The Evolution of Telepathy | Trinity College, Cambridge

ELI5/TLDR

Rupert Sheldrake gives a lecture at Trinity College, Cambridge — the same institution where the word “telepathy” was coined in 1882 — arguing that telepathy is not supernatural but a normal feature of animal biology that evolved alongside social groups. He presents decades of his own experimental data: dogs that reliably wait at the door when their owners decide (not start) to come home, a parrot that says the names of pictures its owner is looking at in another room, mothers whose milk lets down when their infant needs them miles away, and telephone telepathy tests where subjects guess the caller at 45% versus a 25% chance rate. The effect, he argues, is stronger between emotionally bonded people and doesn’t depend on distance. Most of the lecture is less about the phenomenon itself than about why mainstream science refuses to investigate it — a taboo he blames on a specific kind of militant skepticism that, by his account, has never actually looked at the evidence.

The Full Story

How a Cambridge biochemist stopped dismissing telepathy

Sheldrake did not begin curious. He was trained at Cambridge in what he calls the “standard issue mindset” — atheism, materialism, total skepticism toward anything psychic. The mind is the brain, the brain is in the skull, so telepathy must be rubbish and the evidence for it must be either fraud or error. He believed this confidently, in the way young scientists often do, without having looked.

The person who dented this was Sir Rudolph Peters, a retired professor of biochemistry from Oxford working in the Cambridge lab. Over tea, Sheldrake dismissed telepathy in the usual terms. Peters asked if he’d looked at the evidence. Sheldrake said no. Peters said he had, and told him about a disabled, near-blind boy in Cambridge who could read eye charts his ophthalmologist knew he couldn’t possibly see — but only when his mother was present. When they separated mother and son by six miles and a telephone, the boy still named the letters she was looking at. Chance: 3.8%. Actual: 38%, over 100 trials. Magicians reviewed the tapes for fraud and found none.

What struck Sheldrake was less the data than his colleagues’ response when he relayed it. They dismissed it exactly as he had, without wanting to see it. He calls this the diagnostic feature of skepticism about psychic phenomena — unwillingness to examine evidence because of prior certainty it must be wrong.

Morphic fields, termite architects, and coordinated flocks

Sheldrake didn’t pivot to telepathy right away. He was a plant physiologist, specifically a specialist in auxin, the main plant growth hormone. His real interest was in morphogenesis — how a fertilized egg becomes a body. Standard chemistry, by his own admission as an expert, wasn’t answering the question. He came to suspect there was a different kind of field at work in living organisms, which he called a morphic field — a field of form, organizing matter into the characteristic shapes of its species.

From there he extrapolated outward. If fields organize cells into organs, why not animals into groups? Termites construct cathedral-scale mounds while blind. In the 1920s, Eugene Marais cut a termite mound in half with a steel plate so the two sides could not communicate — and on both sides, the insects rebuilt arches and tunnels that lined up across the barrier. Starlings in a murmuration turn faster than nearest-neighbor computer models can explain; the best current simulations treat the flock as a magnetic field, each bird a domain. Fish in a school do the same thing — a flash expansion away from a predator where no one bumps into anyone. Nobody, Sheldrake notes, actually knows how this is coordinated.

“They’re not supernatural. They’re not paranormal. They’re normal. They’re part of the ordinary coordination of animal groups.”

This is the thesis: what we call telepathy is just one visible consequence of a field that already coordinates biology in ways we don’t understand.

Dogs, cats, parrots

Since actual wolves in Canada are hard to study — and no biologist hunting for grants will touch the topic — Sheldrake turned to the animals people live with. He put out a call for unusual pet stories and was buried in letters. Anecdotes aren’t data, but the same patterns kept repeating from everywhere: cats that vanish the moment their owner decides to take them to the vet, even when the carrier hasn’t been touched and the word “vet” hasn’t been spoken. A survey of all 65 veterinary clinics in North London found 64 reporting chronic missed-appointment problems with cats. The remaining one had given up the appointment system entirely.

The testable version of this claim is dogs that know when their owner is coming home. Sheldrake’s most studied dog, JT, was filmed continuously while his owner left the house. The owner came back at randomly chosen times — in taxis, to eliminate familiar car sounds. The measured result: the dog spent about 4% of her absence at the door or window for most of her absence, and over 50% of the time starting roughly when the owner decided to come home, before the taxi had even arrived to pick her up. Richard Wiseman, a professional skeptic, replicated the experiment with the same dog and got essentially identical numbers — then announced in the press that he’d debunked it, on the grounds that JT had made a false trip to the window earlier in the session. Years later, Wiseman concedes the data are virtually identical to Sheldrake’s own.

The most peculiar case is a parrot called N’kisi, an African grey with a 1,500-word working vocabulary. The experiment: the owner sits in one room with a sealed envelope. She opens it, looks at a photograph for two minutes. N’kisi, in another room, is filmed continuously. The tapes go to three independent transcribers. When she looks at a picture of flowers, N’kisi says “that’s a flower, little flowers.” When she looks at someone on the phone, the bird makes dialing noises and says “what are you doing on the phone.” The hit rate is well above chance.

Mothers, telephones, emails

With humans, Sheldrake skipped the card-guessing tradition of the Rhine lab and asked what a useful ability would look like under evolutionary pressure. One candidate: nursing mothers who feel their milk let down when their baby cries — even when the baby is miles away. The let-down reflex is a real, well-characterized oxytocin response, normally triggered by the sound of the infant crying or being held. Sheldrake’s North London study with 20 nursing mothers tracked letdown events against independently monitored infant distress. The synchrony produced odds against chance of a billion to one. He notes that mothers who could sense their baby’s distress across distance would have, in evolutionary terms, better-surviving babies.

The most common modern case is the telephone variant: you think of someone, the phone rings, it’s them. Over 80% of people report this. Sheldrake’s controlled version: the subject sits at a landline with no caller ID, is filmed, and gives four names — people they know. A randomly selected friend is phoned and asked to call. The subject must guess who before picking up. Chance rate: 25%. Observed rate across 400+ trials: 45%. P-value: 10⁻¹². Email version: 47%. The effect vanishes with strangers and stays strong with emotional bonds — it works Britain-to-Australia as well as within a household. A modified version tested whether people were actually precognizing who would be chosen, by asking them to guess before the computer had even selected the caller. That version came in exactly at chance. Something about intention in the moment seems to matter.

The sociology of a taboo

A surprising amount of the lecture is about the people who refuse to look. Sheldrake names them: Peter Atkins of Oxford, who denounced the research in The Times as “the charlatan’s fantasy” and, on BBC Radio 5, admitted he had not examined any of it and would not, because “I’d be very suspicious of it.” Lewis Wolpert, who refused to watch the tapes before calling the research empty on TV, then sat through a public debate visibly not looking at the evidence on the screen behind him. Richard Dawkins, who agreed to an interview on the premise it would be about evidence, then refused to discuss evidence on camera and acknowledged off-camera that it was “another Dawkins polemic.”

Sheldrake’s point isn’t that these men are idiots. It’s that an educated person is permitted — expected, even — to publicly denounce a topic they’ve never studied, provided the topic is this one. No one would do that with quantum field theory or radio astronomy. He argues the fear is that telepathy, if real, would overturn science. He disagrees: Faraday’s fields didn’t destroy Newton, they enlarged him. Consciousness already has a “hard problem” with no accepted solution. Admitting that animals coordinate at a distance via fields nobody has mapped yet wouldn’t break the laws of nature — it would just expand the list.

He closes by noting that in every traditional society he’s lived in or read about — the Kalahari bushmen, rural India — these phenomena are unremarkable. When anthropologists arrive with the premise that they can’t happen, they don’t study them. Which leaves most of humanity’s experience of this stuff outside of science.

Key Takeaways

Sheldrake’s thesis is evolutionary, not mystical. He treats telepathy as a biological feature of social species, selected for because it helps coordinate group behavior across distance. The comparison he leans on is quantum entanglement — an analogy, not a mechanism.
The strongest controlled data is on telephone telepathy and the dog-at-the-door experiments. Both have been replicated independently. The P-values are real; the effect sizes are not subtle.
The effect is strongest between emotionally bonded subjects. Strangers in labs barely perform above chance. Families and close friends perform well over it. Sheldrake argues this is why decades of strangers-guessing-cards gave such weak results.
Distance doesn’t seem to matter. Australia-to-London works as well as room-to-room, which is consistent with the quantum entanglement analogy but not with any ordinary field that falls off with distance.
Pre-cognition vs. telepathy was specifically controlled for. When subjects had to guess before the random selection was even made, performance dropped exactly to chance. Something about the sender’s intention in the moment matters.
Most of the resistance, Sheldrake argues, is social rather than evidential. Skeptical public intellectuals routinely refuse to examine the evidence before denouncing it. Working scientists, in private, are often more open — especially the ones whose dogs wait at the door.

Claude’s Take

Sheldrake is an unusual figure — a properly trained Cambridge biologist who spent the back half of his career on topics his peers treat as career suicide. That makes him both more interesting than a pure fringe figure and harder to evaluate, because the sociology of the field is so warped that neither loud endorsement nor loud dismissal is a reliable signal.

What’s actually defensible here, separated from the rhetoric. The telephone telepathy experiments are a real experimental protocol with a clear null hypothesis, a clear chance rate, filmed subjects, random selection of callers, and independent replications at Amsterdam and Freiburg. The effect sizes he reports are large enough that the usual debunking moves — file-drawer effects, subtle cueing — don’t trivially explain them. If the numbers are what he says they are, something is happening that we don’t have a story for.

What’s weaker. “Morphic fields” and “morphic resonance” are his broader theoretical framework, and they are genuinely not established science. They’re a hypothesis without a mechanism, and the physics community has never found them useful. When Sheldrake uses quantum entanglement as an analogy, he’s careful to call it an analogy — but the fact that he reaches for it at all is a tell: there’s no actual theoretical machinery behind the claim, just a family resemblance to a real phenomenon in a totally different domain. The N’kisi parrot data has had real methodological criticism (the statistical analysis of the transcripts is contested). The mother-and-baby milk letdown study has a sample of 20 and has not been independently replicated at scale. “Odds against chance of a billion to one” from a sample of that size should raise eyebrows about how the analysis was done.

What’s unfair to Sheldrake. The skeptical public intellectuals he names — Atkins, Wolpert, Dawkins — really do have a pattern of refusing to engage with his actual data before dismissing it. That’s bad epistemics regardless of whether his conclusions are right. Dismissing a body of experimental work because you find the conclusion offensive is not science. The specific anecdote of Dawkins admitting on set that the program was a polemic is consistent with how Dawkins operates and should bother people who take Dawkins seriously as a science communicator.

Where this leaves me. The controlled experiments on telephone telepathy and home-coming dogs are the strongest thing here and deserve more attention than they get. The framework Sheldrake wraps them in — morphic fields, resonance, a whole new kind of memory in nature — is speculative and has not done the work of generating novel predictions that pan out. A reasonable position is: the phenomena deserve proper replication by non-affiliated labs, and the theoretical scaffolding can be bracketed for now. That’s not where either camp actually sits, which is why the topic stays in the ditch.

Score: 6/10. The lecture is clearly argued and has real data in it, and Sheldrake is genuinely more careful than his reputation suggests. It loses points because the most interesting experimental claims are presented at a level of detail that makes them hard to evaluate without going to the papers, and because it spends almost as much time on grievance against skeptics as on the science — understandable given his career, but it flattens the case. Worth watching if you’re interested in the sociology of scientific taboos; treat the specific empirical claims as starting points for further reading, not as settled.