Overview

An apparitional experience is when a person, while fully awake, perceives a living being or occasionally an object with no known physical cause. Researchers use the neutral term apparitional experience rather than ghost. The word ghost assumes the person has survived death. Several other explanations are still possible: telepathic hallucination, psychological projection, environmental triggers, or simple misperception.

The sub-type that has drawn the most scientific attention is the crisis apparition. In these cases, someone perceives the figure of a friend or relative. Only later do they learn that person was dying, seriously ill, or in a life-threatening accident at that exact moment and at a distant location. If the timing cannot be explained by chance, prior knowledge, or any normal sensory cue, researchers suspect some unusual channel of information was at work. The most common hypothesis is telepathy. This idea was central to the founding mission of the Society for Psychical Research (SPR), set up in London in 1882. Its early leaders, Edmund Gurney, Frederic W. H. Myers, and Frank Podmore, spent years investigating it.

Research methods

Apparitional experiences happen on their own. They cannot be triggered in a lab on demand. So researchers have relied on three main tools: collecting cases, running large surveys, and comparing results against baseline rates.

The earliest approach, used in Phantasms of the Living (1886), involved asking the public to submit written first-hand accounts. Each report then went through several checks: face-to-face interviews with the person who had the experience, statements from family members or bystanders, and cross-checks of dates against death records. The goal was to filter out fraud, faulty memory, and pure coincidence before treating a case as meaningful.

The second approach was the large-scale population survey. Researchers asked a representative sample of people whether they had ever seen an apparition while fully awake. By estimating how common such experiences are in the general population, and then comparing that to the rate of experiences that coincided with the death of the person seen, investigators built a probability argument. If death-coincident hallucinations happened far more often than chance would predict, that was treated as evidence for something unusual.

Key results and surveys

Phantasms of the Living (Gurney, Myers, and Podmore, 1886) catalogued more than 700 carefully checked cases.^[1] More than half involved an apparition at or near the moment of the perceived person’s death. The authors argued this concentration was far too high to be coincidental.

The follow-up Census of Hallucinations, published in the Proceedings of the SPR in 1894, surveyed 17,000 members of the public.^[2] Of those, 1,684 confirmed having a waking apparitional experience. The investigators concluded that the proportion of death-coincident cases exceeded what chance would predict by enough to rule out random coincidence as a full explanation.

Later community surveys across the twentieth century broadly confirmed these findings. Louisa Rhine collected roughly 8,000 case reports from mid-century onwards. Haraldsson ran national surveys in Iceland, Great Britain, Sweden, and the United States. Both found that apparitional experiences are widespread and not limited to people with psychiatric conditions. One consistent detail across surveys: most reports are visual. This is unlike the mainly auditory hallucinations linked to schizophrenia. Collective cases, where multiple witnesses report the same figure, appear in roughly one quarter of survey returns where a second observer was present.^[4] That proportion is hard to explain if the experience is purely internal to one person’s mind.

Variations and related paradigms

Apparition research overlaps with several nearby areas of parapsychology. Mental mediumship and its research tradition ask whether information about the deceased can be transmitted in unusual ways. The same historical cases appear in both fields. Near-death experiences often include apparitional encounters with deceased relatives, making them a natural companion area. Reincarnation cases research, developed especially by Ian Stevenson, sometimes includes apparition-like announcing dreams and recognitions that suggest a continuing identity after death.

Within the broader apparition category, researchers distinguish several sub-types. Crisis apparitions coincide with death or danger. Post-mortem apparitions (ghosts) recur at a location after the person has died. Collective apparitions are seen by more than one witness. Reciprocal apparitions are cases where both the person seen and the person seeing them report an unusual experience at the same time. Deathbed visions, seen by the dying person themselves, form another sub-category. These were studied especially by Karlis Osis. Widowhood hallucinations were documented by Welsh GP W. Dewi Rees in a landmark 1971 community study of bereaved spouses.^[6] His work shows how closely related post-bereavement experiences are to the classic crisis-apparition category. All of these sub-types also connect to research on clairvoyance and precognition, since some apparitional cases carry information that appears to have been acquired by unusual means.

Skeptical critiques

Critique 1: Selection and reporting bias inflate the coincidence rate

Mainstream critics argue that people are far more likely to remember and report a hallucination if it was later followed by news of a death. Experiences that were not followed by a death get forgotten. This makes the apparent coincidence rate look much higher than it really is.

Proponents point to cases where corroborating documents, such as letters, diaries, or third-party witness statements, were secured before news of the crisis reached the person who had the experience. This narrows the window for after-the-fact reshaping of the story.

This is a strong objection for cases relying solely on memory. It is weaker for cases with pre-crisis documentation.

Critique 2: Memory contamination distorts the timing and details

Accounts are typically recorded well after the event. Retrospective reconstruction may unconsciously reshape the timing or details of the experience to fit the crisis, even without any intent to deceive.

Researchers acknowledge this risk. The strongest cases in the literature are those where written records were made before the news arrived, limiting how much retrospective reshaping could have occurred.

A genuine concern for most case collections. It does not apply equally to all cases.

Critique 3: Prior knowledge provides a mundane explanation

The person who had the experience may have been subliminally aware of the other person’s deteriorating health without consciously registering it. This could produce an apparently meaningful experience with a perfectly ordinary cause.

Investigators attempt to screen for this by checking whether the percipient had any recent contact with or news about the person seen. Cases where no such contact can be established are treated as stronger evidence.

A valid concern that is hard to rule out completely in any individual case.

Critique 4: The statistical argument rests on uncertain base rates

Gurney’s and Sidgwick’s calculations depended on estimates of how common waking hallucinations are in the general population. Those estimates are themselves uncertain. Small errors in the base rate can substantially change the probability estimates.

The Census of Hallucinations was designed partly to address this by providing a much larger sample. Researchers in the late twentieth century re-examined the historical datasets with modern statistical tools and found the death-coincidence excess remains difficult to dismiss on sampling grounds alone.

A serious technical objection. It has been partially addressed but not fully resolved.

Critique 5: The evidence is entirely anecdotal and cannot be reproduced in the lab

The entire body of evidence rests on case reports. These cannot be independently reproduced under controlled conditions. A phenomenon that produces no consistent laboratory result after more than a century of investigation does not meet the reproducibility standard expected of a scientific claim.

Proponents argue that spontaneous experiences by their nature cannot be induced on demand. They point to the consistency of findings across independent surveys in multiple countries as a form of replication, even if it is not laboratory replication.

The strongest standing objection. Cross-cultural consistency is suggestive but does not substitute for controlled experimental evidence.

Responses and evaluation

Proponents take the bias objections seriously. Their strongest response is to emphasise cases where corroborating documents were secured before news of the crisis arrived. This substantially narrows the window for confabulation (unconsciously filling in or reshaping a memory).

Collective cases are harder to dismiss. When multiple independent witnesses report the same figure without prior communication, a purely personal memory error cannot explain all of them. Gurney proposed that group hallucinations could spread telepathically from one mind to another. Myers held that collective apparitions were non-physical but objectively real entities. Tyrrell’s model placed the construction in a shared unconscious layer of mind shared by the witnesses.^[5]

Most researchers today treat the anecdotal record as suggestive but not conclusive. They also acknowledge that these experiences may have more than one cause. Some cases may involve genuine unusual cognition. Others may reflect misperception or faulty memory. The overlap with widowhood hallucinations is relevant here.^[6] Those experiences are statistically common, clinically benign, and found in every culture studied. That supports the view that the experiential category itself is real, even if the ultimate cause is still debated. Ongoing work in near-death experience research and mediumship research continues to produce case material bearing on the same underlying questions about consciousness and survival.^[7]