The Most Undeniable Psi-Phenomena in 1 Paper | Etzel Cardeña

ELI5/TLDR

A psychology professor at Lund University rounded up every meta-analysis he could find on psi research — extra-sensory perception, mind-affecting-matter, that whole drawer — and noticed something inconvenient: they all point in the same direction. Small effects, but consistent ones, with statistical significance comparable to mainstream medicine and biology. In the Ganzfeld experiment (people in a soft sensory-deprived state guessing one of four images) participants land at around 32% instead of the chance 25%. Cardeña’s case isn’t “psi is proven.” It’s that the data exist at the same standard we accept everywhere else, and dismissing them with “this is impossible” is a metaphysical reflex, not a scientific one.

The Full Story

What the 2018 paper actually is

Cardeña wants you to know up front: he didn’t run a meta-analysis. He ran a meta-meta-analysis — what he calls an umbrella review. He gathered roughly twelve to thirteen existing meta-analyses, each itself summarizing dozens or hundreds of individual psi experiments, and asked one simple question. Do they agree?

The answer, awkwardly for everyone, is yes. Every single meta-analysis he could find — across different experimental paradigms, different decades, different labs — pointed the same way: small but real effects above chance.

If you only had one meta-analysis for one kind of studies and all of the others showed no evidence for psi, I think you would be very justified to say perhaps there’s some mistake here in this type of experiments. But in this case every single one was supportive.

The bar he sets isn’t “this proves psi is real.” It’s that the statistical signal is at the same level as findings in medicine, biology, and parts of physics that we accept without theatre. Two reviewers of his paper, he mentions almost in passing, refused to engage with the data on the grounds that the phenomenon is impossible. That move — refusing the evidence because of a prior commitment to what’s possible — is the move he’s pushing back against.

The two camps: ESP and psychokinesis

The field splits into two big buckets. Extra-sensory perception (ESP) is the “knowing things you shouldn’t be able to know” bucket — your friend in Tokyo affecting you with no phone link, a dream that maps onto tomorrow’s news. Psychokinesis (PK) is the “affecting matter with intention” bucket — bending spoons, nudging dice.

The evidence is uneven. ESP has more supportive data overall. Psychokinesis is messier, mostly because the macro version (visible spoon-bending) is the natural habitat of stage magicians and frauds. Cardeña is blunt: there’s basically no good laboratory evidence for macro-PK. The micro version — trying to nudge a random number generator so it spits out slightly more 1s than 0s — does show small effects in well-run studies. We’re talking about 50.05% vs an expected 50.00%. Tiny.

But here he makes an argument that’s quietly elegant. If the effect were any larger, the world would be unlivable.

If I could affect matter by my will, I would have already killed some people. I am not a saint. But before that, I would have been killed by the first one whom I failed.

A universe where willpower readily moves matter is a universe of constant low-grade catastrophe. The smallness of the effect, in his telling, isn’t a bug — it’s roughly what the universe would have to look like for it to keep working at all.

The Ganzfeld effect

This is the headline result and worth slowing down on. “Ganzfeld” is German for “whole field” — coined by Gestalt psychologists in the early 20th century. You put someone in a setting where their senses get nothing structured to chew on. Soft red or white light filling their vision, no edges or shapes. White or pink noise (random sound that washes like waves) playing into their ears. A relaxation induction first. Ten minutes of this and the perceptual filters loosen.

Then a computer randomly picks one of more than a hundred film clips, but doesn’t show it yet. The participant tries to mentally tune in to “the clip that will be selected.” After thirty minutes of free association, they’re shown four very different clips and asked to rank which one matches what they were experiencing.

By chance: 25% correct. One in four.

Across hundreds of studies and thousands of participants: roughly 32-33%.

They vary a lot? Not really. You end up very rarely with less than 25%. Most end up being around 32%.

That’s about a 30% relative bump above chance. Not enough to predict tomorrow’s stock price. Enough that, accumulated across enough trials, it becomes statistically vanishing-rare to attribute to noise.

The qualitative texture of these hits is also strange. People rarely describe the clip directly. They describe something that bleeds in around the edges. One participant, mid-session, talked about boys playing on a meadow with a giant balloon with big eyes. The clip turned out to be a Russian Winnie-the-Pooh cartoon — Pooh and a friend (cartoonish “boys”), a balloon Pooh floats up on, a beehive whose bees stare at him with big eyes. The fingerprints are all there, smudged.

It is not as if they perceived what was seen. It’s much more as if they have been affected by something they heard while they were dreaming.

Why the effects are so small

Cardeña offers two reasons, neither of them mystical.

First, evolution. He flags a common misunderstanding here: evolution isn’t a process that optimizes upward, it’s a process that culls. What survives is whatever was good-enough-not-to-die. For survival you need decent perception of the immediate world — don’t walk into walls, don’t get hit by carts. You don’t need precognition. So the trait, if it exists, never got selected for or against; it just sits at low intensity in the population.

Second, attention. Following a Platonic line, he argues we live inside a perceptual world tuned for adequacy. The signal of “non-local information” is always there in the background, but it gets drowned out by the loud foreground of ordinary sensory input. Quiet the foreground — meditation, hypnosis, Ganzfeld, the edge of sleep — and the weak signal becomes faintly audible.

This is the part of his model that does the most work. Psi, in his framing, isn’t a sometimes-it-comes phenomenon. It’s an always-on but always-flooded phenomenon. The question isn’t “what triggers it” but “what attenuates the noise enough for it to show up.”

The taxonomy problem

A surprisingly large chunk of the conversation isn’t about psi at all — it’s about how sloppy the vocabulary has become. “Trance” is his favorite example. It gets used for spirit possession, for meditative emptiness, for daydreaming, for someone going on a killing spree. When pressed, people sometimes retreat to “trance-like state,” which is somehow worse — now you’re invoking a likeness to something whose definition was never specified.

His response, in a paper co-written with fifteen other multidisciplinary experts, is to ditch trigger-based labels. Don’t say “the hypnotic state” — there’s no such thing as a single state produced by hypnosis. Some people drop deep, some stay completely unaffected, and the variance is huge. Instead, describe the phenomenology: feeling outside the body, vivid imagery experienced as real, dampened conceptual thought, and so on. Then you can compare a hypnosis session, a psychedelic trip, a meditation retreat, and a walk through a museum on the same axes — and notice how much they overlap.

This matters for the psi research too. “Telepathy” sounds like reading a mind, but the Winnie-the-Pooh example shows it isn’t that at all. It’s more like a leaky channel into associative imagination, not a transcript.

Compatibility with physics

The pushback he hears most often is the metaphysical one: this can’t be real because physics says so. His reply is essentially — which physics? He cites Bernard d’Espagnat, the French quantum theorist, whose interpretation lands somewhere close to philosophical idealism: there are no objects in the ultimate sense, no time, no space — at least not in the form our daily experience presents.

How can a person be a materialist or a physicalist when the physicists themselves are saying we do not know what that is?

He’s careful here. He’s not saying quantum mechanics proves psi. He’s saying the confident materialism that rules psi out a priori is itself standing on philosophical ground, not bedrock physics. Once you accept that the foundational layer is genuinely uncertain, the door doesn’t have to swing wide open — but the argument from impossibility loses its teeth.

Skeptics, and the quality of skepticism

He’s read the major critics — Hansel, Alcock, the rest — and his verdict is that the criticism has weakened over time. The strong early critics did their homework: they walked through experimental setups and pointed at specific flaws. The weaker modern criticism, in his view, is either lazy (handwave at the file-drawer effect without doing the math, or accuse parapsychologists of crypto-religiosity without surveying any of them) or dismissive on principle.

The file-drawer problem — the worry that we only see the studies that found something, while the null results sit in researchers’ filing cabinets — has been formally addressed using statistical correction methods. To overturn the meta-analytic effects via file-drawer alone, he argues, you’d need hundreds of unpublished negative studies. Given how few parapsychologists exist worldwide and how thin the funding is, that’s not a serious explanation.

He doesn’t expect a paradigm shift. Drawing on Planck’s old observation that science advances one funeral at a time, he points out that funerals don’t actually settle anything — every dead skeptic leaves behind students who heard the dismissal. What you get instead are slowly shifting percentages: in the last 10-15 years, journal editors have grown marginally more open, partly because of consciousness science’s failure to explain itself by neurons alone.

The actor and the dream

The interview ends on a tangent that’s actually the through-line. Sicco Vermeer brings up Jim Carrey’s transformation into Andy Kaufman during the filming of Man on the Moon — a documentary captures Carrey effectively losing his self for months, with Kaufman’s actual relatives reacting to him on set as if their dead son had come back.

Cardeña’s reaction is calm. You do this every night, he says. In dreams you become somebody else — a woman, an animal, a stranger using your name with a different biography attached. The “narrative homogeneous self” — the named, handed, history-having you — is one of the categories through which you apprehend the world. Like time and space, it’s useful. Like time and space, it’s somewhat optional. Hand someone a mask, or a social media account, and watch how quickly that self gets renegotiated.

We are not this solid person. We are changing. We’re shifting every day. We go into dreams. We become somebody different.

The anomaly the Ganzfeld points at, in his telling, is the same anomaly that ordinary dreaming and serious acting point at: that the boundaries we treat as load-bearing — self, time, space — are softer than the daylight version of the world admits.

Key Takeaways

The 2018 American Psychologist paper is an umbrella review — a meta-analysis of meta-analyses — covering ~12-13 reviews. All of them produced supportive evidence for psi. Consistency across paradigms is the central argument, not magnitude of effect.
Statistical significance in psi studies is comparable to mainstream medicine and biology. Two reviewers refused to consider Cardeña’s data on a priori metaphysical grounds, not methodological ones.
ESP evidence is stronger than psychokinesis evidence. Macro-PK (visible matter manipulation) has essentially no good lab evidence. Micro-PK on random number generators shows small effects (~50.05% vs 50.00%).
Cardeña’s argument for why effects are small: a universe where willpower readily moves matter would be uninhabitable. Smallness is structural, not a flaw.
Ganzfeld setup: soft uniform light, white/pink noise, relaxation induction, 10-30 min, then a 4-way forced-choice. Chance = 25%. Observed across hundreds of studies = ~32-33%. Very low variance across well-run replications.
Hits in Ganzfeld are usually fragmentary and dreamlike, not direct perception. The Russian Winnie-the-Pooh / “boys with a balloon and big eyes” example is the canonical illustration.
His mechanism sketch: psi-relevant information is always present in non-conscious processing but flooded by perceptual input. Altered states (hypnosis, meditation, Ganzfeld, hypnagogia) attenuate the filter, not generate the signal.
Evolution misconception: evolution culls non-adaptive variants rather than optimizing upward. Psi traits, if real, would persist at low population intensity because there’s no strong selection pressure either way.
Taxonomy critique: terms like “trance,” “hypnotic state,” “psychedelic state” are defined by trigger rather than phenomenology, which is backwards. Variance within any trigger is huge — same hypnotic procedure produces deep states in some, boredom in others.
D’Espagnat-style quantum interpretations dissolve the confident materialist objection. Cardeña doesn’t use physics to prove psi; he uses it to disarm the “this is impossible” reflex.
File-drawer problem has been addressed statistically. Overturning the meta-analytic effects via unpublished null results would require hundreds of hidden studies, implausible given field size and funding.
Self is one of the cognitive categories — alongside time and space — through which we organize experience. Dreams, deep acting, and altered states all show how negotiable it actually is.

Claude’s Take

Score: 6/10. This is a thoughtful interview with a serious researcher making a defensible version of an embattled case, but it’s an interview, not a paper, and the format does the heavy lifting on rhetoric while leaving the actual evidentiary work in the footnotes.

Cardeña is the right person to make this case. He’s a Swedish-university chair, his methodological taxonomy work is genuinely useful even if you ignore the psi piece, and his core epistemic move — “engage with the data at the standard you’d apply elsewhere” — is correct in principle. The problem is that “elsewhere” is itself wobbly. Medicine and biology have replication crises that make their statistical thresholds a weaker calibration point than they sound. p < 0.05 across hundreds of studies in a field with strong publication selection, motivated researchers, and small effect sizes near the noise floor is exactly the recipe that produced ego depletion, power posing, and most of priming research. The Ganzfeld 32% has held up better than those, granted — but the same toolkit that produces the effect also produced a lot of effects that are now understood to be artifacts of flexible analysis pipelines.

He addresses the file-drawer problem too quickly. The standard correction methods (fail-safe N, trim-and-fill, p-curve) all rest on assumptions that get aggressively contested by skeptics like Bem critics and replication-movement researchers. “You’d need hundreds of unpublished studies” sounds like a knockout but is actually a calculation downstream of model assumptions about what the underlying distribution looks like. There’s also been a substantial pre-registration push in psi research over the last decade, and the picture from pre-registered Ganzfeld replications is murkier than the historical meta-analytic average suggests.

The argument from quantum mechanics is the part I’d hold at arm’s length. D’Espagnat is a real and serious physicist, but invoking quantum interpretations to soften the metaphysical objection to psi is a move with a long, troubled history. Quantum mechanics doesn’t license non-local mind-to-mind information transfer in any established way — entanglement specifically can’t transmit information. Saying “physicists also disagree about ultimate reality” is true but doesn’t do the inferential work he wants. It moves the conversation from “this violates physics” to “the philosophy is unsettled,” which is fair, but it’s a smaller win than the rhetorical framing implies.

What he gets clearly right: the dismissal-by-impossibility move is bad epistemics, the taxonomy of altered states is genuinely sloppy in the literature, and the “all twelve meta-analyses point the same way” observation does deserve a substantive response from skeptics rather than a dismissal. The smallness-of-effect argument is also more interesting than the usual psi-defender talking points — it’s a structural-physics-of-the-world argument rather than a special pleading.

The Winnie-the-Pooh example is the most evocative thing in the interview and also the most dangerous. Hits like that are exactly where confirmation bias does its best work — the brain pattern-matches loose associations into hits after the fact, and a single dramatic example doesn’t tell you anything about base rates. He’d be the first to say so; the interviewer doesn’t push.

Net: worth your time if you’re interested in the philosophy of how science handles its outer-edge data, in the taxonomy of altered states (which is the strongest contribution and basically uncontroversial), or in the meta-question of what a fair-minded skepticism would look like. Not worth treating as a settled scientific update on whether psi exists. The honest read is “the meta-analytic signal is real and unexplained, and the methodological criticisms haven’t decisively resolved it in either direction.” That’s a smaller and more boring claim than the title promises, but it’s the one the evidence supports.