Kathryn Paige Harden is a professor in the Department of Psychology at the University of Texas at Austin, where she leads the Developmental Behavior Genetics lab and serves as Director of Clinical Training. Her latest book is Original Sin: On the Genetics of Vice, the Problem of Blame, and the Future of Forgiveness.
In this week’s conversation, Yascha Mounk and Paige Harden discuss why twin studies reveal the substantial influence of genetics on human behavior, how genetic effects actually increase rather than decrease over a person’s lifetime, and why acknowledging genetic influences shouldn’t be seen as incompatible with progressive politics.
This transcript has been condensed and lightly edited for clarity.
Yascha Mounk: Thinking about genetics, and particularly the genetic influence on human behavior, is, for lack of a better word, often right-coded. People assume that it’s people on the political right who want to emphasize the genetic determinants of human behavior, and it’s often people on the left who are very allergic to this, who say that genetics doesn’t really explain anything—it’s social forces. Somehow, if we accept that genetics influences human behavior a lot, then that would lead us down all kinds of dark alleyways.
You see yourself, and think of yourself, as a woman of the left. You also think that we have, in many areas of our life, underestimated the role that genetics plays in explaining human behavior. Let’s start with the latter part of this: why should we think that genetics is such a big determinant of human behavior?
Kathryn Paige Harden: We should think about the importance of genetics because we have a lot of scientific evidence that genes matter in shaping our lives. We can see that in twin studies, which were the workhorse of genetic science in the 20th century: how similar are genetic relatives, even if they haven’t been raised together, for certain aspects of their behavior? We see it now with more modern technology, where we can actually look at specific segments of DNA, how they differ between family members, and are beginning to discover how they help shape behavior.
It’s very rare that genes are determining behavior, but my PhD advisor from graduate school is very famous for a paper called “The Three Laws of Behavior Genetics.” The third law of behavior genetics is: everything is heritable—which means that everything that differs between people, when you look at it scientifically, tends to show some influence of the genes that you were born with. I’m a scientist; I’ve been working in this field since I was 18. It’s important to me because I think it’s a scientific fact that is incontrovertible at this point: genes influence behavior.
Mounk: Let’s walk people through these arguments step by step, because I think a lot of people have heard about twin studies and they know that nowadays there’s probably more modern techniques that are slightly different. What exactly do the twin studies show and establish?
Harden: Twin studies typically compare identical twins—who were conceived from one egg and one sperm, and then there was an error in early cell duplication so that now there are two people—to fraternal twins, who are the sort of twins who look like ordinary siblings: two sperm, two eggs, who just happen to share a pregnancy. That logic can be expanded to compare full siblings to half siblings, or adoptive parents and children to biological parents and children.
What all these designs have in common is that they take people who share a social relationship. Both fraternal twins and identical twins have the same social relationship as siblings being raised in the same home with the same parents, but have different genetic relationships.
Mounk: The idea here is to try to hold as many things constant as possible, other than genetics. You’re going to assume that perhaps there are differences between siblings—perhaps the family became a lot richer over the course of three or four years, and so if one sibling is three years younger, you might say it’s because of that. But if you’re comparing identical twins to fraternal twins, there’s no three years in between for the family to become richer.
Harden: Yes, and it’s the same pregnancy—the same maternal smoking, the same neighborhood they happen to live in, the same prenatal supplements that the mother was taking. The only difference between them, in theory—and we can get into whether this assumption holds—is the difference in their genetic relationship. You’re then trying to see: does the similarity in their behavior track the similarity in their genetic relationship, or does it track the similarity in their social relationship?
One of my favorite studies looks at twins where the mother is wrong about whether the twins are identical or fraternal. The mother thinks they’re fraternal because they look different to her—she knows them so well. But when you look at their DNA, it’s clear that they’re identical. The question is: do they look as similar in their behavior as identical twins typically do, or as fraternal twins typically do?
Harden: The study shows that behavior tracks the actual genetic relationship, not what the mother thinks the twins are. It’s not explainable by being raised in the same home, and it’s not explainable by parental expectations of how similar they would be. The best predictor of their similarity is how many genes they share.
I want to go back to what you said at the beginning: is that result right-coded or left-coded? It really depends on what you’re talking about. Some of the earliest twin studies were done around schizophrenia, showing that the best predictor of having schizophrenia is having an identical twin who also has schizophrenia. Subsequent twin studies have shown that you also see this higher similarity in identical twins for addiction, for sexual orientation, for depression, for all sorts of mental health problems.
Mounk: That might be left-coded because it implies that it’s not your fault.
Harden: If we look at whether people with more right-wing political ideologies versus left-wing are more likely to believe twin studies, or more likely to believe that genetics influences behavior, it very much depends on what behavior you’re talking about. Liberals are much more likely to say yes, identical twins are more similar for addiction; yes, sexual orientation is partly shaped by the genes that you inherit. It is the application of this to specifically cognitive ability and achievement that tends to be more right-coded in the United States.
Mounk: Tell us a little bit about the magnitude of these findings. When you compare identical twins to fraternal twins, give us some statistic or concrete measure that allows people to get a sense of how much variation there is between identical twins—obviously there’s still some variation. How much more variation is there once you get to fraternal twins? How does that compare to a half sibling, for example?
Harden: There was a really big meta-analysis where they pooled data from 50 years of twin studies—something like 15,000 studies, millions of pairs of twins. The punchline was that on average, across behavior but also medical traits, about half of the variation in outcomes was due to genetic differences between people. One way to think about that is: if people were all genetically the same, how much could you make outcomes the same? You could shrink people’s differences by a substantial degree if they were all genetically identical.
It varies by trait. For things like schizophrenia or height, in modern populations almost all of the variation between people is due to their genetics. For things like personality or depression, it’s about half. When you start talking about education, it’s a fifth to a third. It depends on the phenotype you’re looking at. It also depends on the population you’re studying. If you’re looking at a nutritionally deprived sample—people who didn’t get enough to eat in childhood—you’re going to see less of a role for genes than if you’re looking at people who all had ample access to food, where most of the differences between them are necessarily about their genetics because there isn’t that much environmental variation.
Mounk: The average height in China has shot up over the last 30 years. There’s a really nice example of that in a writer who was in Sichuan teaching in the 1990s. He’s a little shorter than me—like five-seven or something—and he was the tallest person in the classroom. He went back a few years ago and suddenly he was one of the shorter people in the classroom, and he was really struck by that. Obviously it’s not that the genetics of people in China has completely transformed over the course of 30 years—it’s that they’ve overcome the nutritional disadvantage they had as recently as the 1990s.
If you’re in a place where a family doesn’t have enough money to buy enough food for all children, and one child for one reason or another is favored by the parents, or is better at elbowing their siblings out of the way to get to the food, they might end up being much taller—so it’s not genetically determined, it’s determined by social factors. But once you look at a context where everybody gets enough to eat, genetics is the thing that really drives it.
We hope you’re enjoying the podcast! If you’re a paying subscriber, you can set up the premium feed on your favorite podcast app at writing.yaschamounk.com/listen. This will give you ad-free access to the full conversation, plus all full episodes and bonus episodes we have in the works! If you aren’t, you can set up the free, limited version of the feed—or, better still, support the podcast by becoming a subscriber today!
If you have any questions or issues setting up the full podcast feed on a third-party app, please email leonora.barclay@persuasion.community
The other interesting finding here is that you might think genetics plays the biggest role early in life, and that later on its influence fades as people’s paths diverge, as they make different choices. If I’m understanding it right, the actual finding is the other way around?
Harden: What you see is that twins get more similar over time, and people who don’t share genes get more different over time, such that heritability estimates tend to go up with age rather than down. If you’re looking at the similarity between adopted children and their parents, or adopted children and their siblings, they are most similar to these non-genetic relatives in childhood. Then, as they leave the family home and have more control over their environment, the evidence of the genotype tends to emerge over time.
One way to think about that is that, especially in modern contexts, people have increasing amounts of control over their environments—both conscious and unconscious—as they go through life. They pick their peers, they pick their partners, they pick their daily routines, they pick what city they want to live in. Each of those choices reflects, even in small part, their own personality and temperament, and so there’s going to be a feedback loop where the environment is reinforcing those initial genetic differences. That sense of becoming more like your identical twin, and less like people who are genetically different from you, increases over age rather than decreases.
Mounk : What are the main responses from people who are skeptical of the genetic explanation? Perhaps let’s start with popular objections and then go through some of the most sophisticated scientific objections. I should say that my wonderful podcast producer, Leo, is messaging me in all caps over WhatsApp because she has an identical twin and keeps saying there’s no way they’re going to get more similar as they get older. I think for a lot of people, a lot of their sense of self is bound up in many of these questions.
Harden: There are a couple of objections. One is, I think, just a distaste for even doing the research—and this may be piggybacking on the comment you just made, that people’s identities are bound up with their behavior. It can feel inherently reductionistic, or inherently threatening to human agency, to make really valued parts of people’s identity—or very problematic or darker parts of people’s behavior—the object of scientific study, and especially scientific study at the level of genetic analysis. It collapses across our sense of ourselves as embodied animals and our sense of ourselves as choosing, agentic wills or reasoners. One objection is simply: it gives me the ick to think about genes influencing my sexual identity, my religious activity—that can just be uncomfortable.
Another objection is that regardless of the results, and regardless of how the results are presented, there’s the fear that it is necessarily giving ammunition to the most reactionary, eugenic, or racist elements within a society. There’s a history of people using genetics to justify really horrible things. The objection is: even if you’re scientifically correct, there’s no way to do this work responsibly in our current political climate.
Then there’s the scientific objection, part of which is very correct: twin studies make a ton of assumptions. You’re taking a really complicated developmental process and flattening it into one bucket for genes and another bucket for environments—but genes and environments don’t work like that. They’re always combining over the course of development. So what does it mean to have a statistical model that tries to disentangle them? Often people’s objections can float between these. They’re animated by fear, and that makes them perseverate on some of the assumptions of twin studies. These are not mutually exclusive—they’re fluid objections.
Mounk: It is strange that this is an area in which people have very strong moral convictions, and it’s not always clear what drives them. There are a lot of people who feel a kind of sense of threat from recognizing some of this genetic influence. I’m thinking of one very dear friend of mine—an avid listener to this podcast, so she’ll recognize herself being mentioned here—who for a very long time believed strongly that the most important determinant of behavior is cultural: that the way to understand behavior is to study sociology and anthropology and the cultural determinants of how people behave. Obviously, if you spend time in Italy and you spend time in England, those places shape people in deep ways, and that’s not a genetic difference between Italians and English people by and large—it is a difference in culture, norms, and what personality traits those cultures encourage. There’s a real pull to that view.
She now has two children, and when I was visiting her, she looked at me and at her kids and said: “Yascha, I was totally wrong. The difference between these kids—it’s just clearly like they come to the world with their own personalities and attitudes.”
I wonder whether there’s a slightly mistaken assumption in the background about what it would take to be responsible for our personalities—as if, on the one hand, if my personality is determined by my genes then I have no responsibility for it and it’s out of my hands, and on the other hand, there’s some idea of having totally chosen my own personality, and only then can I be responsible for it and be my own person. That background assumption may simply be wrong. We may just have to accept that if I like somebody as a friend, it’s because of the traits they have, and it’s perfectly appropriate to be grateful to them for their generosity, or to decide that I don’t want someone in my life because they’re mean and not a very good person. Whether or not behavior is in part genetically determined is, in that respect, kind of irrelevant. I have a friendship, a social relationship with the person in my life, and the question of how they came to be that person is not nearly as directly relevant to my gratitude, my anger, or my relationship with the people in my life as many people seem to assume.
Harden: What you’re getting at is just really core to the dilemma of being a person—and I write about this a lot in my new book. We go through our lives and we experience ourselves in relationship with other people. We feel certain things. We have resentments, gratitudes, affections. We’re drawn to certain things and repulsed by others. At the same time, we have this peculiarly human gift and burden of being able to reflect on all of that—and to reflect in this really peculiar way we call scientific, which is to measure it, to put numbers on it, to try to develop theories about what predicts it: in terms of culture, genetics, parental socialization, life events. Behavior genetics sits right at the intersection of our experience of ourselves as loving, behaving, feeling people, and our existence as self-conscious minds that turn ourselves into objects of scientific study.
I don’t think it’s really about someone having assumptions that are wrong versus right. Of course you’re going to come to learning about the science with priors, because you’ve been a human. You’ve had a sense of yourself, and you’ve necessarily had to develop a story about why you are the way you are. Science, on the one hand—and this is fading—has carried a kind of cultural authority. So people so often experience scientists saying “here’s one lens on behavior” as if it were “this is the only lens” or “this is the right lens.” If that conflicts with the narrative you’ve developed about your life, of course that’s going to bring up feelings.
I talk about this with my undergraduates all the time. I teach intro psych here at UT, and I say at the beginning: psychology is not going to be like your calculus class. A professor going into calculus can assume you have no prior knowledge of calculus. Whereas I’m going to be talking about attachment and personality and parental relationships—and you already have thoughts and feelings about all of that, because you’ve been a person. I’m not trying to say that your theory is necessarily wrong. I’m trying to add a layer of information on top of that to refine some of those priors. It’s necessarily a topic that people have feelings about, because the question of why you are the way you are is one you can’t get through life without having constructed a story about, one way or the other.
Mounk: How should we think about a few key questions? I had Emily Oster on the podcast perhaps about a year ago. One way in which she reads the influence of genetics—but also the influence of some basic socioeconomic facts about you that are not really under your control—is to say that parents hugely overestimate the impact of little decisions they make. Do I take the kid to this ballet class or not? Do I take them to this slightly marginally better school, even if one school is much further away? If I allow them to have 20 minutes more screen time, is that going to deplete their ability to pay attention forever? By and large, you would also say: you’re overthinking it. As long as you make sure they have enough food to eat, and they feel loved and safe, and they don’t experience some horrible form of trauma as a child, some mixture of genetics and the very basic facts about their environment is going to determine the vast majority of a life outcome. Torturing yourself over all of these marginal decisions is just a mistake. Would you broadly agree with that? Do you think that’s one of the upshots of this kind of research?
Harden: I broadly agree with that. A lot of things that especially middle-to-upper-middle-class mothers—these are the moms in my cohort—obsess about are just not going to make that much of a difference in terms of measurable child outcomes. They’re not going to meaningfully change your child’s IQ score, or likelihood of becoming depressed, or likelihood of graduating from college, based on whether they go to a Montessori preschool versus a Waldorf preschool.
I would refine that conclusion in two ways. One is that parenting is a relationship, and you do things in relationships not because they’re going to change the other person, but because they seem to be the best thing for how you’re relating to that person in the moment. It would be really strange if someone said it doesn’t matter how you treat your husband because you’re not going to change his IQ. That’s not the point of behaving toward your husband in a certain way—it’s to live out your values and to treat him in a way that he deserves and wants to be treated. A lot of parenting is about that.
I switched my oldest kid’s school this year from public school to private school, and I didn’t do that because I think going to private school is going to change his cognitive ability at 30. It was because he was unhappy at his public school and he’s happier at his new school. The reduction of parenting into can we optimize on these measurable variables—rather than asking what relationship you’re in with your kid, and what makes them happy this year, even if it doesn’t have a long-term effect—I think that can be missing from some of these conversations.
Mounk: You’re presenting that partially as a refinement—or perhaps a small objection—to Oster’s outlook. I would put it the other way around: if you’re obsessed with how every little thing you do is going to impact the IQ score at age 30, or the likelihood of becoming a successful trader at Goldman, or a world-class pianist, you’re not in fact focusing on the relationship in the moment. If you’re freed from that anxiety, and recognize that these choices you’re making are not going to make or break the kid or determine all of those things, it actually frees you up to think more about what your child needs, how to create a family culture where everybody’s thriving. You can actually focus more on that really important question.
Harden: I would agree with that reframing. I talk about this a lot in my second book: genetics can be both liberating and horrifying in equal measure. Thinking about how much your child’s life might be shaped by the role of the genetic dice does confront you with the limits of maternal control. I do not think of my kids as lumps of clay that I have unlimited control to mold, and that frees me a lot from the burden of feeling responsible for everything about them. But it also means confronting the fact that bad things might happen to them, that they might do bad things, and that I have limited control over how much I can prevent or shape that outcome.
Going back to our earlier point, I think that can be part of the implicit negative reaction to genetics: we love control, and thinking about this factor that’s not fully within our control can be, again, both liberating and horrifying in equal measure.
Mounk: Control is a burden, but it is also reassuring. In a very different context, I’ve made this point a few times about conspiracy theories. Conspiracy theories are scary because these evil people hold all the strings and are making the world go wrong in all kinds of ways—but it’s also comforting, because there’s a group of people who have control over the world, and if only we uncover them and replace them, everything is going to be fine. In a strange way, the idea that there is a locus of control somewhere always has this double-edged nature. As a parent, if you have control—well, that’s horrible, because everything depends on you. But it’s also very flattering; it’s a form of ego extension.
Harden: Yes. With the conspiracy theory, one factor has all of the control, and there is something comforting about the cognitive closure of being able to say: it’s this. Not that the world is a messy, complex place where everything is interacting in ways we don’t fully understand, and our ability to steer the ship is always half guesswork—but rather, if we can just get rid of the deep state, everything is going to be better. When you get into “it’s all genetics” or “it’s all culture” or “it’s all capitalism,” that can be comforting because one problem offers the possibility of one solution. Whereas what we’re saying here is that it’s nature and nurture and culture and economies in combination—and knowing where to pull on that sweater is hard. That’s a lot less comforting.
Mounk: Let’s go a little bit further down on this point of education. What do you think drives educational outcomes and success? How do you think that an unwillingness to take the genetic component in educational outcomes seriously has shaped and perhaps misshaped educational policy?
Harden: If we look at the United States in the last 120 years, education has gone up—more people are educated, more people can read, more people can do advanced math. That is not because genetics have changed. It is because we have invested in mass public education and continue to make it a cultural priority to educate every successive cohort of students. The most important thing about education level is that it’s a cultural practice that requires the commitment of adults to educate the next generation. I’m always amazed when I’m helping my kids with their math—my 13-year-old is learning algebra, something that took humanity tens of thousands of years to develop, and now we teach it in a year. That’s an amazing cultural accomplishment.
Within that system, within any cohort of kids moving through an educational system, it’s obvious that kids raised by high-income, more educated parents go further in school. Also, kids who happen to inherit a combination of genes that makes going through school easier for them also get further in school. Those associations—whether we’re looking at a correlation with a genetic indicator or a correlation with socioeconomic status—are of comparable magnitude.
This is not surprising to the average American. If you ask the average American to estimate how important genetics is for education, their estimate is roughly what the meta-analytic twin study estimate is. People have been in classrooms, they’ve been around kids, they’ve noticed their own kids. It is not saying anything revolutionary to observe that some kids have an easier time paying attention, remembering things, manipulating information, and learning new vocabulary words, and that our education system is cumulative and stratified such that those initial advantages compound over the course of education. Who is most likely to graduate from college? Who is most likely to take calculus in high school? Who is most likely to be in the advanced math class in seventh grade? It’s not just the richest kids—it’s also the kids who have a certain combination of genetic variants.
It’s not radical in that sense, but it is radical from the perspective of research on education, because almost none of it considers genetic differences between people.
Mounk: How would this change the consensus in that field among researchers who don’t look at genetics? I can come up with a couple of hypotheses. One is that it may imply the system is somewhat less unjust than it appears. Presumably there’s some genetic correlation between a parent’s academic achievement and a child’s academic achievement, if there’s a genetic component—not a perfect correlation, but some correlation. If you’re only looking at the extent to which socioeconomic or educational background shapes outcomes, you say: the children of college graduates go to college at much higher rates, because if you don’t have parents who went to college, you just never have a chance—you don’t know how to apply, you didn’t go to the right high school, you didn’t read books when you were six years old.
Presumably socioeconomic status does play some causal role. But once you put some kind of genetic factor into the analysis, and it turns out that some percentage of that difference is predicted by genes, that implies that a six-year-old who doesn’t come from a household with those socioeconomic achievements or that educational status actually has a much better chance of making their way than we expected. So interventions designed to make up for socioeconomic disadvantage may not make as much of a difference as we thought.
Harden: A couple of things there. Empirically, it’s kind of worked out the opposite: the research that has controlled for genetics has actually given us some of the strongest evidence of continuing socioeconomic disparities in education. I was just reading a new preprint yesterday where researchers were looking at children at the very top of the distribution of a polygenic score—and we can get into how this is constructed, but it’s a genetic marker indicating whether you have genetic variants statistically associated with going further in school. They were looking at the top end of this genetic distribution for kids from lower-income and higher-income families, and showing that even at the very top of this genetic distribution, there were still big differences in educational outcomes by family income that accrued over the course of education.
I mention that study because it’s a really common fear that if we start to take genetics into account, it’s going to make our evidence for disparities disappear—disparities that need to be addressed with social policy. But in actuality, the studies that have taken genetics into account have made a really strong case for the fact that there are still family economic disparities in educational opportunity in the United States. So empirically, that’s not what we’re seeing.
Mounk: To be clear, empirically, what you’re saying is that the impact of socioeconomic status doesn’t disappear. But it does become smaller when you take into account the genetics.
Harden: It doesn’t disappear. I don’t want to misrepresent the study off the top of my head, so I don’t know exactly how that coefficient changes. But without accounting for genetics, you’re always open to the objection: well, this looks like a socioeconomic effect, but it’s really just differences in kids’ ability.
Mounk: What really frustrates me in these discussions is this: what we actually want to do is give people opportunity. There’s an ideological point of view that says, I care about the outcome, I want everybody to have good opportunity—and then, in service of that, denies the scientific evidence, or feels uncomfortable even engaging with the academic evidence, because engaging with it might undermine the mechanism to which one is committed. But the point of this information is that it might show that the mechanism isn’t in fact going to work. If what is driving disparities were not socioeconomic status at all, then all of the interventions meant to compensate for socioeconomic status are not going to produce more opportunity.
As you’re pointing out, it seems unrealistic that socioeconomic status doesn’t play some independent causal role—and the study seems to confirm that it does. But there is this strange dynamic where people oversimplify their causal model because they care about the outcome, and in doing so actually obstruct their ability to influence that outcome in the way they hope.
Harden: One thing I sometimes say when talking to education researchers is: imagine that someone has snuck into every data set you have and deleted all the information about student socioeconomic status. You don’t know which kids come from poor families and you don’t know which kids come from rich families. You’re still tasked with figuring out which schools are doing the best job of improving student learning. How much harder would your job be?
Your job would be harder because the classic problem in education research is how to disentangle selection—what’s being driven by the student—from what schools and teachers are actually doing. One way to get at that problem is not simply comparing a kid from a rich family to a kid from a poor family without taking that difference into account. And that is the current situation that most education research is in with regards to genetics. If you’re not collecting that information, if you’re not considering it, you’re still trying to disentangle which schools are working, which curricula are working, which teacher practices are working, which policies make a difference. You still have to grapple with the problem of what schools are doing versus what students are selected into those schools—and you have a whole piece of information that’s just not there.
It would be unimaginable to conduct this research without family income, and yet we consider it a matter of course to conduct it without genetics. Going back to your point—do you want to know how the world works or not? If we think that information about what causes what in the actual world is helpful for designing policy, then why would you leave data on the table that could help you understand student developmental outcomes? That doesn’t make sense if you’re truly invested in the ends you claim to care about.
Mounk: There are areas where this dynamic plays out that have nothing to do with genetics. One example is the female wage gap, which is often presented as women earning less money for doing the exact same roles. That is not what the evidence shows—they might earn around 2% less for doing the actual same roles at the same level of seniority. The female wage gap is much larger than that in most countries, actually larger in continental Europe than in the United States. The reason has to do with other factors. If you become a lawyer, the key period in which it’s decided whether you make partner—and make enormous amounts of money—is in your 30s. In academia, it’s decided whether you make tenure, often also in your 30s. If you have children during those years, that reduces your likelihood of making partner or getting tenure, and puts you on a completely different earnings trajectory. The reason female lawyers make less money is not that female partners at the most prestigious law firms make less than male partners—it’s that fewer capable women end up as partners at those firms because of those circumstances.
That doesn’t make it just or fair, but it does mean that if you’re trying to address this problem with anti-discrimination laws, negotiation training, or cultural messaging, you’re not going to get to the root of it. The problem actually requires a more fundamental intervention—perhaps in academia, for instance, there shouldn’t be an up-or-out tenure track where if you make it there’s very little advancement further down the line, and if you don’t, you can never really break in. A more radical intervention may be required, but you can only understand the nature of the intervention once you understand the nature of the mechanism. Yet some people find it somehow counterproductive to complicate the picture, fearing it will make people less committed to addressing the problem. In the process, they obscure the problem in such a way that the measures taken to address it simply can’t work.
Harden: This dynamic you’re picking up on—where a mechanism, a causal story, or a proposed policy is offered as a solution to something people see as genuinely unjust—can become a stand-in for care about the problem, or the perceived injustice, in and of itself. When you say the causal model is different, or the mediating pathway is different, or that the policy didn’t work, it can be perceived as saying there’s no problem to be solved, or no injustice to be addressed. How to avoid our means being conflated with our ends is a really difficult science communication problem, and one where people end up talking past each other.
My friend Jennifer Doleac has done work on ban-the-box—the policy of banning employers from asking whether applicants have an incarceration history—which was a very popular measure. It turns out that it tends to increase job discrimination against the formerly incarcerated, and also against Black men generally, because in the absence of information people simply rely on stereotypes.
Mounk: If you have a racist employer who stereotypes Black men as likely to have been to jail, but you can tell them whether or not an applicant has a criminal record, they can at least reason: this applicant doesn’t have a criminal record, so I’m safe hiring them. If you remove that information—and I’m not approving of that reasoning, but it’s what you would expect given those incentives and background beliefs—then they’re much more likely to think: who knows whether this person has a criminal record, better hire someone else.
Harden: The pushback Jennifer Doleac has gotten is telling—she’s been accused of not caring about second chances for people with an incarceration history. Her response is: she does care, and that’s precisely why she wants to know what actually works to get them employed, not what we think works. I feel similarly about education. If we think a certain math curriculum is going to help people get to Algebra 2, and we care about that as a society, but then you look and see that one school is only performing better than another because it has a higher concentration of kids who’ve inherited genes that make them slightly better at manipulating nonverbal information—I want to know that.
Part of this is being attuned to opportunity costs: anything we do that doesn’t work, and anything that makes it slower to identify what doesn’t work, is not free. That’s time and opportunity that could be used more effectively. There’s a real pragmatic case to be made for incorporating genetics—not because it’s a silver bullet, but because it’s another variable. You wouldn’t leave a variable of comparable statistical power off the table if it weren’t genetics. So why not use this one too?
Mounk: We’ve covered a lot of ground, much of it terrain from your last book. In your new book—in case the last one wasn’t controversial enough—you go for the maximally difficult part of this discussion: how do we think about the problem of evil? How do we think about people who do really bad things, the genetic determinants of that, and how should it influence our social response? Perhaps again, let’s start with the empirical evidence. Is there empirical evidence to suggest that the propensity to engage in serious crime has powerful genetic determinants?
Harden: Again, I would say genetic influence rather than genetic determinants. Going back to the laws of behavior genetics—everything is heritable, which is the third law. The second law, I believe, is that siblings raised in the same family are not usually all that similar—that being raised in the same family doesn’t make you alike, and that most of our environmental influences are much more idiosyncratic and not structured by the family.
Mounk: Is the first law simply the counterpoint—that there is some variation in behavior that is not influenced by genetics, that genetics is not 100%?
Harden: Yes. What we see is that there are very few extremely rare genetic disorders—less than 0.1% of the population, usually running in a few families—where we see what we would call monogenic, single-gene disorders that cause people with them to be seriously antisocial, aggressive, and impulsive. I talk in the book about one study of a Dutch family where the men inherited an X-linked gene that affects an enzyme in the brain, and they were all seriously antisocial. They had committed arson, homicide, and rape; multiple members of the family were in prison. So from the monogenic side, we see evidence that rare genetic variants can really warp your ability to restrain aggressive and impulsive impulses.
Most people who are aggressive do not have a monogenic disorder, but we are increasingly identifying genes that—and again, these are not deterministic, they work in combination with the environment—do increase your risk of becoming addicted to alcohol or drugs, of struggling with serious impulsivity problems from childhood, of behaving in more aggressive ways. This middle ground, where there’s not a single gene but many genes that raise your probability, produces correlations roughly as strong as those we see with a childhood maltreatment history, and that raises real questions. My book is me grappling with that question: how do we make sense of the fact that we didn’t choose to have this body and brain, and yet when we’re adults and we hurt each other, we hold each other responsible for our behavior?
The other theme in the book is that I was raised as an evangelical Christian when I was a young woman, and I’m really interested in how older narratives from Christianity around the inheritance of evil or the inheritance of sin continue to shape our debates about the science. Genetics is a very young science, and I think we are interpreting much of it through the lens of these much older, more Christian stories.
Mounk: You can see again why there might be so much resistance to thinking about genetics as a determinant here. I had a great podcast episode with Abigail Marsh on psychopaths. The fact that you can often diagnose a psychopath very early—at seven or eight years old, you can relatively reliably tell that a child has psychopathic traits—means that for obvious reasons, particularly if you have a child who exhibits those behaviors, there’s a reason why you might not want to accept that your child has a very strong tendency toward that. Abigail Marsh argues that with the right interventions, you can at least moderate the adverse impact it’s likely to have on the life of the child and on the lives of the people around them. But it is very scary to contemplate.
There’s an important distinction we need to make, and perhaps should have addressed earlier. One thing people sometimes mix up is that genes can be hugely influential on our behavior, while at the same time there is going to be significant variation within families. Siblings only share 50% of their genetics, and especially when it comes to particular constellations of genes or recessive genes, it’s very possible that some members of a family have certain traits that others don’t. I wonder whether that’s part of what makes people so instinctively skeptical—whether we’re inclined to say: you’re telling me it’s all genetics, but I’m different from my parents, so it can’t be genetics.
Harden: We tend to think of genetics and heredity through much older ideas of inheritance. If you had a hereditary title, you passed it down to your oldest child intact, in full. If you inherited your parents’ estate, you got all their money. But you don’t get all your parents’ genes—you get a random 50% draw from your mother’s and your father’s. One way to think about it is: your mother has a pantry of ingredients that she’s used to make a recipe, and your father has a pantry of ingredients he’s used to make a recipe. You get 50% of the ingredients from each pantry. What recipe can you now make? It might be very different from either one.
That’s also why parents are always surprised by how different their second child is. They think: this is what happens when you combine my genes with my partner’s—and then they have their second child and it’s a completely different person. It’s a different combination. Getting out of the habit of thinking of genes as a source of continuity, and thinking of them also as a source of discontinuity, shuffling, and change from parent to child, is a really important shift in thinking. You are not your parents, and part of the reason you’re not your parents is that your genes are different from theirs. You’re not your parents’ clones.
After my first book came out—and I write about this a little in the new one—I had so many people reach out to me. Someone who was the oldest of eleven and felt so different from their siblings, the only one who went to college. Someone who was adopted and didn’t find out until young adulthood, and said it suddenly made so much more sense because they had always felt like an outlier. Someone whose brother was an addict who had never been able to stay clean, who had always worried about having children because of what they might carry. People have this intuitive sense that family history matters, but that it can also surprise you. I appreciate being able to describe how that’s actually happening scientifically—this reshuffling that occurs whenever we create a child.
In the rest of this conversation, Yascha and Paige discuss how society should respond to these questions, to what extent we have free will, and what genetic influence on our behavior tells us about humanity. This part of the conversation is reserved for paying subscribers…
