In 1920, one of the longest animal experiments of the twentieth century began at Harvard University. The psychologist William McDougall had rats learn a simple task across 49 generations — and found that each generation learned faster than the last. Two decades later, a team in Melbourne repeated the experiment over 50 generations — with a control group that was never trained. The control group improved just as much. This result has stood in the journals since 1954. It refutes the most obvious explanation, fits no known biological model, and to this day no one has explained it conventionally. The British biologist Rupert Sheldrake considers it the strongest published evidence for what he calls morphic resonance.
The experiment: rats in a water maze
William McDougall (1871–1938) was a British-American psychologist who from 1920 at Harvard University set out to test the Lamarckian hypothesis — the idea that acquired characteristics can be inherited by the next generation.
The design was simple: young rats were placed in a water tank with two escape routes — one illuminated and one dark. The illuminated route delivered an electric shock; the dark one was safe. The measure was the number of errors (shock escapes) until a rat chose the correct, dark exit twelve times in a row.
McDougall bred the offspring of tested rats for 49 generations and measured the error rate in each generation. The result was clear: after 20 generations his rats were learning more than ten times faster than the first generation. McDougall published three reports (1927, 1930, 1933) in the British Journal of Psychology supplements.
To counter the obvious objection that he was unconsciously breeding the cleverest animals, McDougall employed a countermeasure: in many generations he deliberately crossed the slowest learners — and the improvement continued nonetheless.
Edinburgh: a first replication attempt
F.A.E. Crew of the Department of Animal Genetics at the University of Edinburgh was the first to attempt a replication of McDougall's experiment. His results were inconclusive — the rats showed no clear improvement across generations. Crew himself acknowledged methodological weaknesses in his design and declared the question open. His report appeared in 1936 in the Journal of Genetics.
Melbourne: 20 years, 50 generations — and the surprise
The most thorough repetition came from Australia. W.E. Agar and his colleagues F.H. Drummond, O.W. Tiegs and M.M. Gunson at the University of Melbourne ran the experiment for 20 years across 50 generations. Four reports appeared in the Journal of Experimental Biology between 1935 and 1954 — one of the oldest biological journals in the world.
Agar confirmed McDougall's central finding: the trained rat lines did indeed learn faster across generations.
But Agar — unlike McDougall — had maintained a control group: a separate rat line bred over the same 50 generations but never trained. No rat in this line had ever seen the water tank. None had ever received an electric shock. None had ever had to learn. The control rats were kept in a separate room, without any contact with the experimental animals.
When the control rats were finally tested, they learned just as fast as the trained line — and the improvement across generations was the same.
This result has two immediate consequences:
- It refutes the Lamarckian hypothesis. If untrained descendants also learn faster, the improvement cannot be due to genetic transmission of acquired experience — the control rats had no trained ancestors.
- It poses a new, unanswered question: why do rats in Melbourne get better at a task they have never practised, while on the other side of the world rats are training on the same task?
Agar and his colleagues noted the result as puzzling in their fourth and final report (1954). They offered no conventional explanation. The finding disappeared into the specialist literature — too uncomfortable for any current theory, too solid to ignore.
Rupert Sheldrake: morphic resonance
In 1981, British biologist Rupert Sheldrake (b. 1942, Cambridge, PhD in biochemistry, Fellow of the Royal Society of Biochemistry, nine years of research on plant morphogenesis) published A New Science of Life, in which he presented the hypothesis of formative causation — today better known as morphic resonance.
Sheldrake's core idea: organisms are shaped not only by their genes and their immediate environment but also by a cumulative memory of their entire species. Each member of a species draws on a collective form memory and simultaneously contributes to it. The more often a particular pattern — a behaviour, a crystal structure, a learning task — is performed by members of a species, the easier it becomes for all other members to realise the same pattern. Regardless of spatial distance. Regardless of genetic relatedness.
For Sheldrake, the McDougall–Agar experiments are the strongest published evidence for morphic resonance: Harvard rats learn a task progressively better over generations, and simultaneously rats in Melbourne — without training and without genetic connection — get better at the same task. Conventionally, this is inexplicable.
The Nature editorial: a book for burning
Nature, one of the most prestigious scientific journals in the world, published in 1981 one of the most unusual reviews in the history of science. The then editor-in-chief John Maddox wrote that A New Science of Life was "the best candidate for burning there has been for many years." In the same breath, Maddox conceded that the hypothesis was testable and called for experimental testing rather than condemnation.
The review is remarkable because it illustrates the mechanism described by the 1906 pattern: the reflex to reject a hypothesis emotionally first, and then — almost reluctantly — to concede that it is scientifically verifiable. Maddox did both in a single editorial.
Further evidence for morphic resonance
Sheldrake has since the 1980s conducted or inspired further experiments:
- Crystallisation of new compounds: When a new chemical compound is crystallised for the first time, the process is difficult. In subsequent crystallisations — even in other laboratories worldwide — it becomes easier. Sheldrake considers the conventional explanation (seed crystals carried in laboratory air) implausible in many cases.
- Hidden images: A hard-to-recognise image (e.g. a Dalmatian in a dot pattern) was shown to millions of viewers via Thames Television in 1983. Subsequent test subjects who had not seen the broadcast recognised the image significantly faster than a comparison group tested before the broadcast.
- Japanese characters: European test subjects learn real Japanese kanji (known to billions of people) faster than specially invented pseudo-kanji of equal visual complexity. A study on this appeared in 2010 in Explore: The Journal of Science and Healing.
Each of these tests has attracted methodological objections. None has yet been replicated with the rigour required for broad scientific acceptance. But none has been refuted either.
The hundredth monkey — what is true and what is not
In connection with morphic resonance, the story of the hundredth monkey is frequently told: Japanese macaques on Koshima Island supposedly learned from 1953 onward to wash sweet potatoes in the sea. Once a critical mass — the proverbial hundredth monkey — had mastered the behaviour, it allegedly jumped suddenly to monkeys on other islands, without physical contact. This story became world-famous through Lyall Watson's book Lifetide (1979) and Ken Keyes' The Hundredth Monkey (1982) and is still cited today as evidence for non-local information transfer.
What is established: The research on Koshima is real and serious. The Japanese primatologist Masao Kawai documented from 1953 how a young female named Imo began washing sweet potatoes in a stream — and later in the sea, which also salted them. The behaviour spread through the troop over years: first to Imo's playmates, then to their mothers. Kawai published this in 1965 in Primates. It is a milestone in primatology and a well-documented case of cultural transmission in animals.
What is open: Sweet potato washing was indeed observed in other macaque populations in Japan. But these other troops were not studied with the same scientific rigour as Koshima. It is therefore unclear whether the behaviour was independently invented there (researchers also provisioned other troops with sweet potatoes on beaches, which could prompt the discovery of washing), whether individual animals migrated between populations, or whether something else was at play. The question is neither cleanly confirmed nor cleanly refuted — it has simply never been systematically investigated.
What is demonstrably false: The popular narrative of a sudden jump once a critical mass was reached. This dramatic tipping point — the hundredth monkey, after which it suddenly starts everywhere — does not exist in the data. Lyall Watson fabricated it. Ron Amundson demonstrated this clearly in 1985 in the Skeptical Inquirer. Watson himself admitted in a letter to the Skeptical Inquirer that he had embellished the story.
The real lesson: Anyone looking for robust evidence of non-local learning transfer should not stop at the hundredth monkey but read on here. The McDougall–Agar rat study — 20 years, 50 generations, published in the Journal of Experimental Biology, with a never-trained control group that improved just as much — is methodologically stronger by orders of magnitude than the popular monkey story. There the control group is cleanly isolated, the documentation is complete, and the finding has stood unexplained in the journals since 1954.
What mainstream science says
Morphic resonance is not an accepted scientific theory. The mainstream classifies it as speculative. The main objections:
- No mechanism: Sheldrake postulates a field but provides no physical theory explaining how information is transmitted without a known carrier across space and time.
- Experimental weakness: The positive results come from small studies with limited statistical power, often without independent replication.
- Alternative explanations: For many individual findings, conventional explanations can be constructed — seed crystals, unconscious selection, statistical artefacts.
These objections are serious. But they concern the interpretation, not the data themselves. The Agar–Melbourne data still stand in the journals: 20 years, 50 generations, peer-reviewed in the Journal of Experimental Biology, with a finding that the authors themselves called puzzling and for which to this day no conventional explanation exists.
Connections: PEAR and the Global Consciousness Project
The rat experiments do not stand alone. They belong to a broader research tradition that asks whether consciousness and information operate through channels unknown to conventional physics:
- The PEAR lab at Princeton University documented over 28 years that human intention influences physical random number generators to a tiny but statistically significant degree (2.5 million trials, Z ≈ 7).
- The Global Consciousness Project scaled this idea to a worldwide network and found correlated deviations during major events (500+ events, p ≈ 10⁻¹²).
In all three cases — McDougall's rats, PEAR's random generators, Nelson's global network — there is a statistically striking finding that is hard to explain conventionally, and an interpretation that is testable but unproven. The pattern is the same: the data exist; what they mean is open.
What this means for the consciousness question
McDougall's 49 generations at Harvard. Agar's 50 generations in Melbourne. The puzzling improvement of the untrained control group. These are data from peer-reviewed journals, not from esoteric guidebooks.
These data do not prove morphic resonance. They do not prove a collective species memory. But they pose a question that no one has answered since 1954: why do rats in Melbourne learn faster when rats at Harvard are practising the same task — without genetic connection, without direct contact, without any known transmission channel?
Anyone who considers the question absurd must explain the data. Anyone who considers the data artefactual must refute the methodology — across 20 years and 50 generations, in one of the oldest biological journals in the world.
Sources: William McDougall, "An experiment for the testing of the hypothesis of Lamarck", British Journal of Psychology 17, 1927, pp. 267–304; 20, 1930, pp. 201–218; 24, 1933, pp. 213–235. F.A.E. Crew, "A repetition of McDougall's Lamarckian experiment", Journal of Genetics 33, 1936, pp. 61–101. W.E. Agar, F.H. Drummond, O.W. Tiegs & M.M. Gunson, "First–Fourth Report on a Test of McDougall's Lamarckian Experiment on the Training of Rats", Journal of Experimental Biology 12, 1935, pp. 191–211; 19, 1942, pp. 158–167; 29, 1952, pp. 32–38; 31, 1954, pp. 307–321. Rupert Sheldrake, A New Science of Life, Blond & Briggs, London 1981. Rupert Sheldrake, "Rat Learning and Morphic Resonance", essay at sheldrake.org. John Maddox, "A book for burning?", Nature 293, 1981, pp. 245–246. Masao Kawai, "Newly-acquired pre-cultural behavior of the natural troop of Japanese monkeys on Koshima Islet", Primates 6, 1965, pp. 1–30. Lyall Watson, Lifetide, Hodder & Stoughton, London 1979. Ron Amundson, "The Hundredth Monkey Phenomenon", Skeptical Inquirer 9/4, 1985, pp. 348–356.