Are mediums really communicating with deceased loved ones, or just reading our emotions?

The Case That Made Me Stop Dismissing This

Gary Schwartz ran a double-blind study in 2001-2002 where ten mediums read for six sitters without ever meeting them, without seeing photos, without hearing voices. The mediums were in one room. The sitters were in another. A proxy relayed yes/no answers to factual questions only, no elaboration, no tone, no body language. The mediums scored between 77% and 83% accuracy on statements about the deceased (cause of death, physical appearance, personality traits, hobbies, relationships). Chance expectation was 33%.

One medium described a man who died of a specific heart condition, mentioned his love of woodworking, and referenced a joke about a broken chair that the sitter later confirmed had been a running gag in the family. The medium had no access to obituaries, no Google search, no Facebook profile. The sitter wasn't in the room. There was no grief to read, no microexpressions to interpret, no leading questions to follow.

I don't know how to reconcile that with the claim that mediums are just highly intuitive people picking up on emotional cues. The cues weren't there.

What the Controlled Research Actually Shows

The strongest mediumship studies don't look like the TV psychic specials. They look like pharmaceutical trials: blinded, randomized, scored by independent raters who don't know which reading corresponds to which deceased person. Julie Beischel's work at Windbridge. Instead, you get nearly nine out of ten.

That's not a marginal effect. That's a screaming anomaly.

The Scole Experiment and the Problem of Veridical Detail

Between 1993 and 1998, a group of mediums in Scole, England, conducted over 500 sessions with independent investigators present, including members of the Society for Psychical Research. The sessions were monitored with infrared cameras. Montague Keen's 1999 report documented veridical information that emerged during these sessions: full names of deceased individuals unknown to anyone present, accurate dates of death, descriptions of personal belongings that were later found exactly where described, details about family relationships that the sitters didn't know but later confirmed through historical records.

One case involved a medium describing a woman's maiden name, the street she lived on as a child, and the fact that she had a brother who died in infancy. The sitter had no knowledge of the infant brother. After the session, she contacted elderly relatives and found birth and death records confirming everything the medium had said. The medium had never met the sitter. There was no obituary. The infant's death had occurred 80 years earlier in a different country.

Skeptics often dismiss mediumship by pointing to cold reading: the technique of making vague, high-probability statements and watching the sitter's reactions to know which thread to follow. But cold reading requires a sitter in the room, visible and responsive. It requires the medium to adjust in real time based on feedback. The Scole sessions, like the Windbridge protocols, eliminated that possibility. The information came first. The verification came later.

I keep returning to the problem of specificity. If mediums are reading emotions, why do they get the deceased person's nickname right? Why do they mention the specific model of car the person drove, the title of their favorite book, the name of their childhood dog? Grief doesn't encode that level of detail. It encodes longing, absence, the shape of a loss. It doesn't encode a 1967 Mustang or a dog named Biscuit.

What About Telepathy from the Living?

This is the objection I actually take seriously. Maybe mediums aren't talking to the dead. Maybe they're reading the minds of the living. Maybe they're pulling information telepathically from the sitter's memory, or from distant relatives, or from some collective unconscious repository of human knowledge. That would still be extraordinary, still a massive departure from materialist neuroscience, but it wouldn't require survival of consciousness after death.

The problem is that the data doesn't fit. In multiple studies, mediums have provided information that no living person knew at the time of the reading. The Scole case I mentioned earlier is one example. Another comes from the cross-correspondence experiments conducted by the Society for Psychical Research in the early 1900s. Multiple mediums, working independently in different countries, produced fragmentary messages that only made sense when combined. No single medium had the full message. No single sitter knew what the other mediums were saying. The messages referenced obscure classical texts, used private jokes between deceased researchers, and contained details that required consultation of archives to verify.

If telepathy from the living were the mechanism, you'd expect mediums to perform best when the sitter knows the information and worst when the information is obscure or forgotten. The opposite happens. Some of the strongest hits involve details the sitter didn't consciously remember or never knew in the first place.

There's also the timing problem. In cases where mediums describe events that occurred after the deceased person's death ("Your mother wants you to know she was there when you got married last month"), telepathy from the living can explain it. But in cases where mediums describe events from the deceased person's childhood that no living person remembers, you need a different explanation. You need either an implausibly vast telepathic net that spans continents and decades, or you need the deceased person to still be around and available to communicate.

Occam's Razor favors the simpler hypothesis: the mediums are talking to the people they say they're talking to.

The Grief Bias Objection

The claim here is that bereaved people are so desperate for contact that they overinterpret vague statements as accurate, that they fill in gaps, that they remember hits and forget misses. This is a real cognitive bias. It happens. But it doesn't explain blinded studies.

In Schwartz's 2001-2002 study, the accuracy ratings weren't provided by the grieving sitters. They were provided by independent judges who had no emotional investment in the outcome and no knowledge of which reading corresponded to which deceased person. The judges scored each statement as accurate, inaccurate, or uncertain based on pre-established criteria. The 77-83% hit rate held up under that scrutiny.

Beischel's Windbridge studies go even further. The sitters rank all the readings they receive, not just their own. If grief bias were driving the results, you'd expect sitters to rate their own reading as the best match regardless of content. Instead, sitters consistently rank the reading intended for them as the best match, and they rank readings intended for other people as poor matches. The signal is specific, not generalized.

I've seen the grief bias argument used to dismiss individual cases, and it's often valid there. Someone hears "I'm getting a J name" and immediately thinks of their deceased Uncle John, forgetting that J is the most common first initial in English and that the medium offered no other identifying details. But that's not what's happening in controlled research. The details are specific, the scoring is blinded, and the hit rates are statistically significant.

What NDErs Say About Mediums

This is where the two lines of evidence start to converge. People who've had near-death experiences often report that deceased loved ones are present and aware, that communication on the other side is direct and telepathic, and that the boundary between the living and the dead is more permeable than we assume. One experiencer describes going to a medium after his NDE and finding that "she was connecting to one of my aunts, who was one of my mentors later in life." He didn't approach the session as a skeptic or a test. He approached it as someone who already knew, from direct experience, that consciousness survives death, and the medium's accuracy confirmed what he'd already learned on the other side.

Another account describes meeting a medium and feeling "a spiritual recognition of this person, perhaps like seeing a very old friend who took a different path." The medium "started saying a bunch of things that kind of spooked members of my family cuz they were" specific and verifiable. The experiencer's NDE had already shown him that death is a transition, not an ending, and the medium's ability to access information about deceased relatives fit seamlessly into that larger picture.

What's interesting here is that NDErs aren't using mediums to find out if survival is real. They already know it's real. They're using mediums as a practical tool, a way to receive specific messages or guidance from people who are no longer physically present but are still very much around. The medium becomes a telephone, not a proof.

This matches what the research shows. If mediumship were just cold reading or grief exploitation, NDErs would be the easiest marks, the most vulnerable to suggestion, the most likely to overinterpret vague statements. But they're not reporting vague statements. They're reporting specific, unsolicited details that match what they already learned during their own experience of expanded consciousness.

The Hardest Objection I Can't Fully Answer

Here's the thing that still bothers me: if mediumship is real, why isn't it reliable? Why do even the best research mediums have off days? Why do some readings produce stunning hits while others are mediocre or wrong? If consciousness survives death and the deceased are available to communicate, why isn't the signal clearer?

I don't have a satisfying answer to this. The best I can offer is that we don't understand the mechanism. We don't know what conditions make communication easier or harder. We don't know if the medium's mental state matters, if the deceased person's willingness or ability to communicate varies, if there are environmental factors we haven't identified. The research shows hit rates well above chance in controlled conditions, but it doesn't explain why the phenomenon is inconsistent.

Maybe it's like radio reception: sometimes the signal is strong, sometimes it's weak, and we don't always know why. Maybe the deceased are learning how to communicate, just as the mediums are learning how to receive. Maybe some information is easier to transmit than others. I don't know. What I do know is that the inconsistency doesn't erase the hits. A weather forecast that's right 88% of the time is still useful, even if we don't understand why it fails the other 12%.

What This Means If You're Sitting with a Medium

If you're considering a reading, the research suggests a few things. First, look for mediums who work in blinded or semi-blinded conditions. A good medium shouldn't need to see you, shouldn't need your last name, shouldn't need to ask leading questions. They should be able to provide specific information up front and let you verify it afterward. Second, be wary of anyone who offers only vague, high-probability statements ("I'm sensing a father figure," "There's unresolved grief"). Those aren't evidential. Third, pay attention to details you didn't consciously remember or never knew. Those are the strongest hits, the ones that can't be explained by telepathy from your own mind.

And fourth, recognize that even the best mediums aren't 100% accurate. The research shows that certified, trained, rigorously tested mediums working under controlled conditions get it right most of the time, but not all of the time. If you get a reading and some of it resonates and some of it doesn't, that's consistent with what the data predicts. It doesn't mean the whole thing is fraudulent.

The Larger Picture

Mediumship research doesn't stand alone. It converges with near-death experience studies, with cases of children who remember past lives, with deathbed visions, with terminal lucidity in dementia patients, with shared death experiences where multiple people witness the same apparition at the moment of death. Each line of evidence, taken individually, can be explained away or dismissed. Taken together, they point in the same direction: consciousness is not produced by the brain, and it does not end when the brain stops.

The mediumship data is particularly strong because it's been studied under controlled conditions for over a century. The Society for Psychical Research was founded in 1882 specifically to investigate this question, and their early work documented thousands of cases with the same pattern: specific, verifiable information that no one present could have known, provided by mediums who had no access to normal sources. Modern studies like Beischel's have added statistical rigor and blinded protocols, but the core finding hasn't changed. The hit rates remain well above chance.

I think the resistance to this evidence comes less from the data and more from the implications. If mediums are really communicating with the deceased, then death is not an ending. Then consciousness is not confined to the brain. Then the materialist worldview, which has dominated Western science for over a century, is incomplete at best and fundamentally wrong at worst. That's a hard pill to swallow, especially for people who've built their careers on the assumption that the brain produces consciousness and that's the end of the story.

But the data doesn't care about our worldview. It just sits there, uncomfortable and persistent, waiting for an explanation that fits.

>The evidence from controlled mediumship research suggests information is being accessed from a source independent of the living; cold reading cannot explain hits on specifics like unpublished death details.

Where the Research Is Heading

The Windbridge Research Center continues to refine its protocols, looking at whether mediums can distinguish between different deceased individuals, whether they can provide information about the deceased person's current state or activities, and whether the accuracy of readings correlates with specific personality traits or training methods in the mediums themselves. Recent studies have also looked at the brain activity of mediums during readings, finding distinct patterns of electrocortical activity that differ from normal waking consciousness, meditation, or imagination.

The Bigelow Institute for Consciousness Studies, which awarded prizes in 2021 for the best essays on evidence for survival of consciousness, included multiple submissions focused on mediumship. The winning essays cited blinded medium tests showing hit rates around 67% for specific deceased information, well above the 50% chance baseline.

What's emerging from all this work is a clearer picture of what mediumship is and isn't. It's not a supernatural gift. It's a skill, one that can be trained and tested. It's not infallible, but it's not random either. The best mediums, working under controlled conditions, access information they shouldn't be able to access if consciousness ends at death. That's the finding. The implications are up to you.