Original episode:https://www.youtube.com/watch?v=sRKBGVFVYAw · Timestamps are clickable — they seek the player in place
The host interviews guest David Reich, a professor of ancient DNA at Harvard, about a new preprint from his lab. The central topic is the detection of natural selection in human populations in Europe and the Middle East over the last 18,000 years. Reich explains that previous studies struggled to find strong evidence of selection because its signal is overwhelmed by the much larger genetic changes caused by migration. He then details his lab's new methodology, which uses a massive dataset of over 16,000 ancient individuals to statistically isolate the signal of selection. The study's main finding is that natural selection has been rampant, not quiescent, and that it dramatically accelerated during the Bronze Age. This intensified selection affected traits related to immunity, metabolism, and complex polygenic traits like those predicting cognitive performance. The episode also explores broader mysteries in human evolution, including a new model Reich is developing for the relationship between modern humans, Neanderthals, and Denisovans.
[00:00] Introduction: The surprising intensity of human adaptation during the Bronze Age.[00:38] Guest introduction: David Reich, a geneticist studying human history with ancient DNA.[01:22] Context for the new study: The long-held goal of studying biological change with ancient DNA has been unrealized due to small sample sizes.[03:52] Why study frequency changes? To detect natural selection's response to environmental pressures.[05:21] The study's origin and the mainstream view that recent natural selection has been quiescent.[08:26] The core challenge: Signals of selection are a tiny fraction (~2%) of total genetic frequency changes, which are dominated by migration.[09:16] Distinguishing selection within a population from population replacement/admixture.[12:13] The study's top-line findings: Thousands of locations in the genome show signs of selection.[13:38] Patterns in selection: A strong enrichment for immune and metabolic traits, with weaker signals for highly polygenic behavioral traits.[16:30] Reconciling with past views: Selection has accelerated, especially in the last 5,000 years (the Bronze Age).[18:06] The "evolutionary mismatch": Human biology adapting to the shock of agriculture and high-density living.[19:52] Specific examples of selection inflection points, like the TYK2 variant for tuberculosis risk.[23:02] The Bronze Age as a special period of intensified selection across many different traits.[27:27] Comparison with a shorter timescale: The lack of detectable selection in African Americans over ~300 years highlights the importance of time.[29:37] Selection on complex traits: Pigmentation and predictors for cognitive performance also show selection peaking in the Bronze Age.[35:41] Contradicting the "collective intelligence" hypothesis: Data shows selection for traits predicting intelligence in early civilization, not against.[39:57] The complexity of polygenic scores: The "years of schooling" predictor is correlated with many traits, possibly reflecting a general factor like deferred gratification.[44:47] The puzzle of pre-Bronze Age selection: Why weren't traits like intelligence "maxed out" earlier?[52:51] The "thrifty gene" hypothesis: Selection against body fat since the agricultural revolution.[56:51] Speculation on AI and human potential: If intelligence wasn't the main trait under selection, there may be more "room at the top."[58:04] The latent variation in the "Out of Africa" population was sufficient for all subsequent adaptations.[1:03:20] The "cognitive revolution" puzzle: No evidence of a single genetic "switch" for modernity around 50,000 years ago.[1:10:00] The mystery of agriculture's late start, possibly linked to the unique climate stability of the Holocene.[1:18:04] Current research mysteries: The complex relationship between modern humans, Neanderthals, and Denisovans.[1:23:08] Whiteboard session: Reich explains his new model for Neanderthal origins, linking them to a modern human cultural and genetic expansion ~300,000 years ago.[1:54:40] Methodology deep dive: Reich details the technical and statistical innovations that enabled the study.1. - Claim: The study of biological adaptation using ancient DNA has historically been difficult because the signal of natural selection is very weak (~2% of total genetic change) compared to the overwhelming noise from migration and genetic drift (~98%). - Evidence: [08:43] - Type: Fact
2. - Claim: Contrary to the long-held view that natural selection has been quiescent in recent human history, a new large-scale study of ancient DNA shows that selection has been "rampant" and is detectable at thousands of positions in the genome. - Evidence: [09:07], [12:33] - Type: Opinion
3. - Claim: The new methodology isolates the signal of selection by modeling out the effects of population structure (migration, drift) and looking for consistent, directional changes in allele frequencies across many small, temporarily isolated populations, described as an "archipelago of little populations in space and time." - Evidence: [10:51] - Type: Fact
4. - Claim: The strongest signals of recent natural selection are concentrated in genes related to immunity and metabolism, likely as an adaptation to increased population density, new pathogens from domesticated animals, and dietary changes associated with agriculture. - Evidence: [14:14] - Type: Opinion
5. - Claim: The period of most intense natural selection was not the initial shift to agriculture (Neolithic), but the Bronze Age (roughly 5,000 to 2,000 years ago), suggesting this was a period of greater "evolutionary mismatch" and adaptive pressure. - Evidence: [17:39], [26:32] - Type: Opinion - Uncertainty: The speaker notes this is surprising, as the "cartoon picture is that the big transition is farming" [27:11].
6. - Claim: Some genetic variants show reversals in selection. For example, the TYK2 variant, a major risk factor for severe tuberculosis, increased in frequency before the Bronze Age but then rapidly decreased, possibly due to the rise of tuberculosis as an endemic disease. - Evidence: [20:52] - Type: Example - Uncertainty: The explanation involving tuberculosis is described as "speculative" [21:49].
7. - Claim: Polygenic scores for complex traits, including those that predict performance on cognitive tests and years of schooling in modern populations, also show strong directional selection that peaked during the Bronze Age and has been negligible in the last 2,000 years. - Evidence: [30:40], [31:33] - Type: Opinion
8. - Claim: The genetic predictor for "years of schooling" is highly correlated with many other traits (e.g., age at first birth, obesity, walking pace), suggesting selection may be acting on a more general underlying trait like deferred gratification or a life-history strategy toggling between quality versus quantity of offspring. - Evidence: [38:53], [49:11] - Type: Speculation - Uncertainty: The speaker uses phrases like "I’m just waving my hands" [40:55] to indicate the speculative nature of the interpretation.
9. - Claim: The genetic toolkit for complex behaviors like farming was present in humans long before the Holocene. The independent invention of agriculture in multiple locations around 12,000 years ago may be linked to a unique period of global climate stability, a fact Reich finds "unbelievable" but is told is true by colleagues. - Evidence: [1:10:05], [1:11:01] - Type: Speculation
10. - Claim: Reich proposes a new model for human origins where Neanderthals are the product of a modern human expansion ~300,000 years ago that brought Middle Stone Age technology into Europe and mixed with a local archaic population. This would explain why Neanderthals share mitochondrial DNA and Y-chromosomes with modern humans despite being genetically closer to Denisovans overall. - Evidence: [1:23:08], [1:25:13] - Type: Speculation - Uncertainty: Reich repeatedly states the model is speculative and "probably it's wrong, who knows" [1:29:06].
For a long time, the common wisdom among scientists was that human evolution had slowed down dramatically. The idea was that our species hit a kind of sweet spot tens of thousands of years ago, and since then, natural selection has been pretty quiet, or "quiescent" [06:15]. A new, massive study of ancient DNA from David Reich's lab at Harvard turns this idea on its head. Their findings suggest that, far from being quiet, natural selection has been "rampant" [09:12] in recent human history, especially in the last 5,000 years.
So, why was this powerful force so hard to see until now? Reich explains that the signal from natural selection is incredibly faint. Imagine trying to hear a whisper during a hurricane. The hurricane is migration. Over the last 18,000 years in Europe and the Middle East, huge waves of people moved around, replacing or mixing with previous populations. These migrations caused massive, sudden shifts in the gene pool. Compared to this, the slow, steady pressure of natural selection is a tiny effect. Reich estimates that migration and other random factors account for about 98% of all genetic change, while adaptation from natural selection is only about 2% [08:43]. Previous studies, with much smaller datasets, simply couldn't filter out the overwhelming noise of the hurricane to hear the whisper.
Reich's lab overcame this by using two key innovations. First, they assembled an enormous dataset of over 16,000 ancient human genomes [1:59:18]. Second, they developed a new statistical method to analyze it. Instead of looking at the whole chaotic history at once, they modeled the past as an "archipelago of little populations in space and time" [10:51]. Think of it like a series of isolated islands. They'd look at a small group in, say, Hungary for a few hundred years when not much migration was happening, and then another small group in Britain for a few hundred years. In each of these little "experiments of nature" [11:06], they asked: is a particular gene being consistently pushed in the same direction? If the frequency of a gene for, say, disease resistance is inching upwards in Britain, and also in Hungary, and also in Italy, across many different stable periods, that's a strong sign of selection at work, not just random chance.
When they applied this method, the whisper of selection became a clear signal. They found thousands of locations in the genome that have been under selection [12:33]. The strongest signals were clustered in two main areas: immunity and metabolism [14:14]. This makes perfect sense. The period they studied covers the agricultural revolution and the rise of cities. People started living in much denser communities, side-by-side with domesticated animals, which created a perfect storm for new infectious diseases. Their diets also changed dramatically. The human body was under intense pressure to adapt to these new challenges.
Perhaps the most surprising finding is when this pressure was most intense. You might guess it was during the initial shift to farming in the Neolithic period. But the data shows the opposite. The period of most dramatic adaptation was actually much later, during the Bronze Age (roughly 5,000 to 2,000 years ago) [17:39]. As Reich puts it, the "cartoon picture is that the big transition is farming," but the "biological readout" from our genomes says our bodies were reacting much more strongly to whatever was happening in the Bronze Age [27:11]. This suggests the "evolutionary mismatch" [19:12]—the gap between our hunter-gatherer biology and our new way of life—was even more severe during this later period of intensified agriculture, urbanization, and population density.
The study also looked at more complex traits that are influenced by hundreds or thousands of genes. They found that genetic predictors for traits like skin pigmentation and even cognitive performance also show a strong signal of selection that peaked, once again, during the Bronze Age [31:33]. For the last 2,000 years, selection on these cognitive predictors has been negligible.
Reich is careful to point out that we shouldn't take a term like a "predictor for years of schooling" too literally when looking at the ancient past. There were no schools 5,000 years ago. But the combination of genes that predicts more schooling in modern people is also correlated with many other things, like having children later in life and a lower risk of obesity [38:53]. He speculates that what was actually being selected for was a more general underlying trait, perhaps something like an ability to delay gratification or a life strategy that favored investing more resources in fewer children [49:11].
Finally, the conversation touches on some of the biggest unsolved mysteries in human history. One is why agriculture started so late. Genetically, humans were "ready" for it for tens of thousands of years, but it only appeared around 12,000 years ago, and then in multiple places independently. Reich finds it "unbelievable" but says his colleagues in climate science tell him it's true: the last 12,000 years have been a period of unique and unprecedented climate stability [1:11:01]. Perhaps you simply can't rely on planting crops if the climate is swinging wildly from one decade to the next.
The episode ends with Reich sketching out a new, speculative model for the origin of Neanderthals [1:23:08]. The standard view is that Neanderthals and their cousins, the Denisovans, are a sister group to modern humans. But the evidence is messy. Neanderthals share their maternal line (mitochondrial DNA), their paternal line (Y-chromosome), and their advanced stone-tool technology with modern humans, not Denisovans. Reich's new idea is that Neanderthals might be the product of an early modern human expansion out of the Middle East around 300,000 years ago. As this small group of modern humans spread into Europe, they mixed with a much larger local archaic population. Over time, their genome was mostly "swamped" by the local DNA, making them look genetically like Denisovans. However, they retained the culture and key parts of the genetic lineage (the Y-chromosome and mitochondrial DNA) of the original modern human pioneers. It’s a radical idea that elegantly explains a lot of contradictory data.
[10:51] The "Archipelago" Metaphor. This is Reich's core explanation for his lab's new methodology. He describes how they statistically isolate the faint signal of natural selection by treating history as an "archipelago of little populations in space and time." It's a conceptually dense but powerful way to understand how they cut through the noise of migration.[26:32] The Bronze Age Surprise. The host and Reich discuss the counterintuitive finding that the Bronze Age, not the earlier Neolithic period, was the time of most intense selection. Their exchange highlights why this upends the "cartoon picture" of human history and what it implies about the biological shocks of early civilization.[39:57] Interpreting "Years of Schooling". Reich provides a masterclass in scientific nuance, explaining why we shouldn't take modern polygenic scores literally when applying them to the past. He carefully unpacks how a predictor for "years of schooling" is likely a proxy for a more fundamental life-history strategy, a great example of avoiding simplistic conclusions.[1:10:00] The Mystery of Agriculture's Late Start. A fascinating digression on why farming, which humans were genetically capable of for tens of thousands of years, only appeared in the last 12,000 years. The hypothesis about the unique climate stability of the Holocene is a mind-bending idea that reframes our entire modern era.[1:23:08] Whiteboard Session on Neanderthal Origins. In a segment captured after the formal interview, Reich spontaneously sketches out his new, speculative model for where Neanderthals came from. Hearing him work through this complex, elegant idea in real-time—connecting genetics, archaeology, and population dynamics—is a rare and exciting glimpse into the scientific process at the cutting edge.A faithful reconstruction and plain-language retelling of the episode, generated by PodLens.