Thirty-six theories about how we became human

Humans are mediocre sprinters and extraordinary distance athletes. The question is whether that endurance is an adaptation for hunting, or a by-product of something else.

The idea belongs to David Carrier, who argued in Current Anthropology in 1984 that hominins could exploit a thermoregulatory loophole: quadrupeds cool by panting, and panting is difficult at a gallop, so a persistent biped can push a large mammal into hyperthermia. It was formalised twenty years later by Dennis Bramble and Daniel Lieberman in Nature ("Endurance running and the evolution of Homo," 2004, vol. 432, pp. 345–352), who assembled a list of features that make sense for running and for nothing else in particular.

That anatomical list is the strongest part of the case. The nuchal ligament stabilises the head against pitching — chimpanzees lack one, and it appears in Homo erectus. The Achilles tendon and the plantar arch act as springs, returning roughly half the energy of each stride. Gluteus maximus, small in apes, is enormous in us and fires mainly to stop the trunk pitching forward during running, not walking. Semicircular canals enlarged in the genus Homo, improving gaze stabilisation. Toes shortened. Add two to four million eccrine sweat glands and a furless skin, and you get an animal that can dump heat while moving — the specific combination the video correctly identifies.

For two decades the ethnographic evidence was thin enough to be embarrassing: Louis Liebenberg's fieldwork with Ju/'hoansi and /Gwi hunters in the central Kalahari, chasing kudu and gemsbok in the heat of the day, was essentially the exhibit. Critics — Henry Bunn and Travis Pickering foremost — pointed out that a documented handful of hunts among one group is a slender foundation for a species-defining behaviour, and that ambush hunting is far cheaper in calories.

The 2024 revival changed the picture. Eugène Morin (Trent University) and Bruce Winterhalder (UC Davis) published "Ethnography and ethnohistory support the efficiency of hunting through endurance running in humans" in Nature Human Behaviour (2024, vol. 8, pp. 1065–1075). They trawled sources from 1527 onward and found roughly 391 accounts of endurance pursuit across 272 locations worldwide — the Tarahumara, Paiute, Navajo, Australian and southern African groups, Arctic hunters running down caribou over crusted snow. A 1930s–40s Californian survey they re-examined found 114 of 141 societies reporting the practice. They then modelled it: because running compresses the time cost of a hunt, it can pay off even when the runner covers roughly twice the ground a walker would.

Where it stands. Endurance running as a real and widespread human capability: not in doubt. Endurance running as the selective driver that built the genus Homo: still argued. The honest position is that persistence hunting was one tool in a kit that also included ambush, scavenging, trapping and, later, projectiles — and that the anatomy Bramble and Lieberman listed had to evolve for something.

No other animal can accelerate a projectile the way a human shoulder can. The anatomy that permits it appears with Homo erectus — and the mechanism is not muscle.

Neil Roach, working with Madhusudhan Venkadesan, Michael Rainbow and Daniel Lieberman, published the key study in Nature in 2013 ("Elastic energy storage in the shoulder and the evolution of high-speed throwing in Homo," vol. 498, pp. 483–486). Using motion capture on collegiate baseball players — including trials with a brace restricting torso rotation — they showed the throwing arm behaves as a catapult. During the cocking phase, the arm rotates back and stretches the ligaments and tendons crossing the shoulder; that stored elastic energy is then released, producing the fastest motion the human body generates. Muscle alone cannot contract fast enough to explain the speeds.

Three derived traits make the catapult possible, and all three appear together in Homo erectus roughly 2 million years ago: a tall, mobile waist that allows the torso to rotate independently of the hips; low, laterally oriented shoulders (in apes the shoulder is shrugged upward, built for climbing); and reduced humeral torsion, which lets the arm rotate far enough back to load the ligaments. Chimpanzees throw — famously and accurately enough to be unpleasant at a zoo — but at roughly a third of the speed, because their shoulders cannot store the energy.

The genuinely secure projectile evidence is younger and made of wood. The Schöningen spears from Lower Saxony — eight or more, around 300,000 years old, tapered and weighted forward like a javelin — are the earliest unambiguous hunting weapons, attributed to Homo heidelbergensis. Replica throws by trained javelin athletes have shown they are effective to roughly 20 metres. Everything before that is inference from anatomy.

Where it stands. The claim "throwing is an evolved human specialisation with a datable anatomical signature" is well supported. The claim "throwing made us human" is the same category error as the previous entry: a real capability doing more explanatory work than it can carry. The video's own verdict — "probably not, but more important than most people realise" — is close to the field's.

Brains are metabolically ruinous. The classic answer to how we afford ours is that we traded gut for grey matter — and paid for the trade with animal tissue.

Leslie Aiello and Peter Wheeler set out the expensive tissue hypothesis in Current Anthropology in 1995 (vol. 36, pp. 199–221). Brain tissue costs roughly 20% of resting metabolism in an adult human despite being 2% of body mass. Aiello and Wheeler noted that human guts are markedly smaller than expected for a primate of our size, and proposed a compensatory trade: a higher-quality, more digestible diet permits a shorter gut, and the saved energy funds the brain. Meat and marrow are the obvious quality upgrade.

The archaeological correlate is cut-marked bone. Contested marks at Dikika, Ethiopia push back to about 3.39 million years; more secure assemblages come from Bouri, Kanjera South, and the famous FLK Zinjanthropus level at Olduvai. Briana Pobiner's work at the Smithsonian addresses the supply side: how much edible tissue actually remains on a large carcass after a felid has fed, and for how long. The answer is: more than you would guess, especially in awkward places between bones that a hominin with a sharp flake can reach and a hyena cannot. Bone marrow and brain are the real prize — hammerstone-accessible, calorie-dense, and unavailable to any competitor that cannot break a femur.

This does not mean early Homo was vegetarian. It means the specific narrative — erectus appears, meat consumption jumps, brains expand — is not currently supported by the bone record, and that whatever drove encephalisation may have been more diffuse: a broader, higher-quality diet including tubers, marrow, honey and aquatic resources, plus cooperative foraging and food sharing.

Where it stands. Meat mattered. "Meat made us human" is now the weakest of the four "made us human" claims in this layer, because the pattern it rests on has been shown to be an artefact of excavation history. Note also that the expensive-tissue hypothesis itself has been challenged on comparative grounds: a 2011 Nature analysis of 100 mammal species by Navarrete, van Schaik and Isler found no general negative correlation between brain size and gut size, pointing instead to fat storage and locomotor costs.

Cooking gelatinises starch, denatures protein and destroys toxins. It is the single largest increase in food value our species ever engineered. The fight is entirely about when.

Richard Wrangham's Catching Fire: How Cooking Made Us Human (2009) makes the strong version of the argument: cooking is not a convenience but a biological requirement. Humans have small mouths, weak jaws, small teeth, a short gut and a high-energy brain — a suite that makes no sense on raw food. Wrangham points to the Giessen Raw Food Study, in which long-term raw-food dieters in Germany, eating high-quality modern produce with blenders and dehydrators available, still showed chronic energy deficiency and, in about half of women under 45, amenorrhoea.

The evidence for controlled fire runs deep but thin. Burnt bone and ashed plant material at Wonderwerk Cave, South Africa, around 1 million years ago (Berna et al., PNAS 2012) is the oldest reasonably secure case. Gesher Benot Ya'aqov in Israel, around 790,000 years ago, preserves clustered burnt flint suggesting repeated hearth positions — and in 2022, Zohar and colleagues reported in Nature Ecology & Evolution that tooth-enamel crystallography from carp at the same site indicates fish cooked at controlled low temperature rather than burnt: the oldest direct evidence of cooking.

Where it stands. Cooking's effect is not disputed by anyone. Its timing is disputed by nearly everyone, and the two-million-year version of the hypothesis requires fire evidence that does not yet exist. Wrangham's reply is essentially taphonomic — open-air hearths from 1.5 million years ago would rarely survive — which is fair, and also unfalsifiable, which is the problem.

Between roughly 50,000 and 10,000 years ago most of the world's large land mammals disappeared. Two suspects, one alibi each, and sixty years of argument.

Paul S. Martin of the University of Arizona proposed the overkill hypothesis in 1967 and sharpened it into "blitzkrieg" in a 1973 Science paper. The distinction matters and is usually lost: overkill is human-driven extinction over up to a few thousand years; blitzkrieg is the strong form — a rapidly advancing colonisation front that exterminates naïve fauna within centuries. Martin's model had Clovis hunters sweeping from Alberta to Patagonia in about a millennium.

The chronology: the Last Glacial Maximum ends around 19,000 years ago; the Younger Dryas cold snap runs 12,900–11,700 BP; the Holocene begins as the ice sheets collapse. In North America, roughly 35 genera vanish. Australia lost its megafauna far earlier, around 45,000 years ago, shortly after human arrival — which is the single strongest data point for the human side, because the Australian extinctions do not line up with a climate event. Eurasia lost fewer taxa; Africa, where large mammals co-evolved with hominins for millions of years, lost fewest of all. That gradient — least extinction where humans are oldest — is the overkill argument in one line.

Donald Grayson and David Meltzer have pressed the empirical objection for decades: across an entire continent and a millennium, there are only about fifteen sites where humans can be shown to have killed or butchered now-extinct megafauna, representing just a handful of taxa. The video's rebuttal — that fifteen is actually a large number given how bad the fossil record is — is a real argument in the literature, not a rhetorical dodge, but it cuts both ways: absence of evidence is weak evidence either way.

On the other side, Sandom, Faurby, Sandel and Svenning's 2014 PNAS global analysis found extinction severity correlated better with human arrival than with climate variables. Against them, Stewart, Carleton and Groucutt (Nature Communications, 2021) found North American declines tracked climate more closely than modelled human population growth. Ancient sedimentary DNA (Wang et al., Nature 2021) has complicated everything by showing that mammoths and horses persisted in parts of Beringia thousands of years later than their youngest bones suggest — meaning the extinctions were slower and more staggered than any blitzkrieg.

Everyone alive descends from African populations. That much is settled. Almost everything about the shape of the departure is not.

The genetic version began with Rebecca Cann, Mark Stoneking and Allan Wilson's 1987 Nature paper (vol. 325, pp. 31–36), which used mitochondrial DNA — inherited only from the mother, and therefore a clean single lineage — to place the common ancestor of all living mtDNA in Africa. "Mitochondrial Eve" is a misleading name for a real object: the most recent common ancestor of one molecule, currently dated to roughly 150,000–200,000 years ago. She was not the only woman alive, and the Y-chromosome equivalent lived at a different time (~200,000–300,000 years, older once the deep A00 lineage is included).

The video's account of haplogroup L3 is accurate and is the crux of the genetic case. Every non-African mtDNA lineage on Earth belongs to macro-haplogroups M and N, and both descend from L3, an East African lineage roughly 70,000 years old. If modern humans had absorbed substantial maternal ancestry from Eurasian archaic populations, that pattern would be broken. It is not.

The fossil timeline has moved a long way since 2015. Jebel Irhoud in Morocco, redated by Hublin and colleagues in Nature in 2017, pushed recognisable early Homo sapiens to ~300,000 years — and pointedly, to North-West Africa rather than the East African cradle. Omo Kibish in Ethiopia was redated to at least ~233,000 years by Vidal et al. (Nature, 2022) using volcanic ash geochemistry. Misliya Cave in Israel yielded a ~180,000-year-old maxilla. Skhul and Qafzeh in the Levant date to ~90,000–120,000.

Multiregionalism, associated above all with Milford Wolpoff, held that Homo erectus populations across Africa and Eurasia evolved together into modern humans, held in one species by continuous gene flow. In its classical form it is dead: the mtDNA and Y data, and later the whole-genome data, do not permit it. But its ghost has been rehabilitated in two ways. First, archaic introgression is real — a few per cent of Neanderthal and Denisovan ancestry in living people (entries 11, 27, 36) is precisely the kind of regional continuity Wolpoff argued for, at a much smaller dose than he proposed. Second, Eleanor Scerri and colleagues' 2018 "African multiregionalism" (Trends in Ecology & Evolution) argues our species emerged from a subdivided, intermittently connected network of African populations — a braided stream rather than a single source. The 2023 Ragsdale et al. Nature analysis of weakly structured stem populations supports this.

We are the only primate that is functionally naked, and the only one with a full-body evaporative cooling system. Those two facts are related, but they may not be the whole story.

Three explanations compete, and they are not mutually exclusive. Thermoregulation is the oldest: fur blocks evaporation, so an animal that cools by sweating benefits from losing it — which ties this entry directly to entry 01. Sexual selection is Darwin's, from The Descent of Man (1871), and remains difficult to test. The ectoparasite hypothesis is Mark Pagel and Walter Bodmer's, published as "A naked ape would have fewer parasites" in Proceedings of the Royal Society B (2003, vol. 270, pp. S117–S119). Their argument: fur is habitat. Ticks, lice, fleas and the diseases they carry breed in it. Once hominins became more social and began returning to fixed home bases, parasite load would have risen sharply, and reduced hair would both deny parasites shelter and make them visible for removal — with grooming, already a primate social currency, becoming more effective on bare skin.

Retained hair is the part most people skip. Scalp hair shields the head from direct solar load — the body's most heat-sensitive organ sits directly beneath it. Eyebrows and eyelashes divert sweat and detect contact. Axillary and pubic hair sit over apocrine glands and almost certainly function in scent dispersal, with a secondary role reducing skin-on-skin friction. The video's "trip-wire" framing is a reasonable folk version of the sensory function: terminal hairs are innervated and do register crawling insects.

Where it stands. No single-cause account works. The consensus is a compound: selection for evaporative cooling in an active, day-ranging, savanna-mosaic hominin, reinforced by parasite avoidance as group sizes and site fidelity increased, with sexual selection plausibly maintaining and exaggerating the pattern once it existed. The cost — catastrophic UV vulnerability — was paid for with melanin, which is why skin pigmentation and hairlessness are the same story told twice.

The claim is that psilocybin mushrooms growing in ungulate dung drove the emergence of human cognition. The claim is not merely unproven; its central citation is a misreading.

Terence McKenna, an ethnobotanist and public speaker rather than a research scientist, set out the idea in Food of the Gods (1992). As hominins moved onto grassland and followed herds, they would have encountered dung — and Psilocybe cubensis is coprophilic. McKenna proposed a dose-graded ladder of benefits: low doses sharpen vision (better hunting), moderate doses increase arousal and social bonding, high doses produce the glossolalia and boundary-dissolution he took to be the origin of language and religion.

What psychedelics do do is now reasonably well characterised: Robin Carhart-Harris and colleagues have shown psilocybin desynchronises the default mode network and transiently increases global functional connectivity between regions that normally talk past each other. That is interesting, and it is the seed of legitimate therapeutic research. It says nothing about the Pleistocene.

The real archaeology of intoxication

Here the video is on far better ground, and the material deserves more than it got. There is genuine, published evidence for deep-time psychoactive use:

• Raqefet Cave, Israel, ~13,000 years ago. Li Liu and colleagues (Journal of Archaeological Science: Reports, 2018, vol. 21, pp. 783–793) found starch and phytolith residues in Natufian stone mortars consistent with wheat/barley-based brewing — currently the oldest candidate evidence for beer, predating agriculture.
• Göbekli Tepe, Turkey, ~9500 BCE. Oliver Dietrich and colleagues (Antiquity, 2012, vol. 86, pp. 674–695) described limestone vats up to 160 litres with oxalate residues, interpreted as evidence for large-scale feasting and possibly fermentation at a monumental site built by foragers.
• Cannabis, Jirzankal Cemetery, Pamirs, ~500 BCE. Ren, Tang, Spengler and colleagues (Science Advances, 2019) recovered high-THC cannabis residue in wooden braziers from burials — the clearest early evidence for smoking cannabis for its effects rather than growing it for fibre.
• Coca in the Andes from roughly 8,000 years ago; opium poppy in Neolithic Europe.

The "beer before bread" hypothesis itself is older than either find — Robert Braidwood convened a symposium on it in 1953, and Brian Hayden revived it in 2013 — and remains a minority but respectable position. The residues are suggestive; alternative readings (porridge, ritual food preparation) have not been excluded.

Where it stands. Stoned ape: outside science, and its foundational citation is a misquotation. Ancient intoxication: real, well-published, and considerably more interesting.

A highly heritable condition that reduces reproductive success should vanish. It has not, anywhere. That is a genuine puzzle — and it does not licence calling the condition a gift.

The numbers first, because the video's are loose. Heritability is usually estimated at 60–80% from twin studies. Prevalence is remarkably stable worldwide: Saha, Chant, Welham and McGrath's systematic review of 188 studies (1,721 estimates, 46 countries) found a lifetime prevalence around 0.72%, with no significant difference between men and women or between urban, rural and mixed sites. People with schizophrenia have substantially reduced fertility — in Swedish registry data, roughly 0.23 of the population rate for men. High heritability, global persistence, low fecundity: that is the paradox.

Three explanations have been offered. Tim Crow's, from the 1990s, is the boldest: schizophrenia is "the price Homo sapiens pays for language," a by-product of the cerebral asymmetry that made lateralised speech possible. Balancing selection proposes that risk alleles carry compensating advantages in non-affected carriers — most often framed as creativity or schizotypal cognition. Daniel Nettle and Helen Clegg (Proc. R. Soc. B, 2006) found positive schizotypy associated with higher mating success in artists and poets; Simon Kyaga and colleagues' Swedish national registry study (British Journal of Psychiatry, 2011) found creative professions over-represented among the siblings of people with schizophrenia rather than among patients themselves — an important distinction the popular version always drops.

Where it stands. The paradox is real and unsolved. The "schizophrenia built civilisation" version is not supported and carries a cost the video gestures at but does not fully own: it romanticises a condition with a life expectancy gap of 15–20 years and a suicide rate roughly ten times the population baseline. The interesting scientific question — why a highly polygenic, fitness-reducing psychiatric architecture is conserved across every human population — remains open.

A small-brained hominin, alive at the same time as early Homo sapiens, found only in the most inaccessible chamber of one South African cave. Everything about it is strange; the argument is over how strange.

Cavers Rick Hunter and Steven Tucker reached the Dinaledi Chamber of the Rising Star system in the Cradle of Humankind in September 2013 and found bones on the floor. Lee Berger's team described the species in eLife in 2015; over 1,500 specimens from at least 15 individuals came out of that chamber alone, later joined by material from the Lesedi Chamber. Paul Dirks and colleagues dated the deposits in 2017 to 335,000–236,000 years ago — which was the real shock. This was not an australopith-grade creature from three million years ago. It was contemporary with the earliest members of our own species.

The anatomy is a mosaic: a brain between 465 and 610 cc (chimpanzee range), curved fingers and shoulders built for climbing, but a humanlike wrist, a modern-looking foot, and teeth that place it in Homo. The route to the fossils is the other half of the story — the horizontal squeeze called Superman's Crawl, the jagged dolomite of Dragon's Back, and finally the Chute, a roughly 12-metre vertical fissure that narrows to about 18 centimetres. The original excavation team was recruited by advertisement for slim, experienced cavers with palaeontological training; all six who qualified were women.

In 2023 Berger's team published preprints in eLife claiming three things: deliberate burial in dug graves, controlled fire (soot, charcoal, burnt bone), and engraved cross-hatched markings on the cave walls. If true, a creature with a chimpanzee-sized brain was performing behaviours previously reserved for large-brained hominins, and much of what we assume about the cognitive prerequisites for symbolic behaviour would need rewriting.

The alternatives are less romantic and not yet excluded: a natural death trap (animals falling in and dying over centuries), gradual accumulation, or water transport of remains. Against water transport, there is no flood debris mixed with the bones, and the assemblage is almost entirely hominin — which is genuinely hard to explain by any passive process. That anomaly is why the debate is live rather than closed.

Where it stands. The species is real, the date is solid, the deposit is unexplained. The video's "it would be pretty insane if they buried their dead, used fire, and lived in caves" is the correct instinct — it would be, and that is exactly why the evidentiary bar is high and has not yet been cleared.

They disappear around 40,000 years ago, shortly after we arrive. The coincidence is not in doubt. The mechanism is — and the answer turns out to be partly that they did not entirely disappear.

Tom Higham's 2014 Nature chronology, using ultrafiltered radiocarbon dating across 40 European sites, placed Neanderthal disappearance at roughly 41,000–39,000 years ago, with regional overlap with incoming Homo sapiens of up to about 5,400 years. A more granular 2022 study by Igor Djakovic, Alastair Key and João Cascalheira (Scientific Reports) applied optimal linear estimation to French and northern Iberian assemblages and estimated 1,400–2,900 years of overlap in that region — the figure the video cites, correctly.

The violence argument

Shanidar 3, an adult male Neanderthal from Iraqi Kurdistan, has a partially healed lesion on the left ninth rib. Zollikofer and colleagues (PNAS, 2002) analysed it as a penetrating wound; Steven Churchill and colleagues (2009) reconstructed the geometry and argued it was most consistent with a thrown, relatively low-kinetic-energy projectile — i.e. a light spear delivered from distance rather than a thrusting spear. Since there is little evidence Neanderthals used spearthrowers and some evidence early H. sapiens did, the inference of interspecies violence follows. It is a single specimen with an ambiguous wound, and it is the best evidence there is. Treat it as a hint, not a case.

The genomic answer

Richard Green, Svante Pääbo and colleagues published the draft Neanderthal genome in Science in May 2010 and reported the finding that reframed the whole question: people with ancestry outside sub-Saharan Africa carry Neanderthal DNA. The original estimate was 1–4%; current estimates usually run 1–2%. Because that ancestry is found across every non-African population, the admixture cannot have been a one-off.

2024 dated it precisely. Two papers — Iasi et al. in Science and Sümer et al. in Nature, working from early Eurasian genomes including the Ranis and Zlatý kůň individuals — converged on a main admixture pulse centred around 47,000 years ago, within a window of roughly 50,500–43,500 BP, lasting on the order of 7,000 years. That is a single extended episode shortly after the expansion out of Africa, not a scatter of encounters over tens of millennia.

Assimilation, not annihilation. Fred Smith's assimilation model — Neanderthals absorbed into a much larger incoming population rather than exterminated — fits the genetics well. Neanderthal effective population size was small and fragmented, with documented inbreeding (the Altai individual sequenced by Prüfer et al. in 2014 had parents as closely related as half-siblings) and associated skeletal pathologies at El Sidrón and Krapina. A small, structured, low-diversity metapopulation meeting a larger, better-connected one does not need to be massacred to disappear.

On cannibalism, the video's aside is right and worth stating plainly: Neanderthal cannibalism is well documented — Moula-Guercy in France (Defleur et al., Science, 1999), El Sidrón in Spain, and Goyet in Belgium (Rougier et al., Scientific Reports, 2016). Evidence for H. sapiens butchering Neanderthals is much thinner: a single mandible from Les Rois with cut marks, cautiously discussed by Ramirez Rozzi and colleagues in 2009.

One hypothesis explaining a dozen traits at once. That is its appeal, and precisely why it fails.

The German pathologist Max Westenhöfer floated an aquatic phase in Der Eigenweg des Menschen (1942) and largely abandoned it. The modern version belongs to Sir Alister Hardy, a distinguished Oxford marine biologist, who published "Was Man More Aquatic in the Past?" in New Scientist on 17 March 1960 — having first aired it in a talk to the British Sub-Aqua Club. Elaine Morgan, a Welsh television writer, made it famous in The Descent of Woman (1972) and The Aquatic Ape (1982), partly as a corrective to the aggressively male "killer ape" narratives of the period.

The claimed evidence is a list: hairlessness (drag reduction, as in cetaceans), subcutaneous fat (insulation), bipedalism (wading), voluntary breath control (diving), a descended larynx, vernix caseosa on newborns, "webbing" between fingers, and the infant diving response. It is an impressive list until you check each item separately — which is what John Langdon did in the paper that effectively ended the hypothesis's academic life: "Umbrella hypotheses and parsimony in human evolution: a critique of the Aquatic Ape Hypothesis," Journal of Human Evolution 33(4):479–494 (1997). His argument is methodological. A single hypothesis explaining many traits feels parsimonious, but each trait already has a better-supported separate explanation, and the aquatic account requires an entire ecological phase for which there is no fossil, sedimentological or faunal evidence whatsoever.

The infant diving response deserves its own note because it is the argument most often deployed. It is real: submerge an infant and the heart rate drops, the airway closes, and blood is shunted to the core. It is also the mammalian diving reflex, present in rats, dogs, seals and every other mammal tested. It is a shared ancestral trait of the whole class, hundreds of millions of years old. It is evidence for nothing about hominins specifically.

The claim: that as recently as three thousand years ago, humans had no inner voice — and experienced their own decisions as commands from gods.

Julian Jaynes was a Princeton psychologist, and The Origin of Consciousness in the Breakdown of the Bicameral Mind (1976) is one of the strangest books ever taken seriously by academics. His model: in the bicameral mind, one hemisphere generated linguistic instruction which the other experienced as an external auditory hallucination — the voice of a god, a king, an ancestor. Introspective self-awareness did not exist. It emerged, he argued, only when the bicameral system broke down under the stress of population growth, writing, and civilisational collapse in the second millennium BCE.

His evidence is chiefly literary. In the Iliad, he noted, characters rarely deliberate: gods intervene, and heroes act. There is no rich vocabulary of inner states — psyche, thumos and noos refer to something closer to breath, spirit and perception than to mind. In the Odyssey and later texts, introspection appears. He read the same transition in the Hebrew Bible, in Mesopotamian texts, and in the ubiquity of idols and oracles.

What survives is the adjacent, genuinely active research: Charles Fernyhough's work on inner speech and on the varieties of auditory verbal hallucination, including in people who are not distressed by them; the finding that hallucinated voices' content and interpretation vary strongly by culture; and the historical study of how societies have understood involuntary thought. Jaynes asked a good question — what is the inner voice, and has it always sounded like this? — and gave an answer nobody can test.

The most repeated version of this claim is that we lost a tennis ball of brain in 30,000 years. The most recent analysis says the effect may not exist at all.

The raw comparison is real enough. Upper Palaeolithic Homo sapiens crania average roughly 1,500 cc; modern global averages run nearer 1,350. Neanderthals ranged higher still, up to about 1,750 cc. Taken at face value that is a reduction of around 10%.

In 2021 Jeremy DeSilva and colleagues published an ambitious explanation in Frontiers in Ecology and Evolution: assembling 985 crania spanning 9.8 million years, they placed the reduction at roughly 3,000 years ago — startlingly recent — and proposed that cognition had been externalised into social groups and information systems, drawing an analogy with ant colonies, where collective intelligence permits individual brain reduction.

Two confounds matter regardless. Body size fell through the Holocene as populations shifted to agriculture, and brain size scales with body size — some of the reduction is simply smaller people. And domestication syndrome (entry 34) predicts brain reduction as part of a package, which is a different causal story from outsourced cognition.

One of the few genuinely predictive rules in biology, and it applies to us with embarrassing precision.

Joel Asaph Allen, an American zoologist, published the rule in 1877: within a warm-blooded lineage, populations in colder climates have shorter limbs and appendages. Carl Bergmann's earlier rule (1847) is its companion: colder climates favour larger bodies. Both follow from one geometric fact — heat is produced by volume and lost through surface — so a compact form conserves heat and an elongated one sheds it.

In humans the classic contrast is between circumpolar populations — Inuit, Yupik, Sámi — with short distal limb segments and broad trunks, and Nilotic peoples of South Sudan and Ethiopia, notably the Dinka and Nuer, with exceptionally long, thin limbs and narrow bodies. The measure is the crural index, tibia length divided by femur length: it is low in cold-adapted populations and high in heat-adapted ones. Christopher Ruff's work in the early 1990s formalised this by modelling the body as a cylinder whose breadth tracks climate while height varies for other reasons — a neat resolution of why tall Dinka and tall Scandinavians are tall in different ways.

Neanderthals fit the rule almost too well: short forearms and lower legs, barrel chest, wide pelvis — a body plan convergent with cold-adapted modern humans, which is one reason the "Ice Age specialist" reading of them persists.

Wisdom teeth are not a design flaw. They are a mismatch between a tooth count inherited from the Pleistocene and a jaw shaped by soft food.

Noreen von Cramon-Taubadel's 2011 PNAS study is the key reference: analysing global cranial and mandibular samples, she found that mandibular shape tracks subsistence strategy — hunter-gatherers versus agriculturalists — more strongly than it tracks neutral genetic distance, while the rest of the cranium tracks genetics as expected. In other words, the jaw is responding to what it is used for.

The mechanism is developmental as well as evolutionary. Jawbone grows in response to mechanical stress; a childhood of tough, fibrous, mechanically demanding food produces a longer dental arch than a childhood of cooked cereals and processed food. The tooth count did not shrink at the same rate as the bone housing it, and the third molar — last to erupt, at 17 to 25 — is the one left without room. Hence impaction, crowding, and a routine surgical procedure that would have been unnecessary for most of our species' history. Malocclusion in general is largely a post-agricultural phenomenon.

The cleanest demonstration that natural selection is still producing new human physiology — and one of the most exaggerated stories on the internet.

Melissa Ilardo, working with Rasmus Nielsen and Eske Willerslev, published "Physiological and Genetic Adaptations to Diving in Sea Nomads" in Cell (2018, vol. 173, pp. 569–580). The Bajau of Indonesia have lived a marine subsistence life for over a thousand years, free-diving with weights and hand-carved wooden goggles. Ilardo carried a portable ultrasound to Jaya Bakti and imaged spleens.

The spleen matters because of the mammalian dive response: on submersion the heart slows, peripheral vessels constrict, and the spleen contracts, injecting a reserve of oxygenated red blood cells into circulation. A bigger spleen is a bigger oxygen tank. Bajau spleens were about 50% larger than those of the neighbouring, non-diving Saluan — and, crucially, the enlargement was present in Bajau individuals who do not dive. That rules out training and points to genetics.

The selection signal sits on PDE10A, associated with spleen size via thyroid hormone T4, and on BDKRB2, which relates to the peripheral vasoconstriction component of the dive response. Selection appears to have acted within roughly the last thousand years — a human adaptation that arose within recorded history.

The best-documented case of a human adaptation inherited from an extinct species — and it works by doing less, not more.

At 4,000 metres the air holds about 60% of the oxygen available at sea level. The lowlander response is to overproduce red blood cells, which thickens the blood, raises pulmonary pressure and causes chronic mountain sickness, pregnancy complications and premature death over the long term. Tibetans do the opposite: they keep haemoglobin concentration near sea-level values and compensate with higher resting ventilation, higher nitric-oxide-mediated blood flow, and more efficient oxygen delivery.

Emilia Huerta-Sánchez and colleagues identified the mechanism in Nature in 2014 ("Altitude adaptation in Tibetans caused by introgression of Denisovan-like DNA," vol. 512, pp. 194–197). The gene is EPAS1, which encodes HIF-2α, a master regulator of the hypoxia response. The Tibetan version is an unusually divergent haplotype carried by around 87% of Tibetans and fewer than 10% of Han Chinese — and it matches Denisovan sequence. It entered the modern human gene pool by interbreeding and was then swept to high frequency by selection on the plateau. EGLN1, in the same oxygen-sensing pathway, shows a second, independent signal.

Short stature evolved independently in rainforest foragers on three continents. Convergence that strong usually means a real selective pressure — the argument is which one.

A note on language first: "pygmy" is widely regarded as pejorative and is avoided in current literature in favour of specific group names — Baka, Aka, Efé, Mbuti in Central Africa — or the descriptive "rainforest hunter-gatherers." Similar short stature evolved separately among the Andamanese and Aeta in Southeast Asia and in some Amazonian groups, and genomic work confirms these are independent events rather than shared inheritance.

Three explanations compete. Thermoregulation and mobility: a small body sheds heat better in humid forest where evaporative cooling is inefficient, and moves more easily through dense understory. Growth-hormone axis: George Perry and Nathaniel Dominy (Trends in Ecology & Evolution, 2009) reviewed evidence that African rainforest groups show altered IGF-1 signalling and reduced sensitivity to growth hormone rather than deficiency of it. Life history: Andrea Migliano, Lucio Vinicius and Marta Mirazón Lahr (PNAS, 2007) proposed the most interesting version — under high adult mortality, the optimal strategy is to stop growing early and start reproducing early. Short stature is then not an adaptation to the forest at all, but a side-effect of an accelerated life history in a dangerous environment.

The video's phrasing — that they "evolved to literally mature faster than everyone else" — is a fair summary of Migliano's model, and it is worth noting that the model makes testable demographic predictions about mortality schedules, which is why it has held up better than the purely ecological accounts.

Every mammal shuts down lactase production after weaning. Several human populations independently broke the off-switch — and the reason may be famine rather than breakfast.

The mutations do not sit in the lactase gene itself but in an enhancer region within the neighbouring MCM6 gene that controls LCT expression. The European variant is −13910*T. Sarah Tishkoff and colleagues (Nature Genetics, 2007) identified entirely separate African variants — −14010*C in Tanzanian and Kenyan pastoralists, −13907 and −13915 in Sudan and Arabia. Same phenotype, different mutations, at least four independent origins. Bersaglieri and colleagues (2004) had already shown the European variant carries one of the strongest selection signatures anywhere in the human genome.

The textbook case of balancing selection, and the clearest example anywhere of evolution accepting a catastrophic cost for a survival benefit.

Anthony Allison established it in the British Medical Journal in 1954 ("Protection Afforded by Sickle-cell Trait Against Subtertian Malarial Infection"). Carrying one copy of the HbS allele confers strong protection against Plasmodium falciparum — later work by Thomas Williams and others put the reduction in severe malaria risk at roughly 50–90%. Carrying two copies causes sickle-cell disease. The allele's frequency in any population settles where the protection gained balances the disease cost, which is why HbS frequency maps almost exactly onto historical falciparum endemicity.

The same pressure produced several independent solutions, which is the part the video gets right and is genuinely striking:

Duffy negativity is the most complete of these: it does not mitigate vivax malaria, it very nearly abolishes it, and it is close to 100% frequency across West Africa — an evolutionary victory so total that P. vivax is essentially absent from the region. Miller and colleagues demonstrated the mechanism in the New England Journal of Medicine in 1976.

The first documented case of a human population evolving tolerance to a poison in its water supply.

Groundwater in parts of the Andes is naturally arsenic-rich — in the Camarones valley of northern Chile, concentrations exceed 1,000 µg/L, roughly a hundred times the World Health Organization guideline. People have lived there for around 7,000 years.

Carina Schlebusch and colleagues published "Human Adaptation to Arsenic-Rich Environments" in Molecular Biology and Evolution (2015, vol. 32, pp. 1544–1555), studying the community of San Antonio de los Cobres in the Argentine puna. They found a strong selective sweep around AS3MT, arsenic (+3) methyltransferase, the enzyme that methylates inorganic arsenic into forms the body excretes more readily. Carriers of the protective haplotype convert arsenic to the less toxic dimethylated form more efficiently and excrete more of it. The sweep is estimated at roughly 7,000–10,000 years old, matching the occupation history.

Not evolution, strictly — but the clearest illustration in the medical literature of what a founder effect does in an isolated population.

Martin Fugate, a French orphan, settled at Troublesome Creek in Perry County, eastern Kentucky, around 1820, and married Elizabeth Smith. Both, by extraordinary chance, carried a recessive allele causing deficiency of NADH-cytochrome b5 reductase, encoded by CYB5R3. The enzyme's job is to keep haemoglobin's iron in its ferrous, oxygen-carrying state; without it, methaemoglobin accumulates, and blood that cannot bind oxygen properly appears dark, giving the skin a blue cast.

Four of their children were blue. In an accessible community that lineage would have dispersed and the allele would have stayed vanishingly rare. In the hollows of eastern Kentucky in the nineteenth century — no roads, no railway, a marriage pool of a few families — cousins married cousins for generations, and two carriers meeting became routine rather than miraculous. Blue-skinned children appeared across six generations.

The scientific resolution came in the 1960s when the haematologist Madison Cawein III tracked the family down, established the enzyme deficiency, and treated it with methylene blue — which acts as an electron acceptor, providing an alternative pathway to reduce methaemoglobin. The effect was immediate: patients' skin turned pink within minutes. The last widely reported case was Benjy Stacy, born in 1975, whose alarming colour at birth faded within weeks and who carried only a residual blueness in his lips and fingernails under cold or stress.

As roads reached the region and the marriage pool widened, the allele dispersed back into rarity. The video's framing — "if that happened early in our evolution, maybe we'd all be blue" — is a decent intuition about drift in small populations, which is exactly the regime in which most of human prehistory took place.

The hypothesis is a straightforward application of inclusive fitness. It has been tested repeatedly, and it has failed nearly everywhere it has been tested.

W.D. Hamilton's 1964 inclusive-fitness theory established that helping relatives reproduce propagates shared genes — a full sibling shares 50% of your variants, a niece or nephew 25%. E.O. Wilson applied this to homosexuality in Sociobiology (1975): if same-sex-attracted individuals redirect resources toward nieces and nephews, the alleles involved could persist without direct reproduction. This is the "gay uncle" hypothesis.

The better-supported alternative is sexually antagonistic selection: alleles that reduce reproduction in one sex increase it in the other. Andrea Camperio Ciani and colleagues (Proc. R. Soc. B, 2004) reported higher fecundity among the maternal-line female relatives of gay men. More powerfully, Brendan Zietsch and colleagues (Nature Human Behaviour, 2021) used UK Biobank data to show that heterosexual people carrying more same-sex-behaviour-associated alleles report more opposite-sex partners — a direct mating advantage that would maintain the alleles in the population. The genetic architecture itself was mapped by Andrea Ganna and colleagues in Science (2019): highly polygenic, five genome-wide-significant loci, all common variants together explaining perhaps 8–25% of variation, and no capacity whatsoever to predict any individual's orientation.

The broadest recent framing comes from José Gómez, Adela González-Megías and Miguel Verdú (Nature Communications, 2023, vol. 14, 5719), who mapped same-sex sexual behaviour across 261 mammal species and found multiple independent origins correlated with social living and with lethal intrasexual aggression — supporting a role in establishing and maintaining social bonds and defusing conflict. The authors are explicit that this says little directly about human sexual orientation, which is a different phenomenon from behaviour.

Human females stop reproducing decades before they die. Almost no other mammal does this. The grandmother hypothesis is the leading explanation, and it just lost its cleanest supporting case.

George Williams suggested in 1957 that ceasing reproduction could be adaptive if the risk of late-life childbirth outweighed the benefit and the effort were redirected to existing descendants. Kristen Hawkes and James O'Connell turned it into a research programme with the Hadza of Tanzania, observing that post-menopausal women were among the most productive foragers in camp, digging deep tubers and provisioning weaned grandchildren — which allows their daughters to wean earlier and have the next child sooner. Their key papers appeared in Current Anthropology (1997) and PNAS (1998).

The comparative evidence came from whales. Killer whales and short-finned pilot whales have long post-reproductive lifespans, and Darren Croft, Michael Cant and colleagues showed matriarchs lead pods to salmon during scarcity and that reproductive conflict between generations of females favours an early stop. A 2024 Nature study extended documented menopause to five toothed-whale species. For a while the pattern looked clean: menopause occurs in long-lived, socially complex species where older females hold ecological knowledge.

Stone tools on islands that were never connected to the mainland. Something crossed deep water a million years ago. What it used is the open question.

The Wallace Line, running between Bali and Lombok and up between Borneo and Sulawesi, marks the edge of the Sunda continental shelf. Islands east of it — Flores, Sulawesi, Timor — stayed islands even at maximum glacial sea-level drop, separated by deep, fast, cold water. Yet the archaeology is unambiguous: stone tools at Wolo Sege on Flores date to about 1 million years ago (Brumm et al., Nature, 2010), and Mata Menge yielded hominin fossils around 700,000 years old in paired 2016 Nature papers. Sulawesi has tools at Talepu from ~194,000 years. In the Philippines, the Kalinga site produced a butchered rhinoceros dated to 709,000 ± 68,000 years (Ingicco et al., Nature, 2018).

Two explanations exist. Deliberate watercraft — bamboo rafts, log floats, dugouts — attributes planning, cooperation and probably language to Homo erectus. Natural rafting, argued by Graeme Ruxton and David Wilkinson among others, points out that tsunamis and floods routinely launch vegetation mats carrying animals, that Indonesia is one of the most tsunami-prone regions on Earth, and that a single successful accidental crossing over hundreds of thousands of years is statistically unremarkable. The presence of other large mammals on Flores — Stegodon, giant rats — that certainly did not build rafts is a strong point in favour of passive dispersal.

Where it stands. Deliberate seafaring by Homo sapiens at ~50,000 years (Australia) and ~30,000 years (Ryukyus) is well established. Deliberate seafaring by Homo erectus is a live hypothesis with no direct evidence and a viable null alternative.

A human species discovered from a fingertip, defined by its genome before anyone knew what its face looked like — and given one, at last, in 2025.

In 2008, excavators at Denisova Cave in the Altai mountains recovered a fragment of a juvenile finger bone. In 2010, sequencing by the Max Planck team around Svante Pääbo and analysis with David Reich revealed a lineage distinct from both Neanderthals and modern humans. It remains the only hominin group first identified from DNA rather than morphology.

The finds since then have been extraordinary. Denisova 11, published by Viviane Slon and colleagues in Nature in 2018, is a first-generation hybrid: a girl with a Neanderthal mother and a Denisovan father. The Xiahe mandible from Baishiya Karst Cave on the Tibetan plateau, identified by ancient protein analysis in 2019, is ~160,000 years old and explains where the EPAS1 altitude allele of entry 18 came from; a 2024 Nature paper reported a Denisovan rib from the same cave dated to roughly 48,000–32,000 years.

Denisovan ancestry today is concentrated in Oceania and Island Southeast Asia: roughly 4–6% in Papuans and Aboriginal Australians, with the highest known levels in the Ayta Magbukon of the Philippines. Multiple distinct introgression events have been identified, meaning Denisovans were not one population but a deeply structured set of them across East Asia.

On late survival, the video's claim of persistence to ~15,000 years ago traces to Guy Jacobs, Murray Cox and colleagues (Cell, 2019), who identified several distinct Denisovan introgression signals in Papuan genomes, one of which implies gene flow considerably later than the others — potentially within the last 30,000 years, possibly later still. This is an inference from the length and distribution of introgressed haplotypes, not a fossil, and it is contested. There is no Denisovan bone anywhere in Island Southeast Asia.

A published claim in Nature for hominins in California 130,000 years ago — more than 100,000 years before anyone thinks humans reached the continent.

The site was exposed during highway construction in San Diego County in 1992. Steven Holen, Thomas Deméré and colleagues published it in Nature in 2017 (vol. 544, pp. 479–483): mastodon limb bones showing spiral fractures characteristic of fresh-bone breakage, arranged around large cobbles interpreted as hammerstones and anvils, with uranium–thorium dating on the bone giving 130,700 ± 9,400 years.

If accepted, the implications would be enormous — and would require some hominin other than Homo sapiens, since our species was not in northeast Asia at that date. Neanderthals or Denisovans are the only candidates, hence the video's framing.

The genuinely important pre-Clovis record is elsewhere and much younger. The White Sands footprints in New Mexico — human tracks in ancient lakeshore sediment — were dated to 23,000–21,000 years by Matthew Bennett and colleagues (Science, 2021), and independently confirmed in 2023 by radiocarbon on pollen and optically stimulated luminescence on quartz after the original seed-based dates were challenged. Monte Verde in Chile (~14,500 years) and Cooper's Ferry in Idaho (~16,000) are secure; Chiquihuite Cave in Mexico (~30,000, claimed 2020) remains disputed.

Where it stands. There is no credible evidence that any hominin other than Homo sapiens ever set foot in the Americas. The video's "I'll keep my mind open for future discoveries" is the right posture; the current answer is no.

The claim that Neanderthals were fur-covered apex predators who hunted, ate and raped Homo sapiens, and that terror of them forged our intelligence.

Danny Vendramini is a former theatre and film director from Australia with no training in anatomy, archaeology or genetics, who describes himself as an evolutionary detective. Them and Us: How Neanderthal Predation Created Modern Humans (2009) argues that the reconstruction of Neanderthals as broadly humanlike is a sentimental error, and reimagines them as nocturnal, hyper-aggressive, fur-covered carnivores with forward-facing predatory eyes — reconstructions he illustrated himself.

There is no evidence for any of it. No Neanderthal remains show fur; skin and hair pigmentation genes recovered from Neanderthal genomes indicate ordinary variation in pigmentation, not pelage. No Homo sapiens remains anywhere show Neanderthal butchery. The anatomical reconstructions are freehand redesigns of published fossils.

On cannibalism, the direction of the evidence is the reverse of the hypothesis. Neanderthals demonstrably butchered other Neanderthals — Moula-Guercy in France (Defleur et al., Science, 1999), El Sidrón in Spain, Goyet in Belgium (Rougier et al., 2016). And there is a single cautious report of the opposite: a mandible from Les Rois in Charente with Neanderthal-like features and cut marks, in an Aurignacian layer, discussed by Fernando Ramirez Rozzi and colleagues in 2009. The video's closing line — "maybe we were the bad guys after all" — is, on the current evidence, better supported than the theory it is responding to.

Dreams are disproportionately unpleasant. Any functional theory has to explain why an organ that could rehearse anything spends the night rehearsing disaster.

Antti Revonsuo, a Finnish cognitive neuroscientist, published the threat simulation theory as a target article in Behavioral and Brain Sciences (2000, vol. 23, pp. 877–901). The argument: dream content is biased toward threat far beyond what waking life supplies; that bias intensifies in people exposed to real danger; the threats rehearsed are ancestrally typical (pursuit, falling, attack, social exclusion) rather than modern; and rehearsal in a state where the body is paralysed and the cost of failure is zero is a cheap way to tune threat-perception and avoidance responses.

Three competitors are worth knowing. Revonsuo, Tuominen and Valli's social simulation theory (2015) extends the logic: dreams over-represent social interactions and may rehearse alliance, reciprocity and reputation management — arguably a larger survival problem for a hyper-social primate than predators. Erik Hoel's overfitted brain hypothesis (Patterns, 2021) borrows from machine learning: a network trained too narrowly on its own data overfits and generalises poorly, and dreams inject the biological equivalent of noise or data augmentation — which explains why dreams are weird, a feature no other theory really accounts for. And there remains the deflationary tradition running from Hobson and McCarley's activation–synthesis model (American Journal of Psychiatry, 1977) to G. William Domhoff, in which dreaming is the cortex confabulating a story over random brainstem activation, with no function at all.

A single-celled parasite that can only complete its life cycle inside a cat, and that alters rodent behaviour to get there. Roughly a third of humanity is infected.

Toxoplasma gondii is an apicomplexan whose definitive hosts are felids — it can only sexually reproduce in a cat's gut. Oocysts are shed in faeces, contaminate soil and water, and are ingested by intermediate hosts, in which the parasite forms tissue cysts, particularly in muscle and brain, and waits.

The manipulation is real and elegantly demonstrated. Manuel Berdoy, Joanne Webster and David Macdonald (Proceedings of the Royal Society B, 2000) showed infected rats lose their innate aversion to cat odour — the "fatal feline attraction" — and in some tests are actively drawn to it, while their other behaviours remain normal. The parasite carries genes for tyrosine hydroxylase, a rate-limiting enzyme in dopamine synthesis, offering a plausible mechanism.

What is not in doubt: congenital toxoplasmosis is a genuine and serious risk, which is why pregnancy guidance covers cat litter and undercooked meat; and the parasite's spread was almost certainly amplified by us. Agriculture concentrated grain, grain concentrated rodents, rodents attracted cats, and settled human communities became the ideal amplification environment. In the wild version of the same story, Eben Gering and colleagues (Nature Communications, 2021) found infected spotted hyena cubs approach lions more closely and are more likely to be killed by them.

The claim that our lineage was reduced to about 1,300 breeding individuals for a hundred thousand years — and the methodological fight that followed.

Wangjie Hu, Ziqian Hao, Haipeng Li and colleagues published "Genomic inference of a severe human bottleneck during the Early to Middle Pleistocene transition" in Science in 2023. Using a new coalescent method called FitCoal on 3,154 modern genomes, they inferred that the ancestral population crashed from roughly 98,000 breeding individuals to about 1,280 — a 98.7% loss — between approximately 930,000 and 813,000 years ago, and stayed there for around 117,000 years, losing about two-thirds of standing genetic diversity. Strikingly, the fossil record for this window is nearly empty, which the authors noted as independent support.

Toba: an idea that did not survive testing

Stanley Ambrose proposed in 1998 (Journal of Human Evolution) that the Toba supereruption in Sumatra ~74,000 years ago caused a volcanic winter that nearly exterminated humanity, explaining our low genetic diversity. It was popular for two decades and has been dismantled empirically. Continuous sediment cores from Lake Malawi show no ecological collapse across the ash layer (Yost et al., 2018), and excavations at Pinnacle Point in South Africa show continuous occupation through the Toba ash horizon, with cryptotephra shards identified in situ (Smith et al., Nature, 2018).

The genuinely secure bottleneck is smaller and later: the out-of-Africa founder effect, roughly 50,000–70,000 years ago, when a subset of an African population founded every non-African lineage on Earth. This is why non-African populations carry a subset of African genetic diversity, and why African populations remain the most genetically diverse today.

The counterexample is Neanderthal. Their effective population size was small, structured and inbred: the Altai individual sequenced by Prüfer and colleagues (Nature, 2014) had parents as closely related as half-siblings, and skeletal pathologies cluster at El Sidrón and Krapina. The 2024 description of "Thorin," a late Neanderthal from a lineage isolated for some 50,000 years, reinforced the picture of small groups that stopped exchanging genes. The video's framing — that we survived partly because a large, diverse population stayed in Africa as a genetic reservoir — is a fair reading of the contrast.

An adult human skull genuinely does resemble a juvenile chimpanzee's. The trouble starts when that observation is promoted into an explanation of everything.

The Dutch anatomist Louis Bolk proposed "fetalization" in 1926: humans are apes that reach sexual maturity while retaining juvenile morphology. Stephen Jay Gould gave the idea its serious modern treatment in Ontogeny and Phylogeny (1977), setting it within the general framework of heterochrony — evolutionary change by altered developmental timing.

The observations are real. A juvenile chimpanzee has a rounded braincase, a flat face and small jaws; as it matures the face projects forward, the brow ridges thicken and the skull becomes robust. Human skulls never make that transition. Our foramen magnum stays relatively central, our brain-to-body ratio stays high, our dentition erupts late, our childhood is extraordinarily prolonged, and our brains retain synaptic plasticity for decades. The functional pay-off Gould emphasised is real: a long dependent childhood is the developmental window in which language, culture and technical skill are absorbed.

The proposal that we bred ourselves for tolerance — by systematically removing the intolerant.

Richard Wrangham's The Goodness Paradox (2019) states the mechanism most clearly: humans are simultaneously the least reactively aggressive and among the most proactively aggressive of primates. His resolution is that language and weapons let coalitions of subordinate males conspire against and execute a bullying alpha — capital punishment as a selective force. Over time, that removes the genes for hair-trigger reactive aggression. Brian Hare's parallel work approaches it through social cognition and tolerance.

The predicted signature is domestication syndrome: the cluster of traits — reduced facial projection, smaller teeth, depigmentation, floppy ears, curled tails, extended juvenile behaviour, reduced sexual dimorphism — that appear together across domesticated species even when only tameness was selected. Adam Wilkins, Wrangham and Tecumseh Fitch proposed in Genetics (2014) that this is explained by mild deficits in neural crest cells, the embryonic population that contributes to adrenal glands (hence tameness) and to facial cartilage, pigment cells and ear cartilage (hence the morphology). Zanella and colleagues (Science Advances, 2019) added evidence implicating BAZ1B dosage in craniofacial development.

For humans, Robert Cieri and colleagues (Current Anthropology, 2014) documented reduction in brow-ridge projection and facial masculinity over the last ~80,000 years and linked it to reduced testosterone reactivity. Bonobos are the comparative case: split from chimpanzees perhaps 1–2 million years ago, they resolve conflict through social bonding rather than aggression, and show reduced canines, paedomorphic skulls and depigmented lips — self-domestication without a domesticator.

The effect is well known and reasonably well documented. The evolutionary story attached to it in this video is the most speculative claim in the entry.

Masahiro Mori, a robotics engineer at the Tokyo Institute of Technology, published "Bukimi no Tani" in the Japanese journal Energy in 1970 — three pages, no data, an intuition and a graph. Affinity rises with human likeness until, near the top, it collapses into revulsion, then recovers as likeness becomes indistinguishable. Mori added that movement deepens the valley: a moving prosthetic hand is more disturbing than a still one. The essay was largely unknown in English until Karl MacDorman and Norri Kageki's authorised translation in IEEE Robotics & Automation Magazine in 2012.

Explanations compete, and none is settled. Pathogen avoidance (in the tradition of Paul Rozin's disgust research): a near-human that looks subtly wrong resembles a diseased or dead human, and disgust is the appropriate response. Mortality salience (MacDorman): androids evoke unconscious awareness of death. Mate selection: subtle abnormality signals poor genetic quality. Categorical perception / perceptual mismatch (Burleigh, Schoenherr and others): the discomfort arises from a stimulus that sits on a category boundary and disrupts configural face processing — currently the explanation with the most experimental support, and the least evolutionary storytelling.

One genuinely interesting comparative datum: Shawn Steckenfinger and Asif Ghazanfar (PNAS, 2009) found macaques look longer at real and at cartoonish monkey faces than at realistic synthetic ones — an uncanny-valley-shaped response in a species with no exposure to robots, suggesting something about face processing rather than about culture.

Populations we have never found a bone of, detectable only as anomalies in the DNA of living people.

Arun Durvasula and Sriram Sankararaman published "Recovering signals of ghost archaic introgression in African populations" in Science Advances (2020, vol. 6, eaax5097). Analysing 405 West African genomes — Yoruba, Esan, Mende and Gambian — they found segments too divergent to be explained by modern human demographic history, and inferred introgression from an archaic lineage that split from the modern human line before the Neanderthal divergence.

Earlier signals pointed the same way: Michael Hammer and colleagues (PNAS, 2011) found evidence of archaic admixture in African populations from resequencing data, and Lorente-Galdos and colleagues (Genome Biology, 2019) reported similar signals. The picture that emerges is that Africa had its own archaic populations, its own late-surviving lineages, and its own introgression events — the continent was never a single undivided reservoir of modern humans.

The ghosts extend beyond Africa. Alan Rogers, Nathan Harris and Alan Achenbach (Science Advances, 2020) found that Denisovans themselves carry introgression from a "super-archaic" lineage plausibly related to Homo erectus — meaning some of the archaic DNA in living Melanesians arrived by a chain of at least two prior admixture events. This is what the video is describing in its Southeast Asia segment, and it is correct in outline.

Where it stands. The statistical signal is solid and replicated by independent methods. The populations behind it are entirely unknown — no skull, no jaw, no finger bone. The video's closing hope, that a fossil turns up, is the field's too.