David Reich – Why the Bronze Age was an inflection point in human evolution
Geneticist David Reich discusses a major new preprint showing that natural selection has been far more pervasive in the last 18,000 years than previously thought, with a surprising intensification during the Bronze Age (~5,000-2,000 years ago) rather than at the initial transition to farming. The study, using ancient DNA from ~16,000 individuals across Europe and the Middle East, finds strong selection signals on immune, metabolic, and cognitive traits, challenging the long-held view that human evolution has been largely quiescent in recent history. Reich also presents a speculative new model of Neanderthal origins tied to the Middle Stone Age Revolution.
Summary
David Reich, a professor of ancient DNA at Harvard, discusses a major new preprint from his lab led by postdoctoral scientist Ali Akbari. The study uses ancient DNA from approximately 16,000 individuals spanning 18,000 years in Europe and the Middle East — a roughly 14-fold increase over previous datasets — to detect signals of natural selection across 10 million positions in the genome. A key methodological innovation was borrowing a technique from medical genetics: using the correlation between selection signals and known trait-associated variants from genome-wide association studies (GWAS) as an independent calibration tool to validate which selection signals are real.
The central finding challenges the long-held view that human natural selection has been largely quiescent over the last few hundred thousand years. While 98% of allele frequency changes are attributable to genetic drift and migration, the remaining 2% attributed to directional selection is nonetheless pervasive — tugging at positions across the genome. The team identified 479 positions with 99% confidence of being under real selection, and approximately 3,800 with 50% confidence. The genome-wide association with immune and metabolic traits showed a four-to-five-fold enrichment among high-confidence selection signals, while behavioral/cognitive traits showed weaker enrichment due to their highly polygenic architecture.
A striking and unexpected finding is that the intensification of natural selection peaked during the Bronze Age (roughly 5,000–2,000 years ago), rather than at the Neolithic transition to farming (~10,000–8,500 years ago). This is visible across multiple traits: skin depigmentation in Europeans, lactase persistence, metabolic traits related to obesity and Type 2 diabetes, immune traits like the TYK2 tuberculosis risk variant, and even the polygenic score for cognitive performance/years of schooling. Reich argues this reflects a qualitative wrenching of human biology caused by the dramatic shift to high-density, urban, agriculturally intensive, and zoonotic-disease-rich Bronze Age environments.
On cognitive traits specifically, Reich reports that the polygenic score predicting performance on IQ tests and years of schooling in modern white British people shows a roughly one standard deviation increase over the last 10,000 years, with the strongest selection signal between 5,000 and 2,000 years ago and virtually no detectable selection in the last 2,000 years. The validity of this signal is bolstered by a cross-population replication: the same genetic variants that predict years of schooling in Chinese people in China today show a five-to-six standard deviation correlation with the ancient European trajectory, making chance an implausible explanation.
Reich also discusses how European hunter-gatherers score approximately three standard deviations below the modern mean on this polygenic predictor, which he notes is primarily a migration artifact rather than selection. He reflects on the 'collective intelligence hypothesis' — the idea that ancient people may have been smarter — noting the data point in the opposite direction, though he cautions that what's being measured may be a broader executive function or 'deferred gratification' trait correlated with many outcomes, not raw intelligence per se.
On the question of why farming wasn't invented before the Holocene despite genetic readiness, Reich endorses the climate stability hypothesis: the last 12,000 years represent an unusually stable climatic period on a two-million-year timescale, which enabled independent agricultural invention across multiple continents.
In a post-recording whiteboard session, Reich presents a speculative new model of Neanderthal origins. He argues that the standard model — in which Denisovans and Neanderthals are genomic sisters diverging from modern humans ~700,000–800,000 years ago — is increasingly strained by the fact that Neanderthal mitochondrial DNA and Y chromosomes cluster with modern humans rather than Denisovans, pointing to a ~200,000–300,000-year-old interbreeding event. He proposes that a population originating in the Caucasus or Northeast Africa invented Levallois/Middle Stone Age technology and expanded both into Europe (creating Neanderthals via ~95% genomic replacement with local archaics while culturally persisting) and into Africa (contributing ~20% to anatomically modern humans). This model would explain the shared cultural toolkit and uniparental marker clustering of Neanderthals and modern humans, and draws an analogy to the Ptolemaic epicycle problem — a standard model patched with increasing complications rather than replaced by a simpler alternative.
Key Insights
- Reich argues that natural selection intensified most strongly during the Bronze Age (5,000–2,000 years ago), not at the Neolithic transition to farming, suggesting the shift to high-density, zoonotic-disease-rich urban Bronze Age life was a more biologically wrenching transition than the initial adoption of agriculture.
- Reich reports that the polygenic score predicting IQ test performance and years of schooling in modern white British people shows a roughly one standard deviation increase over 10,000 years, with the strongest selection between 5,000 and 2,000 years ago — and the signal is validated by a five-to-six standard deviation cross-population correlation with the same variants' effects on schooling in Chinese people today.
- Reich presents a speculative model in which a population that invented Middle Stone Age/Levallois technology ~300,000 years ago expanded into Europe, interbreeding with local archaics to produce Neanderthals via ~95% genomic replacement — explaining why Neanderthal mitochondrial DNA and Y chromosomes cluster with modern humans despite the rest of their genome clustering with Denisovans.
- Reich explains that across the entire 18,000-year dataset, 98% of allele frequency changes are due to migration and genetic drift rather than directional selection, making natural selection signals extremely difficult to detect — yet the new methodology reveals selection is 'everywhere,' tugging at nearly every position in the genome.
- Reich notes that the TYK2 variant — the major genetic risk factor for tuberculosis — first rocketed up in frequency from 8,000–6,000 years ago to ~9–10%, then reversed and declined sharply in the last 3,000 years, likely reflecting a shift from protection against some prior pathogen to net harm once tuberculosis became endemic.
Topics
Transcript
[0:00] Humans, at least in this part of the world, were wrenched into a way of living that was so different from how their hunter-gatherer ancestors lived that the organism had to adapt very strongly. Maybe the degree of that wrenching process moving into the Bronze Age was qualitatively greater than the degree of the wrenching process that happened from the initial transition to growing plants, which is surprising, because our cartoon picture is that the big transition is farming. But the genetic data, the biological readout, is [0:32] saying our genome is reacting much more strongly to these events that happened 5,000 years ago. I am back with David Reich, who is a professor of ancient DNA at Harvard.…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Dwarkesh Patel
The Skill Great Teachers Have That LLMs Completely Lack - Grant Sanderson
Grant Sanderson discusses a critical limitation of LLMs compared to great human teachers: the inability to reframe or redirect flawed student thinking while validating the creative reasoning behind it. Great teachers can recognize when students approach problems incorrectly and guide them toward better frameworks without dismissing their underlying logic.
The Real Advantage AI Has Over Human Geniuses - Grant Sanderson
Grant Sanderson discusses how AI systems can overcome cognitive biases by systematically adopting different contexts and approaches, using multiple agents with conflicting objectives. He illustrates this with an IMO problem where the elegant intuitive solution was incorrect, and argues that AI's ability to deliberately introduce entropy and diversity could be a key advantage over human thinking patterns.
Why smarter AI models could drive up compute prices 10x
As AI labs like Anthropic scale revenue 10x year-over-year while compute capacity only grows 3x, compute prices must rise significantly to bridge the gap. The speaker argues that as AI models become more capable, they can monetize the same compute at much higher rates, potentially driving prices up 10-15x, creating winner-take-most dynamics and pricing out current AI applications.
Grant Sanderson's Advice for Students
Grant Sanderson advises students pursuing mathematics to think critically about the value they create and where money flows in their chosen career path, rather than simply following their aptitude. He argues that teaching is one of the most stable post-AGI careers due to its relational, mentoring nature that goes beyond just explaining concepts.
The Problem With How LLMs Generate Text - Grant Sanderson
Grant Sanderson explains that autoregressive text generation in LLMs is fundamentally misaligned with human writing processes. Rather than composing with foresight and deep connections, LLMs are forced to predict tokens sequentially with memory wiped between predictions, making it difficult for them to form the unlikely but substantive connections that characterize good writing.