified would be useless in conversation, traffic, sport and danger. It must settle the scene quickly enough to guide action. Most of the time, context and further evidence correct the first draft. Ghost conditions remove or weaken that correction. Darkness limits detail. Fear shortens inspection. The supposed figure disappears. A recording cannot be questioned. The witness leaves the room. The provisional person survives because the evidence ends before revision.

The evolutionary shorthand that humans possess a single overactive agency detector is tempting and too neat. The relevant capacities are distributed: face processing, biological motion, voice recognition, threat attention, social prediction and learned concepts of invisible agents. Their contribution changes by setting. A bereaved spouse hearing a familiar cough is not solving the same problem as a hiker mistaking a stump for a crouched person, though both involve person knowledge guiding uncertain perception.

The useful conclusion is narrower. Seeing is controlled inference, not receipt. The more person-like the possible cause and the poorer the conditions, the less safe it is to treat immediacy as independence from expectation. A ghost feels discovered because perception hides the work that made one.

Presence Can Be Felt Without a Visitor

An apparition is easy to picture because it has a shape. A sensed presence is stranger: the certainty that another being is nearby without a clear face, voice or touch from which that certainty could have been read. The person may be behind the shoulder, beside the bed or just outside sight. Turning to look often reveals nothing, yet the absence does not immediately dissolve the feeling.

This experience is not confined to one belief system or diagnosis. It has been reported in bereavement, sleep transitions, extreme environments, neurological illness, anxiety, spiritual practice and ordinary life. Mountaineers and endurance travellers have described an unseen companion during isolation and exhaustion. People with Parkinson's disease sometimes report a nearby presence. Others experience one once, remain healthy and never receive an explanation more precise than a tired brain under unusual conditions.

The best working account begins with the body. To act, the brain must keep track of where the self is, what produced a movement and which sensations belong to one's own action. That account draws on touch, proprioception, vision, balance and motor predictions. When those signals agree, selfhood feels located and singular. When they conflict, a movement or touch generated by the person can be misassigned, as though another agent occupied nearby space.

Neurological cases helped locate this possibility. Researchers then built a robotic arrangement in which participants moved one device while another delivered corresponding touches to their backs. With synchronised feedback, the motion and touch belonged together. When feedback was delayed, the mismatch increased reports of an alien presence in some participants. The experiment did not create a detailed Victorian ghost, and many people did not report one. It induced a stripped-down component: somebody else seems to be here.

That distinction protects the result from becoming another myth. There is no established ghost centre in the brain and no single lesion or timing error that explains every encounter. Felt presence can arise through several routes. Threat may bias attention towards an unseen agent. Grief may keep a familiar person's model active. Sleep can mix dream agency with the bedroom. Neurological disruption can disturb body ownership or spatial mapping. Suggestion can make an uncertain sensation legible as a presence. Similar experiences need not have identical causes.

The finding matters because presence often carries stronger conviction than a vague image. A witness may say they did not see the dead person but knew who it was. That knowledge can feel too specific to be constructed. Yet recognition is routinely produced without a complete sensory portrait. You know who entered a room from gait, timing, perfume or the pattern of keys in a lock. After years with somebody, the model is dense enough to run from tiny cues, and perhaps sometimes without a current cue strong enough to identify.

Presence also explains why fear can rise before a ghost acquires form. The order may be body first, story second. A surge of arousal, immobility, spatial unease or the sense of being watched demands a cause. The dark corner then supplies a figure. From memory, the sequence may reverse: the figure frightened the witness. Human recall tends to preserve the intelligible event rather than the untidy construction that preceded it.

A visitor normally causes presence. That does not make presence a detector that can never fire alone. The feeling is evidence about the state of an experiencing system. Whether it is evidence about another being still requires another step.

Sleep Lets Dream and Room Overlap

The bedroom ghost has a recurring anatomy. The person wakes in darkness, cannot move or call out, senses an intruder and may see a figure near the bed. Pressure seems to bear on the chest. The event lasts seconds or minutes, then movement returns. The witness has not merely imagined being frightened. They have been awake enough to register the room while part of the physiology and imagery of sleep remained active.

During rapid eye movement sleep, vivid dreaming is common and large skeletal muscles are normally inhibited. This protects the sleeper from enacting most dream movements. Sleep paralysis occurs when awareness and perception of the surroundings return before that inhibition has fully released, or when it arrives as the person falls asleep. Breathing continues, but the combination of immobility, lying supine, altered respiratory sensation and fear can be experienced as weight or constriction.

The intruder is not required by the physiology, yet it fits the problem the brain has been given. A person is conscious, unable to act and highly aroused. Dream imagery may still supply a voice, footsteps, touch or a visible form. The sense of presence supplies an agent. The room supplies the stage. The experience therefore feels unlike an ordinary dream because the actual bedroom is part of it.

Cultures give the visitor a name. Newfoundland tradition spoke of the Old Hag. Other accounts involve demons, witches, dead relatives, shadow people or attacking spirits. In some modern settings, the figure becomes an alien. The bodily pattern can recur while the identity changes, which is strong evidence that physiology constrains the experience and culture interprets it. It does not prove that every traditional night being was an episode of sleep paralysis. Stories can collect several experiences under one name, and later readers can force a medical label onto accounts too thin to diagnose.

Sleep also produces less dramatic crossings. Hypnagogic experiences occur as wakefulness gives way to sleep; hypnopompic experiences occur during the return. A name may be called with startling clarity. A figure may stand briefly in the doorway. A touch may seem external. These events can occur without full paralysis and may be recognised immediately as products of sleep. They become ghosts when the content is identified as a dead or invisible person and the sleep boundary is forgotten, unknown or rejected.

The memory conditions help. An event at three in the morning is rarely documented at once with light levels, posture, timing and prior sleep. By breakfast, the most memorable elements are the figure, fear and certainty. The half-dreamt transitions that made them possible leave little trace. Retelling gives the sequence a cleaner beginning and end. A visitor entered, acted and departed.

Sleep paralysis is therefore a powerful explanation with a strict boundary. It fits episodes tied to sleep, immobility and the bedroom. It does not explain a daytime sighting, a repeated sound while several people are walking around, or a figure caught only in a photograph. Even within a matching episode, stress, irregular sleep, sleep deprivation and other factors alter risk without determining the content.

The practical value is relief, not reduction. Knowing that the body can be awake and paralysed while dream imagery overlaps the room can remove the supernatural threat from a terrifying event. Recurrent episodes causing distress or substantial sleep disruption deserve proper assessment. The person does not need to be told that nothing happened. Something did. The explanation is that waking arrived in pieces.

The Dead Stay in the Social Model

A person does not vanish from the mind at the moment the body dies. Their voice, routines, likely reactions and place in the day remain available. You still know which chair they used, what time they called and what they would say about a decision. The relationship has lost reciprocal contact, but the model built through years of interaction does not receive the news in every system at once.

This helps explain sensory and quasi-sensory experiences of the deceased. A bereaved person may hear a familiar voice, glimpse the person in a crowd, smell their scent, feel a touch or sense quiet companionship. The experience may be fleeting and recognised as unusual. It may occur during sleep or waking. It may comfort, unsettle or do both. Reviews of bereavement research find that such experiences are often benign and should be understood in relation to the person's life, relationship and culture rather than automatically classified as symptoms of disorder.

Expectation works at a fine scale. After decades of shared life, a spouse may predict the sound of the key, movement in the next room or a body beside them. Habits continue to generate readiness after the expected event has stopped. A coat in a crowd matches posture and gait for one instant. A common phrase appears in ambiguous sound. The recognition system completes what the changed world no longer supplies.

Memory contributes without functioning as a recording. Recalling a person reactivates features of their face, voice and manner. Grief also directs attention towards reminders and coincidences connected with them. None of this requires deliberate wish fulfilment. The experience can arrive against expectation, and a person who rejects an afterlife can still have it. Wanting contact is neither necessary nor sufficient.

The social meaning matters as much as the mechanism. Modern bereavement theory has moved away from the assumption that health requires severing every inner tie. Continuing bonds can include talking inwardly to the dead, keeping objects, consulting remembered values and sensing an ongoing connection. A presence experience can fit within that relationship rather than interrupting recovery. It becomes concerning through distress, danger, loss of function, broader symptoms or a pattern requiring clinical interpretation, not through the bare fact that the dead seemed close.

Two opposite errors follow. The first treats the event as direct evidence that consciousness survived. The experience may support a person's faith or private understanding, but its existence alone cannot distinguish survival from memory, expectation, sleep, perception or other causes. The second treats any natural account as an attack on love. Explaining how a familiar voice can be generated does not show that the relationship was shallow. It shows how deeply another person had been learned.

Culture can reverse the emotional sign. In one setting, seeing a dead relative may be welcomed as reassurance or guidance. In another, it may be feared as failure to rest, danger to the household or evidence that ritual obligations were neglected. The same gentle presence can therefore produce comfort, urgent ceremony or medical concern depending on what a returning dead person is understood to mean.

Bereavement also supplies the strongest challenge to the claim that ghosts are products of primitive fear. Many encounters are affectionate, ordinary and brief. The figure does not rattle chains or threaten punishment. It sits in the known chair, says the person's name or appears at the edge of a familiar routine. Horror cinema has trained the public to expect invasion. Grief more often supplies continuity.

The dead stay everywhere because human relationships are predictive systems as well as memories. They prepare actions, fill spaces and generate expectations. Death ends new input from one person. It does not instantly dismantle the form that person acquired in everybody else.

Culture Gives the Presence a Biography

An unexplained presence does not arrive with a passport. Culture supplies one. It decides who can return, why they would return, how they appear, what they want and which response is proper. That is why ghost experience is widespread while ghosts are not interchangeable.

Ancient Mesopotamian texts treated the dead as members of a continuing household economy. Proper burial and offerings mattered; a neglected or violently dead person could trouble the living. The problem was less whether spirits existed than whether relations between living and dead had been maintained. In parts of medieval Europe, returning dead might request masses, confess hidden wrongs or warn about punishment. Their appearance fitted Christian ideas of sin, penance and purgatory. In Chinese Buddhist and local traditions examined through the medieval Ghost Festival, hungry or suffering beings were linked to ritual feeding, kinship and the transfer of merit. These systems differ in theology, social structure and historical setting. The shared fact is that the dead retain claims.

Stories turn an ambiguous event into a social diagnosis. Chains suggest wrongful confinement. A child figure points towards lost innocence. Repetition on an anniversary makes time meaningful. A room becomes the site of murder, betrayal or improper burial. The explanation often concerns a broken obligation because obligations are what death leaves behind. Ghosts make absence behave as agency.

The form also follows local media. Oral communities can anchor a presence to a path, tree or threshold through repeated telling. Manuscript and print preserve named cases and circulate standard motifs. Theatre and fiction supply visible conventions: the shroud, pale face, unfinished sentence, cold entrance. Photography creates translucent doubles. Cinema teaches figures to appear in mirrors and at the edge of frames. Internet culture gives them closed-circuit footage, voice notes and looping clips. People do not need to believe a story consciously for it to guide what a vague event resembles.

This is not proof that expectation creates every detail. Culture works through selection as well as production. A person may have an odd experience first and search for a category later. Among its many features, those matching an available ghost type will be remembered and repeated. Features without a place in the story fall away. A colour becomes white, an uncertain age becomes a child, a movement becomes beckoning. Retelling does not have to be dishonest to become more conventional.

Names alter action. Call a noise plumbing and someone checks pressure. Call it a burglar and they leave or confront. Call it an ancestor and they may speak, offer food or seek ritual help. The label changes attention during the next event, creating a loop between interpretation and experience. Once a house has a named woman on the stairs, future visitors watch the stairs for a woman, and each ambiguous sighting strengthens the name.

The phrase “every culture has ghosts” therefore needs care. Many societies have concepts of ancestors, spirits, souls, revenants or dangerous dead, but forcing them all into the English ghost can erase important distinctions. Some beings were never human. Some dead remain helpful members of a lineage rather than trespassers. Some traditions care more about ritual obligation than visual apparition. The comparison is useful only if difference survives it.

What does recur is the human problem. Bodies disappear before relationships, debts, fears and social roles are finished. Perception can produce person-like experience without enough evidence to settle its source. Communities inherit both the absence and the report, then give them a form that can be acted upon. That is close enough to everywhere and always to explain the subtitle, and far enough from universality to keep it honest.

Evidence Machines Make New Ghosts

A machine seems to solve the witness problem. Cameras do not grieve. Recorders are not frightened. Meters display numbers. If a trace appears there, it feels independent of belief.

That confidence is useful and incomplete. Instruments answer questions defined by their design. A camera records light reaching a sensor through a lens during an exposure, then software transforms the signal into an image. It does not label the causes. Dust near a flash can become a bright defocused orb. Long exposure turns a moving person translucent. Reflections, insects, sensor noise, compression and image processing create forms that were invisible to the eye. The result is objective in one sense: a file exists. Its interpretation remains human.

Spirit photography established the pattern early. Nineteenth-century plates could retain earlier images, receive double exposures or be deliberately combined. William Mumler built a business around portraits in which pale dead relatives appeared beside sitters. At his 1869 New York hearing, photographers showed ways such effects could be produced, but the prosecution did not prove exactly how he made every image and he was discharged. The legal failure became persuasive ambiguity. A technique existed, a specific act remained unproved, and the ghost occupied the gap.

Audio produces a harder version because listeners can search repeatedly. Electronic voice phenomena are words found in hiss, radio fragments, reverberation or other recordings not recognised during capture. Once a phrase is suggested, speech perception organises the sound around it. Repetition increases fluency and confidence. Removing the label, testing independent transcriptions and preserving the unedited recording often reveal that the voice was partly supplied by the listening procedure.

Electromagnetic field meters create another translation. They are legitimate tools for detecting fields produced by wiring, appliances, transmitters and other sources within their specifications. Ghost hunters may treat an unexplained change as a responsive presence, especially when a display rises after a question. Yet a meter has no spirit setting. Without a mapped baseline, shielding, source checks, time synchronisation and a rule fixed before observation, the number cannot distinguish a dead person from a phone, cable, lift motor or investigator moving through a spatial gradient.

Thermal cameras record surface temperature or infrared radiation according to the instrument and conditions. They can reveal insulation failures, warm footprints or a person behind some obstructions. They can also be misread through reflections, changing emissivity, airflow and automatic scaling. “Cold spot” is a description before it is a cause. Measuring one does not show who made it.

The deeper problem is selection. A team may record for six hours, collect several devices and later display the strangest ten seconds. The denominator disappears. With enough channels, a coincidence becomes likely: a knock follows a question, a meter shifts, a face appears in noise. If the search rule is flexible, the best fragment will seem designed. Proper evidence needs the full recording, control periods, synchronised instruments, blind analysis and predictions that could fail.

New technology therefore does not drive ghosts from the world. It changes where uncertainty lives. The naked-eye apparition depends on memory and testimony. The digital ghost depends on pipelines, thresholds, metadata and interpretation. Artificial intelligence adds another layer by enhancing faint audio, filling missing pixels or generating a deceased person's likeness and voice. An output can become clearer as its connection to the original event becomes weaker.

This repays the first idea. A ghost is an event plus an explanation, and evidence machines multiply events while concealing how they were made. The machine provides the fragment. Human perception restores the person. Unless the chain from signal to claim is visible, objectivity becomes another costume the ghost can wear.

How It Actually Works

The dead remain members of the household

A dead person in ancient Mesopotamia could still be neglected. Cuneiform ritual and literary texts describe the dead as dependent on burial, remembrance and offerings from descendants. A person denied proper rites, killed violently or forgotten might become troublesome. The returning dead were therefore connected to household maintenance. Feed them, name them and keep the relationship in order, or disorder could cross back into the living world.

This was not a primitive version of a modern haunted-house programme. The dead belonged to a different social and religious system. Their appearance made sense through kinship, inheritance, ritual duty and the unstable boundary between corpse, spirit and ancestor. What survives on clay is also selective: incantations, omens, stories and administrative traces, not a representative survey of ordinary experience.

The Greek and Roman record gives the ghost a narrative job familiar now. In a letter written around the start of the second century CE, Pliny the Younger tells of an Athenian house so notorious that it stood empty. At night came chains and an emaciated old figure. The philosopher Athenodorus rented the bargain property, followed the apparition and marked the place where it vanished. Excavation uncovered bound bones; after proper burial, the haunting stopped.

The story contains most of the later machine. A place has a reputation before the witness enters. The ghost repeats one sign. A brave investigator refuses panic. The apparition points towards concealed history. Physical remains confirm a moral wrong, ritual repair ends the disturbance and the property becomes usable again. Pliny presents it as a story he had heard while asking whether apparitions are real. It is evidence that Romans told and argued about such cases, not a case file whose excavation can now be checked.

The ghost becomes a moral messenger

Medieval European ghosts often returned into an organised moral geography. Christian teaching distinguished the dead, saints, angels, demons and resurrected bodies, while local stories preserved walking corpses, household presences and restless dead. As ideas of purgatory developed, a returning soul could request prayer, masses, restitution or confession. The apparition made an invisible condition visible and gave the living a task.

The social work is easy to miss when the story is read only for horror. A ghost could reveal an unpaid debt, expose improper burial, support a monastery's claims or warn that behaviour carried consequences beyond death. The witness's identity mattered. A monk, widow, child or local notable brought different authority. So did the place and the chain by which the account reached a chronicler. A private encounter became useful only after a community decided what kind of message it contained.

Other regions developed different arrangements. The medieval Chinese Ghost Festival, shaped through Buddhist teaching, local practice and family ritual, addressed suffering beings and obligations between generations through offerings, merit and public ceremony. It cannot be reduced to Europe's purgatorial visitor. Both show why the returning dead survive institutional change: they connect cosmic order to duties that can be performed now.

Reform divides the possible dead

The Reformation did not make northern Europe secular. It changed the authorised explanations. Protestant rejection of purgatory removed one major reason for ordinary souls to return seeking masses. In English polemic, an apparent dead person might therefore be reassigned to fraud, error or a devil impersonating the dead. Catholic writers could defend a wider traffic between worlds. The experience did not disappear when the official map changed; its permitted biography changed.

Print sharpened the contest. Pamphlets circulated wonders, providences, apparitions and exposures beyond the village in which they began. A local noise could become evidence in a national religious argument. Sceptical explanation also developed inside this traffic. Dreams, melancholy, misperception and manipulation were not inventions of twentieth-century psychology. Early modern writers already knew that a forceful apparition might come from the witness, an impostor or a misunderstood event.

By the eighteenth century, ghost-seeing could be discussed through nerves, imagination and hallucination while remaining culturally active. Enlightenment did not replace belief with one settled natural account. It created rival authorities: clergy, physicians, antiquarians, philosophers, entertainers and eyewitnesses. The ghost moved from a known category within a shared cosmology towards a disputed event requiring explanation.

That move made testimony central. A respectable witness, daylight, close distance and several observers became assets. So did reluctance: the person who insisted they had never believed in ghosts sounded less prepared to see one. The modern formula was forming. Conviction would rest less on the old doctrine of what the dead do and more on the apparent quality of the encounter.

The dead learn to knock

In late March 1848, Maggie and Kate Fox reported mysterious raps in their family's house at Hydesville, New York. The noises appeared to answer questions and count numbers. Neighbours came. The alleged communicator acquired a history. The sisters were soon demonstrating spirit communication in public, and the events became a founding episode of modern Spiritualism.

The movement grew because the method was portable. A spirit did not need to appear. It could knock, tilt, write, play an instrument or speak through a medium. Communication could be coded and repeated before an audience. The séance converted haunting from an accident tied to one location into an appointment. The dead could be called.

The timing mattered without supplying one cause. Telegraph networks had made meaningful signals travel invisibly. Photography preserved faces after the person had gone. Industrial cities separated families while war and disease produced enormous loss. Religious authority was contested, yet interest in progress and experiment was high. Spiritualism could present itself as faith tested through phenomena: a message arrived, witnesses observed it and a medium repeated the conditions.

It also rearranged authority. Women who were restricted in church, politics and public speech could speak as mediums before mixed audiences, though their authority was often framed as passivity to a spirit. Séances crossed entertainment, religion, mourning and commerce. Sincere conviction, performance skill, unconscious action and fraud could occupy the same room. A sitter might receive comfort from a practised trick without knowing it was a trick. A medium might believe an automatic movement came from elsewhere. Exposure of one method did not settle every motive or experience.

Maggie Fox publicly confessed in 1888 that the sisters had produced rappings with their bodies, then later recanted. The confession is important and not a master key. It bears on the Fox performances, not every table, trance or private encounter that followed. By then Spiritualism had institutions, newspapers, churches, celebrities and a large transatlantic following. A movement born through knocks no longer depended on the origin story being clean.

The table moves before the spirit does

Table-turning swept through Europe and North America in the early 1850s. Participants rested their hands on a table and waited. It trembled, rotated, tilted or tapped out letters. The force felt external because no individual intended to push, and the group produced more movement than any one participant could attribute to themselves.

Michael Faraday treated the claim as an experimental problem. In 1853 he built layered indicators that could reveal whether hands moved before the table or the table pulled the hands. The hands led. When participants could see an index showing their pressure, the movement diminished. The effect did not require conscious deceit. Small expectations and muscular adjustments accumulated across several people until the table moved, while each person experienced themselves as following rather than causing it.

This family of effects became associated with ideomotor action: thought, expectation or attention producing small movement without a felt decision to move. It remains relevant to pendulums, dowsing rods and talking boards. The board can spell a surprising answer because several hands guide a pointer without any participant experiencing authorship over each motion. Shared action diffuses agency.

The important lesson is not that one apparatus defeated the dead. It is that a control must interrupt the proposed mechanism. Asking sitters whether they pushed cannot work if the process is unconscious. Watching the table cannot establish which direction force travelled. Faraday changed the setup so rival causes made different predictions.

The response also shows why evidence disputes persist. Believers could accept that some tables moved through muscular pressure while reserving other cases. Fraudulent mediums could adopt methods that defeated simple observation. Investigators could become advocates, and advocates could expose fraud. The category remained mixed, so no single demonstration could empty it.

The camera inherits the séance

William Mumler's first spirit photograph appeared in Boston in 1861 when a faint second figure emerged in a self-portrait. He soon sold portraits in which deceased relatives or guides hovered beside living sitters. Photography carried a new authority. A plate seemed able to register what the eye missed, and the image could be inspected after the medium had left.

It could also be manipulated through the ordinary techniques and accidents of early photography: reused or insufficiently cleaned plates, double exposure, prepared negatives, staged figures and printing choices. In 1869, New York prosecutors accused Mumler of fraud. Photographers demonstrated ways to produce comparable effects, and P. T. Barnum supplied a fabricated spirit photograph to show how easily Abraham Lincoln could be made to appear. The court discharged Mumler because the prosecution had not proved its case against his particular process.

That outcome captures the evidential trap. Demonstrating a possible normal method is not proof that this operator used it in every image. Failing to establish the exact method is not evidence that a dead person stood before the plate. The gap between those propositions became the product. Mumler continued to attract sitters, including Mary Todd Lincoln, whose portrait with a pale Abraham became his best-known image.

A photograph also changes memory. Instead of recalling a brief figure, the viewer can return to one frozen ambiguity and add detail. Faces emerge through repeated inspection. Family resemblance becomes persuasive after a candidate is proposed. The image feels permanent while its exposure history, plate handling and selection process disappear.

Science counts the apparitions

The Society for Psychical Research was founded in London in 1882 to investigate mesmerism, thought-transference, apparitions, mediumship and related claims with organised inquiry. Its members included committed believers, sceptics and people willing to suspend judgement. That mixture produced both serious methodological work and enduring dispute about how much evidence should move a prior belief.

Early researchers collected detailed cases and looked for apparitions reported near a person's death, hoping that coincidences beyond chance might support telepathy or survival. Phantasms of the Living appeared in 1886. A larger Census of Hallucinations, published in 1894, gathered more than seventeen thousand responses to a question about waking impressions of a person or thing when no physical cause was present. It showed that apparition-like experiences were not confined to psychiatric hospitals or avowed Spiritualists.

The project also exposed the limits of retrospective testimony. Memories of timing can shift. Dramatic coincidences are reported and preserved more readily than non-coincidences. Sampling was not modern probability sampling, interviewers and respondents varied, and the definition gathered several kinds of experience. Counting experiences did not by itself identify their causes.

Later psychical research moved between field cases, laboratory tests, medium investigations and fraud detection. Results rarely ended the argument because controls altered the phenomenon, replications failed, or a statistical anomaly did not specify an agent. The strongest legacy may be methodological rather than supernatural: preserve original testimony, record timing before checking the target, blind judges to the desired answer and count misses as well as hits.

A place learns its story

A haunting becomes easier to maintain when it belongs to a location. Buildings generate ambiguous events without assistance: pipes contract, floorboards carry distant weight, air pressure moves doors, animals enter cavities, traffic produces low vibration and voices cross walls through routes nobody notices. Old buildings are not uniquely supernatural. They are unusually legible as stories because their causes are hidden and their surfaces display age.

Reputation changes the sampling. Visitors told that a room is active spend longer monitoring bodily sensations and small environmental changes. They may enter at night, move quietly and turn off the ordinary sources of orientation. A cold draught, momentary dizziness or sound from another floor becomes relevant in a way it would not during a daytime viewing. Expectation need not manufacture a sensation. It changes which sensations are selected and how quickly an agent is assigned.

Field experiments have tried to separate place from story. At reported haunted sites, unusual experiences sometimes cluster in particular areas, which can reflect lighting, airflow, acoustics, magnetic variation, architecture or the route visitors are encouraged to take. Clustering alone does not identify a ghost. It tells investigators where to measure and what alternatives to test.

The Haunt project built an artificial chamber where seventy-nine participants spent fifty minutes under combinations of infrasound and complex electromagnetic fields. Many reported unusual sensations, but the number of reports did not track which physical condition they received. It was more closely related to individual differences associated with suggestibility and anomalous experience. A separate double-blind test of weak complex magnetic fields likewise found that suggestibility predicted sensed presences better than whether the field was active. These studies challenge claims that one special field reliably manufactures a haunting. They do not prove that every building report is suggestion, and their artificial settings limit what can be generalised.

A place also accumulates testimony. Each visitor is not an independent detector if all have heard the same account, followed the same guide and inspected the same staircase. Later reports can converge because the site supplies a template. Convergence is stronger evidence when descriptions were recorded separately before witnesses learned what others saw.

The house then feeds back into commerce and identity. A ghost can threaten a sale, attract visitors, justify preservation or give a town a story. Owners, investigators and sceptics all have incentives, though an incentive is not proof of dishonesty. The methodological demand is to make the interest visible and preserve the boring denominator: how many nights, rooms, visitors and instrument readings produced nothing.

The haunted house becomes a studio

Twentieth-century recording and broadcasting made hauntings reproducible as media. A voice could be found on tape. A photograph could be enlarged. Radio and television turned investigation into a genre with a familiar sequence: enter at night, announce a history, ask questions, monitor devices and isolate the most suggestive response.

Digital equipment multiplied channels. Night-vision cameras turn insects and dust into bright moving forms. Automatic gain lifts faint room noise. Compression reshapes low-level audio. Electromagnetic meters respond to ordinary sources. Thermal cameras convert temperature differences into dramatic palettes. None is useless. Each becomes weak evidence when its operating conditions, baselines and false-positive rate are ignored.

Television adds a social pressure the private witness did not face. Nothing happening is a failed episode. Investigators speak, move, react and review until a usable event appears. The audience then receives a compressed sequence in which question and knock sit together, while hours of unmatched questions vanish. The form trains viewers to treat temporal proximity as reply.

Online circulation completes the loop. A clip is detached from the room and examined by thousands who can pause, annotate and suggest identities. Algorithms promote material that holds attention, not material with the best controls. The ghost becomes collaborative: one person records a blur, another sees a face, a third supplies a local death and repetition turns the interpretation into the clip's title.

Artificial intelligence now makes the boundary more difficult. Enhancement systems can remove noise by estimating what should be present. Generative tools can reconstruct faces, animate photographs and imitate voices. These outputs may be moving memorials or deliberate fiction. They cannot serve as untouched records of the source once new detail has been inferred. The newest ghost may be a person-model made explicit by software.

How we know

Ghost research joins kinds of evidence that answer different questions. Historical texts show what people recorded, feared and used, but rarely let us reconstruct the original event. Surveys estimate reported experiences only within their wording, sample and culture. Clinical and neurological evidence can identify mechanisms capable of producing figures, voices or presence without showing that one mechanism caused a past case. Laboratory studies gain control by stripping away the setting that gave an encounter its meaning. Haunted-place studies preserve context but struggle with expectation, selection, environmental variation and low repeatability.

Confidence is highest where a proposed mechanism makes a risky, repeatable prediction: delayed sensorimotor feedback increases some presence reports; visible feedback reduces ideomotor table movement; labels alter what listeners hear in ambiguous noise. Confidence is lower when one explanation is assigned retrospectively to a single story. No controlled body of evidence has established that the surviving personality of a dead person causes apparitions, recordings or instrument changes. That does not make every case solved. It fixes the present evidential boundary between an unexplained experience and proof of survival.

What People Get Wrong

“Either the witness lied or the ghost was real”

This false choice survives because an encounter can feel too detailed and personal to be called a mistake. If the dead did not return, the witness must have invented the account.

Human experience offers many alternatives. A person can accurately report seeing a figure while being wrong about what caused it. Memory can preserve the central conviction while changing sequence and detail. Sleep imagery can overlap the room. A real sound can receive a false identity. Suggestion can shape an event without conscious agreement. Several people can influence one another while each remains sincere.

Fraud belongs in the range, especially where money, status or performance is involved. It is not the default explanation for a private bereavement presence with no audience. Nor does sincerity establish survival. Honesty describes the witness's relation to the report, not the report's relation to an external cause.

Contempt destroys evidence. People hide timing, sleep, medication, grief or prior expectations when they expect ridicule, leaving investigators with a dramatic summary and none of the conditions needed to understand it. Respect is method: take the report seriously enough to separate what was experienced from what is being claimed.

“Ghosts look the same everywhere”

White, translucent people who drift through walls are a successful modern convention, not a universal human template. Earlier European ghosts could be solid, corpse-like, shrouded, animal-shaped or indistinguishable from living people. Some traditions centre on voices, dreams, smells, possession, ancestor presence or dangerous dead rather than a visible figure.

The error comes from translation and media. English labels many different beings as ghosts. Global film and television then export a small visual vocabulary, while museums and popular books select examples that resemble it. Similarity increases after circulation.

There are recurrent problems beneath the difference: improper burial, violent death, broken obligation, longing, dangerous places and uncertainty about where the person has gone. Those themes do not make the beings identical. A purgatorial soul seeking prayer, a hungry ghost receiving offerings and a family ancestor protecting descendants occupy different moral systems.

Cultural variation is evidence. If the same biological capacities operate across societies while identities and expected behaviours change, experience cannot be understood apart from interpretation. It also prevents a familiar category from swallowing other people's concepts and reporting the loss of detail as universality.

“Sleep paralysis explains every night visitor”

Sleep paralysis is one of the strongest natural explanations in this subject because its fit can be exact: awakening, immobility, sensed intruder, chest pressure, bedroom imagery and abrupt release. That success encourages overreach.

The mechanism applies to transitions into or out of sleep. It does not explain a figure seen while walking downstairs, knocks heard by awake people moving through a house, or an image discovered later on a camera. Even at night, a witness may have been fully mobile or awake for a long period. A similar story is not enough to assign the state retrospectively.

Historical creatures should not be diagnosed from a motif alone. A tale of a witch pressing the sleeper may preserve sleep paralysis, nightmare, illness, folklore, social conflict or several layers added through retelling. The old account rarely records the clinical detail needed to decide.

Getting the scope right matters in both directions. Recognising a matching episode can reduce terror and direct recurrent distress towards appropriate assessment. Refusing to turn it into a universal solvent keeps other mechanisms visible. A good explanation earns scope by matching conditions, not by being famous.

“Seeing a ghost means mental illness”

Hallucination is often heard as a diagnosis, though it describes an experience: perception-like activity without a corresponding external stimulus. Such experiences can occur around sleep, bereavement, sensory impairment, fever, drugs, neurological conditions and in people without a psychiatric disorder.

Context changes meaning. A brief sensed presence after a spouse's death is not equivalent to persistent threatening voices accompanied by severe confusion or loss of function. A person who recognises an image as unusual differs from someone whose wider beliefs and behaviour have become dangerously disorganised. One experience cannot carry the whole clinical judgement.

The stigma grew partly because unusual perception was historically used to mark irrationality, while psychical researchers and spiritualists claimed the same reports as evidence of hidden faculties. Both camps turned a varied phenomenon into one identity.

None of this means ignoring risk. New, frequent, distressing or impairing experiences, sudden confusion, neurological symptoms, substance effects or commands to harm require proper help. It means that fear of being labelled should not stop accurate reporting. A ghost account can be culturally meaningful, medically relevant, benign or several of those at once. Diagnosis depends on the person and pattern, not the spooky noun.

“Haunted places emit a special physical signal”

Infrasound and electromagnetic fields are often offered as the missing physics of haunted rooms. Both are real physical phenomena. The leap comes when any unusual reading is treated as a cause of apparitions or as energy produced by spirits.

Controlled studies have not shown that weak complex magnetic fields reliably induce sensed presences. In an artificial haunted-room experiment, participants reported many unusual sensations, but reports did not track whether infrasound or electromagnetic manipulation was active. Suggestibility and individual differences explained more of the variation.

That does not prove environments are irrelevant. Lighting, airflow, acoustics, vibration, temperature, mould, pollutants, building movement and ordinary fields can alter sensation or produce puzzling events. Carbon monoxide is a safety concern because it can cause headache, dizziness, weakness and confusion. A detector alarm requires leaving the danger and following emergency guidance, not collecting a better ghost reading.

Measurement should narrow causes. An unexplained field change is an invitation to map wiring, appliances, transmitters and instrument limits. Calling it paranormal at the moment it becomes interesting stops the investigation where it should begin.

“A camera or recorder removes human bias”

A device can preserve a signal more faithfully than memory while introducing its own transformations. Exposure, focus, gain, compression, filtering, automatic enhancement and file handling decide what survives. The human then chooses where to point, when to record, which segment to review and what label to give it.

Ghost photography has exploited this gap since the nineteenth century. Double exposures and reused plates produced pale figures that looked independent of the sitter's belief. Digital flash turns nearby dust or insects into bright orbs. Audio gain raises hiss and distant speech. Repeated listening after a suggested phrase can make noise sound stable and intentional.

The strongest test is blind interpretation. Preserve the raw file, include control recordings, remove the proposed words or identity and ask independent listeners or viewers what they detect. Record all trials, not the selected hit. An impressive fragment becomes weaker when thousands of empty frames are restored to the denominator.

The lesson reaches far beyond ghosts. Machines do not abolish inference. They relocate it into settings, software and selection. A timestamp proves when a file was made, not who is inside it.

“Unexplained evidence counts as evidence of survival”

A case can resist every ordinary explanation offered and still fail to identify its cause. This feels unsatisfying because mystery is treated as a contest: once fraud, sleep, reflection and error have been challenged, the dead appear to win by elimination.

The logic works only if the list of alternatives is complete and each has been tested adequately. In open human settings, it rarely is. Missing information, equipment behaviour, unnoticed people, memory change and unknown normal causes remain possible. “Unexplained” often means the evidence is too poor to choose, not that it is too extraordinary for nature.

Survival is also a specific claim. It requires evidence that a deceased person's continuing mind caused the event, not merely that a face, voice or coincidence occurred. Information apparently unknown to the witness is stronger than a vague figure, but it still needs secure records, exclusion of ordinary access, independent judging and replication.

None of this requires declaring survival impossible. It assigns the burden properly. Possibility belongs to philosophy and personal belief until observations discriminate it from rivals. An honest unknown is not a defeat. It is the only answer that keeps the question open without awarding it to the preferred cause.

Use It

Split the claim into layers

When somebody reports a ghost, resist the demand for one immediate answer. Write four lines: what happened in the environment, what the person experienced, who or what they thought it was, and what larger cause they inferred.

“I heard three knocks” is different from “the knocks answered me”, which is different from “my father answered me”, which is different from “personal consciousness survives death”. Each step may be understandable. Each needs additional evidence.

This separation improves both sceptical and believing accounts. A sceptic cannot erase the experience by naming a possible cause. A believer cannot convert recognition into proof without showing how the identity was established. Unknown details can remain unknown rather than being filled by whichever side speaks first.

Use the same method on photographs and recordings. The file exists. A shape resembles a face. The face is said to be a former resident. The former resident is said to have acted intentionally. Those are four claims, not one object called evidence.

Capture before interpretation

The first version of an encounter is perishable. Record it before researching the building, comparing accounts or replaying a clip with suggested words. Note time, light, position, duration, wakefulness, movement, other people, devices, recent expectations and what ended the event. Preserve uncertainty: “dark upright form” is better evidence than a later “Victorian woman” if clothing and sex were not clear at first.

Independent accounts matter only when they are independent. Ask witnesses to write separately before discussion. Do not tell the second person what the first saw. Record disagreement rather than smoothing it into a shared story. Convergence that appears after a tour guide names the child in the corridor says more about transmission than about the corridor.

For media, keep the original file, metadata and full recording. Save the minutes before and after the selected clip. Enhancement should create a copy, never replace the source. The stronger the later interpretation, the more important the untouched first state becomes.

Test the room before the dead

Begin with causes capable of harming somebody or producing a repeatable event. If a carbon monoxide alarm sounds or exposure is suspected, go outside and follow current emergency guidance. Investigate electrical faults, unstable structures, intruders, animals, blocked ventilation and other hazards before spending a night provoking an apparition. Mystery does not reduce physical risk.

Then map ordinary behaviour. Visit in daylight. Listen from neighbouring rooms. Run water, heating, lifts and appliances. Watch doors under changing pressure. Photograph reflective surfaces from the witness position. Note traffic, trains, outside lights and radio sources. A cause that appears only at night may be a cause that becomes noticeable only when the building is quiet.

Do not expect every case to end with a theatrical reveal. Old buildings contain intermittent systems. A pipe may knock once a week; an animal may leave; a reflection needs one angle. “Could not reproduce” is not the same as “has no ordinary cause”. It means the test did not capture one.

Make the machine show its work

Before using a device, state what it measures, its range, its known error and what result the ghost hypothesis predicts. An electromagnetic meter detects fields, not agency. A thermal camera maps radiation or inferred surface temperature, not the dead. An audio recorder stores pressure variations transformed by hardware and software, not certified speech.

Establish baselines with the room occupied and empty. Synchronise clocks. Place control devices where no effect is predicted. Move phones and radios. Log every reading rather than filming only the display that jumps. Set thresholds before looking at the data.

For audio, ask blind listeners to transcribe without a proposed phrase and include ordinary control clips. For images, present alternatives with matched contrast and resolution. A result that appears only after aggressive processing must be reported as the output of that processing. Artificial intelligence enhancement should be treated as reconstruction, because it may add probable detail that was never present in the source.

The question is not whether a machine can be fooled. It is whether the chain from event to conclusion is visible enough to locate the uncertainty.

Respect meaning without certifying cause

A bereaved person may describe a presence as the most important event of the year. Correcting them before understanding what it meant is poor inquiry and poor company. Ask whether the experience comforted, frightened or changed them, and whether they want explanation, witness or practical help.

Meaning does not settle mechanism. A natural account can coexist with value. Hearing a dead partner's familiar phrase may express how deeply that voice remains embedded in expectation and memory. Calling it memory need not demote it. Calling it communication may fit a religious framework while remaining beyond what the event can establish publicly.

Use language proportionate to the relationship. “I do not know what caused it” is different from “nothing happened”. “This can occur in grief” is different from “you imagined it”. Where a person is distressed, impaired, confused or at risk, compassion includes encouraging appropriate medical or psychological support rather than protecting a paranormal interpretation from examination.

Design a result that could fail

Before investigating, ask what each explanation predicts. A reflection should change with angle or light. A sound transmitted through pipes should track plumbing. Suggestion should alter reports when visitors receive different histories. A claimed responsive intelligence should answer pre-set questions more accurately than control periods and independent judges would expect.

Avoid moving the target after the result. If one knock means yes, decide that first. If any noise, silence or equipment failure can be interpreted as a reply, the test has no failure condition. If a named spirit supplies only vague words that fit many people, identity has not been demonstrated.

Count opportunities. One relevant word in ten recordings differs from one in ten thousand. A knock after a question differs from hundreds of questions with no knock. Search methods that allow endless replay, filtering and phrase selection need stronger controls because they generate chances for a match.

A good test may leave the larger metaphysical question untouched. That is acceptable. It can still show that this table moved through hand pressure, this photograph arose through exposure, or this claimed information was available normally. Local answers accumulate. Grand verdicts rarely arrive from one dark room.

The limits

The framework cannot reconstruct every past encounter. Reports lose timing and conditions; witnesses influence one another; buildings change; original files disappear. Some cases will remain unexplained because the evidence is gone, not because the cause exceeded nature.

Laboratory mechanisms also have limits. Producing a sensed presence through delayed touch shows capability, not identity with every spontaneous presence. Suggestion effects in selected groups do not make witnesses interchangeable or prove that expectation caused a given sighting. Cultural analysis explains available meanings without reducing belief to ignorance.

No method here proves that survival after death is impossible. It shows what present evidence has and has not established. The same restraint applies to diagnosis. A book can describe routes to unusual perception; it cannot decide the cause of one person's recurring experience without context and assessment.

The one thing to keep

Keep the gap between presence and person.

Your mind is built to close that gap quickly. It takes incomplete light, sound, movement, memory and bodily feeling and produces somebody. Usually that speed is what lets you recognise a friend, catch a voice in noise and know who entered without looking. In grief it lets a relationship remain active. In darkness and sleep it can put an agent where no body can be found. Culture names the agent. Technology gives it a trace.

The mistake is not having the experience. The mistake is forgetting that the experience arrived before its explanation.

A ghost is therefore a lesson in disciplined wonder. Take the witness seriously. Check the house. Preserve the file. Test the ordinary causes. Admit what remains. Do not call every mystery a spirit, and do not call every spirit report foolishness. The permanently different view is this: certainty that somebody is there tells you something important about a human mind and relationship, but it does not yet tell you who, if anyone, is in the room.

Terms

Ghost. A dead person or disembodied agent believed to be present or active among the living. The word also labels experiences and traces whose cause has not been established. Context decides which meaning is intended.

Apparition. A person-like figure or appearance without an identified ordinary body. Apparitions may be visual, multisensory or recognised through presence rather than detailed sight. The term does not specify a cause.

Haunting. A pattern of unusual events attributed to an agent attached to a place, object or person. Repetition and location distinguish it from a single encounter. A reputation can become part of the pattern.

Sensed presence. The feeling that another being is nearby without clear sensory evidence. It occurs in several settings and does not by itself identify an external visitor. The experience can be vivid and spatially precise.

Bereavement experience. A sensory or presence-like encounter involving someone who has died. Such experiences can be benign, comforting or distressing and require context rather than automatic diagnosis. Relationship, culture and wider functioning change its significance.

Pareidolia. Perceiving a familiar form, especially a face, in ambiguous material such as stains, shadows or noise. The stimulus exists; its person-like organisation is constructed.

Apophenia. Detecting meaningful connections or patterns in unrelated or weakly related events. Flexible searching across many recordings or coincidences increases opportunities for an apparently designed match. Counting all opportunities is therefore essential.

Illusion. A misperception of a present external stimulus. A coat interpreted as a person is an illusion; a figure experienced without a corresponding stimulus is described differently.

Hallucination. A perception-like experience without the corresponding external stimulus. The term describes a phenomenon, not a complete diagnosis, cause or judgement about the person reporting it. Clinical meaning depends on the wider pattern.

Source monitoring. The process of deciding where a memory, image or thought came from. Errors can make internally generated material feel heard, seen or received from elsewhere. Sleep, grief and suggestion can affect this judgement.

Predictive processing. A family of accounts in which perception combines incoming signals with expectations and updates through mismatch. It is useful here without being one settled theory of every ghost.

Agency detection. Sensitivity to signs that an intentional being caused an event. Knocks, movement and coincidence become more compelling once they are interpreted as responsive rather than mechanical.

Threat bias. Preferential attention to possible danger under uncertainty. In darkness or arousal, an ambiguous shape may be resolved towards a watcher because the cost of delay feels high. Fear then strengthens the selected interpretation.

Sleep paralysis. Temporary inability to move when falling asleep or waking, often with awareness of the room. Sensed intruders, chest pressure and vivid imagery can accompany it.

Hypnagogic experience. Imagery, sound, touch or presence arising during the transition from wakefulness into sleep. It can feel externally located while the person still knows the bedroom.

Hypnopompic experience. A perception-like event during emergence from sleep. Dream material may overlap the waking environment briefly, with or without paralysis or complete later recall. The transition may last only briefly.

REM atonia. The normal reduction of large skeletal-muscle activity during rapid eye movement sleep. Its persistence into awareness helps produce the immobility of sleep paralysis.

Ideomotor effect. Small movement produced by expectation, attention or imagery without a conscious decision to move. It helps explain table-turning, pendulums, dowsing and talking boards.

Séance. A gathering intended to communicate with spirits, often through a medium, raps, table movement, writing or voices. It combines ritual, expectation and group interaction.

Medium. A person who claims or is believed to transmit information or effects from spirits. Mediumship may involve speech, writing, physical phenomena, private consultation or public performance.

Spiritualism. A modern religious and cultural movement centred on communication with the dead, conventionally dated from the Fox sisters' 1848 rappings and developed through churches, séances and publications.

Psychical research. Organised investigation of telepathy, apparitions, mediumship and related claims. It emerged in the nineteenth century and has included believers, sceptics, field collectors and experimentalists.

Society for Psychical Research. The London organisation founded in 1882 to investigate disputed mental and spiritual phenomena through scholarly inquiry. Its members have never held one collective verdict.

Poltergeist. A reported haunting dominated by noises, object movement or physical disturbance, often around a person rather than one dead identity. Cases can include error, agency and fraud.

Revenant. A dead person believed to return bodily or visibly, especially in medieval and early modern accounts. The term often implies troublesome action rather than a faint transparent apparition.

Ancestor presence. Continued involvement of the dead within kinship, ritual and family identity. It may be protective and ordinary, showing why the English category ghost can mislead across cultures.

Electronic voice phenomenon. Speech-like content found in recordings without an agreed speaker at capture. Noise, radio intrusion, suggestion and repeated listening must be tested before identity is assigned.

Orb. A circular bright feature in a photograph, commonly produced when flash illuminates dust, moisture or insects close to the lens and renders them out of focus.

Electromagnetic field meter. An instrument measuring aspects of electric or magnetic fields within its design limits. A reading can locate wiring or transmitters; it does not measure spirits.

Infrasound. Sound below the conventional lower limit of human hearing, though strong vibration may still be sensed. Proposed haunting effects require measured exposure and controls, not a mysterious label.

Go Deeper

The social history

Owen Davies, The Haunted: A Social History of Ghosts (Palgrave Macmillan, 2007). Davies follows English ghost belief from the medieval period into modern media, with wider European and American comparisons. Read it for the part this book can only compress: how theology, class, law, print, architecture and commercial entertainment changed what a returning dead person could do. It is organised historically rather than as a catalogue of famous houses, though its geographical centre is Britain and it should not be mistaken for a global survey. Its strongest lesson is that a ghost changes when the institutions authorised to explain it change, even if reports continue.

The oldest evidence

Irving Finkel, The First Ghosts (Hodder & Stoughton, 2021). Finkel is a British Museum cuneiform specialist and works directly from Mesopotamian tablets describing troublesome dead, burial duties, ritual remedies and the practical problems of drawing a ghost. The book is learned, funny and accessible, and it makes clay tablets feel like communications from households rather than museum labels. Use it for ancient evidence and translation, while remembering that written ritual and literary texts cannot estimate what every Mesopotamian privately experienced or believed. The drawings and ritual instructions also show how physical form was already a problem for writers trying to represent an invisible dead person.

The psychology

Chris French, The Science of Weird Shit: Why Our Minds Conjure the Paranormal (MIT Press, 2024). French gives the clearest current general account of anomalistic psychology: sleep paralysis, memory, suggestion, coincidence, mediumship, near-death experience and the reasons intelligent people can form paranormal beliefs. His sceptical conclusions are explicit, but witnesses are treated as psychologically interesting rather than foolish. Read it to understand experimental method and ordinary mechanisms. The scope extends far beyond ghosts, so it supplies a toolkit rather than a social history of this one subject. Its account of why evidence feels compelling is especially useful before reading any spectacular case collection or watching a paranormal investigation.

The medium behind the spirits

Simone Natale, Supernatural Entertainments: Victorian Spiritualism and the Rise of Modern Media Culture (Penn State University Press, 2016). Natale explains Spiritualism through performance, celebrity, newspapers, photography and commodity culture without reducing every participant to a dupe or every medium to a theologian. It is the best continuation of the séance and spirit-photography story, showing why new media made invisible communication feel modern. The book is scholarly and concentrated on the nineteenth-century Atlantic world, which is precisely why its account of technology and belief has depth. Read it beside modern ghost media and the supposedly new mixture of mourning, spectacle, commerce and technical display begins to look familiar.

Notes and Sources

The Whole Thing in One Page and Why You Should Care

Experience, attribution and evidence. The manuscript separates an event, the experience built from it, the identity assigned to it and the wider causal explanation. This is an editorial model assembled from source-monitoring research, anomalistic psychology, bereavement scholarship and the history of apparition inquiry. It is not a formal diagnostic taxonomy. The separation prevents two invalid moves: treating a natural cause as evidence that no experience occurred, and treating the force of an experience as evidence that its first attribution is correct.

Robot-induced presence. Blanke and colleagues used delayed sensorimotor feedback to induce presence-like experiences in healthy participants and to intensify related symptoms in neurological patients. The study establishes that disturbed integration of movement, touch and body position can contribute to a sensed presence. It does not establish one universal mechanism for apparitions, bereavement experiences, sleep-transition visitors or culturally interpreted encounters. Bernasconi and colleagues later developed related robot-induced hallucination work in Parkinson's disease and linked vulnerability to altered sensorimotor processing in a defined clinical population. The manuscript does not generalise that result to healthy witnesses.

Faces, voices and person inference. Liu and colleagues examined behavioural and neural responses to face pareidolia, including the rapid extraction of face structure from non-face objects. Johnson, Hashtroudi and Lindsay provide the wider source-monitoring framework used to distinguish the qualities of an experience from knowledge of where it came from. Powers, Mathys and Corlett showed in a conditioning task that strong learned expectations could increase hallucination-like reports under ambiguous sensory conditions. That laboratory result concerns a defined task and selected groups. It is evidence that prior expectation can alter perception, not that all voices or apparitions are caused by one Bayesian imbalance.

Carbon monoxide. The safety wording follows current NHS guidance checked on 4 September 2026. Carbon monoxide exposure can produce headache, dizziness, nausea, weakness, tiredness, confusion, chest or muscle pain and shortness of breath, with symptoms varying by dose, duration and person. These signs are not a ghost profile and are not specific to carbon monoxide. An alarm or suspected exposure requires ordinary emergency and safety action. The manuscript avoids the popular habit of assigning carbon monoxide retrospectively to every historical haunting.

Artificial intelligence and memorial media. The text makes a narrow technical claim: enhancement and generative systems can infer missing detail, alter noise, reconstruct images and imitate voices. NIST's synthetic-content review documents image, video and synthetic-audio generation, including imitation-based systems, and treats provenance, disclosure and detection as separate technical problems. A processed result may therefore be visually or acoustically clearer while being less direct evidence of the original event. No claim is made that every enhancement invents its content or that synthetic memorials are inherently deceptive. The evidential issue is provenance: the original file, transformation steps and intended status must remain distinguishable.

Core Idea 1: report, experience and explanation

What counts as a ghost report. Historical and contemporary sources group together visual figures, voices, touches, smells, moving objects, sensed presences and traces found later in media. Davies, McCorristine, French and the early Society for Psychical Research collections show why a single object definition would hide more than it clarifies. The book uses ghost for an experience or trace attributed to a dead or disembodied person, while marking that the attribution may remain disputed.

Layered certainty. The claim ladder is methodological rather than metaphysical. A file can exist while the figure within it remains uncertain; a witness can experience a familiar voice while its external source is absent; an identity can feel exact without being independently established. Each level needs evidence suited to it. This rule follows ordinary causal inference and the source-monitoring distinction between content and origin.

Fraud and mixed cases. The text refuses both universal fraud and universal innocence. Historical mediumship contained deliberate methods, unconscious movement, performance, sincere religious practice and changing motives, sometimes within one career. Braude and Natale discuss mixed phenomena from different evaluative positions. Demonstrating fraud in one act is evidence about that act and the reliability of its producer; it is not a logical proof about every private experience or every other medium.

Core Idea 2: perception under incomplete evidence

Face pareidolia. Liu and colleagues found that face-like patterns in ordinary objects recruit parts of the face-processing system and can be detected rapidly. The manuscript uses this to explain why a face may appear before reflective judgement, not to claim that every apparition begins as three face-like marks. Vision, hearing, biological motion, threat attention and person knowledge contribute differently across cases.

Low light and peripheral vision. The description follows standard visual physiology: rod-dominated vision favours sensitivity over colour and fine spatial detail, while peripheral vision is comparatively effective at detecting change and poorer at resolving precise form. Adaptation, glare, eye movement and uncertain contrast can make faint objects unstable. The book does not claim that a figure vanishing under direct gaze diagnoses one retinal mechanism, only that a change in viewing conditions can alter a marginal percept.

Voice perception. Ambiguous audio becomes easier to parse after a phrase has been proposed. Wilson and French show how suggestion, social influence and paranormal belief can alter eyewitness reports around an ostensibly paranormal performance. Wiseman, Greening and Smith found suggestion effects in a staged séance. These studies support controls such as blind transcription and preserving the unsuggested first response. Their participants and settings do not represent all witnesses or all cultures.

Agency detection. The popular phrase hyperactive agency detection compresses several capacities into one supposed device. The manuscript retains the defensible core, that people infer persons and intentions quickly from incomplete evidence, while rejecting a single-module explanation for bereavement presences, sleep paralysis, sensed presence, table movement and media artefacts.

Core Idea 3: sensed presence and the body model

Sensed presence. A sensed presence is the compelling impression that another being is nearby without sufficient ordinary sensory evidence to identify one. Blanke and colleagues link some cases to disturbed multisensory and sensorimotor integration around the body. The experimental robot effect was variable, and the apparatus did not reproduce the personal history, fear, grief or cultural identity of a spontaneous encounter.

Neurological and extreme settings. Presence experiences have been reported in neurological illness and unusual environments as well as in healthy people. Bernasconi and colleagues supply one disease-specific mechanism study. Suedfeld and Mocellin reviewed reports from polar travel, mountaineering, solo sailing, diving and other dangerous or isolated settings, treating the sensed companion as a recurring experience under severe environmental and psychological strain rather than proof of one cause. The manuscript does not use either literature to diagnose a historical witness or imply that an unusual presence by itself establishes neurological disease.

No single ghost circuit. Face perception, voice recognition, body ownership, spatial mapping, memory, threat and social expectation are distributed functions. A person-like result can emerge from several routes. Calling this construction does not make it voluntary. Most perceptual construction is inaccessible to conscious inspection.

Core Idea 4: sleep and the night visitor

Sleep paralysis. Sharpless and Barber's systematic review found substantial variation in reported lifetime prevalence across general, student and psychiatric samples, with measurement and sampling differences. The manuscript gives no universal percentage. Cheyne, Rueffer and Newby-Clark describe recurrent clusters around sleep paralysis: sensed intruder, fear, pressure, floating or movement sensations and vivid imagery during transitions into or out of sleep.

REM atonia and awareness. During rapid eye movement sleep, skeletal muscle tone is strongly reduced through brainstem mechanisms. In sleep paralysis, awareness returns or persists while voluntary movement remains inhibited. Dream-like imagery and room perception can overlap. The event can be terrifying and is often brief, but recurring or distressing episodes may need clinical assessment. Full sleep physiology and treatment remain with Sleep in a Hurry and Dreams in a Hurry.

Cultural construction. Cheyne and colleagues used night-mare to connect neurological pattern with cultural interpretation. The Old Hag tradition in Newfoundland is one documented naming system among many. The manuscript does not diagnose every witch, demon, ancestor or alien narrative as sleep paralysis. Historical stories often lack the state, timing, mobility and recall information needed for retrospective diagnosis.

Core Idea 5: bereavement and continuing social models

Sensory and quasi-sensory experiences of the deceased. Kamp and colleagues reviewed interdisciplinary evidence covering visual, auditory, tactile, olfactory and presence experiences after bereavement. Such reports can occur without psychiatric disorder and may be comforting, neutral or distressing. Terms, methods and samples vary, so the manuscript avoids a single prevalence estimate and does not present benignity as universal.

Continuing bonds. The idea that a relationship can remain active after death does not require a claim that the deceased is physically present. Habits, predictions, internalised dialogue and social identity can persist. The book uses this as a cognitive and relational explanation for why a familiar person may be experienced under ambiguity. It does not turn grief into the cause of every apparition or offer treatment guidance.

Clinical boundary. A brief bereavement presence is not equivalent to a psychiatric syndrome. Clinical significance depends on context, frequency, distress, insight, function, associated symptoms, substance or medication effects and physical health. New confusion, neurological signs, dangerous commands or major impairment require appropriate professional help. The wording is a safety threshold, not a diagnosis.

Core Idea 6: culture gives the encounter an identity

Everywhere and always. The subtitle is treated as a claim about recurrence across many recorded societies and periods, not a literal statement that every culture has one equivalent ghost category or that every individual believes. Surviving evidence is uneven. Literate institutions preserved some voices and erased others; absence from the record can reflect preservation, translation or scholarly selection.

Mesopotamia. Finkel translates cuneiform material concerning the dead, burial obligations, troublesome ghosts and ritual responses. The evidence supports the claim that relations with the dead were treated as practical household and ritual problems in surviving texts. It cannot establish the private beliefs of every Mesopotamian or convert ritual prescriptions into prevalence data.

Medieval Europe. Schmitt analyses the living-dead relationship, monastic narratives, dreams, purgatory and the social uses of revenant accounts. The manuscript's moral-messenger model is a compression of varied periods and genres. It does not make every medieval apparition a purgatorial soul or treat church teaching and local narrative as identical.

Medieval China. Teiser's study of the Ghost Festival traces Buddhist, ritual, economic and kinship dimensions in medieval China. The comparison with European purgatory is functional, not theological: both could link the condition of the dead to duties performed by the living. Hungry ghost is not used as a translation for every Chinese dead being.

Reformation England. Marshall shows how rejection of purgatory changed the authorised status of the dead without removing concern for them. Apparitions could be reassigned to demons, deception, melancholy, dreams or providential signs. The book confines this claim to its English and northern European setting and does not turn the Reformation into a universal sequence.

Translation and global media. The English word ghost often absorbs ancestors, revenants, spirits, demons and local beings with different relations to body, place, kinship and ritual. Similarity may reflect shared human capacities, historical contact, translation choices or modern media circulation. The text therefore uses cross-cultural recurrence as evidence for a recurring problem, not for one hidden universal creature.

Core Idea 7: evidence machines

Spirit photography. Kaplan and Natale support the account of William Mumler, the authority attached to photography and the 1869 New York proceedings. Period photographers demonstrated ordinary routes to comparable effects, and P. T. Barnum supplied a fabricated Lincoln spirit image. Mumler was discharged because the prosecution did not prove its case against his process. That legal result neither authenticates every plate nor proves one method was used in all of them.

Photographic mechanisms. Reused or incompletely cleaned plates, double exposure, prepared negatives, staged figures, focus and printing can produce pale secondary forms. The list gives possible mechanisms known within early photography. It is not a retrospective finding about every surviving Mumler photograph.

Electronic voice phenomena and digital traces. French supplies the main modern synthesis for ambiguous audio, expectation, selective replay and paranormal interpretation. A recorder preserves acoustic pressure after transformation through microphone, gain, filtering, compression and software. It does not label the source. The recommended blind-transcription control tests whether a phrase is independently recoverable before the proposed words organise perception.

Orbs, electromagnetic meters and thermal cameras. Dust or insects close to a flash or infrared illuminator can appear as bright, out-of-focus discs. Electromagnetic meters respond to fields from wiring, appliances, transmitters and other physical sources within their range. Thermal cameras map infrared radiation and infer surface temperatures under model assumptions. No instrument listed is designed to detect identity or survival. The claim is about measurement compatibility: a reading cannot answer a question the instrument does not operationalise.

Artificial intelligence. Enhancement can denoise, interpolate or reconstruct by applying learned statistical expectations. Generative systems can create faces, motion and voices. These tools can serve restoration, access, art or memorialisation. For evidence, each transformation must be disclosed because inferred detail is not an untouched observation.

Historical sequence

Ancient households. The ancient opening relies on Finkel for Mesopotamian evidence and on Pliny the Younger's Letter 7.27 for the Athenian haunted-house story. Pliny presents the case within a letter asking whether apparitions have real existence and independent power. The manuscript treats it as evidence for Roman storytelling and argument, not as an archaeologically verified excavation report.

Purgatory and moral work. Schmitt supports the account of medieval European ghosts requesting prayer, restitution or ritual action. The text also marks the diversity of corpse-like revenants, household beings and demonic interpretations rather than presenting a neat church system.

The Fox sisters. Late March 1848 at Hydesville is retained as the conventional founding episode of modern American Spiritualism. Braude and Natale support the movement's religious, political, performative and commercial development. Maggie Fox's 1888 confession and later recantation are included because both shaped the public record. The confession bears directly on the sisters' performances, while its use as a verdict on all Spiritualism would exceed its scope.

Women and Spiritualism. Braude documents the close relation between American Spiritualism and women's public religious and reform activity. Mediumship could grant a speaking role while framing the speaker as a passive vehicle. This was an opportunity and a constraint, not a claim that all women mediums were feminists or socially liberated.

Table-turning and ideomotor action. Faraday's 1853 investigation used layered indicators to determine whether the hands preceded the table. His report supports the causal sequence and the reduction of movement when sitters could see the pressure relation. Ideomotor action is used for movement shaped by ideas or expectations without a consciously felt decision. It does not mean the muscles move without physical force or that every séance effect shares one mechanism.

Psychical research. The Society for Psychical Research was formed in London in 1882 to investigate disputed mental and spiritual phenomena through organised inquiry. Gurney, Myers and Podmore's Phantasms of the Living appeared in 1886. The 1894 Census of Hallucinations reported more than seventeen thousand responses to a waking-hallucination question. The sample was assembled through a distributed network rather than modern probability sampling, and the categories combined several experiences. The material is valuable for history and phenomenology; its numerical results should not be converted into a current population estimate.

Haunted sites and expectation. Wiseman and colleagues' site work examined reports at Hampton Court Palace and the South Bridge Vaults and considered environmental and psychological correlates. Such studies preserve place but cannot hold every visitor, sound, cue and environmental condition constant. The manuscript treats clustering as a reason to investigate a location, not as identification of a ghost or of one physical cause.

The Haunt project. French and colleagues placed seventy-nine participants for fifty minutes in a purpose-built chamber and manipulated infrasound and complex electromagnetic fields. Reports of unusual sensations did not vary as predicted by active condition and were associated more strongly with individual differences. The result challenges the claim that those exposures reliably create haunted-room experiences under the tested conditions. It does not show that all environmental factors are irrelevant.

Weak complex magnetic fields. Granqvist and colleagues' double-blind study found suggestibility, rather than field activation, predicted mystical and presence experiences in their sample. Persinger and colleagues published a critical response and reported different results under their procedures. The manuscript retains the cautious conclusion: no one weak-field setup has become a reliable general mechanism for haunting, and disputes about apparatus, procedure and replication remain relevant.

Suggestion and witness interaction. Wiseman, Greening and Smith found that verbal suggestion in a staged séance altered some reports, including reports that a stationary table had moved. Wilson and French used conjuring to examine suggestion, social influence and memory around an apparently paranormal event. These studies justify recording accounts separately and before discussion. They do not imply that groups always converge or that all corroboration is contamination.

What People Get Wrong and Use It

Hallucination is not a diagnosis. The term describes perception-like experience without a corresponding external stimulus, though clinical definitions and boundaries vary. Sleep transitions, bereavement, neurological conditions, sensory loss, substances, fever and psychiatric illness can all be relevant. The book avoids diagnosing from content alone and avoids telling readers to ignore distress or impairment.

Environmental investigation. Lighting, reflection, airflow, plumbing, acoustics, vibration, animals, neighbouring activity, electrical equipment and pollutants are offered as testable classes, not as a checklist guaranteed to solve a case. A failed reproduction leaves the cause unknown. It does not make the excluded cause impossible or award the event to survival.

Blinded mediumship research. Sarraf, Woodley of Menie and Tressoldi's 2021 meta-analysis reported above-chance accuracy across selected modern mediumship experiments and concluded that some mediums may obtain information through unknown means. The literature is disputed over study selection, blinding, judging, analytical flexibility, publication processes and what an anomalous score would identify. Even an accepted information anomaly would not by itself distinguish a surviving deceased personality from information leakage, living-agent mechanisms or another unknown cause. Braude argues that some historical and mediumistic evidence favours survival while taking rival explanations seriously. The manuscript's sentence that no controlled body of evidence has established surviving personality as the cause is therefore an evidential judgement, not a claim that no positive study exists or that survival is metaphysically impossible.

Standards of proof. Survival is a stronger proposition than unexplained experience. A useful test requires secure target information, exclusion of ordinary access, pre-set scoring, independent and blinded judging, full accounting of misses, replication and a route for distinguishing the dead-person hypothesis from rival sources. Few spontaneous apparition cases retain enough evidence for such a test after the fact.

Respect and meaning. Kamp and colleagues support the non-pathologising treatment of many bereavement experiences. Historical sources show that ghost accounts can carry ritual, moral and social functions. Neither line of evidence certifies an external dead agent. The practical language deliberately allows personal, cultural or religious meaning to remain while public causal claims receive separate scrutiny.

Terms and Go Deeper

Terminology. Apparition, hallucination, illusion, pareidolia, sensed presence, sleep paralysis, source monitoring, ideomotor action and the instrument terms are defined for general reading rather than clinical coding or engineering specification. Revenant, ancestor, hungry ghost, purgatory, Spiritualism, séance, medium and psychical research remain historically and culturally situated.

Further reading. Davies is the broad social history used for institutional change. Finkel provides direct access to ancient Mesopotamian material. French supplies the current sceptical psychological synthesis and experimental toolkit. Natale supplies the deep account of Spiritualism as religion, entertainment and media culture. Together they extend history, primary evidence, mechanism and technology without pretending to settle the metaphysical question by recommendation.

Bibliography

Primary and original evidence

Faraday, Michael. “Experimental Investigation of Table-Moving.” The Athenaeum, 2 July 1853, 801-803.

Gurney, Edmund, Frederic W. H. Myers, and Frank Podmore. Phantasms of the Living. 2 vols. London: Rooms of the Society for Psychical Research and Trübner & Co., 1886.

Pliny the Younger. Letters, Volume I: Books 1-7. Translated by Betty Radice. Loeb Classical Library 55. Cambridge, MA: Harvard University Press, 1969.

Sidgwick, Henry, Alice Johnson, Frederic W. H. Myers, Frank Podmore, and Eleanor Mildred Sidgwick. “Report on the Census of Hallucinations.” Proceedings of the Society for Psychical Research 10 (1894): 25-422.

Society for Psychical Research. “A Brief History of the Society for Psychical Research.” Institutional history. Current page checked 4 September 2026.

History, culture and interpretation

Braude, Ann. Radical Spirits: Spiritualism and Women's Rights in Nineteenth-Century America. 2nd ed. Bloomington: Indiana University Press, 2001.

Braude, Stephen E. Immortal Remains: The Evidence for Life After Death. Lanham, MD: Rowman & Littlefield, 2003.

Davies, Owen. The Haunted: A Social History of Ghosts. Basingstoke: Palgrave Macmillan, 2007.

Finkel, Irving. The First Ghosts. London: Hodder & Stoughton, 2021.

Kaplan, Louis. The Strange Case of William Mumler, Spirit Photographer. Minneapolis: University of Minnesota Press, 2008.

Marshall, Peter. Beliefs and the Dead in Reformation England. Oxford: Oxford University Press, 2002.

McCorristine, Shane. Spectres of the Self: Thinking about Ghosts and Ghost-Seeing in England, 1750-1920. Cambridge: Cambridge University Press, 2010.

Natale, Simone. Supernatural Entertainments: Victorian Spiritualism and the Rise of Modern Media Culture. University Park, PA: Pennsylvania State University Press, 2016.

Schmitt, Jean-Claude. Ghosts in the Middle Ages: The Living and the Dead in Medieval Society. Translated by Teresa Lavender Fagan. Chicago: University of Chicago Press, 1998.

Teiser, Stephen F. The Ghost Festival in Medieval China. Princeton, NJ: Princeton University Press, 1988.

Perception, presence, sleep and bereavement

Bernasconi, Fosco, Elodie Blondiaux, Jevita Potheegadoo, et al. “Robot-Induced Hallucinations in Parkinson's Disease Depend on Altered Sensorimotor Processing in Fronto-Temporal Network.” Science Translational Medicine 13, no. 591 (2021): eabc8362. DOI: 10.1126/scitranslmed.abc8362.

Blanke, Olaf, Polona Pozeg, Masayuki Hara, et al. “Neurological and Robot-Controlled Induction of an Apparition.” Current Biology 24, no. 22 (2014): 2681-2686. DOI: 10.1016/j.cub.2014.09.049.

Cheyne, J. Allan, Steve D. Rueffer, and Ian R. Newby-Clark. “Hypnagogic and Hypnopompic Hallucinations during Sleep Paralysis: Neurological and Cultural Construction of the Night-Mare.” Consciousness and Cognition 8, no. 3 (1999): 319-337. DOI: 10.1006/ccog.1999.0404.

Cheyne, J. Allan, Ian R. Newby-Clark, and Steve D. Rueffer. “Relations among Hypnagogic and Hypnopompic Experiences Associated with Sleep Paralysis.” Journal of Sleep Research 8, no. 4 (1999): 313-317. DOI: 10.1046/j.1365-2869.1999.00165.x.

Johnson, Marcia K., Shahin Hashtroudi, and D. Stephen Lindsay. “Source Monitoring.” Psychological Bulletin 114, no. 1 (1993): 3-28. DOI: 10.1037/0033-2909.114.1.3.

Kamp, Karina Stengaard, Edith Maria Steffen, Ben Alderson-Day, et al. “Sensory and Quasi-Sensory Experiences of the Deceased in Bereavement: An Interdisciplinary and Integrative Review.” Schizophrenia Bulletin 46, no. 6 (2020): 1367-1381. DOI: 10.1093/schbul/sbaa113.

Liu, Jiangang, Jun Li, Lu Feng, Ling Li, Jie Tian, and Kang Lee. “Seeing Jesus in Toast: Neural and Behavioral Correlates of Face Pareidolia.” Cortex 53 (2014): 60-77. DOI: 10.1016/j.cortex.2014.01.013.

Powers, Albert R. III, Christoph Mathys, and Philip R. Corlett. “Pavlovian Conditioning-Induced Hallucinations Result from Overweighting of Perceptual Priors.” Science 357, no. 6351 (2017): 596-600. DOI: 10.1126/science.aan3458.

Sharpless, Brian A., and Jacques P. Barber. “Lifetime Prevalence Rates of Sleep Paralysis: A Systematic Review.” Sleep Medicine Reviews 15, no. 5 (2011): 311-315. DOI: 10.1016/j.smrv.2011.01.007.

Suedfeld, Peter, and Jane S. P. Mocellin. “The ‘Sensed Presence’ in Unusual Environments.” Environment and Behavior 19, no. 1 (1987): 33-52. DOI: 10.1177/0013916587191002.

Haunted places, suggestion and disputed physical mechanisms

French, Christopher C. The Science of Weird Shit: Why Our Minds Conjure the Paranormal. Cambridge, MA: MIT Press, 2024.

French, Christopher C., Usman Haque, Rosie Bunton-Stasyshyn, and Rob Davis. “The ‘Haunt’ Project: An Attempt to Build a ‘Haunted’ Room by Manipulating Complex Electromagnetic Fields and Infrasound.” Cortex 45, no. 5 (2009): 619-629. DOI: 10.1016/j.cortex.2007.10.011.

Granqvist, Pehr, Mats Fredrikson, Patrik Unge, et al. “Sensed Presence and Mystical Experiences Are Predicted by Suggestibility, Not by the Application of Transcranial Weak Complex Magnetic Fields.” Neuroscience Letters 379, no. 1 (2005): 1-6. DOI: 10.1016/j.neulet.2004.10.057.

Persinger, Michael A., and Stanley A. Koren. “A Response to Granqvist et al. ‘Sensed Presence and Mystical Experiences Are Predicted by Suggestibility, Not by the Application of Transcranial Weak Magnetic Fields’.” Neuroscience Letters 380, no. 3 (2005): 346-347. DOI: 10.1016/j.neulet.2005.03.060.

Wiseman, Richard, Caroline Watt, Paul Stevens, Emma Greening, and Ciarán O'Keeffe. “An Investigation into Alleged ‘Hauntings’.” British Journal of Psychology 94, no. 2 (2003): 195-211. DOI: 10.1348/000712603321661886.

Wiseman, Richard, Emma Greening, and Matthew Smith. “Belief in the Paranormal and Suggestion in the Séance Room.” British Journal of Psychology 94, no. 3 (2003): 285-297. DOI: 10.1348/000712603767876235.

Wilson, Krissy, and Christopher C. French. “Magic and Memory: Using Conjuring to Explore the Effects of Suggestion, Social Influence and Paranormal Belief on Eyewitness Testimony for an Ostensibly Paranormal Event.” Frontiers in Psychology 5 (2014): 1289. DOI: 10.3389/fpsyg.2014.01289.

Mediumship and the survival dispute

Sarraf, Matthew, Michael A. Woodley of Menie, and Patrizio Tressoldi. “Anomalous Information Reception by Mediums: A Meta-Analysis of the Scientific Evidence.” Explore 17, no. 5 (2021): 396-402. DOI: 10.1016/j.explore.2020.04.002.

Current technical and safety guidance

National Institute of Standards and Technology. Reducing Risks Posed by Synthetic Content: An Overview of Technical Approaches to Digital Content Transparency. NIST AI 100-4. Gaithersburg, MD: National Institute of Standards and Technology, November 2024. DOI: 10.6028/NIST.AI.100-4.

National Health Service. “Carbon Monoxide Poisoning.” NHS conditions guidance. Current page checked 4 September 2026.