Summary
For most of the twentieth century, human evolution was drawn as a tree: a trunk with branches, species splitting off and going their own way, replacement clean and final. The genomic revolution has replaced this picture with something messier, more interesting, and more relevant to neurodiversity than it first appears. Homo sapiens did not simply replace the other hominins. We absorbed them. Our genome is a palimpsest, written over with fragments of Neanderthal, Denisovan, and at least two unidentified archaic populations whose fossils we have never found. We are, as the anthropologist John Hawks has argued, less a tree than a braided stream: rivulets of human lineage that split, run separately for tens of thousands of years, then merge again, carrying genetic material back and forth across species boundaries.
This matters for a neurodiversity wiki because some of those archaic fragments are neurological. Neanderthal-derived variants influence brain shape, visual processing, and connectivity patterns in living people. A subset of these variants is enriched in autistic individuals. The deep history of human cognition is not a story of one brain plan winning and the rest disappearing. It is a story of different cognitive architectures braiding together across hundreds of thousands of years, with consequences that are still playing out in every consulting room and classroom.
The braided stream
The tree model assumed that once populations separated, they stayed separated. Neanderthals branched off from the common ancestor roughly 600,000 years ago, evolved independently in Europe and western Asia, and were replaced when Homo sapiens arrived. Clean. Linear. Wrong.
Svante Pääbo’s sequencing of the Neanderthal genome in 2010 proved that interbreeding had occurred: 1–4% of the DNA of every non-African modern human derives from Neanderthals. Subsequent work has revealed that this was not a single event. A Princeton study published in 2025 mapped at least three distinct waves of contact between sapiens and Neanderthal populations spanning approximately 200,000 years: one around 200,000–250,000 years ago, another around 100,000–120,000 years ago, and a final major episode around 50,000–60,000 years ago. Interglacial climate shifts created shared habitats in Europe and central Asia, bringing the populations together repeatedly.
The braided stream metaphor, developed primarily by Hawks, captures what the genetics reveal. Populations diverged for tens of thousands of years, developing distinct anatomies, tool traditions, and — this is the point — distinct cognitive architectures. Then they converged again, exchanged genes, and diverged once more. The resulting genome is not a clean inheritance from a single lineage. It is a weave.
Who interbred with whom
The picture is more complex than “sapiens met Neanderthals.”
Sapiens and Neanderthals. The best-documented case. Non-African modern humans carry 1–4% Neanderthal DNA, acquired primarily during the major interbreeding episode around 50,000 years ago. A 2024 study in Science narrowed the window to 45,000–50,500 years ago, with interbreeding lasting roughly 7,000 years. A 2026 study confirmed a sex bias: the predominant pattern was Neanderthal males and sapiens females. The original authors interpreted this as reflecting social dynamics, but a simpler explanation may be biological. Haldane’s rule predicts that when hybrids between two species have reduced fertility in one sex, it is always the heterogametic sex — in mammals, males. The evidence fits: the Neanderthal Y chromosome is completely absent from modern human genomes, and the X chromosome shows fivefold lower Neanderthal ancestry than autosomes, both consistent with male hybrids being subfertile or sterile. Neanderthal mitochondrial DNA is also absent, suggesting that lineages where the mother was Neanderthal (and children inherited her mitochondria) did not persist. The sex bias we see in the modern genome may reflect not who chose to mate with whom, but which pairings produced offspring who could themselves reproduce.
Sapiens and Denisovans. The Denisovans are known primarily from a finger bone, a jawbone, and a great deal of DNA. Melanesian populations carry 4–6% Denisovan DNA, the highest proportion in any modern population. The Tibetan EPAS1 gene, which enables survival at high altitude by regulating haemoglobin, was acquired from Denisovans. A gene acquired 50,000+ years ago still determines who can live above 4,000 metres.
Neanderthals and Denisovans. In 2018, Viviane Slon and Svante Pääbo sequenced a bone fragment from Denisova Cave and discovered “Denny”: a girl, roughly thirteen when she died around 90,000 years ago, whose mother was Neanderthal and whose father was Denisovan. A first-generation hybrid, proving that interbreeding between these groups was not vanishingly rare — you don’t find a first-generation cross unless it happened routinely.
Ghost populations. At least two archaic hominin populations left DNA in living humans without leaving identifiable fossils. Four West African populations (Yoruba, Esan, Mende, Gambian) carry 2–19% ancestry from an unknown archaic group that diverged from the sapiens lineage 360,000 to over a million years ago. Candidates include late Homo erectus or Homo heidelbergensis, but we don’t know. Separately, evidence points to “super-archaic” ghost populations that interbred with ancestors of Neanderthals and Denisovans before either encountered sapiens.
The picture that emerges is not a family tree. It is a tapestry of repeated contact, gene exchange, and reabsorption across species boundaries, over timescales that dwarf recorded history. The different threads of the human lineage were never as separate as the taxonomy implied.
Different brains, not worse brains
Neanderthal brains were larger than ours, averaging 1,400–1,700 cc against the sapiens average of roughly 1,350 cc. Size alone tells you little. The organisation was different.
CT scanning of Neanderthal and sapiens crania reveals a consistent shape difference. Sapiens brains are more globular — rounder, basketball-shaped. Neanderthal brains were more elongated, rugby ball-shaped. Philipp Gunz and colleagues at the Max Planck Institute developed a “globularity index” and found zero overlap between species. Newborns of both species start with similar elongated brain shapes, but sapiens undergo a “globularisation phase” during early development that Neanderthals did not. The genes associated with this phase (UBR4 on chromosome 1, PHLPP1 on chromosome 18) are involved in neurogenesis and myelination, and people carrying the Neanderthal variants of these genes today have measurably less globular brains.
What did the shape difference mean functionally? Research from Eiluned Pearce and Robin Dunbar at Oxford found that Neanderthal brains allocated proportionally more neural tissue to visual processing and somatic (body-sensing) systems, at the expense of the areas sapiens use for social cognition and extended group coordination. Neanderthals evolved at higher latitudes with lower light levels, and their larger eye sockets and expanded visual cortex reflect adaptation to those conditions. They likely had superior visual acuity, particularly in dim environments, and finer proprioceptive awareness.
The trade-off was social, and the archaeological evidence is more dramatic than Dunbar’s theoretical predictions suggest. Dunbar’s social brain hypothesis would predict Neanderthal group sizes of 120–150 based on neocortex ratio. The genetic and archaeological record tells a different story. At Chagyrskaya Cave in the Altai Mountains, a landmark 2022 study in Nature sequenced DNA from multiple Neanderthal individuals and found they belonged to a community of roughly 10–20 people. Among them were a father and his teenage daughter, and several second-degree relatives, all living at the same time. Genetic diversity was extremely low, consistent with a tiny, closely related group. Broader estimates for Siberian Neanderthal populations suggest communities of fewer than 60 individuals, and effective population sizes across the species of just 3,000–12,000.
These were not isolated findings. A late Neanderthal individual nicknamed “Thorin,” discovered at Grotte Mandrin, belonged to a population that had remained genetically isolated for approximately 50,000 years. The picture that emerges is of very small bands, sometimes living in proximity to other small bands (the same raw materials appear at neighbouring caves, suggesting connected groups), but with far less inter-group mixing than sapiens populations maintained. Sapiens groups were also small in absolute terms, but they were embedded in wider networks of exchange, alliance, and cultural transmission that Neanderthal social organisation apparently did not support at the same scale.
The cognitive implication: Neanderthal intelligence was optimised for intimate, close-knit groups operating in familiar territory with exceptional sensory and spatial awareness. Sapiens intelligence was optimised for maintaining larger, looser networks across wider geographical ranges. Different social architectures, built on different neural allocations, producing different ways of being human.
What the tools tell us
Ludovic Slimak, the palaeoanthropologist directing the Grotte Mandrin project in Mediterranean France, has argued that Neanderthal cognition was not inferior to sapiens cognition but differently organised. His excavation at Mandrin revealed that sapiens arrived in Europe around 54,000 years ago, 10,000 years earlier than previously accepted, occupied the Rhône Valley for roughly 40 years, and were then replaced by returning Neanderthals. The cave records at least nine alternating occupations. Two species, sharing territory, neither permanently displacing the other.
The tool record supports a picture of parallel sophistication. Neanderthal Mousterian technology employed the Levallois technique: a planned, multi-step process requiring the knapper to prepare a stone core through a sequence of strikes before removing the final usable flake. This is not crude work. It demands planning, spatial reasoning, and precise motor control. Late Neanderthal Châtelperronian industries independently developed blade technologies and bone tools previously attributed only to sapiens.
The sapiens Neronian tools found at Mandrin show a different kind of sophistication: standardised microblades under a centimetre long, precision-engineered projectile points consistent with archery technology. Sapiens excelled at miniaturisation and standardisation. Neanderthals excelled at robust, adaptable tool systems for close-range hunting. Different solutions to different ecological problems, reflecting different cognitive strengths.
Slimak’s broader argument, laid out in The Naked Neanderthal (2023), is that Neanderthals should be understood on their own terms rather than measured against a sapiens yardstick. He is notably cautious about claims of Neanderthal symbolic behaviour — sceptical, for instance, of confident assertions about burial practices and religion — while insisting that the cognitive sophistication visible in the tool record, the hunting strategies, and the ecological adaptations constitutes a form of intelligence that does not need to look like ours to count.
Other researchers are less cautious. Evidence from Spanish caves (La Pasiega, Maltravieso, Ardales) dates hand stencils and schematic marks to beyond 64,000 years ago, before sapiens reached Europe. Neanderthals at Cueva de los Aviones collected and processed ochre pigments, creating complex mixtures in shell containers, 115,000–120,000 years ago. Across multiple European sites, Neanderthals collected eagle talons and feathers — not for any practical purpose, but apparently for adornment. Whether these constitute “symbolic thought” depends on what you mean by the term, and this is where the debate sits.
The neurodiversity connection
The Neanderthal variants that persist in modern human genomes are not distributed randomly. They cluster in specific functional domains: immune system genes (HLA variants, still adaptive), skin and hair pigmentation, and — critically for this wiki — neurological development.
Pauly and Casanova (2024), publishing in Molecular Psychiatry, examined Neanderthal-derived polymorphisms in autistic individuals across multiple US populations. Their finding was specific: autistic people do not carry more total Neanderthal DNA than non-autistic people, but a subset of 25 Neanderthal-derived variants is significantly enriched in autistic probands and their siblings. These variants are associated with increased activity in visual processing regions (intraparietal sulcus, occipital cortex, fusiform gyrus) and decreased connectivity in the default mode network — the same pattern Gunz found in people carrying Neanderthal brain-shape variants, and the same pattern documented in autistic neuroimaging studies.
The implication is not that “Neanderthals gave us autism.” It is that archaic variants contributing to enhanced visual processing and reduced default mode connectivity — traits that were adaptive in a Neanderthal cognitive ecology — persist in modern populations, where they contribute to the neurological variation we call the autism spectrum. The traits are ancient. The diagnostic category is modern. The mismatch between the two is informative.
Separately, research has found that Neanderthal-introgressed alleles are enriched in ADHD risk variants (Scientific Reports, 2020). The genetic architecture of ADHD and autism shows considerable overlap (seven shared loci), consistent with the overlap problem documented elsewhere in this wiki (see The overlap problem). Both conditions may draw partly on the same pool of archaic cognitive variation.
The FOXP2 gene offers a more nuanced picture. Neanderthals carried the same two FOXP2 mutations as modern humans — the mutations once thought to be uniquely human and uniquely responsible for language capacity. But a regulatory difference (a POU3F2 binding site) altered FOXP2 expression in Neanderthals, suggesting a similar but not identical language capacity. Other language-associated genes (CNTNAP2, ROBO1, ROBO2) show human-specific changes in non-coding regions that arose after the sapiens–Neanderthal divergence. Language capacity did not switch on at a single point. It was assembled incrementally, with different hominin lineages reaching different configurations of the same underlying architecture.
The deeper picture
What does all this mean for how we think about neurodiversity?
The braided stream model shows that the human genome is not the product of a single lineage’s optimisation process. It is a composite, assembled from contributions by multiple hominin species over hundreds of thousands of years. Some of those contributions enhanced visual processing at the expense of social network capacity. Some shifted the balance between prediction and sensory precision. Some altered how attention distributes across channels.
These are not errors. They are ancient cognitive adaptations, tested across tens of thousands of years in ecological contexts very different from a modern classroom or office. The nervous system of a person who processes visual detail with unusual precision, who finds large social groups draining, who attends monotropically rather than polotropically, may be running neural architecture with deep evolutionary roots — architecture that was adaptive in one umwelt and is labelled disordered in another.
This is not a claim that autism is “just” Neanderthal cognition. The genetics are far more complex than that. But it is a claim that the variation we pathologise today has a deep history, that it was shaped by evolutionary pressures acting on multiple hominin lineages, and that the braided stream model — in which different cognitive architectures merge, coexist, and recombine across hundreds of millennia — is a better frame for understanding neurodiversity than the medical model’s assumption of a single correct brain plan from which some people deviate.
We are a deeply braided species. Our brains reflect that braiding. Neurodiversity is, among other things, the living record of it.
Open questions
How many distinct archaic contributions to modern human neurological variation are there? The ghost populations in West African genomes suggest contributions we cannot yet trace because we lack reference genomes.
Neanderthal cognition was optimised for very small groups — bands of 10–20, genetically isolated for millennia. What happens when cognitive traits shaped for that ecology are carried, via introgression, into a species organised around larger, more interconnected social networks? The traits that made a Neanderthal a superb member of a tiny band — intense perceptual focus, deep local expertise, preference for familiar people and routines — may sit differently in a sapiens social world that demands broad networking, flexible group membership, and rapid social inference across large populations. The parallel to autistic experience in neurotypical-majority environments is suggestive.
Can the specific Neanderthal-derived variants identified by Pauly and Casanova be traced to functional differences in living people beyond the autistic population? If these variants enhance visual processing generally, they should be detectable in non-autistic populations too.
Key sources
- Hawks, J. The braided stream analogy for human evolution. (johnhawks.net)
- Pääbo, S. (2010). A draft sequence of the Neandertal genome. Science, 328, 710–722.
- Slon, V. et al. (2018). The genome of the offspring of a Neanderthal mother and a Denisovan father. Nature, 561, 113–116.
- Huerta-Sánchez, E. et al. (2014). Altitude adaptation in Tibetans caused by introgression of Denisovan-like DNA. Nature, 512, 194–197.
- Gunz, P. et al. (2019). Neandertal introgression sheds light on modern human endocranial globularity. Current Biology, 29(1), 120–127.
- Pearce, E. & Dunbar, R. (2013). Neanderthals, their visual systems and the limits of cognitive adaptation. Proceedings of the Royal Society B.
- Slimak, L. et al. (2022). Modern human incursion into Neanderthal territories 54,000 years ago at Mandrin, France. Science Advances, 8(6).
- Slimak, L. (2023). The Naked Neanderthal. Pegasus Books.
- Pauly, R. & Casanova, E.L. (2024). Enrichment of a subset of Neanderthal polymorphisms in autistic probands and siblings. Molecular Psychiatry.
- Durvasula, A. & Sankararaman, S. (2020). Recovering signals of ghost archaic introgression in African populations. Science Advances, 6(7).
- Princeton (2025). Mapping three waves of Neanderthal–sapiens contact over 200,000 years. Science.
- Hoffmann, D.L. et al. (2018). U-Th dating of carbonate crusts reveals Neandertal origin of Iberian cave art. Science, 359, 912–915.
- Zilhão, J. et al. (2010). Symbolic use of marine shells and mineral pigments by Iberian Neandertals. PNAS, 107, 1023–1028.