David Reich — Bronze Age shock, the Neanderthal puzzle, & the sudden spread of farming

Summary

David Reich, professor of ancient DNA at Harvard, returns to Dwarkesh’s podcast to discuss a new paper with collaborator Ali Akbari that overturns a long-standing consensus in human evolution: that natural selection has been quiescent in our species since the agricultural revolution. By scaling ancient DNA sequencing two orders of magnitude (now 22,000 genomes from Europe and the Middle East going back ~10,000 years) and developing a new statistical methodology, Reich and Akbari find that selection has actually intensified over the last 10,000 years — and was strongest during the Bronze Age, roughly 3,000-4,000 years ago. Their selection statistic shows a 5-fold enrichment of mutations affecting traits, and the genetic predictor of cognitive performance moved by roughly one standard deviation over the period (from the median to the 85th percentile of today’s distribution), with most of the motion happening between 4,000 and 2,000 years ago.

The middle of the conversation explores what kinds of traits were selected. Immune traits (including tuberculosis resistance), body fat reduction, and traits associated with cognitive performance and schizophrenia risk show strong signals. Reich offers a speculative model: shamanistic societies, oral traditions, and early religion may have created selection pressure on neuroticism, vivid mental imagery, and religiosity, while reducing the fitness of traits like schizophrenia — a kind of “social selection” specific to civilization. The body fat trajectory contradicts the popular narrative that agriculture made people unhealthier; Reich notes that on the time scale of the data, the genetic shift is toward less body fat, suggesting agricultural populations adapted to a peculiar nutritional regime.

The final stretch is on the Neanderthal puzzle. After the recording ended, Reich spontaneously walked Dwarkesh through a heretical new model he’s been working on. The standard story (Neanderthals are an archaic lineage we interbred with a little) doesn’t fit the evidence: mitochondrial and Y-chromosome divergence times between Neanderthals and modern humans are wildly inconsistent if you assume a clean split. Reich’s alternative: Neanderthals are essentially genetically-swamped modern humans. A small population near the Caucasus invented Middle Stone Age technology ~300,000 years ago and expanded outward. Those moving into Europe interbred with archaic locals, got genetically swamped (5% modern, 95% Neanderthal genetics) but retained their culture. The same expansion went into Africa, met more diverged archaic Africans, and became us. The implication: Neanderthals and modern humans share cultural ancestry — the difference is which archaics they mixed with.

Highlights

”The median has gone to the 85th percentile in 10,000 years”

“The effect of selection has been so strong that compared to 10,000 years ago versus now, the median has gone to the 85th percentile. And that’s a huge effect over the last 10,000 years on something like intelligence or the thing that predicts household income.” — Dwarkesh Patel, summarizing the paper’s headline finding, 33:44

yt-dlp --download-sections "*33:44-35:00" "https://www.youtube.com/watch?v=sRKBGVFVYAw" --force-keyframes-at-cuts --merge-output-format mp4 -o "median-to-85th-percentile.mp4"

”You’re more closely related to a Neanderthal on your mother’s side than to your father”

“If you look at any bit of your DNA that you get from your mother and that same bit on the same chromosome copy you get from your father, typical time they share common ancestor is 1 or 2 million years ago. That’s before the split from Neanderthals and Denisovans.” “So there’s many places in your DNA where you’re more closely related to a Neanderthal on your mother’s side than you are to your father.” — David Reich, 1:21:31 and 1:21:47

yt-dlp --download-sections "*1:21:25-1:22:25" "https://www.youtube.com/watch?v=sRKBGVFVYAw" --force-keyframes-at-cuts --merge-output-format mp4 -o "neanderthal-mother-father.mp4"

”The marvel is just hauled without wheels”

“It’s even more impressive because it’s not only without metal, but it’s without animals and without wheels. Which is crazy, like the marvel is just hauled without wheels.” “Right! Take any person who has an old-world superiority and take them to these places and they will not have it anymore.” — Dwarkesh Patel and David Reich, on Teotihuacán, 1:17:11

yt-dlp --download-sections "*1:17:11-1:18:00" "https://www.youtube.com/watch?v=sRKBGVFVYAw" --force-keyframes-at-cuts --merge-output-format mp4 -o "teotihuacan-no-wheels.mp4"

”I’ve really been almost traumatized by this”

“The work that I’ve been involved in has consistently shown that I was wrong in my biases coming into the work. And I’ve really been almost traumatized by this. Again and again, I’ve come into a project with some kind of guess about what the data was showing, and then the data doesn’t show that… I spent several years trying to make these results go away, as did my colleagues, and we just couldn’t make the results go away.” — David Reich, on discovering Neanderthal interbreeding evidence and Bronze Age selection, 1:54:35

yt-dlp --download-sections "*1:54:35-1:56:30" "https://www.youtube.com/watch?v=sRKBGVFVYAw" --force-keyframes-at-cuts --merge-output-format mp4 -o "traumatized-by-data.mp4"

”The Neanderthal puzzle Reich can’t stop thinking about”

“I remain very, very confused about the relationships between archaic and modern humans. Genetically, the Denisovans and the Neanderthals are sisters. They descend from a common ancestral population five or 600,000 years ago. And that group descends a couple hundred thousand years before, 7 or 800,000 years ago, from the common ancestors of modern humans.” — David Reich, 1:18:03

yt-dlp --download-sections "*1:18:03-1:20:30" "https://www.youtube.com/watch?v=sRKBGVFVYAw" --force-keyframes-at-cuts --merge-output-format mp4 -o "neanderthal-puzzle-setup.mp4"

”Genetic swamping: Neanderthals are modern humans who got absorbed”

“The idea I’m playing with — and probably is wrong, who knows — is that there’s a landscape, this is maybe in the Caucasus area, and as this population spreads, there’s a wave front of expansion and they’re interacting with the local archaic humans. These pioneers at the wave front will sometimes interbreed with the local population. Their DNA will get swamped by the local group so that by the time they make it to the other side they’re largely local.” — David Reich, sketching his Neanderthal swamping model on the whiteboard after recording ended, 1:30:08

yt-dlp --download-sections "*1:29:08-1:31:30" "https://www.youtube.com/watch?v=sRKBGVFVYAw" --force-keyframes-at-cuts --merge-output-format mp4 -o "neanderthal-swamping-model.mp4"

Key Points

• The mainstream view: natural selection has been “quiescent” since ~300,000 years ago (6:08) - The status quo Reich’s paper overturns
• Scaling: 22,000 ancient DNA samples (16,000 European/Middle Eastern) (2:02:25) - Up from ~200 in 2015 — two orders of magnitude
• Methodology innovation: GRM + per-locus correction (1:58:49) - Use a genetic relatedness matrix to control for ancestry, then look for per-locus deviations
• 7,200 positions in the DNA at 50% confidence; 3,600 are real (12:33) - The actual selection hits
• 5x enrichment in trait-associated mutations under selection (2:06:00) - Strong signal
• Selection strongest 2,000-4,000 years ago (30:00) - Bronze Age
• Polygenic score for cognitive performance moved ~1 standard deviation over 10,000 years (32:23) - Within ancestry-controlled populations
• Tuberculosis resistance: a major recent selection signal (21:00) - The biggest infectious disease killer in the world today
• FADS1/FADS2: vegetarian-to-omnivore adaptation (24:19) - Converts plant fatty acids; selected as diet shifted from meat to cereals
• Body fat: genetic shift is toward less body fat since farming (52:58) - Contradicts the “agriculture made everyone unhealthier” narrative on long time scales
• Schizophrenia risk variants under negative selection (46:10) - But may have been advantageous in shamanistic/visionary religious contexts before
• The “collective intelligence” hypothesis (35:42) - Joe Henrich’s theory: division of labor lets each person know less, but Reich’s data argues the opposite — selection drove cognitive predictors up
• Migration effect is huge compared to selection (34:32) - Selection alone explains ~2% of frequency changes; population admixture is ~98%
• Mature human genetic variation already existed 300,000+ years ago (1:07:36) - The clay needed for selection was already in place
• Population size as a constraint on selection (1:00:51) - Beyond ~10,000 people, time-span dominates, not population size
• Why didn’t farming start before the Ice Age? (1:09:41) - Mystery: same population biology, very different cultural trajectory. Climate? Cognitive? Reich doesn’t know
• Yamnaya as “tracer dye” in genetic migrations (1:36:15) - Trackable into Indo-European language family even when DNA is diluted
• Teotihuacán: agriculture without metal, animals, or wheels (1:17:11) - Old World superiority doesn’t survive contact with Mesoamerican cities
• Mitochondrial vs Y-chromosome paradox in Neanderthals (1:27:21) - Mitochondrial divergence is 300-400k years; Y is 700-800k years. The standard model implies sex-biased gene flow events occurring twice with low probability
• The “matrilineal expansion” hypothesis (1:38:55) - One sex-biased expansion + one sex-biased survival event could explain the mt/Y paradox
• Sima de los Huesos site (Spain, 300-400k years ago) (1:28:27) - Evidence of an early population pulse related to modern humans into Neanderthal territory
• The Copernican analogy (1:46:42) - Current archaic-modern human model has Ptolemy-style epicycles; Reich proposes a simpler heliocentric alternative
• Archaeological observation: technology consistently moves with modern humans (1:50:46) - Middle Stone Age technology appears wherever modern human DNA expands. Either learned transmission was easy, or there were genetic differences we don’t yet see
• Anatomically modern humans by ~300k years ago, no cultural quickening for 250k years (1:53:18) - The cultural acceleration only begins ~50k years ago

Mentions

Companies

• Harvard (0:39) - Where David Reich is professor of ancient DNA
• Jane Street (43:38) - Sponsor; hive bucks internal currency for compute allocation
• Crusoe (1:08:31) - Sponsor; fastokens open-source tokenizer with 9x speedup
• Cursor (55:00) - Sponsor; Dwarkesh used it to prep by running parallel LLM queries

Products & Technologies

• Genetic Relatedness Matrix (GRM) (1:58:49) - The methodological backbone; controls for ancestry change over time
• Polygenic scores (32:23) - Aggregate trait predictors built from many small-effect alleles
• Selection statistic (2:05:32) - Gaussian-distributed test for non-zero allele change over time
• Middle Stone Age / Levallois technology (1:31:43) - Stone-tool tradition that spreads with modern humans ~300k years ago
• Mousterian (1:36:00) - Neanderthal-associated stone tool culture
• FADS1 / FADS2 (24:19) - The vegetarian-vs-meat-eating allele complex
• Whole-genome capture / target enrichment (2:10:03) - Method that economically extracts the 1% of skeletal DNA that’s human

People

• David Reich (0:39) - Ancient DNA geneticist at Harvard
• Ali Akbari (5:22) - Reich’s former postdoc; co-author of the new natural selection paper; developed the GRM-based methodology
• Iain Mathieson (1:55:30) - Co-author of the 2015 paper that first analyzed ~200 European/Middle Eastern ancient genomes
• Joe Henrich (35:42) - Author of the “collective intelligence” hypothesis Reich’s data argues against
• Julian Jaynes (51:48) - Origin of Consciousness in the Bicameral Mind — speculative framework for shamanistic-era cognition
• Bhatia (27:24) - Co-author on 2014 African American genome analysis
• Aristarchus of Samos (1:49:03) - Ancient heliocentrist, parallel to Reich’s alternative model
• Copernicus (1:48:00) - The Ptolemy/Copernicus analogy for the Neanderthal puzzle

Sites and Cultures

• Sima de los Huesos (Spain) (1:28:27) - 300-400,000 year old skeletal site showing early modern-human-related DNA
• Teotihuacán (Mexico) (1:16:54) - Mesoamerican pyramid complex; agriculture without metal/wheels/animals
• Yamnaya (1:36:15) - Bronze Age steppe people; Indo-European tracer dye
• Khoisan (1:35:08) - Modern African group carrying ancient archaic ancestry

Surprising Quotes

“I remain very, very confused about the relationships between archaic and modern humans.” — David Reich, on the Neanderthal puzzle, 1:18:03

“These pioneers at the wave front, they’ll sometimes interbreed with the local population, there’s so many of them around that their DNA will get swamped by the local group so that by the time they make it to the other side they’re largely local.” — David Reich, on his Neanderthal genetic-swamping model, 1:29:54

“The work that I’ve been involved in is consistently shown that I was wrong in my biases coming into the work. And I’ve really been almost traumatized by this.” — David Reich, 1:54:35

“Like, you know, this is the power of data. The mainstream view in human evolution in the last several decades has been that natural selection has been pretty quiescent.” — David Reich, 6:08

“Take any person who has an old-world superiority and take them to these places and they will not have it anymore.” — David Reich, on visiting Teotihuacán, 1:17:18

Transcript

Dwarkesh Patel: 0:00 Humans, at least in this part of the world, were wrenched into a way of living that was so different from how the hunter-gatherer life used to be.

Dwarkesh Patel: 0:39 I am back with David Reich, who is a professor of ancient DNA at Harvard. How do you describe what it is that you study?

David Reich: 0:46 I’m a geneticist, and I work on human history and how ancient people relate to each other and to people living today.

Dwarkesh Patel: 0:53 We did an interview two years ago, which ended up being one of the most popular interviews on the show. Today we’re going to talk about your new work on natural selection.

David Reich: 1:25 The dream was that when this ancient DNA field started 16 or 17 years ago, we were going to use it to figure out the experiment of nature — the population genetics of the last 10,000 years.

David Reich: 3:00 You can position that individual exquisitely with respect to other people from whom you have data. But when you are interested in studying selection, you really need many samples at each point in time and the right statistical methods to detect frequency changes.

Dwarkesh Patel: 3:52 Why are frequency changes especially indicative of selection?

David Reich: 4:00 What we’re interested in is the experiment of nature that’s occurred in our history over the last tens of thousands of years.

Dwarkesh Patel: 5:20 So what did you guys find?

David Reich: 5:22 Seven years ago, Ali Akbari, who was at the time a postdoctoral scientist in my laboratory and a few years later became a faculty member, started working on this.

David Reich: 6:08 The mainstream view in human evolution in the last several decades has been that natural selection has been pretty quiescent over the past several hundred thousand years.

David Reich: 6:58 This population had a set of gene frequencies, and these variants popped around randomly, the process known as drift.

David Reich: 7:52 So this made it seem that just selection has been quiescent. Maybe a few hundred thousand years ago, the ancestral human population was under more selection.

David Reich: 9:00 In adaptive natural selection, there are tiny fraction of the total frequency change. The vast majority of it are these migration events.

Dwarkesh Patel: 9:15 Can I ask a clarifying question? Why are we discounting population admixture or replacement as selection?

David Reich: 9:52 It could count, and may count, and probably should count in some respects. But it could also be that this population replacement is not adaptive.

Dwarkesh Patel: 12:00 In this next section, David explains the nitty-gritty of the methodology of this paper.

David Reich: 12:29 7200, which where we’re 50% confident.

Dwarkesh Patel: 12:32 Oh, sorry.

David Reich: 12:33 So I think we’re getting about 7,200 positions in the DNA that have 50% confidence of being real.

Dwarkesh Patel: 12:40 So only half of those are real?

David Reich: 12:48 So 3,600 of them are real.

Dwarkesh Patel: 12:51 Does that also mean that outside of those 7,200, you’re confident the other locations are not under selection?

David Reich: 12:58 No.

David Reich: 12:59 If you look at the 25% probability cutoff there will be tens of thousands. There will be many real ones there too.

Dwarkesh Patel: 14:49 Just to make sure I understand, this is not to say that behavioral or psychiatric or cognitive traits are not under selection.

Dwarkesh Patel: 15:00 Actually likely to be among the locations you’ve identified as under selection.

David Reich: 15:05 That’s exactly right. So it might seem from the analysis that immune traits are highly selected, but that’s because they have the biggest individual-locus effects.

David Reich: 16:31 We’ve been able to prove there’s two reasons.

David Reich: 18:00 Over the time period we’re analyzing, plausibly the whole time period is increased compared to previous periods.

Dwarkesh Patel: 19:46 In the paper you have many examples of intensification of selection around specific positions.

David Reich: 19:58 One of the things we do in this work is we look carefully at many, many positions in the DNA.

David Reich: 21:00 Which is the major infectious disease killer in the world today — tuberculosis. If you look at the trajectory of those variants, they go up sharply.

Dwarkesh Patel: 21:52 The thing it was protecting against was probably another disease?

David Reich: 21:55 Maybe.

Dwarkesh Patel: 22:59 One of the big takeaways for me was just that something weird happened in the Bronze Age.

Dwarkesh Patel: 23:52 …FADS1/FADS2, which helps convert plant fatty acids into long chain fatty acids that your body needs. That’s relevant as you go from a diet of meat as a hunter-gatherer to a diet of cereals. Why is that under special selection in the Bronze Age?

David Reich: 24:19 FADS1-2 — this variant is a vegetarian/meat-eating adaptation. Already in work prior to this we’d shown selection on it.

David Reich: 26:27 That happens in the Neolithic, beginning 10 or 11 or 12,000 years ago, and spreads into Europe after 8,500 years ago. But what we’re seeing is that the wrenching process moving into the Bronze Age was qualitatively greater than the degree of the wrenching that happened from the Neolithic.

Dwarkesh Patel: 27:24 You did some work with Bhatia in 2014, looking at 20 or 30,000 African American genomes.

David Reich: 28:29 That may be the case. It also may be that the period is just too short to see much effect.

David Reich: 30:00 Look to see in what time is the natural selection strongest, and the time period is really 2,000 to 4,000 years ago.

Dwarkesh Patel: 32:15 The standard deviation here is how much the polygenic score for the trait itself moves?

David Reich: 32:23 How much the polygenic score moves over a 10,000-year period within a population that is held constant in terms of its ancestry.

David Reich: 33:00 We’ve divided up our whole data set into an archipelago of little populations in different places in space and time.

David Reich: 33:33 What we’re doing here is we’re measuring the strength of selection at each point in time after correcting for the big population.

Dwarkesh Patel: 33:44 Okay, so the effect here is huge then — if you’re saying one standard deviation, a standard deviation above the median would be somebody in the 85th percentile. So you’re saying that the effect of selection has been so strong that compared to 10,000 years ago versus now, the median has gone to the 85th percentile. And that’s a huge effect over the last 10,000 years on something like intelligence or the thing that predicts household income.

David Reich: 34:32 One thing you can see in the data is the migration impact is huge.

Dwarkesh Patel: 35:42 There’s this person who has this theory called collective intelligence hypothesis — this idea that selection has been against intelligence because of division of labor.

Dwarkesh Patel: 36:00 Only need to understand a small and smaller part of the world, therefore the ancients were much smarter than us.

David Reich: 36:58 This is the power of data. I think Joe, if you asked him before this work, what he would have predicted is the opposite of what we found.

David Reich: 39:00 If you control for that for number of years of schooling, all of the signal of years of schooling goes away.

Dwarkesh Patel: 39:41 Years of schooling was not itself a meaningful thing in the past.

David Reich: 40:06 Years of schooling is correlated with many other things.

David Reich: 42:00 Worlds where the populations have been essentially completely disconnected, and there’s no way by chance that the trajectory in Europe and East Asia would be the same.

Dwarkesh Patel: 42:52 Okay, just to make sure I understood — you’re looking at ancient DNA in Europe and saying selection happened here. And then you separately go and look at modern East Asian populations and the same trajectory is present.

David Reich: 43:37 Correct.

Dwarkesh Patel: 43:38 [Jane Street sponsor segment]

Dwarkesh Patel: 44:44 Stepping back, what does this tell us about what actually changed culturally?

Dwarkesh Patel: 45:00 It’s surprising to me that things like intelligence or lack of schizophrenia were under selection.

David Reich: 46:10 Great question. The human adaptation — selection is very effective.

Dwarkesh Patel: 46:59 When the Bible is being written, especially the Old Testament, is exactly when selection for intelligence is at the highest?

David Reich: 47:08 Yeah, exactly. But there are values about strength or courage or religiosity.

Dwarkesh Patel: 47:47 Obviously the thing we’re referring to is not direct performance on an IQ test.

Dwarkesh Patel: 48:00 IQ test intelligence is not that correlated with “here is a new world environment and go figure out how to process food.”

David Reich: 48:36 I’m being very speculative. Let me give you two examples.

David Reich: 51:00 Being anxious or being imaginative or being neurotic might be helpful in a shamanistic tradition, in a religious tradition where you’re expected to communicate with the divine.

Dwarkesh Patel: 51:48 Julian Jaynes has this famous theory in the origins of consciousness in the bicameral mind.

David Reich: 52:10 You’re being asked to believe in visions.

David Reich: 52:31 It’s super interesting to imagine — to ask the question why certain traits are not always advantageous.

Dwarkesh Patel: 52:49 Another trait you find under selection is the trend away from body fat since the agricultural revolution. Why?

David Reich: 52:58 What you see is a reduction in the combination of genetic mutations that make you at risk for obesity.

David Reich: 54:00 Maybe what you’re seeing is that this group of people in Europe and the Middle East over the last 10,000 years has moved into a peculiar nutritional regime.

Dwarkesh Patel: 54:35 The data goes against the common story — the common story is that hunter-gatherers were actually healthier.

David Reich: 55:20 Time scale issue. You’re absolutely right.

Dwarkesh Patel: 56:35 A random question — given that you mentioned other things mattered much more than IQ test intelligence, we’re currently building AI systems to make as smart as possible.

Dwarkesh Patel: 57:00 The reason that’s relevant is we’re currently building AI systems trying to make them as smart as possible.

David Reich: 57:11 We have lots of energy to spend on it.

Dwarkesh Patel: 57:13 And at the same time making sure they’re not schizophrenic. If intelligence is just one thing humans are evolving for…

David Reich: 57:31 There’s more room at the top for a lot of these traits.

Dwarkesh Patel: 58:01 So all of this evolution since out of Africa is acting on alleles that already existed.

David Reich: 58:53 That’s a rich question. The human population has, within for complex traits, a tremendous amount of variation.

David Reich: 1:00:00 Wait for the mutation to occur in some people. And when the populations are relatively small, only 10,000 people, you might have to wait a long time.

Dwarkesh Patel: 1:00:51 How likely is it that the thing that changed at the Bronze Age is just that the human population was big enough?

David Reich: 1:01:13 I think that’s not likely to be true, but it’s an extremely interesting thing to think about. Already when population sizes were ~10,000 people.

David Reich: 1:03:00 Work in small populations or large populations — it’s not going to be affected by population size.

Dwarkesh Patel: 1:03:05 Interesting — you’re saying that once you hit a given threshold of population, the dominant factor is time span, not population size.

David Reich: 1:03:10 Correct. It’s very interesting. And it’s actually not widely understood.

Dwarkesh Patel: 1:03:17 Speaking of data contradicting what you might have otherwise assumed, one of the papers you sent me beforehand…

David Reich: 1:03:58 50,000 years ago or so, or 100,000 to 50,000 years ago, there’s a quickening of the pace of change in culture.

Dwarkesh Patel: 1:05:56 And this group, 50,000 years ago, they are the ancestors of everybody out of Africa or also some Africans?

David Reich: 1:06:02 This is the population ancestral to West Africans, to most East Africans, to all non-Africans.

Dwarkesh Patel: 1:06:50 The differences we see between different groups of people, especially if this group 50 to 100,000 years ago is essentially identical to modern people genetically — everything we see now was already there.

David Reich: 1:07:01 Yeah.

Dwarkesh Patel: 1:07:02 So basically everybody in the world or almost everybody in the world — the variants we see between different humans today were already in this 50,000-year-ago group.

David Reich: 1:07:36 A lot of us in human genetics think that our population contains within it the clay that’s needed to make almost any kind of trait.

Dwarkesh Patel: 1:08:31 [Crusoe sponsor segment]

Dwarkesh Patel: 1:09:41 So we were talking earlier how there’s no fixed differences between humans 50,000 years ago and humans today. If there’s no genetic difference, why didn’t farming start before the Ice Age?

David Reich: 1:10:06 That is such an interesting question. Genetically we’re there. The common ancestor population has all the ingredients for it.

Dwarkesh Patel: 1:11:56 So you increased the range there. 100,000, 200,000, 300,000 years ago.

Dwarkesh Patel: 1:12:00 Based on the genetic differences between modern people and people from even 300,000 years ago, you think basically there’s no genetic difference?

David Reich: 1:12:10 I don’t know — I’m thinking about this all the time right now.

Dwarkesh Patel: 1:14:11 The climate thing seems surprising given there were so many different environments in which agriculture was independently developed.

David Reich: 1:14:39 It’s very, very surprising and we accept it, but it’s a crazy observation that most normal people don’t realize.

David Reich: 1:15:00 These people all have in place the cognitive, behavioral, intellectual ingredients for the farming revolution.

Dwarkesh Patel: 1:16:23 Is it possible that agriculture existed but you didn’t have modern metallurgy or whatever allowed populations to explode?

David Reich: 1:16:41 I think we would see that archaeology. The extraordinary developments in the Americas, which are entirely Stone Age.

Dwarkesh Patel: 1:16:52 You would see them today even if they had gone to completely vanish.

David Reich: 1:16:54 Oh yeah, we should go for a trip to Teotihuacán in Mexico and it’s so impressive.

Dwarkesh Patel: 1:17:11 And it’s even more impressive because it’s not only without metal, but it’s without animals and without wheels. Which is crazy, like the marvel is just hauled without wheels.

David Reich: 1:17:18 Right! Take any person who has an old-world superiority and take them to these places and they will not have it anymore.

Dwarkesh Patel: 1:17:55 What are other questions you have that you’re either investigating right now or want to investigate?

Dwarkesh Patel: 1:18:00 These kinds of big-picture questions of human history.

David Reich: 1:18:03 I’m perplexed. I don’t know if we’ve talked about it before, but I remain very, very confused about the relationships between archaic and modern humans. We have genome sequences now from archaic humans who lived in Europe and Central Eurasia and the Neanderthals. We have archaic sequences from these enigmatic Denisovans who we now have a skeleton for. So genetically, the Denisovans and the Neanderthals are sisters. They descend from a common ancestral population 5 or 600,000 years ago. And that group descends a couple hundred thousand years before, 7 or 800,000 years ago, from the common ancestors of modern humans.

Dwarkesh Patel: 1:20:18 And then connecting them would be interbreeding events or being in the same place at the same time that we missed?

David Reich: 1:20:24 There’s a known interbreeding event from the lineage leading to modern humans into Neanderthals, but it’s supposed to be only 5%.

Dwarkesh Patel: 1:21:00 So when was this interbreeding event?

David Reich: 1:21:02 300,000 to 200,000 years ago.

Dwarkesh Patel: 1:21:04 So the common ancestor between Neanderthals and most humans alive today is potentially more recent than the common ancestor between modern humans?

David Reich: 1:21:14 Oh for sure.

Dwarkesh Patel: 1:21:15 Which is crazy.

David Reich: 1:21:15 Not the divergence of all archaic humans including Denisovans is within human variation.

Dwarkesh Patel: 1:21:24 Wait what?

David Reich: 1:21:25 The average time to the common ancestor of any two human genes is 1 or 2 million years ago.

David Reich: 1:21:31 If you look at any bit of your DNA that you get from your mother and that same bit on the same chromosome copy you get from your father, typical time they share common ancestor is 1 or 2 million years ago. That’s before the split from Neanderthals and Denisovans.

David Reich: 1:21:47 So there’s many places in your DNA where you’re more closely related to a Neanderthal on your mother’s side than you are to your father.

Dwarkesh Patel: 1:21:55 And I’m sure there’s a simple explanation but how?

David Reich: 1:21:58 It’s the same reason that if you have a sister, in some places in your DNA you’re more closely related to her than to your other relatives.

Dwarkesh Patel: 1:22:21 That’s insane.

Dwarkesh Patel: 1:22:23 What are the other big questions?

David Reich: 1:22:25 I think that’s the main thing I’m thinking about a lot these days.

Dwarkesh Patel: 1:22:59 After the recording ended, David started spontaneously explaining a new theory he’s working on about Neanderthal genetics on a whiteboard.

David Reich: 1:23:08 The thing I’m thinking about a lot recently is the possibility that maybe we’re not thinking in the right way about the relationship between archaic and modern humans.

David Reich: 1:23:28 These two — Neanderthals and Denisovans — separated earlier, maybe 700 to 800,000 years ago, from the ancestors of modern humans.

David Reich: 1:24:00 That happened maybe 200 to 300,000 years ago. And that resulted in modern humans contributing DNA to the ancestors of Neanderthals.

David Reich: 1:25:22 I’m interested in the possibility that actually the right way to think about Neanderthals is as somehow culturally modern human.

David Reich: 1:27:00 Landscape in Europe — mostly picks up the local DNA and becomes mostly archaic genetically, but retains its modern human culture.

David Reich: 1:27:21 If you look at the mitochondrial DNA — people get from their moms — Neanderthals and modern humans share an ancestor.

David Reich: 1:27:40 If you look at the Y chromosome you get from your dad, Denisovans and modern humans share an ancestor more than 700 or 800,000 years ago.

David Reich: 1:27:51 If you look at the Y chromosome, it’s only 3 to 450,000 years. So what the current genetic work is asking us to believe is that events involving sex-biased gene flow happened twice.

David Reich: 1:28:05 The probability of this occurring by chance is low, maybe 5% times 5%.

David Reich: 1:28:27 There’s amazing data from this site in Spain that’s like 2 to 400,000 years old.

David Reich: 1:28:57 It really looked like there was a population related to modern humans that pushed into this Sima de los Huesos-like population.

David Reich: 1:29:08 The idea I’m playing with — and probably is wrong, who knows — is that there’s a landscape, this is maybe in the Caucasus area.

David Reich: 1:29:26 As this population spreads, there’s a wave front of expansion and they’re interacting with the local archaic humans.

David Reich: 1:29:54 These pioneers at the wave front, they’ll sometimes interbreed with the local population, there’s so many of them around.

David Reich: 1:30:00 That their DNA will get swamped by the local group so that by the time they make it to the other side they’re largely local.

David Reich: 1:30:08 Maybe what we’re seeing is this is what’s happened. You have a modern human population that’s matrilineal, for example.

David Reich: 1:31:24 This same group then mixes with Neanderthals and it’s 5% modern here and 95% local here.

David Reich: 1:31:43 So you actually have this key population that makes the Middle Stone Age or Levallois technology.

David Reich: 1:32:40 It expands into archaic humans in Europe, it mixes with the local population, it gets 95% replaced, but still retains its cultural traits.

David Reich: 1:33:00 …so on and so forth. That sort of first appeared in Africa and the Middle East and spread beyond.

Dwarkesh Patel: 1:33:55 I have so many questions. Do you have 15 more minutes?

David Reich: 1:33:57 Yeah.

Dwarkesh Patel: 1:33:58 First of all, what is going on with this group of archaic Africans 1.5 million years ago?

David Reich: 1:34:14 The genetic data suggests — and this is analysis only of modern DNA — that…

Dwarkesh Patel: 1:35:08 This includes the Khoisan and whatever other groups?

David Reich: 1:35:11 All of these groups have this. Maybe it’s in slightly different proportions. Where are these people living? Who knows.

Dwarkesh Patel: 1:36:00 Set in Neanderthals the first time around 300,000 years ago — our ancestors share culture with them, they share the Mousterian stone tools.

David Reich: 1:36:15 Well, it spreads through genes too. If you look at Yamnaya in India, there’s almost no Yamnaya ancestry in India.

Dwarkesh Patel: 1:36:22 Huh?

David Reich: 1:36:23 It’s just diluted, diluted, diluted down.

David Reich: 1:36:52 But it is the tracer dye. It tracks into European languages and culture.

Dwarkesh Patel: 1:37:21 If things are transmitted more through women… let me back up. I don’t understand the matrilineal thing.

David Reich: 1:37:39 The reason I’m talking about these matrilineal or patrilineal expansions is I’m really troubled by the mt/Y mismatch.

David Reich: 1:37:59 If you invoke and assume that there was a matrilineal or patrilineal expansion.

Dwarkesh Patel: 1:38:54 But not both.

David Reich: 1:38:55 It won’t solve the other one so you need to solve the other one. You can solve either by natural selection or by social selection.

David Reich: 1:39:00 Patrilineality and matrilineality are the rule, not the exception in human communities.

Dwarkesh Patel: 1:40:36 I guess I still don’t understand how the maternal — okay, the group spreads and it gets to the next front, and from the existing people, they will have the mitochondrial DNA of the archaic humans.

David Reich: 1:41:03 Yeah.

Dwarkesh Patel: 1:41:04 Why are the people with the archaic mitochondrial DNA not surviving?

David Reich: 1:41:08 It’s a question. There’s multiple possible explanations.

Dwarkesh Patel: 1:41:33 Interesting.

David Reich: 1:41:38 The weakest link in this argument — this argument is probably wrong, but I’m just telling you what I’m thinking about.

Dwarkesh Patel: 1:41:44 Okay. So Neanderthals — 300,000 years ago, our lineage interacts with them, but mostly their lineage survives.

David Reich: 1:42:03 The genetic ancestry doesn’t survive.

Dwarkesh Patel: 1:42:05 So presumably there was also other contact between 300,000 years ago and 70,000 years ago.

David Reich: 1:42:12 Probably. But these are the ones we’re detecting currently.

Dwarkesh Patel: 1:42:15 Is it just sort of like there’s not really an answer? It’s just contingent?

David Reich: 1:42:28 This is not at all surprising given the context. This is 700 or 800,000 years.

David Reich: 1:43:22 Here it’s like 1.2 million years. These are at the edge of not being able to produce children.

David Reich: 1:44:19 This one almost certainly has substantial biological incompatibilities because 300,000 years later we are doing it again.

David Reich: 1:45:00 At the same time. So it really feels like the signature of an explosion of people from one place.

Dwarkesh Patel: 1:45:51 Because otherwise you’d have to believe that Neanderthals independently developed Stone Age technology.

David Reich: 1:45:55 Which is not inconceivable, but like it’s a little bit like believing that farming independently developed in multiple parts of the world.

Dwarkesh Patel: 1:46:00 Right. But it did.

David Reich: 1:46:01 It did.

David Reich: 1:46:03 As I said this is probably wrong. We don’t really know the world we live in.

Dwarkesh Patel: 1:46:34 Do you want to recapitulate the thing you’re saying about the analogy to the Ptolemaic epicycles?

David Reich: 1:46:42 The model we’ve put together collectively about the relationships between modern and archaic humans is similar to the Ptolemaic model.

David Reich: 1:48:00 Convoluted model. Then when Copernicus and colleagues suggested instead that actually what’s happening is everything’s revolving around the sun.

David Reich: 1:48:30 But this is not as fantastic as proposing that everything revolves around the sun rather than the earth.

Dwarkesh Patel: 1:48:41 What is counter-intuitive or unexpected or hard to accept about this alternative model?

David Reich: 1:48:51 I don’t know. Nobody’s thinking about this model right now.

Dwarkesh Patel: 1:48:54 Oh.

David Reich: 1:48:55 It seems like a very natural model to me.

Dwarkesh Patel: 1:49:03 Aristarchus, ancient Greek, had the heliocentric theory because he had done observations about stellar parallax.

David Reich: 1:49:58 That’s a great question. We have to assume that there’s a linkage between the cultural transformations in Africa and Eurasia at this time.

David Reich: 1:50:29 Nobody’s put together the extensive work on modern human substructure with the now extensive work on archaic-modern relationships.

Dwarkesh Patel: 1:50:46 It also seems significant that different groups of humans at this time were capable of adopting Stone Age technology.

Dwarkesh Patel: 1:51:00 Lineages was not so big that you could not show people how to use stone.

David Reich: 1:51:04 Who knows. It could be that this was genetically driven.

Dwarkesh Patel: 1:52:03 And then there’s no fixed differences between modern humans and humans 50,000 years ago. Are there fixed differences with 200,000 years ago?

David Reich: 1:52:23 If you look at the genetic variation going back three or four hundred thousand years.

Dwarkesh Patel: 1:52:58 But anatomically modern, cognitively modern humans exist by the beginning of the Middle Stone Age and before we’re breeding with the Neanderthals?

David Reich: 1:53:18 Anatomically modern humans occur exactly here. It’s the same moment.

Dwarkesh Patel: 1:53:46 But we don’t know what exactly happens if anything between 200,000 years ago and 50,000 years ago.

David Reich: 1:53:56 My understanding is no. They begin to be busy making Levallois stone tools like Neanderthals for 200,000 years.

David Reich: 1:54:00 Stone tools like Neanderthals for 200,000 years. Not more impressive than Neanderthals in any way.

David Reich: 1:54:05 Then there begins to be in the archaeological record a quickening.

Dwarkesh Patel: 1:54:31 Interesting. Thanks for the digression.

David Reich: 1:54:35 The work that I’ve been involved in has consistently shown that I was wrong in my biases coming into the work. And I’ve really been almost traumatized by this. Again and again, I’ve come into a project with some kind of guess about what the data was showing, and then the data doesn’t show that. So when I got involved in the Neanderthal genome project, I was part of a group of scientists who had established that non-Africans were a simple subset of African variation and that there was no evidence at all of Neanderthal interbreeding. I found this very strong evidence of Neanderthals being more closely related to non-Africans than to Africans, and it was very surprising. I thought it must be a mistake. I spent several years trying to make these results go away, as did my colleagues, and we just couldn’t make the results go away.

David Reich: 1:57:00 We compared those ancient people who were the sources of modern Europeans to people in Europe today and we looked at frequency changes.

David Reich: 1:58:49 What we did is we deployed a few innovations to try to improve our power to detect natural selection.

David Reich: 2:00:00 Places in the world where we have particularly fantastic data. The other thing we did is we developed an entirely new methodology.

Dwarkesh Patel: 2:01:25 Just to summarize — you’re trying to make a model that predicts allele frequency changes over time.

David Reich: 2:01:31 Right.

Dwarkesh Patel: 2:01:32 One part is this genetic relatedness matrix which captures how similar different genomes are to each other.

David Reich: 2:01:44 Correct.

Dwarkesh Patel: 2:01:45 And then you have this separate thing — if we look at specific locations, can we just say that oh, this location is more variable than we would expect.

David Reich: 2:02:22 That’s precisely right.

Dwarkesh Patel: 2:02:23 Okay, so what have we learned?

David Reich: 2:02:25 When we analyzed the data this way, we looked at 10 million positions in the DNA that in these 22,000 people, 16,000 of them were European/Middle Eastern.

David Reich: 2:03:00 When you piece them out and count them only once…

Dwarkesh Patel: 2:05:30 So what is a selection statistic?

David Reich: 2:05:32 This is the statistic we use to measure whether a mutation is changing over time significantly.

David Reich: 2:05:44 It can be approximately thought of as a normally distributed Gaussian statistic.

David Reich: 2:06:00 This statistic is about five — we see about a 5-fold enrichment in mutations that affect a trait. Instead of 15% of the mutations under selection…

Dwarkesh Patel: 2:06:46 Just to make sure I understood — in order to figure out what alleles have been under selection, you look at trait-associated alleles.

David Reich: 2:07:31 That’s exactly right. This is a brilliant idea that Ali had. It abandons the traditional approach.

David Reich: 2:08:29 A major concern is that what we’re seeing is not really mutations under selection, but rather both trait-association and selection coming from the fact that we’re looking at coding regions.

David Reich: 2:09:00 And in genes — genes are also the parts of the genome that tend to be associated to traits.

Dwarkesh Patel: 2:09:49 So the thing that’s allowed you to increase the amount of sequences by two orders of magnitude…

David Reich: 2:10:03 There’s been a whole series of improvements. The huge drop in sequencing cost.

David Reich: 2:12:00 It became economically efficient to do it.

Dwarkesh Patel: 2:12:02 Sorry, what was the other 99% of the DNA?

David Reich: 2:12:05 It’s mostly microbial — from bacteria and fungi that colonize a person’s body after they die.

Dwarkesh Patel: 2:13:46 Awesome. David, thanks for your time.

David Reich: 2:13:48 Thank you. Thanks, Dwarkesh.