Joseph Henrich——人类凭借文化演化击败了更聪明的物种

Today, I have the pleasure of chatting with Joseph Henrich, who is a professor of human evolutionary biology at Harvard University and an author of two of my favorite books, The WEIRDest People in the World and, before that, The Secret of Our Success.

I was just mentioning to you that I remember reading this many years ago when I was in college, and at the time, I didn’t think I would get a chance to ask you questions about it. But the most proximal reason I wanted to interview you is that I recently had your colleague, David Reich, on, and we were discussing certain things in the record of human history where he said, “Eventually, you’re just going to have to have Joseph Henrich on and ask him these questions, because he’s the one who would know.”

So let me ask you one of the questions I was super intrigued by, which he raised, and we didn’t come up with an answer to. One of the things he’s discovered through his genetic evidence is that 70,000 years ago, across Eurasia, there were so many different human species, from the Denisovans to the Neanderthals to the Hobbits. And then, apparently, there was this one group, potentially the size of 1 to 10,000 people, in the Near East, which subsequently exploded, and now everybody descended from Eurasia descends from this one group.

So I guess the question is: What happened? What did they figure out?

A typical assumption, when people think about this—if you put it in the Paleolithic—is that it has to do with some kind of genetic changes. Now, in Reich’s lab, there are no obvious big changes in the DNA, so it’s a little bit of a puzzle.

Neanderthals, for example, had larger brains, and in primates, larger brains usually go along with more computational abilities and more ability to solve problems. So the expanding population out of the Middle East, out of Africa, might’ve actually been less able at an individual level to process information.

But if you look back over the more recent period of human history, you can see that it’s a story of expansions of different populations. For example, in Africa, we have the Bantu expansion about 5,000 years ago, which actually eliminates a whole bunch of hunter-gatherer populations that previously existed in Africa. We have the remnant populations in parts of the Congo, in the Kalahari, and the Hadza, for example, in Tanzania.

If you look at the Austronesian expansion—that’s the peopling of the Pacific—that was the expansion of one group of people at the expense of others. And, of course, the Neolithic expansion into Europe is another example.

So really, human history is a story of these different expansions. And it could be that this expansion across Eurasia, which then led to interbreeding—we know it’s the same species. Humans interbred with Denisovans and Neanderthals, as well as probably other species. There’s a ghost species in there.

This could be just institutional changes. If you have institutions, for example, that interconnect your population, you can maintain more sophisticated technology. Some paleoanthropologists, for example, have speculated—with some evidence—that the expanding populations had projectile weapons, such as bows and arrows.

And humans have periodically gained and lost bows and arrows in different parts of the world. In Australia, for example, bows and arrows are never invented. In the New World, populations probably didn’t have bows and arrows, but then later developed them.

It’s also really interesting how some technologies you would just think of as extremely—I don’t know what the word is—but the New World, not having the wheel or something. I guess it kind of makes sense with no domesticated animals, but again, it’s such a—

Right. Although the New World, everybody has dogs, and you can pull carts with dogs. So I’ve never really bought that. And, of course, you can use llamas in the New World to pull carts as well. People do that today.

So what’s your explanation for why there are no wheels in the New World?

There were wheels on Mayan carts or on Mayan toys. My explanation is just the collective brain.

Almost every single first invention of something big that we think is important for humans was invented in Eurasia. And Eurasia, as Jared Diamond famously pointed out, builds on other people’s work. It’s the largest continent by far, and it has the biggest population. It’s also oriented along an east-west axis, which allows ideas and people to flow more easily.

There’s a belt, which the historian Ian Morris calls the Lucky Latitudes, that runs from basically southern China all the way through to the Mediterranean. Ideas are just flowing back and forth there, across the center of Eurasia.

But you also ended up with more complex state bureaucracies and the kinds of things that allow you to organize and move people around and whatnot.

And what’s the explanation for why the collective brain leads to state capacity? Or is it the other way around?

Well, you can think of institutions that eventually lead to state capacity as just part of the innovative process of the collective brain.

If you have more groups experimenting with more different ways of governing groups of people, gradually, you get the accumulation of the pieces that you can put together into different kinds of states.

Got it. Okay, so going back to what happened 70,000 years ago, is the basic answer that it could be something like bows, but we don’t know exactly what it was? Or is it—

Yeah, we definitely don’t know exactly what it was. We know that this population expanded, and there does seem to be some tool indications to suggest more complex technology. Probably, technology usually goes along with social organization.

For example, if you look at Australia, which is a continent of hunter-gatherers, there’s an expansion about 6,000 years ago out of northern Australia that eventually takes up 7/8 of the continent. They had a new social organization, including rules about who you marry. You had linguistic exogamy and rituals that interconnected populations.

Rather than having local rituals and local myths, many communities would get together periodically to initiate the young men. This would help bond that whole group. There’d be an exchange of technology and teaching that goes on at these gatherings. They’d spend a few months in the same place, so there was a lot of time for transmission.

And then, in terms of mechanism, if it’s the case that you have these sort of coalescing waves of expansion, and David Reich’s lab’s evidence suggests that this expansion was quite violent, based on whether the genes are passed down through the maternal line or the paternal line, when you think about the mechanism by which technologies or these social institutions develop, how much of it is, “Look, there are all these different groups that are trying different things, and one of them maybe figures something out, and then they just kind of explode?”

And how much of it is that, within each group, they’re experiencing accumulations of learning over time, and it’s not necessarily a selective process? It’s more of an accumulative process.

Well, eventually, the groups have to meet, and they’re going to compete over territories, and a bunch of different things can occur.

Groups can copy each other. We know that from the ethnographic record and historical cases. Sometimes a group will say, “Those guys have really good technology and tools.” Maybe they’ll get some migrants or something like that, and then they can adopt the practices. So that definitely happens.

But there’s also plenty of evidence of violent conflict. The Reich lab’s evidence, and lots of other ancient DNA, suggests that there are these dead-end genetic lineages. The Neanderthals are a dead end in a sense, although we interbred with them, so they’re in us in some sense. But they don’t have their own pure lineage or anything like that.

And if you had to guess, whenever this big wave happens, is it more of a sort of concrete technology? Especially if you consider the range of expansion that can often happen, right? It’s like literally that same group goes from Siberia to England. It’s hard to imagine that a single technology lets you explore this wide range.

But it’s also the idea that some cultural artifact is enough that it gets transmitted over tens of thousands of years, and that’s what gives you the edge.

I’m curious how you think about what could possibly be this driving engine?

Well, the way I describe it in The Secret of Our Success is that it’s a package of things. Another good example would be the Inuit expansion out of the North Slope of Alaska. They expand all the way across the Arctic and eventually get to Greenland.

They have a whole package of social practices which help keep them interconnected. Then they have bows and arrows, which the group they are exterminating as they go along—the Dorset—doesn’t have. They have dogs and sleds, and along the coast they have boats, so they’re doing whaling. There’s a whole package put together that allows them to outcompete and eventually exterminate the Dorset.

One of the things that happens, though, is that the Dorset probably had better technology. But they expanded, they spread out, their languages diversified, they lost contact, and they began losing technology. So you want to see this as a dynamic pulse. There were probably expansions and then collapses, another expansion, another collapse. It’s not one giant long march to victory.

And then this process, which has happened many times, where the population gets cut off or something—when you say it’s a sort of cycle, is there some reason why, over time, that knowledge gets fragmented and breaks apart and populations disperse?

I think there’s a cultural evolutionary dynamic that’s part of this, because languages will naturally diversify, and then as they diversify, there’s less contact. People sometimes get inclined to marry endogamously, and especially if there are enough people around locally, why go all the way over there? Those people are getting more culturally different, so they seem a little bit like outsiders.

So the tough part that humans have always had is staying unified, because the natural effect of geography and learning locally is going to tend to fragment us.

1. How cultural evolution works

In The Secret of Our Success, you described this interesting startup problem where, if you don’t have that much accumulated cultural knowledge, developing the ability to do social learning isn’t as valuable. But if you don’t have the ability to do social learning, you don’t have that much accumulated cultural knowledge in your tribe or group. So how is this problem solved?

Before I get to solving the problem, I just want to sketch for the listeners that the question is: Why is this cumulative cultural evolutionary process, which is so important for humans, relatively uncommon in the natural world? It seems like it’s just our lineage. There were a bunch of split-off lineages, but now it’s just us.

To understand that, there’s this idea that you just mentioned, where you’re imagining an increase in brain tissue that’s going to be costly, and I can put that towards individually figuring the problem out for myself, or I can put it into learning from others. In a world without very much cumulative culture, there’s not going to be very much useful information in the minds of everybody else, so I should use that brain tissue for individually solving problems. And so it’s hard to get this runaway process where their brains get bigger for the purposes of learning from others. The question is, how do you get past that valley?

The case I make in The Secret of Our Success is that, say, 3 million years ago or 2 million years ago, there were several factors that came together. The first factor is the rate of change of environments. You get this increase in the fluctuation of environments, so you’re getting more environmental changes. In cultural evolution, a lot of theory shows that there’s a certain rate of change which is favorable to cultural evolution. It’s got to be slow enough so that the information of your parents and the previous generation is useful, but not so slow that you might as well just encode it in the genes.

The second thing is, we’re a ground-dwelling ape, which means we have hands like chimpanzees and gorillas, and we can potentially use tools and whatnot. But unlike them, our ancestors may have been savanna-dwelling apes, which meant we may have lived in large groups. Mammals live in larger groups when they have to deal with predators, and the predator guild in Africa at that point was quite thick. There were a lot of deadly predators. So paleoanthropologists think that our ancestors may have lived in large groups of savanna-dwelling apes.

If you have a lot of individuals, and the culture is sparse, the bigger the group, the more chance that there’s someone doing something useful in the larger group. And so that means it’s easier to get across the threshold. So those are 3 of the main factors that might have allowed our lineage, as opposed to all the other lineages around, to cross this. We were already big-brained primates with hands, we were living in these large groups on the savanna, and the climate changed during this period.

And how big were these groups?

Nobody’s really sure, but maybe a few hundred individuals.

And before the agricultural revolution, was there still this transmission of information across different groups?

Yes. Lots of different evidence suggests that groups were moving trade goods. Well back into the Paleolithic, we see trade. Often we see some genetic transmission.

Right, the expansions and so forth.

Yes. And so, talking about the expansion, that’s something where ancient DNA is useful, because the Neanderthals in Europe, the DNA suggests, lived in very small groups. But the DNA of the expanding groups suggests a larger population. So those would’ve been 2 different collective brains there as well.

So in The Secret of Our Success, you discuss a lot of these lost European explorers with modern technologies, as of at least a couple of centuries ago, encountering peoples who have, for tens of thousands of years, discovered ways of hunting and processing foods and so forth—without which, even these people with modern technology will starve.

As I was reading that, I was wondering: I’m not sure I understand the process by which you learn that, if there’s a 10-step process to making sure this bean is actually nutritious, without any one of those steps you might poison yourself or something. And at no point do you understand why this process works. You don’t have a scientific explanation. How do you even learn that in the first place?

Right. One of the things we know is that even young children have a tendency to preferentially learn from healthier, more successful individuals. Something like bitter cassava has cyanide in it. If you just eat bitter cassava, it won’t taste great. If you then rinse it somewhat, it’ll taste better and maybe you could eat it, but you’re going to accumulate cyanide over the long run. But if you do this whole long process that the populations in South America developed, then you’re totally fine; you never get any accumulation.

So you can imagine that initially this is going to be very strong and people are going to do sensible things, but then it gets a little more mysterious. We know this because bitter cassava gets transported to Africa, and Africans immediately begin eating it improperly processed and getting goiter and the cyanide poisoning that goes along with it.

So then that’s going to be a slow evolution, where groups that do this are going to be more successful and individual families are going to be more successful. For example, you might have a household where they process the cassava more seriously than another family, and they’re going to have more kids and they’re going to be like, “Oh, that family’s really good.” Then people copy what they do in all kinds of ways, but one of them could be copying recipes.

I guess it depends on the mechanism of selection here, because when you consider the different ways in which 2 different individuals might be different, or 2 different households, or even 2 different groups, I guess it makes sense why cyanide poisoning is such a deleterious effect: It is a noticeable, or quite a strong, signal.

But I think you discussed some other ideas in The Secret of Our Success, where the spices in a region match the antimicrobial or antifungal properties you need to stay clean in that particular region. But it sounds like a small effect, and how could such a small effect actually create a strong enough signal that, when you’re deciding who to copy, you notice that this family is healthier because they spice their food in a certain way?

It has to be a big enough effect to matter, but I think if you go back to a world where there’s a lot of improperly processed meat, people don’t have refrigerators, and the leading cause of death in children is diarrhea.

People carry high pathogen loads. If you can knock that pathogen load down, we know from modern research that if you want to make people healthier and even smarter, IQ goes up if you knock the pathogen load down.

Got it. So this process of cultural accumulation you’re talking about: Do you think it is not strong enough to pick up minor increases in fitness, and it actually has to be quite a significant thing for it to matter?

I think it operates a lot like natural selection, in the sense that natural selection will pick up tiny things if you give it long enough. Cultural evolution will do the same thing if you give it long enough, because just small differences in who you’re paying attention to will affect things, but it might take 1,000 years as opposed to decades, right? But we know culture can spread adaptive traits super quickly if it’s a really big effect.

Yeah. So the situation in which our ancestors found themselves, in some ways, it’s sort of like epistemic hell, in the sense that you don’t know why certain things are working, but you do know that if you break with tradition, you might just doom yourself and your family. And that maybe, as you discuss, causes these religious beliefs and taboos and mystical understandings of the world to arise, where you just think, “Look, you’re gonna burn in hell if you don’t do this 10-step process for refining your beans or something.”

In some sense, this requires you to abdicate reason because you have to just be like, “I don’t understand this. This is the way it’s been done, so we’re gonna do it.” Basically, how much reason do you necessarily have to abdicate to survive in the ancestral environment?

One of the ways we research this is by going out to societies that make bows and arrows or have food taboos that protect them from dangerous marine toxins, asking them questions about why they do it, and seeing if they understand. They don’t understand the underlying causal stuff.

Right. But then why do they say they do it?

Well, a typical answer is, “It’s our custom,” or “It’s important to people around here to do it this way, so we do it this way.”

So, just Hadza bows, for example—I didn’t do this research; Jacob Harris, Kim Hill, and Rob Boyd did it—but they asked the Hadza about their bows, how they work, and what the mechanics are, so they understood some stuff. You use the bow and you get some mechanical understanding. But if you asked them, “What if you used a different wood, different materials?” they had never tried anything else but what they learned, how to make this bow, so they couldn’t speak to a lot of that stuff. The things about the compression of the wood are very important, but they didn’t understand that.

Okay, so that’s quite interesting, right? Because you would think that you have to transmit cultural knowledge over time, but you also need to experiment in order to innovate. But it sounds like because of the belief in these customs, you would be less inclined to innovate.

Right. And the thing is, once something gets good, doing it differently almost always makes it worse. And then there’s also the fact that some things are just different because people do them randomly differently, like they make a mistake. So I think people often underestimate the power of error in generating novelty.

One question I have is why this process of cultural learning—so humans have extended childhoods, presumably to give us more time to accumulate culture, and then we live after menopause, presumably because older people, our grandparents, can teach us about the situation we find ourselves in. And 18 years is such a long time.

When I think back to what I was doing as a teenager, I don’t think I was learning that much. I was just less productive for no obvious biological reason. So I guess I don’t intuitively understand why this process of cultural learning takes decades, and why it can’t happen more rapidly.

Well, you probably, as an adolescent, were in school, right? So you were probably accumulating some stuff in school—

Maybe.

And here’s an interesting fact about hunters. Anthropologists who studied hunters—hunter-gatherer societies in different places—found that the physical peak that at least males have is in their early 20s, right? That’s when they run the fastest. They’ve got the best eyesight. But the best hunters in the community are 36 to 40, right? They’re not as fast as those guys, but they just know the track, they know the animals, they know exactly what to do.

Then hunting skill begins to decline because basically the physicality of things begins to catch up, and so there’s this kind of cycle. At 18, young hunters aren’t even producing enough food to feed themselves. So it’s not until they get into their 20s that they’re actually in surplus and bringing food home for everybody else.

And you need to know hundreds, at least, of animals. You need to know animal behavior, you gotta be able to spot tracks and spoor, and we’re just very knowledge-dependent in terms of our hunting and gathering.

2. Why is brain size declining?

Yeah. Given what you just said, and also the experience of these European explorers, it suggests that you just gotta know a lot of shit to make it in any sort of environment on Earth. But then there’s all these other animals, and they seem so dumb and they seem able to get by. So why did humans have such a hard time of it?

Yeah, well, I think we offloaded a lot of stuff into culture. So in one of my lost European explorer examples, there’s a case where they’re in Australia and they have camels. The camels escape, and now Central Australia has lots of feral camels.

The camels survived the lost European explorer challenge because they have innate instincts. They can smell water a mile away. They can detoxify foods on their own. They have a complex digestive system that detoxifies food. We’ve lost all that, and we’re worse than chimpanzees at detoxifying foods because we have all these cultural practices that do the work. So we’ve externalized detoxification and a lot of digestion, actually.

I had this independent researcher and Internet writer called Gwern Branwen on my podcast a couple months ago, and we were discussing AI. I asked him—his theory is very much that the brain became bigger and more intelligent—“Look, if it’s this simple, why did it take so long for evolution to discover intelligence in the first place?”

And he had this interesting answer, which is that, in terms of the signal that evolution is giving you, there’s a very narrow gap between skills that are so useful that they should just be distilled as an instinct into your genome, and skills that are so worthless that they’re not worth learning in the first place. And so this narrow gap means you need that generalization ability. It’s not so primal that you’re going to culturally learn hunger; it’s just going to be in your genome. I don’t know if that sounds like an interesting explanation or helps explain anything.

Well, in cultural evolutionary models, a typical thing would be to let genes compete with individual learning and with cultural learning. And it turns out the rate of change of the environment affects that. So if the environment is changing slowly, then you should put it all in the genes. If it’s changing at moderate speed, then culture is the best way to go, and if it’s fast, then individual learning.

The idea is that individual learning is favored because if it’s changing too quickly, the previous generation doesn’t know anything worth knowing, because the world you’re dealing with is so much different from their world. And so it’s this intermediate range.

Myself and others have argued that the increase in the frequency of change during the Plio-Pleistocene transition 2.5 million years ago is a change that increases the value of cultural transmission, of learning from others. I’m not sure what your guest was thinking. Our brains are not very good at solving problems; otherwise, the lost European explorers could survive. And human brain size has been declining for the last 10,000 years. So we’ve actually been getting dumber. Fewer neurons, less computational power.

Why has brain size been declining?

Well, the collective brain argument suggests that at a certain size, you’ll begin farming out specialists. So because there’s a store of knowledge in the society, we can all be generalists and learn how to do all the different skills—

But at some point, it makes sense, for efficiency, for us to specialize in different skills. In order for that to be the case, though, we have to have social agreements of some sort that allow us to trade or exchange things like that. But then, once we’re specializing, we don’t necessarily need as large a brain, because we distribute the overall brain power amongst the society.

So it could be that we’re becoming more of a superorganism. You see the same thing in ants. When ants get specialized occupational castes, their individual brains shrink.

Yeah. I think David Reich’s lab had a result a month or two ago where they showed that the selective pressure, at least in the European samples—which were the samples they studied—was for greater intelligence over the last 10,000 years. I don’t know if you saw that or what you make of that.

Well, they didn’t say anything about intelligence. They did use the polygenic score for education, but there was no education 8,000 years ago.

Which is correlated, right?

Yeah, it’s correlated. But then the question is, we don’t know what goes into that, right? Is that actually computational software? I mean, people do well in school because they stick to it and they can sit in the same place for long periods of time.

I guess it would be interesting for them to deduct the docility-related polygenic score from the education score, because we don’t have a polygenic score for intelligence yet, right? That’s why you need some proxy.

Well, there are IQs, so there are polygenic scores for IQ. I think they actually did that one as well. Like you said, it’s correlated with education.

But the things that go into giving you an IQ—I mean, the correlation between brain size and IQ is only about, well, across populations it’s 0.24, so it’s not very big. The other thing is, IQ is massively misunderstood. The way to think about IQ, which has been going up over the 20th century by quite a bit—if you rescale modern scores to 1900, it’s about 70—is called the Flynn effect.

I think those are cognitive abilities for navigating the institutional world that we’ve constructed. A huge mistake would be to assume that those are the right cognitive abilities for the next century. So it’ll be a different constellation.

We study herders and the cognitive abilities of herders in northern Namibia. If you’re in northern Namibia, you have to be able to move through the landscape, so being able to pick a direction and know where you’re going without getting lost is super important. Different suites of cognitive abilities are favored in different environments. This idea that there’s a generalized thing applicable across all human environments just doesn’t hold. I mean, education massively increases your IQ, right? If you’re uneducated, you have a totally different IQ.

And then, is the way in which your horsepower is directed a thing that you think is basically set by the time you’re an adult? Or, if somebody’s 30 now and the AI thing is going to happen next and they have to totally reorient away from knowledge work, is that a thing?

Well, I think it’s a continuous scale. If you look at human brains, they’re developing and continuing to add new connections and stuff at least into the mid-20s. Now, there could be even more plasticity after that. Unlike chimps, we’re still not totally myelinated throughout our whole lives, so there’s definitely room. It’s just that there’s less room.

The kind of social learning for which our learning biases are fit, right? Where you pay attention to elders because they survived somehow and they’ve accumulated the knowledge from past generations—how much do you think that actually applies to the way in which the modern knowledge-work economy accumulates knowledge? And how much do you think is just an artifact of the kind of environment our ancestors found themselves in?

Right now, we’re in Silicon Valley, or in San Francisco, and it’s very common here for a 20-something to make a new product or service. They don’t have that much context for how the world works, but they can make big innovations and big changes. At the same time, people on average may be more productive in their 40s and 50s in terms of wages or something. So this social learning you’re describing—how much does it actually describe the world as it exists today?

Yeah. This is super important. We’ll take the patent database, and if your same-sex parent—it has to be the same sex—has patents in a particular domain, you’re 9 times more likely to patent in that same domain.

And how do you separate the genetic versus cultural effect there?

These are very fine domains, so we’re talking about natural adhesives versus synthetic adhesives. So unless you think there are genes for that. The same thing applies to transistors and electronics, so these are very fine domains.

If you grow up in Silicon Valley, you’re much more likely to patent in general, but you’re probably going to patent in computers. If you grow up in Boston, it’s going to be biotech. Having a father or mother who did biotech and then growing up in Boston makes you even more likely to patent in biotech.

And this is true even if you move to New York. If you look at people in New York who are from Silicon Valley and from Boston, the Bostonians are more likely to patent in biotech, and the Silicon Valley kids are more likely to patent in computing. All the same rules apply.

Well, I’m not sure, because maybe in the location, you learn a lot from the environment. But the fact that it’s just kids—if you think these old rules of cultural accumulation and collective brains apply—you would think, “Ah, you’re 50 by the time you’re writing your first useful line of Python code or something.”

In fact, a lot of these big innovations come from people who are much younger, before they should have accumulated much of the know-how that this theory implies.

So the model here is what to focus your efforts on, so it’s the throwing model. If you grow up in Silicon Valley, you focus your efforts on learning how to code or whatever. And so that means that’s where you’re likely to make the innovation.

Right. But in terms of there being 2% growth a year, and a lot of the technologies that are dominant in the economy today—or a lot of the industries—didn’t even exist many generations ago, does that suggest we should just mistrust our instincts about who to give prestige to and what kinds of people we want to pay attention to, whether it’s professors or elders or whatever?

That’s definitely something I’ve thought about, and I wrote about it in The Secret of Our Success. As the rate of cultural change gets faster—and we sort of talked about this with environmental change—the value of older members of the previous generation declines, because the world that they grew up in, and that they honed their skills for, is quite different from the current world.

So, optimally, you would look less far back, or you would look to relatively younger individuals to get your inspiration from, because the world they adapted to is closer to the world you’re going to need to adapt to.

3. Will AGI have superhuman cultural learning?

And speaking of Silicon Valley, let me ask you a little bit about AI. One of the reasons to suspect some incredibly sharp discontinuity from the world as it exists today to a world with AIs—and by AI, I don’t mean just GPT-4; I mean at least a replacement for anything you can do on a computer screen, the AI can do—is that they potentially have a step-function increase in social learning and the ability to accumulate knowledge, perhaps of the magnitude that humans had when the gap opened up between them and nonhuman primates.

In particular, you can just copy everything you know. You don’t have to teach a young person, right? The constraints of biology mean you can’t just replicate brains. You have much more efficient communication. You don’t have to communicate through words; you can just shoot your brain state across. The population size can grow arbitrarily large.

So, to the extent that the collective brain is determined by the size of the population and how interconnected it is, do you expect we wouldn’t even recognize the kind of world these AIs could make as a result of their cultural skills?

Yeah. No, I definitely think it’s pretty interesting, and it holds great potential for expanding the collective brain.

There are little things in there which might make one worry. So if you study the history of innovation, you find out that, for example, serendipitous meetings are super important. There’s a great paper on Silicon Valley showing that companies will cross-reference each other’s patents more likely when the people at those companies tend to frequent the same coffee shops, and they track people on their cell phones and stuff to figure this out. So serendipitous meetings are important, as is improper copying. A huge number of innovations are mistakes where somebody copied incorrectly and then got something better.

I read this interesting theory that maybe evolution designed transcription and translation to have more errors than it could otherwise have, just so that you can have this sparse reward when you’re close to the right sequence.

Yeah, yeah. I think that’s a very interesting area of research, and it makes good sense to me. I’m not sure of the current state of the evidence, but different parts of the genome are more or less susceptible to mutation, which is kind of interesting.

So, going back to AI, maybe another way to phrase it is: look, you’re talking about these serendipitous meetings where you can learn something another person knows. And the great advantage these AIs have is that they can sort of meet with everybody at once. Future versions of AIs, you could really imagine holding the whole internet in context, right?

We’ll be the equivalent of those people isolated in Tasmania, according to the AIs, right? Because they just have everything in context. You can get a PhD in every single field, and you can amortize that across all your copies. So are you sort of banking on, in the next 10 years, you’re just going to be living in the singularity or something because of your belief in the value of cultural knowledge?

Yeah, I see there are various potentials, and I’m particularly interested in using AIs to augment problem-solving in human groups. So you can imagine getting the humans together because the humans have the big advantage of having stuff they care about, right? There’s stuff they want to invent. So the AI is still a tool at some point.

So I’m interested in that, but I’m interested in how things like running out of training data are going to be dealt with—the value of making mistakes and serendipity, if you get rid of all those things or how you’re going to reintroduce them, those kinds of things.

Yeah. It’s quite funny because until recently, people were saying the big issue with LLMs is that they hallucinate and make mistakes. And at some level, hallucination is no different than creativity. Well, maybe there’s a way to harness that, right? But maybe we just didn’t know how to harness it.

That’s right.

What do you make of the idea that you could have AI firms or AI tribes, whatever way to think about it? If the effective population size that can communicate with each other is such an important contributor to how much progress a group in history was able to make, what if you could just literally run billions of AIs and they had this immediate ability to communicate with each other?

Again, I’m not talking about current models; I’m talking about future, human-equivalent models. Basically, I guess I’m just throwing out a bunch of different intuition pumps, and I’m curious which one you find most promising or most interesting. Or do you just find all of them not as convincing?

Yeah, no, I don’t have strong opinions about any of this, but I do think that one thing to make all that work is that humans are constantly getting hit with shocks, right? There are economic shocks, weather shocks, conflicts with other groups, and pandemics. The shocks have big effects on how things go. They inject new information into the system. So I’d worry about a system that is too homogeneous and doesn’t have enough noise, adding shocks and new challenges and things like that.

One of the things missing from our conversation is the kind of creativity that cultural evolution has figured out. To give you an example, at some point in human history, religions invented big, powerful, moralizing gods. And this may have increased people’s ability to cooperate. So forget about the technological element. Some of the most important features of innovation over human history have been institutional—things that get people to cooperate.

So in my work, I argue that one branch of Christianity resulted in the transformation of the families in European societies into small, monogamous nuclear families. I’m not sure an AI would have thought of that. Cultural evolution thought of it because of how it affected how the societies operated, so it sorted this out over historical time.

Can I play with that idea? Because that’s quite interesting. Again, I think this might even reinforce the advantages of AIs in this sense. To the extent that this sort of random fluke by the church led to modernity and the Great Divergence, or whatever, and it was a result of there being a bunch of variation, and then you can select over some group that’s doing the right thing, the advantage you have with AIs is that you can have much higher-fidelity replication of culture and explore a wider range of potential cultures.

It sounds really vague when you talk about AIs in this way. So what do I mean? One problem companies have today is this: suppose a company’s working really well. Maybe it’s early Apple, early Tesla, or SpaceX or something. Suppose it has to do with the culture, the people, or something. It’s not clear how you replicate that culture, not only across different institutions, right? If I’m running a company, I don’t know how to copy SpaceX.

Not only can you not replicate it across institutions, you can’t even replicate it across time. Over time, many institutions tend to decay; their culture fades as people leave or die or something. And then imagine if all of SpaceX—or at least SpaceX at the time you thought it was most effective—were just AIs where you knew every single byte, you could make a copy of it 1,000 times, and you could put it against every single hardware problem we have in the world.

And then, if you think there’s another team that might have some different culture which is better, you can do the same thing with them. Culture here includes not only how they think about their work, but also how the board makes decisions, right? Do they do this Monte Carlo tree search? There are so many different things you could do here.

So because of the wider range of possibilities you can explore and the fact that you can have higher-fidelity transmission of cultural information, maybe the ability to do this random evolution also increases. But I’m curious to get your take on that.

Yeah, as a kind of broad sketch, I think that is pretty interesting. I worry about too-high-fidelity replication just because it’s important to take the details of historical context and time into account. So that same thing, if you fix it, might not really work a decade later when everything else around it has changed, right? It’s kind of a moving target.

But you could fix that just by having lots of different variants that are different versions of that. So, yeah, that would be cool. That gives us a good opportunity to go into WEIRD. But before we get there, one more question about before the agricultural revolution.

Obviously, many groups around the world were different, and many of those differences were probably inspired by the fact that they were living in different environments. They needed to come up with different technologies and adaptations to survive best there. How much random difference that had nothing to do with their environment do you expect there to have been?

I don’t even know what the right category would be, but in the modern world, you might imagine something like whether a society has slavery or not. And maybe pre-modern societies all had slavery, so that’s not the right way to think about variation, but something like that kind of variation.

Yep. So I think there’s lots of reason to think there was lots of variation. One of the ways that researchers study this is they look at ethological variables and they compare it to phylogenetic variables. Here they mean cultural phylogeny.

So if your ancestral populations had matrilineal, matrilocal social organization, how likely are you to have it, controlling for ecology? It turns out both of these things matter.

People adapt to their ecology, but you can still see the signal of past society: some degree of fidelity in the transmission of social institutions, how we make our baskets, and those kinds of things. There are lots of examples in the modern world, and we see this all over the place.

If you look at gender inequality in the modern world, you find that a history of the plow leads to greater gender inequality. Males had a particular advantage in using the plow because it requires upper-body strength. This meant males became the dominant force in economic production at the household level. Even today, in populations where most people aren’t farmers, this persists. So this is cultural persistence that has to do with whether you had the plow.

This is work done by Alberto Alesina and Nathan Nunn. A former postdoc of mine, now at the Harvard Business School, Anke Becker, has done a similar thing with pastoralism. If your ancestors were pastoralists—and pastoralists, for various reasons, have quite strong gender norms—those norms persist into the modern world and still shape, for example, female entrepreneurship.

4. Why Industrial Revolution happened in Europe

Interesting. The reason I’m especially curious about this is because it informs the following question: If the Industrial Revolution didn’t happen in Europe but started somewhere else, how different would the world look today?

Obviously, you discussed the fact that breaking apart these kin ties was necessary for the Industrial Revolution. But to the extent that it was necessary, whichever place had the Industrial Revolution first would have had that, right? Separate from the technologies or cultural practices that were necessary for the Industrial Revolution in the first place, how different would the world look? How much variation was possible given our level of technology?

I have a lot of trouble trying to answer that because China or some place in the Middle East might have been the obvious alternative place for the Industrial Revolution to happen. But I feel like you have to give those places a lot of things that developed only in Europe.

For example, universities began spreading in the High Middle Ages. You’d want them to have universities. You’d want them to have universal schooling, which began to spread and wasn’t present in these places prior to that. That begins to spread in the 16th and 17th centuries. By the time you add all this stuff, it basically starts to look a lot like Europe.

These are all things that are global now, right? Universal schooling, which we find around the world today, begins with the Protestant Reformation in Germany and then later in England. University models are based on the European university and have globalized.

I guess one obviously salient example of variation that might not have been replicated—but I’m curious if you think it might have been—is that it seems like the British Empire was the first major institution that decided slavery was morally wrong and was going to throw its weight around to eliminate it.

I don’t know if you think that naturally follows the development of the social technologies that the Industrial Revolution would have brought about, but that seems super contingent. I’m curious if you disagree.

My story is that the rise of the Industrial Revolution in Europe has to do with the consolidation of Europe’s collective brain. One of the things that requires is trust in strangers, and at least the beginnings of moral universalism. It’s that moral universalism that eventually causes the British to say, “We’ve got to stop with the slave trade thing.” It’s a moral decision that they made because it’s no longer consistent with the changing moral values over time.

Okay, so we’ve been dancing around the thesis of your book following The Secret, which is The WEIRDest People in the World. Before we really jump into it, maybe you can give me a summary of what the thesis is.

The first observation is that there’s a great deal of global psychological variation around the world. European, American, and Australian populations tend to be highly individualistic. They’re inclined toward analytic thinking over holistic thinking. They have a lot of impersonal prosociality, such as trust in strangers. They’re against conformity and have a willingness to cooperate with strangers.

The question is, how can we explain the global variation in these features of psychology? Toward the end of the book, I actually connect these features of psychology to economic differences, including the Industrial Revolution that happened in Europe, which reshapes the world.

The story is that the key event is the spread of a particular form of Christianity into Europe, where the Catholic Church—what becomes the Catholic Church—systematically dismantles the intensive kinship systems in Europe, leading to small, monogamous nuclear families.

This transformation leads to the creation of new institutions. By the High Middle Ages, you get the rise of guilds, which are voluntary groups of craftsmen, and self-help societies, because people don’t have their families to rely on. People begin moving around. There’s occupational sorting into different occupations. Urbanization rises. Chartered towns are on the rise. Universities pop up. New kinds of monasteries pop up.

Europe begins to urbanize, and you get new kinds of law that are based around the individual: contract law. Eventually, this leads to a lot more innovation because ideas are flowing around Europe, and then eventually to the Industrial Revolution. So that’s the argument in a nutshell.

Can you explain again what exactly the Church did that led to the kin-based existing system breaking down?

The Church outlaws polygyny, and so that stops elite males from having multiple wives and concubines and creating a giant family through that. It outlaws cousin marriage, going all the way out to sixth cousins at some point, and that included spiritual kin and other kinds of non-genetic relatives as well as cousins.

It frees up inheritance, so you have inheritance by testament rather than normal patrilineal inheritance. A simple example here is that, in most societies, you inherit access to land corporately, which means you and your brothers and so forth all own the land. It might be that your uncle is actually in charge of the land—your father’s brother. You can’t sell it, you can’t move it around, and you’re also tied to it. It’s where your ancestors are buried, and so there’s this big importance of land.

The Church allows people to give land to the Church, and the Church becomes the largest landowner in Europe because you can do it by testament. Those are just some examples of the ways it transforms the family structure. Eventually, you get these monogamous nuclear families, which are basically unheard of around the world, at least if you look at the anthropological record.

You can really see this in the comparison: individuals can move to European cities as individuals or nuclear families. In China, when you move to the city, you maintain contacts with the folks back home in the village, and people flow back and forth. You get these little enclaves of different clans and stuff in the cities because of those connections.

This is really important because of your ancestors and the rituals that have to be done back in the home community. Christianity does away with all those ancestor rituals.

Okay, great. Let’s jump into it by starting before—much before—even the Church. I wonder how much of the things that make Europe weird existed even in antiquity.

If you read about how Roman society worked, or how these Greek polities worked, already you have an emphasis on nuclear families. You have these universal norms around, “Hey, we believe in republicanism,” or, “We believe in democracy.” The structure of the system matters more, necessarily—at least before the empire, right?

Was Europe already weird before the things that its Church did?

There were certainly some interesting things going on in Greece and Rome. But I don’t think there’s good evidence that you had the kinds of monogamous nuclear families that you would find later. Even European law is built around patrilineage. If your father’s still alive, you are not a full citizen.

You’re in the paterfamilias’ house in these large families. Definitely intensive kinship, for sure: patrilineal, patrilocal. Women, of course, don’t have any rights.

And republicanism—but the formation of Rome is built around a series of elite families. So it’s a clan operation, and they call it republican because the clans all have some say in what’s going on. People sometimes see representative government where it’s not really individuals being represented, but families.

And so this fools a lot of people into thinking there was a lot more individualistic voting and things like that, whereas there’s actually no voting. So in Greece—in Athens—Athens is unique because it does a bunch of things in the laws of Solon that break down the intensive kinship.

For example, in Greece, you get monogamy for the first time, and Athens is considered unusual. Males can only marry 1 Athenian woman, which has potentially positive effects on the competition among Athenian men, but they can have as many slave concubines as they want.

So when Christianity spreads, it ends the whole slave-concubine thing, which was also common in Rome, and that’s another effect on this whole thing.

Yeah. But I guess the extent to which these practices were already formed hundreds or thousands of years before the Church’s intervention maybe suggests that Europe was already on this trajectory, and the Church didn’t necessarily move the needle that much. Or what do you make of that thesis?

Well, we know that. In the book, I talk about data where we can look at contemporary kinship practices, and we can look at the number of centuries that that part of Europe was under the Church. We have a database of the diffusion of bishoprics, and there’s a clear connection between intensive kinship practices and the number of years under the Church.

Hmm. And then if you think about what the Church basically did, I think your work is often used to justify this idea of Chesterton’s Fence, where if you don’t necessarily understand a cultural practice, you should keep it around because—the stuff we were talking about in The Secret of Our Success, where you don’t understand why the 10 steps lead to the bean being edible, but you should trust the sort of wisdom of the ages.

If you think about what the Church did here, right? Isn’t this sort of like—not necessarily a justification, but at least an example of—just breaking down Chesterton’s Fence? The Church doesn’t really understand why these kin-based networks that have existed for thousands of years might have utility. It still just gets rid of all of them, and that results in this lottery ticket that leads to the Great Divergence.

So is there evidence for anti-Chesterton’s Fence from The WEIRDest People in the World?

Sure. Sure. So what you’re talking about is just the idea that culture has imbued institutions and various practices with a logic that we might not understand. So it’s not calling for never changing the institutions. It’s saying, make sure you figure out the logic and the costs and benefits.

I was an expert witness in Canada for the attorney general of British Columbia, who asked the Supreme Court of British Columbia whether the statutes against polygyny were legal. My job was just to inform the court that polygyny has this unfortunate effect: the elite and high-status males tend to get a disproportionate share of the wives, and that creates this pool of low-status unmarried men.

Now you can decide what you want to do with that. But you need to understand the kind of underlying social dynamics that you’re gonna unleash if you legalize or decriminalize polygyny.

Right. But again, going back to the Church, the Church didn’t understand the positive or negative effects they’re gonna have, right? They just said, “Let’s rip the Band-Aid. Let’s see what happens.”

That’s right. And it actually benefited the Church because it released people from their responsibilities to their families and allowed them to migrate in and join the Church, and more heavily invest in the Church.

These later prohibitions against priests marrying and stuff were all an effort to get greater investment in the Church, because you weren’t torn between helping your son and then investing in the Church.

Hmm. And then that end result, which benefited the Church—was that their conscious intention in breaking apart these kin-based ties? Or was that just the accidental byproduct?

Yeah. In the record, I’ve never been able to find much evidence that they were thinking about the destruction of kinship ties. There is a great quote from Saint Augustine where he talks about the benefits of forcing people to marry more distantly.

But this is something that was debated repeatedly in church councils all across Europe, and it doesn’t seem to pop up very often.

And the Church doing this, for obviously selfish reasons—was that in any way related to why the Church spread through Europe in the first place? Or did it end up spreading for separate reasons, and it won out over the other potential religious competitors for separate reasons?

Basically, was this part of the selective mechanism which allowed the Catholic Church to become dominant in Europe in the first place?

Yeah, I think so, although it’s hard to know. There was a lot of randomness going on in the diffusion of the Church. But also, as the Church arrived, these places, over a period of time, became more successful.

You can see rising urbanization in the wake of the Church’s arrival. That would have meant more trade. Their citizens’ rights began developing because they were trying to attract citizens to the towns. That would have given the Church more cachet and ability to move.

There’s also the appearance of these voluntary associations, which are new kinds of monasteries. The monasteries begin spreading throughout Europe, but they’re all linked in a network because, like the Cistercians, they’re all connected and they have big meetings.

So they’re actually spreading a lot of knowledge around Europe as well. So that’s part of this kind of collective-brain story.

5. Why China, Rome, India got left behind

And if you compare what’s happening in Europe at this time to the rest of the world, starting with China: in 1500, the population of England, where a couple centuries later the Industrial Revolution starts, is 3 million. In Ming-dynasty China, it’s somewhere between 100 million to 160 million.

If we take your previous perspective that the size of the collective brain really matters a lot, the size of the collective brain in China just seems so much bigger. So what is the best way to understand what went wrong here?

Right. So when you’re thinking about China, the first thing to remember is that for a lot of history, we can actually see that size difference mattering a lot.

A lot of European inventions, or a lot of stuff used in Europe, are flowing in: gunpowder, paper, the printing press—stuff is flowing into Europe. So, okay, then what happens after about 1000 CE?

The argument is that the destruction of the kinship group opens the floodgates to people moving around. A recent analysis that we’ve done is that, after a bishop arrives at a 1.5 by 1.5 grid cell in Europe, people begin flowing in and out of that grid cell more.

And what we do to calculate that is, we have a big database of a few million famous people, and we have birth and death locations. So we find out that the Church arrives and suddenly people are free to move around.

So you have a flow of individuals around, and you have rising urbanization. Europe passes China around 1200 in the percentage of the population that lives in cities. Cities are where a lot of the action is—cities and towns.

And you have the diversification of occupations. Normally, clans would specialize in different crafts, and you’d learn from your clan brothers and clan fathers how to do the occupation.

In Europe, you get guilds, and you get masters and apprenticeships developing, where strangers will become apprentices to masters, learn from them, move somewhere else as journeymen, learn from that master, and then eventually set up a separate shop. Lots of opportunities for a flow of ideas.

So it has to do with how the kinship system transformed the movement of people, the rising of urbanization, the nature of guilds, and then eventually you get universities and things like that.

So it greatly intensifies the interconnectedness of the collective brain and the amount of cognitive diversity.

What do we know about India during this period? From what I’ve read, from the perspective of how much written history we have, it’s kind of a black box. But we know that there was trade between India and other parts of the world, and we know it had a big population. What was the effect of not having maybe Abrahamic gods, or of having the kind of other cultural practices India had?

Yeah, definitely not an expert on this one, but some of David Reich’s evidence suggests that the caste system is quite old. You can actually see it in the genetic data. And the caste system is not good for innovation, because if you happen to be good at another skill but your caste doesn’t do that skill, you can’t switch over. So it’s going to prevent the sort of available genetic diversity.

Complex families, intensive kinship—there’s reason to think that those things were all important.

Yeah, so, a patchwork of polities.

But there were lots of interesting ideas, actually, that were developed in India, and they moved into Central Asia, eventually ended up in the Islamic world, and then got into Europe. So, for example, Arabic numbers are actually Indian numbers. Zero was probably developed—I mean, Indians were huge with numbers. It’s kind of interesting. Perhaps related to the religion.

Not that this would be a justification for the caste system, but is there any way in which the specialization it engendered would be good from the perspective of the collective brain?

So specialization is good. And the interesting thing about most human societies is that specialization automatically—or not automatically—it tends to evolve along some kind of kinship line. So in Oceania, there were different clans. In each clan, there’d be a canoe-building clan and a warrior clan. So that allows specialization, but it doesn’t allow you to harness genetic diversity, because it’s not like the canoe-building clan had special canoe-building genes. It’s just that you would pass down this cultural knowledge.

Now, a better system, but one that seems hard to evolve, is one in which we all select into occupations in which we think we’re good at. But for that world to exist, you need a world of voluntary associations, with emphasis on the individual rather than on the group.

So this is the world that evolves in Europe once you demolish the intensive kinship units. Otherwise, you’re going to get castes and clans and all these things we see elsewhere, where we see a division of labor, but the knowledge is transmitted through a kin network.

And why wasn’t there another sort of stultifying—because we began the conversation by talking about this—there does seem to be this cycle where, over time, because of cultural differences, because of kin, whatever else, groups will tend to diverge. Languages, culture, and other things you mentioned tend to diverge over time. Why didn’t the same thing happen in Europe where, let’s say, in the 3rd century or 5th century, the Church starts doing this stuff, and then by 1,000 years later, if you’re part of this guild or whatever, now that’s become like the new kin, and now there’s actually less mobility again?

Well, there was a lingua franca, Latin, and intellectuals would all write and communicate in Latin for a long time. So even though they were speaking different dialects of French and German and whatnot, they were able to communicate in Latin. And Christendom basically formed an overarching network that helped.

One of the key things that this world religion does—and other world religions do it too—is that you have to marry other Christians, but it dissolves the tribal line. So Europe has tons of tribes in the pre-Christian world, but because of how you have to marry other Christians, you had Celts marrying Franks. And in fact, the early arrival of Christianity into England was when a Frankish princess married an Angle in Kent. And so they’re marrying back and forth. And this is going to dissolve the tribes because of intermarriage: “What’s the kid?” So that’s how you get rid of the tribes, but you need a world religion to do that.

Right. And so it seems like there were 2 very important features that were relevant here. 1, it seems like, obviously, the competition between different European groups, which gave incentive for monasteries or universities or cities to attract the best people and make all these advancements. But secondly, the fact that they were descended from a common empire, from the Roman Empire. Basically, in a world where the Roman Empire never existed, do you still have this collective brain in Europe emerge?

So I think the Roman Empire plays a big role because, for example, there were Roman roads in parts of Europe which allowed people to flow. And you have communities which were part of Rome and relatively sophisticated. Now, they go into a bit of decline, but there’s still the memory of the empire. So, you know, Charlemagne wants to be crowned emperor of the Holy Roman Empire in 800, even though there’s not very much left of it. But it was still an ideal, and that Carolingian Empire actually had real effects because they worked with the pope in order to enforce a lot of the marriage and family programs.

The Carolingians had their own agenda because they were trying to use these marriage and family practices to weaken some of the other aristocratic families. But they had the religious tools that they could put to work, and if you didn’t have those religious beliefs, then you couldn’t put them to work.

Sometimes people defend ancient conquerors based on a similar idea: that they spread knowledge around the world, and they set up these lines of communication. People say that about Alexander and the Hellenistic expansion and trade and so forth. They say that about Genghis Khan and the Silk Road. They say—I mean, obviously, this led to the Black Death, so it’s tough where that nets out. They say that about Napoleon and the Napoleonic Code and so forth. Where do you basically come out on that story? Have the great conquerors been good for the collective brain or bad for the collective brain?

Well, I’ve never actually focused on that. My immediate reaction is that there must be a better way to transmit the knowledge than through conquest. And then I guess there’s also the open question of how much the disruptions caused by them set things back.

And population decline, spreading disease and stuff, so certainly it could have been done better.

Right. And going back to David Reich’s work, if your theory implies that there should have been a decrease in genetic similarity in Europe after the 5th century or something, as a result of the culture and the Church’s practices, do we see that in the genetic record?

Yeah. So Europeans are quite well mixed compared to other populations. And so we were talking about India. If you do a principal components analysis of this 2-dimensional plot of genetic similarity, you can actually see the castes, and people are pretty spread out. If you put the Europeans on the same plot, they’re well mixed.

Right. But do we see the period of greatest mixing starting in the 5th century?

Yeah, good question. I don’t know the answer to that question. That’s a nice piece of evidence. There are a couple of groups working on medieval ancient DNA, so hopefully we’ll have more answers on that question.

There’s enough now from pre-Roman burials—Bronze Age-type stuff—showing that early Europeans definitely had complex, intensive kin groups: patrilineal, patrilocal residence kinds of stuff, and polygyny.

And then, stepping back, I’ve read there are so many different theories of the Industrial Revolution and of modernity. There’s Robert Allen, the economic historian, who thinks it’s a result of higher wages because of the Black Death. There’s Gregory Clark, who thinks it’s because of this positive eugenics in England, where the upper classes were having more babies.

I mean, there are 20, 30, probably hundreds of different theories of why it happened, where it happened, and when it happened. And so each one of these, when I read them, sounds plausible. I don’t have a knockdown argument against any of them. But I’m not sure how to think about which one is correct, and yet, if we’re stacking all of these up against each other, why do you think this story is any more accurate than any other story that people tell? Is it overdetermined?

Well, the first thing to think about is that those other authors—possibly Gregory Clark is an exception—but they don’t think about the psychological variation that exists.

There’s good reason to think there are all these psychological differences, and you have to believe they don’t cash out. If you think they exist, then you have to explain them, so you need a theory that explains them. Robert Allen has a blind spot when it comes to psychological variation.

Let me give you some evidence that I think is pretty good. We have a database of the diffusion of Roman bishoprics through Europe. We look at what happens when a bishopric arrives. I mentioned before that you get a flow of people.

Another fact is that if you look at the production of famous people, these places where the bishopric has arrived start producing more creative individuals relative to non-creatives. There’s an uptick, and it keeps going up for centuries, suggesting that you’re producing more authors, inventors, writers, and whatnot. If you take each of those and correlate it with modern patents, places that spent more time under the church produced more patents between 1980 and 2014, based on European patent data. This holds if you just compare regions within the same country. You can see long-term effects of this, as well as immediate effects in the historical record on the production of creatives.

What would you make of a theory that says, “Look, all of these theories are basically correct. All of these stories are basically correct”? It’s their stacking up that leads to the Industrial Revolution, rather than any one of them being the most important proximal cause.

Maybe there are reasons for this. Let’s say there are 10 of these stories and you had to stack them all up, and China had 5 of them, but it also had perhaps 3 others independently, and 2 stories that went against them or something. No one of them is that proximately important; it’s their combination, right? I guess your theory seems compatible with Robert Allen’s—that coal being cheap and wages being high was important.

I think one of the mistakes is to focus too narrowly on England, because England was benefiting from ideas that were flowing in from France. There was a lot of great science being done in France. I don’t think Gregory Clark’s right. I can explain almost everything he explains with cultural evolution, and he doesn’t even really take that seriously.

Oh, interesting. Yeah, patience is one of the things he argues for.

In WEIRD, I use the exact same data—I actually get the data from him—to show that you have this increase in patience. But we know that people can culturally learn patience, and this can all be culturally transmitted, and in this world, it leads to more success. If you’re the kind of person who doesn’t waste a lot of your money, and we begin to value thrift and stuff, which Protestantism does, then we should expect an increase in patience, fewer murders, and lower interest rates.

What’s the reason to think that—I mean, his main explanation, I think, is genetic, right? Is there some reason to think that that is less likely to be as important a factor as culture?

We know that migrant populations in the US and Europe shift their psychology over a few generations of being there, and we know there have been natural experiments done by economists like Chris Blattman, where you actively try to teach people—train them, essentially—to discount the future less. That seems to be culturally transmittable. This suggests that—I can’t rule out that there’s been any genetic evolution, but I can show that culture can operate this quickly and powerfully.

Right. I think one of his key pieces of evidence—I don’t remember the exact numbers or even the particular detail—was something like this: if you look at the inheritance of wealth or titles—I forget what the exact thing is—the correlation across generations points to the kind of pattern you would see with genetic inheritance rather than cultural spreading.

Yeah. He doesn’t actually fit any cultural evolutionary models, and if you include that people are learning from parents and learning from their social milieu, your parents determine where you are in the social milieu. You’re surrounded by people who are being more patient and behaving in certain economic ways, and that’s going to have a huge effect.

I just told you about how if you grow up in Silicon Valley, you’re going to patent in computers. It’s not because you have genes for computers.

Given the persistence of these cultural effects, how should that change how we think about whether it makes sense to have more immigration from societies that are already WEIRD, or whether that matters or not?

We have a paper under review right now in which we show that from roughly 1850 to 1940, a big driver of US innovation—both the quality and quantity of patents—is the cultural diversity of counties. We actually use immigration as a kind of way of showing the causal effect of this. What you want is a lot of cultural variation.

If you get people from societies that are more distant from whatever the current US culture is at the time, it’s going to take more time until they’re able to fully integrate. If they’re coming from a very distrusting society, they’re not going to be able to latch into the collective brain immediately.

You see this around 1900. There’s data on people, mostly coming from Southern Italy and from Germany and Britain, and the Southern Italians are coming from a society that has a high distrust of strangers. So their immediate effect on the patenting system is low. But the strength of these cultural traits seems to go down by an order of magnitude each generation.

In the second generation, you can still see where they came from, but they’re a lot more like the majority culture than their parents. Then it goes down another order of magnitude. You just have to do more work, essentially, to assimilate people, enculturate them. But the potential value of having diverse ways of thinking and different ideas can be really valuable, I think.

What do you make of this? Garrett Jones is an economist who has this—he calls this the spaghetti theory of assimilation. I don’t know if you’ve heard of this.

The idea is, look, immigrants do assimilate, but they also assimilate us. Whatever cultural practices they have will, in proportion to their size of the population, have an effect on wherever they migrate to. His example is when Italians came to America: spaghetti, which is traditionally an Italian dish, also became an American dish, right? So they assimilated American culture as well.

Substitute traits like trust or whatever there. This sort of reciprocal assimilation has implications for whether we want immigration from non-WEIRD societies. What do you make of that idea?

I think the spaghetti example—or pizza; pizza or spaghetti, both good examples—is a great thing, right? American food is a fusion of cuisine from lots of societies, and that leads to quite good food. You can see the same thing in music. Things like jazz come from African rhythms, and rock and roll is along that lineage.

But do we want diversity in low trust and high trust?

The question is whether social-interactional traits operate the same way as these things. If you have ethnic enclaves, then you’re going to get low trust in those ethnic enclaves. You get things like the mafia, right? So you don’t want that.

But if you had an immigration policy that distributed people, and in a high-trust society you benefit by being higher-trust because you can do things like collaborate and start companies and stuff, which you have to be high-trust to do. Otherwise, you don’t do it in a very effective way. So there is a pressure for low-trust people to become more high-trust.

But if you’re among mostly high-trust people, there’s not an incentive for you to move down. Social-interactional traits are different from things like food types.

In psychology, there’s been this problem of the replication crisis, where a lot of the main results are hard to replicate, and it’s not clear how much of this is real science. Have you looked into how many of the WEIRD results—the differences in psychology between different populations and so forth—actually replicate? How much of that actually replicates?

Yeah, my sense is that it replicates quite well. We published our paper in 2010, and in the late 2010s, Armin Falk, an economist, and a bunch of others—Anke Becker, Ben Enke—published a paper where they measured economic preferences in 80,000 people around the world. That just showed big variation in things like patience, various kinds of reciprocity, and altruism—the kinds of things we would expect.

And then other large-scale projects have similarly shown lots of variation.

The groups we have today, which are not WEIRD, when you study them, do you think that the fact that they have resisted modernity for so long suggests that they actually are weird in their own way? Which is to say, this is not representative of the way that Europeans might have been thinking in the 2nd century; rather, it’s like the Hadza’s particular weirdness leads them to be averse to modernity in a way that is unique?

Well, that’s always a concern, and it’s especially a concern when you’re studying hunter-gatherers, because a lot of the hunter-gatherers that are left in the world today are in places where agriculturalists couldn’t easily move to. So that’s something we’re thinking about all the time.

The ways around that are to go to places where agriculturalists didn’t go or couldn’t go. So we have a lot of good ethnographic history on people in Australia, and then we have the Arctic populations, which—big sections of Paleolithic Europe were probably kind of like Northern Canada is today. So at least environmentally, it’s not crazy, because Ice Age Europe.

And places like the Aleutian Islands and the West Coast of California—we have lots of ethnographic evidence that there was no agriculture there. So just putting together all these different lines of evidence helps us develop a picture. I wouldn’t want to bank everything on the Hadza. Of course, we can’t do experiments with some of those groups, but we can see whether the Hadza look like these other hunter-gatherer groups.

Speaking of the Ice Age, do you have a take on why, before the Ice Age, agriculture didn’t develop, even though genetically there probably weren’t that big a difference between 20,000 years ago versus 10,000 years ago?

Yeah, I don’t have a clear picture of that. It is the case that the Holocene was a particularly long stretch, a particularly long interglacial. There were some long interglacials, but the one before that was 120,000 years ago. Other than that, they were getting broken up.

I sometimes wonder if we may someday figure out that there was actually a little bit of agriculture, and it would’ve all been destroyed by the Ice Age, right? So it’s not clear that there would be any heritage of it left. We would have to find some remnants of some domesticated crops.

Although I was asking David Reich about this, and I forget the reason he gave, but he did say that if there were societies before the Ice Age, we would have evidence of them in the archaeological record or their genetic record or something like that.

Yeah, if they had gotten to any scale. But we know from modern societies that there are groups, for example, in the Amazon that have root-based agriculture that wouldn’t leave any archaeological record.

This is a question I also asked David Reich. With modern LLMs, and just generally with newer techniques for processing information and understanding evidence, there’s a potential for answering questions that maybe we couldn’t have answered before, especially from a cultural perspective, where you can actually consider lots of different cultural artifacts at once. In that context, you can run cultural simulations or something.

Maybe those are more fanciful ideas; maybe there are more practical ideas. But if there’s one question you have that’s sort of bottlenecked by having the right data or being able to process the data the right way, and future LLMs or AIs could help us with, does something immediately come to mind?

Well, this gives me a chance to tell you something we’ve been working on. We were talking about this kinship hypothesis, the idea that kinship intensity affects psychology. And when I presented the ideas that are in The WEIRDest People in the World, I get a hard time sometimes from historians who will have some kind of very Europe-specific story about why this happened that has to do with European royalty or coal or something like that.

So my move is not to try to get into the weeds and start reading Latin texts, but to go test it somewhere else. If it’s true that kinship intensity should work like this, then we should be able to test it somewhere else.

So we have 2 different corpora from China. We have late-imperial China, something called the gazetteers, which were produced across the prefectures, and they had stylized content. So this is the same genre, thousands of books, and then we have about 7,000 books going back 2,000 years in Chinese history.

The techniques that AI is allowing us to use let us get measurements of psychology from the texts. We take a questionnaire. This is a technique developed by my postdoc, Mohamed Attari. We take a psychological questionnaire that’s been validated, is in English, and we want to get an equivalent in ancient Chinese.

So we run it through a semantic similarity comparison, and we look for quotes from the ancient Chinese corpora that match each sentence in the English corpus. And then we rebuild the questionnaire in ancient Chinese. So that’s our psychological measure for something like individualism or collectivism or moral universalism.

And then we take each book, or each paragraph in each book, and we do a cosine similarity comparison between the 2 sets of text there. That allows us to stamp each paragraph with a measure of individualism, for example. And then we do that for the entire corpora.

This allows us to track psychological change across space and time in China. And then we can correlate that with kinship intensity, and we get the same correlations that we do in Europe.

6. Loss of cultural variance in modern world

Interesting. How worried are you about the fact that, if you think about modernity, it was the result of finding this one cultural variant—at least according to your story—which then helps us develop better technologies, all these new institutions, and so forth? How worried are you about the fact that the modern world, because of the spread of WEIRD, has far fewer cultural variants, and because of this monoculture, a potential variant that might be useful in the next transition just wouldn’t emerge?

Yeah, I am worried about that, because of the destruction of languages. We’re losing languages left and right. English has a particular form that you don’t find in lots of languages, so that’s just an example of the kind of cultural diversity we’re losing. So that’s definitely a worry.

I do think that we’re getting new variants. Japan adopts a lot of WEIRD institutions, but it’s really creating a new, third, unique thing. It’s got a bunch of WEIRD elements, but Japanese law, despite being the same as the US, operates very differently.

So unlike Americans, Japanese don’t tend to sue each other, and they tend to use mediation and things like that, even though the superficial law is pretty similar. So I do think there is a creation of novelty which could be useful. But, yeah, in general, I worry about the loss of novelty.

I also should say that I think that the impersonal world of impersonal institutions in WEIRD psychology is quite fragile. For example, in the US, in rural areas, you see moral universality declining since 2008. This is based on John Haidt’s YourMorals.org data. And so there’s an increasing difference in the morality of urban areas versus rural areas in the US.

I’m sure you get asked about this a lot, but fertility seems to be declining around the world, and there doesn’t seem to be any existing cultural variance, other than maybe the Amish, who can resist it. Do you have takes from a cultural anthropologist’s perspective on what’s happening here?

And a follow-on question is: Should we really encourage these cultural enclaves, like the Amish, given that this anti-fertility meme is so viral that they can’t be part of the common culture and still preserve fertility? Do you really need these closed-off societies in order to preserve fertility? How seriously do you take that idea?

Well, I definitely take declining fertility pretty seriously. And especially if you’re a collective-brain guy like me, just having fewer people is going to be a big problem.

And I think that there will be a spread of ideologies or religions or something that are pronatalist, and those groups will have a big advantage in cultural evolution, because a community that is pronatalist tends to produce more pronatalist babies. Christianity spread using that.

Mormonism spread in the 19th century using that trick. So I just feel like cultural evolution is gonna find the combination, and there’s gonna be some pronatalist groups spreading.

There are a lot of different societies in the world today. Is there some explanation of why none of them have the existing variants necessary to keep fertility high?

Well, from a cultural evolutionary perspective, this is relatively new, right? The demographic transition is only really from the late 19th century. Lots of the world is just getting hit after 1970.

We also have things like rising female labor force participation, which is gonna stifle it, and rising female education. Once that stuff maxes out, then you’ll see variation among different groups in terms of the number of babies they produce, right?

This thing unfolds on demographic timescales, so we’re not gonna see it for a few generations. But there’ll be some group somewhere that will be producing more babies than everybody else.

The reason why I think religion is the likely one is because people do things because they think God wants them to do it. If people come to think that God wants them to have more babies because it’s a way of worshiping him or getting to heaven, or whatever their religious configuration is, then that’ll be a group that produces more babies. Catholics were defying the demographic transition for a while. They just seem to have stopped.

Right. How worried are you about the fact that, because of a decline in war between great powers, there’s this buildup in cultural mutational load that you were talking about? Or maybe things just get less efficient over time and there’s no selective pressure to get rid of that inefficiency.

How worried are you about that just making governments or nations less efficient over time because there’s no outer-loop loss function that says, if you mess up enough, your country might not exist, or your group might not exist? That was what we were talking about with the waves of the Yamnaya or the Anatolian hunter-gatherers just conquering everything. If that doesn’t happen, will we just see culture, nations, and governments degrade over time?

I think the answer to that is yes. The 1 way that countries, or whatever the political institution is, could address that would be doing this thing that I’ve curiously called “domesticating the competition.”

Sports teams constantly renew themselves by competing. Firms live and die and renew themselves over time. Like you said, companies will start off being super great, and then they get too big, and then they get kind of inefficient, and then eventually they disappear. None of them last forever.

That seems to be true of political units. But in principle, you could have at least some renewal processes. Democracy potentially provides a renewal process, although there are things like the institutions that the government builds that are hard to renew.

One simple political idea is that when you create a new department, it should have the same thing that cells have, where at a certain time, they time out and you gotta make a new one. The reason is that just the way human bureaucracies and institutions work is they kind of corrode from the interior, from the inside, just the way cancer spreads in a cell. So you just gotta kill it and make a new one.

We could institute that, but I guess the idea hasn’t quite caught on yet. There are bits and pieces of it around, so it’s not like it’s unused.

Yeah, I guess maybe the thing that’s less understood or is not appreciated is that the reason institutions have improved over time in history is because of the selective process that, at least at certain levels of selection, no longer exists.

Yeah. The Roman Empire didn’t say, “You know, we gotta redo everything. Why not adopt democracy and we could have a republic?” Nobody does that. Things fall apart.

The researcher Stuart Armstrong has this idea called Chesterton’s meta-fence. Here’s what it states: “In our current system, democratic market economies with large governments, the common practice of taking down Chesterton’s fences is a process which seems well established and has a decent track record and should not be unduly interfered with.”

Can you say any more about what the author has in mind? I would like an example or something.

Yes. So I think this is a rebuttal against this common idea that we shouldn’t be arbitrarily changing things. Suppose a new technology comes about and we’re worried about the risk it might pose to society. Suppose young kids are doing something different culturally and we’re worried about what effect that might have on the culture and so forth.

He’s basically saying, “Look, this has been happening for 5 centuries, and this process by which new cultural variants enter the mix and so forth has worked really well for us, even if it’s happening at a rapid clip, and we shouldn’t interfere with it.”

Well, I don’t think I have strong feelings on this. I’m always focused on trying to understand the process of cultural evolution, and it is true that people often resist cultural change that, in retrospect, we think is good.

But of course, when you begin to make cities, you lead to all kinds of epidemic diseases. If you’d anticipated that, you might have worked on some of the public health procedures before you built the cities. So foresight can be good.

Right. I guess the sort of bigger question here is this: It makes sense, in the kind of environment our ancestors faced, why they’d have intuitions against progress, because if you already have technologies that are well-fitted to the environment, any change is likely to be deleterious.

Whether we find ourselves in a world that’s different enough that we can just very intentionally disregard our worries about changes in culture or technology, or whether it’s similar enough that we should actually take those worries seriously, is the bigger question.

There are 2 ways you can interpret The Secret of Our Success, right? 1 is, “Oh, look, this is why we should care more about culture,” and so forth. Another is just, “Oh, wow, they’re living in such a different world. I understand where my intuition against progress comes from, and now I can totally disregard it.”

Right. So I would say that the lesson from The Secret of Our Success is not to disregard it, but just to be cautious, because we could be dismantling things that are really important for the structure of society.

You shouldn’t just dismiss valuable cultural practices as the relics of a medieval age, or the relics of a pre-enlightened age or something like that.

Ritual is a good example because rituals seem to have real psychological benefits in binding people in community and helping develop self-regulation. But it’s easy to be an atheist like me and say, “Ah, rituals are stupid. We gotta stop doing them.”

It turns out they’re doing a bunch of stuff, and if you don’t wanna lose that stuff, you gotta figure out another solution.

7. Is individual genius real?

A question that arises in my head is this: We began this conversation talking about all these societies in the past. Even when they figured out something successful, they did not themselves know why it was successful, and so they just had to ascribe it to custom. They’d say, “This is the way we always made bows,” or hunted or something.

Why think that we have an answer for why the Industrial Revolution happened? Why think it’s any different from the Inuit trying to explain why they make their bows a certain way?

Oh, you mean how what we’re up to is different?

Yeah.

Well, we’re trying to apply science, which has been very successful in all these different areas. So, with evidence, we all put out our arguments, we go through the process, and some evidence is better than others. That’s just our epistemology.

I’m interested in the cultural evolution of epistemologies. Societies have varied over time in what counts as good evidence and what counts as a good argument.

One of the psychological changes that we see emerging in Europe compared to other places is how important what the ancients say is. In lots of societies, if someone says something and you can say Aristotle or Confucius didn’t believe that, then it’s like, “Oh gosh, I guess I’m sunk.”

Whereas at some point, Europeans decided, “Actually, who cares what Aristotle thought?”

We know a lot more than he did. So this is a big epistemological change. And the emphasis on empirical evidence in science is not something you find in earlier traditions within Europe, and it’s quite variable across different societies. So you should think about the very standards by which we count a good argument and good evidence as themselves culturally evolved.

Yeah. And how much do you worry about the lack of cultural variation in science in particular, where we have one big institution, the NIH, which funds most science? There are similar processes of consensus and peer review and whatever that determine whether you get a grant, whether work is accepted, whether it’s considered worthy scholarship, and so forth.

Should we think that maybe we should have more variation in those methods because maybe the scientific method isn’t complete? There are epistemic tools that could be improved, or it could be different, and so forth?

Yeah, absolutely. So, polycentricity. There should be lots of difference and competition amongst these groups, no single funding source, and lots of different priorities. One thing I point out in this book I’m working on is that scientific papers tend to be more impactful when the authors are from more diverse societies.

But interestingly, people are biased to work with people from their own society. So what we actually do is different from what would maximize innovation.

I’m sure you’re going to address this in the book, but I’m already curious now. It seems like the internet hasn’t been as big a deal as somebody thinking from this zoomed-out collective-brain picture might have anticipated in the ’90s. Maybe it’s boosted the rate of innovation somewhat, although it’s not clear, just looking at growth rates or productivity growth rates or something, that it has.

Why didn’t the internet cause the equivalent of another round of urbanization in our ability to do research, economic productivity, or so forth?

Yeah. And that’s something I’m really interested in figuring out. One thing that I still don’t have a full answer to is that there’s something special about face-to-face interactions.

Something like whether, even in the 21st century, two cities have direct long-haul flights between them increases the flow of ideas between those places. But we know these places are connected by the internet, whereas that effect doesn’t seem as strong. Part of this is that people probably have to build trust before they start sharing ideas.

The more different someone is from us, the more you need that trust. Some of the research suggests that this effect is even bigger when places are more culturally different. And this is interesting because they’re probably more valuable to each other when they’re more culturally different. But then face time and being in the same place is even more valuable.

So that’s the kind of direction I’m going. And I still think that there should be exactly what you said: the internet should have caused more innovation. I’m just not satisfied that we’ve fully determined the best way to figure that out.

You had a very interesting chapter in *The WEIRDest People in the World* where you talk about how a lot of innovation through history is more serendipitous than we imagine, and it’s the result of the collective brain being big enough to discover these things.

But if you look at the track record in science and research, there do seem to be certain individuals who make independent discoveries across many different fields that are each quite important—your Newtons and Einsteins and John von Neumanns and so forth. If it were just the result of the right person being in the right lab, seeing the right observation at the right time, this repeated excellence by certain people seems less explainable.

So I’m curious if the collective-brain theory can explain what’s going on there.

Right. Sometimes I’m interpreted as saying that there are no genetic differences among individuals. I definitely think there are genetic differences among individuals that affect their likelihood. But when we take the individual, we often import the person’s life history into our thinking about them.

For example, when Einstein was a patent clerk—so he wasn’t succeeding as an academic—he and his friends got together in something called the Olympia Academy. They called themselves that, and it was just a group of 5 people who would get together and read the interesting books of the time.

If you look at the books they read—and historians have done a lot of work on this—all the major ideas that go into special relativity were read and processed by the Olympia Academy before Einstein developed it. The idea that people think time is relative—people were talking about that, and they were talking about multiple dimensions.

Henri Poincaré, at the same time, came up with the same equation as Einstein, but he didn’t give it as radical an interpretation as Einstein did. He was thinking of the equations as fudge factors, trying to make the math work. So there was an almost simultaneous invention of special relativity, and the ideas were all circulating.

Einstein happened to be in a place that allowed him to put all those things together. As a patent clerk, he was actually processing devices to synchronize trains in different parts of Europe. So he had to think about the amount of time it took a signal coming from different parts of the world. There were a bunch of these.

We know the patents that went through, and we know that he had to look at them. There were only 2 guys. So he happened to be in a particular place and time, in terms of the ideas circulating and what he was doing on a day-to-day basis, that really did give him an edge.

Now, you could then say, “Well, what about general relativity?” There were a couple of other guys who probably would have gotten general relativity. Einstein himself was worried that he was going to get scooped, because once you get the one, it’s just a matter of—I mean, it’s not “just” a matter of figuring it out; the math is really hard. But there were a number of other people who probably would have gotten the math.

Yes. I guess there’s also still the question of why, in 1905, Einstein himself discovered not only special relativity, but Brownian motion and the 2 other things, right? But I guess you would just say it’s the result of him having the right reading group? I mean, obviously, he’s a special individual.

Although he did spend the whole second half of his career trying to show why quantum mechanics was wrong. His Brownian motion paper is foundational in quantum mechanics. But then he decides God doesn’t play dice with the universe, and he doesn’t like the stochastic nature of quantum mechanics, which has more or less proved true and is part of our modern technology and stuff like that.

He spends the whole second half of his career fighting it. So that’s a case where his intuitions didn’t serve him well. He turned out to be wrong for decades.

And, sorry, I think that’s also a case in which your point about having the right ideas available to you is important. I don’t know much about this, but from what I understand, the people who believe in the many-worlds interpretation believe that Einstein was on the right track, and he just didn’t have the idea of the ever-branching multiverse available to him.

That there’s this branching thing, and it’s just a matter of figuring out which branch you’re on.

Right. And so there’s no probability. Yeah, yeah. Okay, so we were just discussing how Einstein was reading the right things and had enough context to come up with these big discoveries.

You’re somebody who has connected ideas across many different fields. How would you describe the input function for the way in which you come up with new ideas? You started off as an aerospace engineer, and you’ve done anthropology across so many different societies, through fieldwork and so forth.

How do you basically think about the Joseph Henrich production function?

Yeah. I try to implement the ideas of the collective brain. So in my lab at Harvard, I have postdocs and graduate students with backgrounds in evolutionary biology, psychology, anthropology, and economics.

Some of my favorite collaborators are economists. We’re just bringing ideas from very different places, but across the social sciences and even the biological sciences, we’re often trying to explain the same kinds of phenomena: economic decision-making, cooperation, things like ethnicity—why does it exist? Stuff like that. So the way we silo science is a big problem, and especially in the social sciences, I think it’s an even bigger problem.

I’ve been a professor of anthropology, psychology, economics, and evolutionary biology, and even the standards of evidence, what constitutes good research, and how you tackle a problem really vary. So the academic disciplines are like different cultural worlds, different tribes, and just pulling the best from them has kind of been my approach.

Among the more polymathic scholars, do you guys have—I don’t know—do you and David and a couple of the other ones in this category have some shared group you guys have, or is it—

No, it’s a series of different networks. David Reich and I, for example, are always talking about how we want to read culture from the genes. If humans have had a long history of gene-culture coevolution and we want to understand what Paleolithic populations were doing, we might be able to see imprints of it in the genes.

In my own work, for example, we’ve measured cousin marriage across different populations and shown that it correlates with runs of homozygosity in the genes. Something like polygeny can be revealed by looking at the ratio, the variance in Y chromosomes to X chromosomes. With fire, we probably have some special genetic adaptations for dealing with the toxins in smoke. So if you want to know when humans got fire, and you see the gene that gives us immunity to toxins in fire, then we can infer that fire is older than that, right?

Interesting. So obviously, the input from different scientific fields in your work is obvious. But from a more philosophical perspective, people compare your work to the same tradition as Burke or Chesterton or something. Do you personally find that your philosophy has been informed by reading them, or is it just independently converging on some of the same themes?

Any convergence is completely independent, because I haven’t really spent much time at all. I’ve gone back after people said, “Your work is kind of like Hayek,” and read some Hayek. I did read The Fatal Conceit in graduate school, but aside from that, I didn’t read very much.

The same thing with Adam Smith. I tackled The Wealth of Nations, and I eventually read The Theory of Moral Sentiments, but it was pretty far into the process. I had picked up on themes they developed before I knew they had developed them.

I want to suggest that maybe you should add a couple of AI people in your rotation.

8. IQ and collective brains

Okay, just because I think this perspective is incredibly informative. What I think is maybe the most important question of our period is: What does this transition toward a society of AIs look like?

Sorry to bring it back to AI, but one other consideration here that your work has really informed me on is that I was maybe hyping up how big a deal I expect societies of AIs to be. But something your work informs is the idea that a single superintelligence is not the place in which to expect these big impacts of John von Neumann times 10.

Then again, there is a key advantage in the scope of social learning they can do as a society. But the idea of a single superintelligence being super powerful is maybe less likely as a result of your work. I don’t know what you think about that interpretation, but—

Yeah, I think that’s right, because the assumption—the sort of model that people seem to carry around in their heads—is that humans have done all these things because of our individual brain power, when really it’s been the power of cultural evolution and a network of minds gnawing away at problems over long periods of time and gradually accumulating not just the obvious tools, but also the cognitive tools for addressing these things.

I buy the idea that individual IQ isn’t the most relevant factor in understanding why discoveries are made. It’s not about having the right genius or something. But what’s the reason for thinking that the average IQ of a society isn’t super important in how much progress a society is going to make, not just the population size?

The first thing that we need to do, though, is to zoom back out and think about what we mean by IQ. Michael Muthukrishna and I have made the case that IQ is just a set of cognitive abilities that leads to success in 20th-century contemporary institutions that have come to dominate the world. It’s a set; it’s a culturally evolved system. We talked before about the Flynn effect, which illustrates that, and the fact that IQ is associated with all these positive outcomes now, but certainly wouldn’t be in pastoralist societies.

So that would mean that if you raise the average IQ of your society, it might lead them to have more abstract thoughts and do better science. That certainly fits. But the interesting thing for me is that the world of the future is going to be quite different from the one now. So the set of cognitive abilities that is going to be favored in the new AI world is not going to be the same set that was favored in 1900, when they designed the IQ test.

For example, in the IQ test they ask you to remember digits backward. I’m not sure how useful remembering lists of digits backward is. That was useful in a previous world, when we had to write everything down and remember a lot of stuff. But in some sense, we’re interfacing with our technology, and we’ve got to figure out what set of cognitive abilities is going to make people best able to solve problems. As we talked about, it’s even the case that the most creative people aren’t the highest-IQ people.

I guess one of the things I’m trying to say is that the minds that might lead us into the new world might not be the ones that have the highest IQ. Once you’re augmented with AI and all these kinds of technological aids we have, the specialized thing that leads someone to do something creative probably isn’t going to be the same set of abilities that did it in 1910.

Yeah. I guess my real opinion is that if we’re in a world where that’s true—and I think that’s probably going to be true—then it really doesn’t matter at all, because AIs will be doing everything. But in the world where, let’s say, AI plateaus or something, then I would expect roughly the same kinds of skills that have mattered for the last couple of centuries to keep mattering for a modern technological society, which is analytical thinking and the ability to understand science and technology and so forth.

But maybe the AI world you’re imagining is different from the one I’m imagining. I still think that people are going to figure out what problems we need to solve, unless we’re just going to tell the AIs to figure out what the problems are and then solve them. They might not be that good at that. I don’t know. We’ll see.

You shared with me this draft of work about collective brains, where you show that ants have developed many of the technologies that humans have, like farming, livestock, division of labor, and so forth. So maybe there’s some amount of blind selection, and it doesn’t matter if it’s natural selection or cultural selection; it’s the size of the group that can go through this learning process that matters, and how many people are available to experience the learnings or have the selection process act upon them.

But I guess the big difference—where maybe individual IQ comes back into the picture—is that I don’t expect there to be any amount of natural selection that will allow ants to land on the moon or make a computer chip or, you know, make a nuclear fusion plant or something. And so is the kind of broad generalization we see with humans—and we might see to an even greater extent with AI—a product of the fact that you’re really bottlenecked by the IQ of the individual? There’s no amount of collective learning that can get you to the moon if you’re an ant?

Right. Well, one thing to keep in mind is that most human societies, over most of human evolutionary history, didn’t get to the moon. So this is one particular group of humans.

And one of the things we talked about earlier in this conversation was the cultural evolution of epistemology. So it was the improvement in what constitutes evidence, what constitutes a good argument, that allows us to get to science and accumulate this kind of knowledge to do these kinds of things. So I see that as part of a continuous trajectory. But it’s just that we have new cognitive tools.

Right. I guess, but there’s a really interesting point that even the epistemic tools which let us get to the moon are themselves a product of cultural evolution. But again, that seems bottlenecked by the fact that if you’re just dealing with chimpanzees or something, there’s no amount of cultural evolution that will result in the scientific method—

Right. It needs some gene-culture coevolution.

Right, right. So you mentioned in The Secret of Our Success that tool use started something like 2 million years ago, and fire—we started domesticating fire closer to 800,000 to 1 million years ago. Intuitively, it doesn’t seem like tool use is that much simpler than fire, but is that a misunderstanding? Why was fire so much more recent?

Well, what we know from other species is that lots of animals use tools, and particularly chimpanzees use tools. So we can assume tools in the common ancestor. Now, what we see in the paleoanthropological record is the increasing use of stone tools. These are pretty simple stone tools. You can see a cutting edge there, but not very much fancier than that.

And then fire is a lot different. A lot of animals are afraid of fire, and they have to run from wildfires and stuff like that. So whatever your story about humans is, you have to overcome the fear of fire in order to tame it. Probably humans first found wildfire, and somebody approached the fire instead of running away, which is the usual thing to do, and then got some of it and put it to use, I guess. So I think it’s the innate fear that animals have of fire. We don’t hang around when things go off.

Interesting. Okay, final question: what’s the next thing you’re working on?

Well, the big thing is this book on collective brains, so I’m gradually working through that. One of the ideas that I really am excited about in this book is that we evolved to think in collectives. There’s this assumption that psychologists have had to understand human decision-making and how our minds solve problems: that we should put people by themselves in experiments and see how they do.

But real human societies—hunter-gatherers, for example—they actually work in groups. And when we want to solve a problem, the first thing we want to do is check with our friends or ask other expert members of the community. So it’s the idea that we think collectively and solve problems collectively in a kind of naturally distributed brain.

Hugo Mercier and Dan Sperber, for example, have argued and shown with various lines of evidence that many of the irrational biases we have are mitigated or disappear entirely when we solve problems as groups. So it’s almost like we evolved to have that positive interaction and correct each other’s errors.

Right. That’s quite interesting, that you care more about the portfolio of intelligences than any one person being calibrated. Right. Excellent. This is super fun. Thanks so much for coming on the podcast.

All right. Great to be with you.