Overview

Mental mediumship research sits at the crossroads of parapsychology, consciousness science, and bereavement studies. A medium claims to receive information from a deceased person and pass it on to a living recipient, called the sitter. The Society for Psychical Research (SPR) began studying mediums scientifically as early as 1882. One of their most thorough subjects was Leonora Piper, a medium who was investigated over many years.^[8] By the 1930s, researchers grew frustrated. They could not determine whether a medium’s accurate information came from a deceased person, from psychic ability, or from fraud. The field went into a long decline.^[9]

Modern lab research restarted around 2001. Landmark double-blind studies by Roy and Robertson and by Schwartz and Russek launched what researchers called the “resurrection” of mediumship science. Since then, studies have included British Spiritualist mediums, Brazilian and Puerto Rican Spiritist mediums, and secular American mediums who have no religious affiliation.^[1] The main challenge is cutting off every normal way the medium could get the information. To do this, protocols evolved from single-blind to double-, triple-, and quintuple-blind designs.

Two goals drive current work. Proof-focused research asks: can mediums give accurate, specific information about the deceased when all normal explanations are ruled out? Process-focused research asks: how do mediums get that information? This second line looks at altered states of consciousness, EEG brain signatures, and the psychology of mediums themselves.^[7]

Key results and meta-analyses

The most important studies since 2001 are described below.

Roy and Robertson (2001, 2004). Scottish astronomer Archie Roy and researcher Tricia Robertson designed a series of experiments to test a key skeptical claim: that medium statements are too vague to be meaningful. They ran experiments under triple-to-quadruple blind conditions. Intended recipients accepted significantly more statements as personally relevant than non-recipients did. The tightest protocol produced odds against chance of about one million to one.^[5]

Beischel and Schwartz (2007). Julie Beischel and Gary Schwartz at the University of Arizona published a triple-blind study. Mediums read absent sitters through a go-between experimenter, which cut off most normal ways of getting information. The results suggested that some mediums can receive accurate information about deceased individuals through no known normal means. The design could not tell whether this came from a surviving consciousness or from powerful psychic ability in living people.^[6]

Beischel et al. (2015) — Quintuple-blind replication. Twenty Windbridge Certified Research Mediums took part in over 86 readings. The study used five separate layers of blinding to prevent any normal information transfer. Here is what those layers mean in practice: the medium did not know the sitter, the sitter was not present, the experimenter running the session did not know which deceased person was being read, a separate analyst scored the results, and the analyst did not know which reading matched which sitter. All analyses returned statistically significant results. The p-values ranged from 0.05 to 0.0001. A p-value of 0.05 means: if there were no real effect, results this strong would occur by chance only about 5 times in 100 tries. A p-value of 0.0001 means only about 1 time in 10,000. The study replicated prior findings and ruled out cold reading, rater bias, experimenter cueing, and fraud.^[4]

Delorme et al. (2013) — EEG study. Arnaud Delorme and colleagues recorded brain activity (EEG) from six Windbridge Certified Research Mediums during masked readings. First, sitter-rated accuracy was again above chance. More importantly, the brain state during claimed communication with the deceased was distinct. It did not match brain activity during ordinary memory, perception, or imagination. This raises questions about what is actually happening in the medium’s mind during a reading.^[3]

Sarraf, Woodley of Menie, and Tressoldi (2020) — Meta-analysis. This was the first published meta-analysis of modern accuracy research. It pooled 18 experiments from 14 papers published between 2001 and 2019. Both standard and Bayesian statistical models found an overall effect size of about 0.18 above chance (95% confidence interval: 0.12–0.25). An effect size of 0.18 is small but consistently detectable: it means the average medium score was about one-fifth of a standard deviation above what chance predicts. The results survived two tests for publication bias, meaning the positive signal is not simply a result of studies with negative results going unpublished.^[2]

Taken together, these studies point to a replicable anomaly. Individual medium performance varies a lot. This suggests that mediums who can repeatedly show accuracy under strict scientific controls may be relatively rare.^[10] Mental mediumship research shares some features with Ganzfeld research and remote viewing work. All three use blinded judging and all three show small but replicable effect sizes across meta-analyses.

Skeptical critiques

Critique 1: Early studies had serious flaws in how they were run.

Wiseman and O’Keeffe (2005); Hyman (2003)^[11][12]

Studies by Schwartz and colleagues from the early 2000s drew sustained criticism. Critics pointed to poor control of sensory leakage, insufficient experimenter blinding, inappropriate statistics, and inadequate descriptions of methods and results.^[11] The Kelly and Arcangel (2011) study was also criticized. It lacked pre-screening of participants. Its blinding was incomplete: mediums were given photographs of the deceased, which could allow visual inference rather than genuine psychic reception.

Evidence strength: These criticisms apply mainly to early studies. Later quintuple-blind protocols directly addressed each of these concerns.^[4]

Critique 2: Accurate readings can be explained by cold reading, warm reading, or hot reading.

Hyman (2003)^[12]

Skeptics point to three normal explanations for apparently accurate readings. Cold reading means inferring information from a sitter’s appearance, dress, and behavior. Warm reading means using knowledge of common human experiences that apply to almost anyone. Hot reading means finding out information about the sitter or deceased in advance through research.^[12] These are real concerns for uncontrolled readings. However, the quintuple-blind protocols were specifically designed to make all three impossible.

Evidence strength: Valid for uncontrolled settings. The tightest modern protocols rule out all three by design.^[4]

Critique 3: Even if the results are real, they might come from psychic ability in living people, not from the deceased.

Beischel and Schwartz (2007)^[6]

Even when fraud and cold reading are ruled out, a major interpretive problem remains. The survival-of-consciousness hypothesis says the information comes from a deceased person’s continuing mind. The super-psi hypothesis says the medium uses extraordinarily powerful telepathic or clairvoyant abilities to gather the information from living sources. Some philosophers argue these two explanations are impossible to tell apart. If that is true, the survival hypothesis cannot be falsified in this context.^[6]

Evidence strength: This is a genuine and unresolved philosophical problem. No current experimental design can fully separate the two explanations.

Critique 4: Too few independent research groups have tested this.

Beischel (2018)^[1]

Fewer than ten groups have published original peer-reviewed research on modern mental mediums since 2007.^[1] This limits independent replication. Funding is scarce, and mainstream science still stigmatizes mediumship research. Both factors slow the growth of the field.

Evidence strength: A real structural limitation. The meta-analysis helps by pooling results across groups, but more independent replication is needed.

Responses and evaluation

Researchers have responded to methodological criticisms by steadily tightening controls. The move from single-blind to quintuple-blind protocols at the Windbridge Research Center directly addresses concerns about sensory leakage, experimenter cueing, and rater bias. Windbridge Certified Research Mediums must complete an eight-step certification process before taking part in accuracy research. This helps ensure that participants can deliver consistent readings under strict lab conditions.^[13]

On the super-psi debate, process-focused researchers take a different approach. Instead of trying to rule out super-psi through accuracy tests alone, they study the neuroscience and phenomenology of mediumship itself. If a medium’s brain shows a distinctive pattern during a reading, that is independent evidence about what is happening, regardless of where the information comes from. The EEG finding that communication with the deceased corresponds to a brain state distinct from imagination or ordinary memory is one such data point.^[3] Survey studies also show that mediums describe reading for the deceased differently from doing psychic readings for the living. The two experiences feel qualitatively different to the mediums themselves.^[14]

Research has also moved into practical applications. Studies on after-death communication and bereavement suggest that mediumship readings may help people who are grieving. This opens a socially relevant line of inquiry that stands apart from the metaphysical debate about survival.^[7] For a deeper look at the underlying phenomenon, see the Mental Mediumship phenomena page. For comparisons with other anomalous information-reception research, see clairvoyance research, telepathy research, and precognition research.