165 | Kathryn Paige Harden on Genetics, Luck, and Fairness

It’s pretty clear that our genes affect, though they don’t completely determine, who we grow up to be; children’s physical and mental characteristics are not completely unrelated to those of their parents. But this relationship has been widely abused throughout history to underwrite racist and sexist ideas. So there has been a counter-reaction in the direction of removing any consideration of genetic heritage from how we understand people. Kathryn Paige Harden argues in favor of a more nuanced view: DNA does matter, we can clearly measure some of its effects, and understanding those effects is a crucial tool in fighting discrimination and making the world a more equitable place.

Support Mindscape on Patreon.

Kathryn Paige Harden received her Ph.D. in psychology from the University of Virginia. She is currently a professor in the Department of Psychology at the University of Texas at Austin. She is the leader of the Developmental Behavior Genetics Lab and co-director of the Texas Twin Project. She was the recipient of the Award for Distinguished Scientific Early Career Contributions to Psychology from the American Psychological Association. Her new book is The Genetic Lottery: Why DNA Matters for Social Equality.

• Web site
• University of Texas web page
• Google Scholar publications
• Amazon author profile
• Wikipedia
• Twitter

[accordion clicktoclose=”true”][accordion-item tag=”p” state=closed title=”Click to Show Episode Transcript”]Click above to close.

0:00:00.9 Sean Carroll: Hello everyone, welcome to the Mindscape Podcast. I’m your host, Sean Carroll. In recent years, has been a lot of discussion and controversy about the idea of a scientific racism, the use of scientific data or techniques to purportedly justify racist policies or attitudes or thoughts about other kinds of human beings.

0:00:22.1 SC: Now, as soon as I say the phrase “scientific racism” and tell you what it is, hackles get raised, right? There’s gonna be some people listening who say, “It’s not racism, it’s just science, we’re just doing science, we’re classifying people. We’re purely understanding the world, why are you trying to ruin it by calling it racism?”

0:00:41.4 SC: Other people are going to say, “It’s not scientific. This is not… This is pseudo-science, this is just a perversion of science.” And on either side, people become very emotional very, very quickly. Even if you’re in the middle, you tend to get very wary about this kind of discourse just because other people are so passionate, and let’s be frank, it lowers your cognitive abilities when your emotional valence goes way up like that. So whatever feeling you have when I start talking about this issue, I’m doing that intentionally.

0:01:12.7 SC: Savour that feeling, get an idea of what that feeling is, because our guest today, Paige Harden, who’s a psychologist at the University of Texas, wants us to move beyond that feeling. She wants to be able to talk about issues like genes and DNA, and how they influence whether strongly or weakly, life outcomes. Your educational attainment, what kind of job you’re going to get, whether you’re gonna get a mental illness, become homeless, things like that.

0:01:40.5 SC: And the whole project of relating genetic information to later life outcomes is very fraught with the danger that it can be misused in racist or other discriminatory ways, and what Paige wants to do is say, “That’s no reason not to use it.” Right? There’s a sort of counter-reaction that says, “Therefore, we shouldn’t mention DNA or genes at all when we talk about human beings, we should treat every human being equally and not worry about their genetic heritage.”

0:02:09.5 SC: Paige’s argument is, that if we really want to make life better for people, if we really want to fight for social justice in effective ways, we need to use all of the information, all the knowledge, all the scientific insight we can get. And there’s no doubt from the data that there is a relationship between genes and outcomes, but what we can try to do is use information we get about that relationship to bring that kind of equality to human life that Elizabeth Anderson talked about in a podcast a while back.

0:02:41.7 SC: Not equality of opportunity, not equality of outcome, but equality of dignity, treating people in ways that lets them lead their lives in society with of equal amounts of dignity for every person, and that’s something where understanding how people are different, even genetically, is going to be important.

0:03:01.9 SC: She has a new book out called, The Genetic Lottery: Why DNA Matters For Social Equality. The idea of the lottery being that we don’t choose our DNA. Our DNA affects who we grow up to be, but we don’t get credit or blame for it, we shouldn’t anyway. So how do you live in a world where people are given unequal amounts of talents from the start?

0:03:24.1 SC: So I was really happy with how this podcast came out. I think the Paige does a very good job, both at explaining the science and in making the case that we have to take that science seriously. Even if it means that we have emotional reactions, let’s look beyond them. Let’s really think hard about it, let’s try to get at the truth. That’s something we can all get behind here at Mindscape. So, let’s go.

[music]

0:04:04.3 SC: Paige Harden, welcome to the Mindscape Podcast.

0:04:04.3 Paige Harden: Thank you for having me.

0:04:06.7 SC: You have written a book about an interesting combination of topics. You have genetics and biology in there, but also psychology, sociology, politics and even philosophy, political philosophy is in there. So there’s a lot of details, I wanna go through them, but just so we don’t miss the point, I thought it’d be best to start with a summary of the point.

0:04:30.8 SC: And so I have a quote here from your publisher’s website, I guess, Princeton University Press, and then I’ll just state it out loud and you can comment on what that means. It says here, “Harden shows why are refusal to recognise the power of DNA perpetuates the myth of meritocracy, and argues that we must acknowledge the role of genetic luck if we are ever to create a fair society.” Alright, that’s it. That’s big stuff. Is that accurate?

[chuckle]

0:04:57.5 PH: That is big stuff. You can leave it to publishers to put it in.

0:05:00.5 SC: I know, right.

0:05:03.0 PH: In very dramatic ways. But it’s not wrong. I do think that is the point of the book. So I’m trying to describe both the science of behavioural genetics, but also trying to think about, what does it mean, how can we make sense of it? Why are people uncomfortable with it? And why our discomfort with it can in fact actually get in our way, if we’re worried about more equal policies and interventions.

0:05:32.8 SC: Right, so I guess the background to this is that, on the one hand, there’s a science, I mean we’re figuring things out with biology and sociology, but on the other hand, there’s a history of misuse of these kinds of ideas. So there’s a tendency to either continue to misuse it, or to say, “We shouldn’t use it at all, because it would be misused.” And you’re trying to go against both of those tendencies.

0:05:58.7 PH: Yeah. Well, I guess I would slightly disagree with you, which is that I never feel like there is in reality just science. There’s always science in context, scientists are always people who are coming into the scientific enterprise with their own sets of preconceptions and ambitions and motivations, and our interests as scientists don’t come out of nowhere.

0:06:23.6 PH: I think that’s more or less important for different fields, but I think particularly when we’re talking about research that’s connecting DNA, genetic differences between people, to socially valued behaviours and psychological characteristics, so things like intelligence or things like how far someone goes in school, there is this history.

0:06:46.8 PH: Which I think most people are somewhat familiar with, of how that research has been appropriated and misappropriated historically, but I think even continuing today. And so, in many camps in psychology and in the social sciences, there’s been the pendulum swinging in the opposite direction, of saying, “This has been so historically intertwined with people who are trying to allege the inferiority or superiority of some people, that we really need to avoid this work entirely.”

0:07:24.4 PH: There’s a scholar that I quote in the book, who says, “There’s no way to study the genetics of something like intelligence without being racist or classist.” And I disagree with that perspective too, and so, in many ways, the book is trying to unravel or disengage the science from the politics in which it’s been entwined for such a long period of time.

0:07:49.5 SC: Yeah, no, I’m very much on board here, I think that you’re completely right in your amendment to the way that I said, that there’s just science. There’s never just science. Completely 100% agree, and that’s a very important point, especially in this context. But it’s also, there’s this human tendency to not want to balance two things, to say that, to just go to extremes and say one thing or the other. So if you’re talking about genetics at all, you’re a racist.

0:08:16.1 SC: And that’s a caricature, but there are people who are pretty close to that attitude, and we should be reaching out to them and saying, “No, look, here we can learn new things about the world and use them to make the world a better place. ‘Cause refusing to learn them is not gonna change the world.” Is that fair?

0:08:33.8 PH: I think that’s fair. I think for people, my personal strategy is when people really object to the idea of doing this type of work at all. And by “this work”, I mean connecting DNA differences between people, DNA sequence variation, to order ultimately social behaviours and social phenotype, social characteristics.

0:09:02.0 PH: Is to remember that often, those objections are coming from a very real place of their own lived experiences with racism and classism, with being familiar with the history of it. And so I think, particularly for people from marginalized communities, they’re coming into this work with a very different set of priors about the cost-benefit analysis.

0:09:24.4 PH: A major goal of the book then for me is to try to take those concerns seriously, but also articulate why I think the cost-benefit analysis is different than many people who fear this work might imagine it to be, from my perspective as a psychologist.

0:09:44.3 SC: So let’s actually start the substantive discussion with the sociology, politics side of things. What are the kinds of outcomes, what are the situations where equity and equality arise, or inequities and inequalities arise, that you care about? What are the sort of ultimate things we’re trying to connect to the underlying genetic component?

0:10:06.5 PH: That’s such a good question, ’cause I think we… “Inequality” is one of those words that can be used in so many different ways, and the differences in their implicit meaning can cause people to kind of talk past each other. At the simplest level, I’m just talking about, how do people’s lives end up differently, in goods that we care about?

0:10:33.2 PH: So these can be material goods, things like how much money you make, how much wealth you accrue over your lifetime. These can be what I would consider more psychological goods, like your opportunity to get an education, how much have you learned? They can be things like your subjective well-being, so if you rate, how satisfied are you with your life, or, “My life is the best possible life I can imagine for me,” is a common way that people get at, just subject… People’s one item rating of how well their lives are going.

0:11:07.5 PH: We can think about actual lifespan, like how long do you live? We can think about more psychiatric things, so depression, suicide, freedom from physical pain, freedom from psychiatric disorders. And I think what we’re seeing right now pretty uniquely is how tied up those different inequalities and outcomes are with one another.

0:11:37.1 PH: So if you look back in history, the wealthiest people, they maybe had better food, but they weren’t necessarily buffered from infectious diseases, they didn’t actually live that much longer than commoners did. But if you look now, even before the Coronavirus pandemic, you’re seeing that the wealthy Americans were also the most educated, also lived the longest, but also reported the greatest subjective well-being, the greatest freedom from pain. They even report that they like their weekends more.

[chuckle]

0:12:11.4 PH: So you’re having all of these different differences in how people’s live end up, that are really entangled with one another. And a major axis in which they all tend to separate is by education level, so the gap between people who have a college education versus don’t, in all of these different forms of inequality, is large and mostly getting larger, in the latter half of the 20th century and into the 21st.

0:12:39.4 SC: That’s a really interesting point because people talk about wealth inequality, it’s very easily measurable, and it’s growing along many measures, but the fact that wealth inequality also correlates with other kinds of inequality very strongly, so the effect is even bigger than you might guess, is one that I hadn’t really thought about before.

0:12:57.3 PH: Yeah, so it’s not just that people are having more money to spend on consumption, but they have longer to spend it, they’re living longer, and it seems to be translating into utility, into happiness in a different way. So you mentioned that the book touches on a lot of different subjects, and I think part of the reason why it does, why do I end up going into political philosophy, for instance, is that as a psychologist, I think about, “Okay, well, I can see these patterns of correlations in our data, but who are the people debating which of these we should take most seriously, what should be our currency of justice, which terms of inequality do we care about?” And that really seems to be the province of philosophers.

0:13:42.2 SC: Well, I think that you mentioned somewhere, either in the book or in a talk, the work of Elizabeth Anderson, one of our favorite philosophers, and she was a previous Mindscape guest, probably the leading person talking about equality these days. Most people enjoyed the podcast, etcetera, she was great, but there is just a slice of people who see the word “equality” in the title of a podcast episode and assume that you’re insisting on equality of outcomes for everybody, that everyone has the same wealth and the same education and everything.

0:14:15.1 PH: And even though no one is asserting that, or looking for it, certainly Elizabeth Anderson wasn’t. So there is some subtlety, like you already said, in what we mean by our goals, when it comes to justice and equality and fairness.

0:14:31.2 PH: Yeah. I loved that episode with Elizabeth Anderson.

0:14:32.8 SC: Thank you.

0:14:33.7 PH: I find her writing on this to be so lucid and persuasive, and she’s been really influential in my own thinking about this. Precisely because I think she is offering people a way out of a kind of tired old “equality of opportunity”, by which people often mean treating everyone exactly the same and whatever results we don’t have to care about.

0:14:59.9 PH: Or “equality of outcome”, which also people use as kinda this bugbear to be scared of. Do we mean like that we’re leveling people down to the same… “Equality”, meaning “identity of outcome and sameness”. Something that I found really interesting about Anderson’s work is she’s really thinking more about what, do we mean by “relational equality”, how do we relate to each other as equals?

0:15:30.1 PH: She has this really great paper that I might be slightly misrepresenting the title, but it’s called, Human Dignity As A Concept For The Economy. And she’s really talking about, well, in what ways do we treat people as equals in the sort of human respect sense, which is different from making sure that everyone has the exact same amount of money in their bank account, but is also different from, we’re gonna treat all people exactly the same and we don’t care about differences in their life outcomes that result from there.

0:16:06.8 PH: So an example that I use in the book is thinking about the American With Disabilities Act as an equal opportunity legislation, but what the Americans With Disabilities Act is doing is it’s not saying, “Well, everyone has the same staircase, and if there’s differences in your enjoyment of the building, like we don’t have to care about that.”

0:16:32.3 PH: It’s saying, “We recognise that there are human differences, and what we’re trying to equalise is people’s ability to participate in our common public space.” And I think that is something that’s really been lost in a lot of the conversations about educational inequality.

0:16:47.9 PH: When I think about equality vis-a-vis education, me personally, I’m less interested in equalising everyone’s chances of getting a PhD in Physics. I’m interested in how do we equalise people’s ability to participate politically and economically in a way that makes them feel respected as equals, regardless of their levels of higher education? And I think that’s something that America has really falling behind on.

0:17:14.3 SC: And something that you emphasise, and Anderson also emphasises, is the role of luck in all of these considerations. The word “meritocracy” appeared in that quote that I started with, and there is this sort of American myth, or maybe it’s a broader myth than that, but the idea that we work hard and we deserve what we get, etcetera, etcetera.

0:17:38.6 SC: But the reality is that in genetics or in other areas of life, luck has a lot to do with it. Maybe say a little bit about how you think about that aspect? Because morally, it’s a tricky thing. Do we take away from people who just get lucky? Or do you just live with that?

0:17:51.1 PH: Yeah, yeah, it’s such an interesting, an interesting rabbit hole to fall down. Even just considering the multiple meanings of the word “merit”, right? So I think one of the ways that we use it is in this very moral sense, what we deserve, the content on someone’s character, like a merit badge, like you’ve done something to earn it. And it’s an accolade that we give out as a function of deserved.

0:18:25.4 PH: And then there’s a completely instrumental definition of merit, I think, which is… This is something I’ve talked about with my father a lot, ’cause he used to be involved in hiring for FedEx, he’s a pilot. And there’s a number of things that pilots are hired on that are obviously unearned, right? Like you have to be not too short to fit into a cockpit, but not too tall to fit into a cockpit, and you need to have correctable vision to 20-20.

0:18:53.2 PH: These are morally arbitrary human functionings, and yet we do think that we should hire pilots on their merits, which is their ability to fly a plane. I think what gets so lost in our discourse about this is people kind of shuffling back and forth between these two definitions.

0:19:11.7 PH: There’s a really wonderful essay on merit and meritocracy by the philosopher, Amartya Sen, that I really appreciate, that kind of describes this distinction between merit as earned, versus merit as instrumental. But I think it infects so much of the discourse about genetics.

0:19:28.5 PH: So this is why in the book, I try to spend a lot of time talking about the difference between valuable, something inherently valuable about a person, versus genetics as associated with traits that are valued right now, given how we’ve constructed society, which I think matches on to these or two kind of working definitions that people have of merit in their heads in these conversations.

0:19:55.8 PH: I would say a similar conceptual confusion surrounds the phrase “equality of opportunity”, what exactly… “What is opportunity? What makes it equal?”, is also something that kind of gets fuzzy in these conversations.

0:20:10.3 SC: Well, with that groundwork laid, let’s talk about the DNA, let’s talk about our genomes here.

[chuckle]

0:20:16.0 SC: So we’ve been having fun with the philosophy, but let’s crunch some numbers and collect some data. So, what is it, for the people out there who are not experts, and I include myself there, what do we measure when we measure what is going on in a person’s genome? Are we looking… I know that the state of the art is rapidly advancing, more reply than I can keep up with. So are we looking at literally the strands of DNA and counting base pairs? Or are we dividing up into genes and looking at frequencies? What’s going on?

0:20:49.9 PH: I think what’s so interesting is that for most of the history of behavioural genetics, we weren’t measuring anything about DNA at all. The idea of connecting genetic differences between people to differences in how their lives turned out, predates anything about our knowledge of… It predates Watson and Crick’s, it predates the word “gene”. It certainly predates our ability to measure something specifically about people’s genes.

0:21:21.8 PH: And the opacity of that approach that we were, in the era of doing twin studies or adoption studies, making inferences about the fact that people’s DNA made a difference for their life outcomes without actually measuring anything about people’s DNA, contributed to that work being really controversial.

0:21:40.0 PH: There’s been a huge sea change with the ability to sequence. And by “sequence”, I mean read people’s DNA letters directly. So everyone’s genome is made up of a molecule for… What are we called? Base pairs, G, C, T and A. Amongst other things, people can differ in their sequence of DNA letters, so you might have a T in one spot, whereas I have a G in that spot.

0:22:16.4 PH: So most commonly now what people are measuring are exactly that, these one DNA letter differences between people in their DNA sequence, and those are called single nucleotide polymorphisms, and they’re commonly abbreviated “snips”. Because of the structure of the genome, chunks of our genome are sort of co-inherited with one another, which means that we measure a snip, but that is “tagging” a number of different variants that are likely to be co-inherited with that one variant that you’ve measured.

0:22:50.2 SC: So just because you’ve measured that a single base pair differs from one person to another, so number one, let’s just pause and reflect on the fact that it’s really amazing that you can measure that one base… [chuckle]

0:23:00.3 PH: It’s amazing, it’s so cool. I think about that all the time. I was actually just giving a lecture to a number of graduate students and PhD students in mostly economics and sociology, that are trying to learn about genetics. And I just wanted to pause right there and be like, “It is wild. It is completely wild.”

0:23:22.2 PH: And that we can do it, that we can do it cheaply. We genotype participants in our lab, and it costs us about $55 a person. And we can do it non-invasively, so you don’t need blood, you can use saliva or cheek swabs. I worked in animal labs when I was in college, sort of laboriously doing PCR with rat blood, and it’s wild to me how much genetic information we can get so cheaply and so easily now, compared to what it used to be. I’m glad that you paused there, ’cause it’s definitely important.

0:23:58.5 SC: I want people to be impressed by this, ’cause it’s been changing a lot. But what you just said is provocative, because you’re saying that we see that there is a difference, presumably of one base pair between this person’s DNA and this person’s DNA, but it is associated, presumably with extremely high probability, with changes elsewhere.

0:24:19.1 SC: So we’re not lining up the two DNA strands and measuring every single base pair on them, even at this advanced level. We’re doing something a bit more core-screen than that?

0:24:31.5 PH: Yes, yeah. So there are studies that are moving towards whole genome sequencing, which is a more fine-grain read of the entire, higher genome. But most commonly, currently, people are using what are called “SNP arrays”, which are these single DNA letter differences that we know are correlated with other variants that are mostly nearby on the genome, and that are reasonably common in the population. By “common”, it’s usually more than 1% of the population or more than 5% of the population.

0:25:07.7 PH: And within that population, we’re talking about people who share recent genetic ancestors. So if we’re thinking about the global pool of human genetic diversity, we’re missing a lot, ’cause we’re missing rare variation, and most of our studies are also missing variation. That is maybe uncommon or absent in European ancestry populations, but more common elsewhere.

0:25:31.5 PH: So we’re zooming in on a pretty narrow slice of that genetic diversity, and then we’re trying to measure it directly and then see if it’s essentially correlated, and we’ll get back to how do we get from correlation to cause maybe later in this podcast, correlated with things we’ve measured about people. And what we measure can be their height, or it could be their, how far they went in school. Or it could be their income, or it could be their glaucoma. Pick your phenotype, depending on your discipline, probably.

0:26:07.8 SC: We actually had Joe Henrich on the podcast…

0:26:08.5 PH: Oh yes, okay, great.

0:26:11.8 SC: Who makes a big deal about the psychology of weird populations. And so, you’re saying there’s a similar thing going on in genetics, where almost all of our information is about a tiny subset of human diversity?

0:26:21.9 PH: Oh, definitely. The genetics of weird population. There’s a sociologist turned metascist, Melinda Mills at Oxford, whose work is really excellent, and she’s published a couple of of metascience papers on this. And I’m forgetting the statistic, but it’s really a shocking amount of what we know about the genetics of behaviour comes from White people in the UK, Iceland, and White people in the US, so it’s three countries that contribute the predominance of our information here.

0:26:56.6 PH: So there’s real drawbacks to that, because in terms of all of science works on variation, and by limiting ourselves in terms of variation, we’re limiting ourselves to what we can find. And many of the problems that we see in social science of weird populations is also found in the genetics of weird populations.

0:27:17.1 SC: Well, that’s good, because one of the goals of Mindscape is to let young people who will eventually be graduate students know that there’s a lot of work remaining to be done. So it sounds like… [chuckle]

0:27:26.4 PH: Oh definitely, yes.

0:27:28.8 SC: Much of the world is remaining to be genetically understood in this way. Okay, so we can, as you said, get a correlation between these features of someone’s DNA, and what? What are we trying to correlate them with? You gave some lists, but in your work particularly.

0:27:46.0 PH: So in my work, we’ve worked primarily with things related to education, so how far you go in school, and then things related to what psychologists call “externalising”, what economists call “risk tolerance”, what epidemiologists might call “health risk behaviour”, which are things like ADHD, conduct disorder, risk for alcohol problems, opiate use, that sort of thing. So those are the two domains of GWAS work, Genome-Wide Association Study work, that my group has worked on. And there’s groups…

0:28:26.5 PH: It’s really actually an amazing field in terms of being dominated by this kind of international team science model. So there’s teams all over the world who are attacking various medical phenotypes, psychiatric phenotypes, and behavioral phenotypes.

0:28:41.5 SC: So it’s… Right, we already treading into murky waters here.

0:28:45.5 PH: Yeah, yes. 100%. [chuckle]

0:28:48.3 SC: It’s not just height or obesity, it’s behaviours. And that’s just a harder thing. So you already mentioned the sticky issue of there’s a correlation that’s easy enough to plot, but the causation is what we care about. What are the kinds of techniques you use to ask whether or not there is really a causal relationship between what you’re measuring in the DNA and someone’s educational attainment, for example?

0:29:10.5 PH: So I think there’s two major classes of problems in terms of, okay, so you’ve done a study, you’ve measured these snips in people. You’ve done a study a million people, you’ve correlated these snips with something you’ve measure about them. You’ve observed these correlations. What do they mean?

0:29:31.0 PH: So the first class of problems has to do with what I’ve already talked about, which is that you’re not actually measuring every single aspect of the genome, and the part of the genome that you’ve measured might be… It’s association might be driven by another genetic variant that’s just been co-inherited with it.

0:29:47.6 PH: That is a problem where people, you might… Basically, in the book, I say it’s like a badly drawn treasure map, you know that the X is in this jungle, but your aerial view of the jungle when you’re flying over it versus like now you’re actually walking through the jungle trying to find the exact treasure spot, are kind of two different things. So you kind of localise it maybe to a region of the genome, but you don’t know the variant, the causal variant.

0:30:15.7 PH: A lot of times people talk about these studies in terms of fine mapping studies, which are, “Okay, I think it’s in this area.” Now I’m gonna measure that part of the genome more closely, more reliably with more specificity, to try to figure out where in this area is driving the effect. So that’s a kind of first class of problems.

0:30:38.1 PH: The second class of problems, is that people’s genetics are correlated with their culture, ’cause people have sex with people who are close to them, not everyone gets to have sex with everyone else.

[chuckle]

0:30:53.6 PH: Repeat over generations, and you get a genome that is structured by a multigenerational history of our social rules about who has sex with who. And so you could see a correlation between a gene and an outcome that’s not because the gene is causing something in my biology that’s causing the outcome, but just because that gene happens to be more common in people from this culture or this particular part of the world, and they also differ in whatever I’m studying for entirely environmental or cultural reasons.

0:31:33.7 PH: Historically, people have tried to get at that problem by essentially trying to use information from across the entire genome to estimate what are called “principal components of ancestry”, which are basically statistical measures of how similar people are by virtue of sharing recent genetic ancestors, and controlling for those in genetic studies.

0:32:03.7 PH: So, instead of comparing people who are very diverse in terms of their genetic background, I’m trying to find people who are fairly homogenous in terms of their, who their recent genetic ancestors are. I’m trying to quantify their similarity based on what I’m measuring about their genome, and then I’m trying to statistically control for that.

0:32:24.4 PH: That’s partially successful, but not fully successful. The best strategy is to not try to compare unrelated people, but actually try to compare family members. So the title of the book is the Genetic Lottery, which is a metaphor I like for lots of reasons, but one of the reasons is when we’re thinking about the genetic differences between two siblings whose parents have the same genes, the genetic differences between them are kind of unbraided from that larger package of culture and environment and geography.

0:33:01.8 PH: Because which genes you happen to inherit from your parents is random, and so to really try to get at, “Is it a genetic cause?”, versus just an aspect of the genome that’s correlated with your environment, you need this kind of natural experiment of the fact that your parents could have given you either one of two copies of their genes and you happen to get one.. And it’s that randomness that gives you some causal purchase.

0:33:28.7 SC: I know that in social sciences and computer science, there’s been a greatly improved sophistication in how we think about causation, people like Judea Pearl, etcetera, working with Bayesian networks and massive probability distributions. Do you need that level of sophistication for what you’re doing here? Or is it just, we look at the controlled experiments we are given access to and work with what we have?

0:33:53.1 PH: I would say, there’s kind of no more controversial word maybe in social science genetics than “cause”, right? The C word, “cause”. “Cause” and “predict” are words that really get us into trouble. Which is… I spend time in the book, a whole chapter in fact, as you know, really defining what I mean by “cause”.

0:34:20.1 PH: So I think to make sense of this, it’s not so much about the sophistication of the analytic techniques, so much as being very clear about what is the model of what a cause is, that’s kind of going into this type of research. And also what does that not entail, what is a cause not, in this.

0:34:43.4 PH: In this case, the best thing that we have access to in humans is this kind of natural experiment of children being randomised to genotypes conditional in their parents genotypes, and so it fits really naturally into the framework of causation that’s arisen around a randomised control trial, which is really trying to peek at the counterfactual, “what would have happened if”. A cause is a difference maker, essentially.

0:35:18.6 PH: What’s improved about that is that it doesn’t necessarily mean that the mechanisms are biological, it doesn’t necessarily mean that the cause is deterministic. In social science, we think about chancy causes. Chancy causes is chancy difference makers all the time.

0:35:32.4 SC: All the time, sure.

0:35:35.2 PH: Like, “Does use of iPhones amongst 12-year-olds increase their risk for depression?”, does not mean that if you got your 12-year-old an iPhone, they would necessarily become depressed. I think the problem is that it’s difficult for us to take that kind of chancy, indeterministic, average difference maker type of framework for causes, when we were talking about genes. We tend to port in a bunch of other assumptions about what genetic causes are, relative to social science causes.

0:36:02.0 SC: Well, my wife, Jennifer Ouellette, actually she’s a science writer, and she wrote a book on the science of self. And so she looked a little bit into these questions, and I remember very vividly how, at least at the time, ’cause the state of the art is changing a lot, but there were very, very few individual genes that mapped cleanly onto an actual trait of a person. Right? Like earwax is one of them.

0:36:23.4 PH: Yes. And that’s still the case.

[chuckle]

0:36:25.4 SC: Yeah, okay, good. So it’s not like you’re looking for a base pair or a gene that makes you tall, or makes you live long, or makes you grumpy. Right? It’s a much more subtle kind of nuanced thing.

0:36:34.2 PH: There is no gene for everything that we are looking at here. And I think this is another thing that’s hard about thinking about genetic causes, is everything we’re talking about is polygenic, influenced by many, many, many, many, many genes, each of which have a small effect. So when we think about genes influencing something, I think many people think of like Mendel’s pea plants, like if you got this version, then you were a wrinkly pea, versus a smooth pea.

0:37:07.2 PH: Or like the early 2000s pop science where people talked about “the gauging”, which that didn’t turn out to be scientifically accurate at all. Instead, what we’re talking about is thousands or even hundreds of thousands of variants, each of which have minuscule probabilistic effects, but in aggregate, add up to something that starts to make a difference on a population level.

0:37:39.9 PH: And I think that that kind of thinking about vast duplicity of small chancy things, is different, I think, than how many people are originally taught about genetics.

0:37:52.3 SC: The other complication, I’m not sure how relevant it is here, but I know from previous podcasts and talking to friends that there’s not just the genome, there’s how it gets expressed. And that’s something that could be environmental as well as genetic. Although your parents, your mum in particular, do influence it. Is that something you can keep track of, or control for? Or is that just a noise in your data?

0:38:18.4 PH: So, people definitely do, and our lab also does this kind of work. DNA is relatively inert molecule. What I describe it as, you can have a cookbook that’s sitting on your shelf, but that does not mean you have dinner on the table. Something has to happen in order for there to be a product to be created. And that is very dynamic.

0:38:42.6 PH: And so there’s many different processes that people talk about in terms of getting from this genetic recipe, which is, we can talk about whether that metaphor is useful or not, to a protein dish, something that’s made. In our lab, we look at, for instance, DNA methylation, which is one of molecular biomarker that’s giving you some information about which parts of the genome are being expressed at a certain time in a certain tissue.

0:39:13.9 PH: What I think is amazing is that, essentially despite that, despite the fact that having a gene or genetic variant doesn’t mean that it’s being expressed in your body or in your brain, we nonetheless are seeing these associations between just the gene sequence variation in the thing that we’re measuring.

0:39:32.1 PH: So part of what makes that relationship between sequence variation and outcome probabilistic, is this kind of more like epigenetic and environmental interaction that’s happening, and yet on average, we still see that kind of genetic signal coming through despite all the complexity that’s layered on top of it by the other levels of our biological and social systems.

0:40:01.2 SC: Okay, so when we get into the details of making these, I almost wanna say predictions, but anyway, identifying tendencies or chances or prospects on the basis of…

[chuckle]

0:40:14.8 PH: Yeah, I will say when I was doing copy edits for my book, I went through control F to look for every instance of the word “predict”…

0:40:25.1 SC: Predict. [chuckle]

0:40:25.2 PH: And scrutinised it about whether or not there was a better word there. So you can call them predictions, but we should talk about what we mean by that. [chuckle]

0:40:30.7 SC: Well, yeah, good. And also, we should talk about how exactly they’re made, because I know that at some point, we talk about the idea of a polygenetic score, which is somehow taking this enormous amount of data in a DNA and making into one number and predicting things on the basis of that number. So what is that? Why is there one number? Why would we ever think that one number was good enough? And is it just like the first step toward a future where we’re much more multivariable?

0:41:02.3 PH: So, a polygenic score is, you’re right, it’s one number that aggregates our best guess of your likelihood of showing a phenotype, of showing a particular outcome, based entirely on information about your DNA sequence. And the way that it’s constructed is that researchers, and it might have been a different group of people or might have been me, who have done a large, what are called “discovery studies”, where you have maybe 50,000 or 100,000, or a million people, and they have estimated the correlation between all of these measured snip genetic variants and the outcome of interest.

0:41:48.8 PH: And now you have a huge dataset that has, every row is the genetic variant measured, and the column is the estimated correlation between that genetic variant and let’s say height in this example. And so I take those and I measure DNA and a new group of people, and I use the results of the previous study as a way to add up genetic information on this new group of people.

0:42:21.8 PH: So if they have inherited two copies of this particular snip from their parents, then it would be two times whatever that estimated correlation is, if they’ve gotten zero, it would be zero, and then I just literally sum that up over that [0:42:38.8] ____. So its incredibly coarse, right? It’s a huge biologically nonsensical grab-bag, if you think about it, right?

0:42:49.6 PH: So in the case of education, it could be genes that are correlated because of this uncontrolled population stratification. It could be genes that make you better at doing math, it could be genes that make you more of a morning person, it could be genes that changed your risk of going through puberty earlier, and we know that girls who go through puberty are discouraged from more difficult Math classes ’cause they feel weird and they get more attention from boys. It could be any number of processes that are all collapsed together into this one number.

0:43:27.1 PH: I think the question is like, “Well, why would you do that? Why would you make this kind of biological grab-bag?” One of the… And I do think that people will… One way that the science will be moving, will be trying to have less crude, gross kind of crunchy measures, [chuckle] compared to polygenic hard scores.

0:43:52.0 PH: One reason that you do that is that even though each of the individual snips that go into a polygenic score have these infinitesimally small correlations, their aggregation turns out to be as strongly correlated with some of the outcomes that we care about, as our other really gross crunchy variables, right?

0:44:12.2 PH: So if I measure a family’s socioeconomic status, it’s their education or their income, their occupational status. That’s also aggregating a huge number of processes that are differing between affluent children and poor children in America, a cacophony of different mechanisms. But it’s telling me something meaningful about differences in the population. And so I think of really polygenic scores as being another sort of clunky aggregate measure of our measures of SES. It’s telling us that that people differ, it’s allowing to quantify those differences, but it’s in buying some predictive power, you’re sacrificing this mechanistic specificity.

0:44:58.5 SC: Sure.

0:44:59.2 PH: And so I think the challenge then is to go back and trace out some more specific mechanisms, in many ways, following the arc of social science in terms of, we observed that poor children did worse in school long before we had really clear mechanistic stories about why.

0:45:16.7 SC: I think I called it a polygenetic score, but that was wrong. It was polygenic score.

0:45:23.6 PH: Polygenic. Which is, it’s a very common error. And I don’t know why when we were… When the field was coming up with this, we just dropped a syllable. It would make more sense for it to be a polygenetic score in place of polygenic score.

0:45:34.6 SC: Alright. [chuckle] You’re being very kind to me. Thank you for that. Thank you for assuaging my guilt. But just so I’m super clear, is the idea that for every outcome we’re interested, we develop a different polygenic score?

0:45:48.0 PH: Yes. So a polygenic score is estimated based on a set of GWAS results that have been conducted for different phenotypes. So if I calculated your polygenic score “for height”, what is my best guess of the genetic variants you have that are statistically correlated with being taller. That would be different than your polygenic score for educational attainment.

0:46:14.4 SC: Got it, good.

0:46:16.0 PH: At the same time, what we see is that you might have done a study of educational attainment, how many years have you gone through school, which is something that is determined in your teenage years or your 20s, or if you’ve gotten a few Cs, sometimes not until your 30s when you’re done with school.

0:46:37.2 PH: And that polygenic score “for educational attainment” is also associated with all the intermediate spots in that trajectory of education. So the educational attainment polygenic score isn’t just correlated with how far people go in school, but also their grades in high school, and whether or not their teacher thought they had attention problems in elementary school.

0:46:59.8 PH: And so you see… It’s different for different domains, but I think a mistake to think of a polygenic score as being too narrowly about the one thing that we search for, study in the original study.

0:47:12.8 SC: Okay. But just making sure, it’s not like you have N IQ, you have A polygenic score. You have different polygenic scores for different things?

[chuckle]

0:47:19.9 PH: Yeah, yeah.

0:47:21.8 SC: Too bad. We could rank people on their polygenic score, that would be the end of it, right? Then we would just know how worthwhile people were as human beings.

0:47:27.1 PH: At the same time, you do see… You see what are called “genetic correlations”, and they can sometimes be surprising and they give us clues. So if you do a genetic study of height and then you do a genetic study… Well, actually, I’ll give you a real example. If you do a genetic study of educational attainment, and then you do a genetic study of schizophrenia, you end up with some of the same genes. And some of them work in the same direction, and some are opposite. And that has actually turned out to be kind of a puzzle for researchers to figure out.

0:47:59.1 PH: I think in academia, we like our like our neat silos. Let the medical geneticists study medical phenotypes like lung cancer, and the psychologists study maybe nicotine rejection, and the economists study labour markets. But in real life, those things go together. We know your success in the labour market and your likelihood of smoking, and your likelihood of developing lung cancer.

0:48:29.3 PH: And so, as a result of life being messy, the genetics are messy too. You get some of the same genetic associations for things that people think of as being in very different camps, medical versus behavioral.

0:48:43.0 SC: Well, also my impression is that some of the things that you’re correlating with are things like the probability of becoming homeless. And if we say that there’s a genetic relationship… Sorry, let’s back up. The idea of a home wasn’t invented when the gene was invented.

[chuckle]

0:49:01.7 PH: Yeah, yeah.

0:49:02.2 SC: So if you say that there’s the genetic relationship between… Or relationship between your genes and let’s say the proclivity for mental illness, or the susceptibility to become a drug addict, you can sort of see the causal path in your mind. Whereas the causal path to homelessness is a bit removed. So how sneaky do you have to be when you even contemplate correlating DNA information with outcomes like homelessness or getting a PhD or something like that? I mean, there must be a million confounding variables in the way.

0:49:34.9 PH: Yes. Well, I use the example of homelessness in the book, really because I’m trying to give the vivid example of something that is obviously a social problem, but is responsive to local social policies. Like here in Austin, we’ve had this huge debate about criminalising camping for unhoused populations. And so there was very large homeless encampments under freeways for about two years, and then the repeal, the camping ban was unrepealed, and so now it’s criminalised again.

0:50:10.2 PH: So it’s clearly a social problem that we deal with with social policy. And within a society, not everyone is equally likely to become homeless. The sorts of vulnerabilities that set you up for risk, things like mental illness, things like doing poorly in school, ultimately, as I say in the book, being homeless is being unable to afford housing, it is about an intersection of how you’ve exchanged your skills for money in the labour market and the affordable housing in your society.

0:50:50.4 PH: So I use that example because I’m really trying to make a point about how our biology is affecting our embodied traits, and that might be a temperamental thing, or that might be our risk for serious mental illness. And then those embodied traits are refracted through this political economic social context, in ways that matter, in ways that we see, we see when we’re driving around.

0:51:23.6 PH: And both as a scientist, but also as someone who’s trying to make sense of my moral responsibilities in a complex world, I really want to think about that whole picture. Both, what can we understand about why some people are born with a higher risk of becoming schizophrenic than other people? Which I actually don’t think is a very controversial statement. And then, how does society act on those embodied differences in ways that create these forms of social inequality?

0:51:57.5 PH: So from a scientific perspective, it seems like such a strange thing, to, at first, maybe a counterintuitive thing, to connect DNA to income, income is clearly social. But if we observe genetic patterns that are correlated with income, what is that telling us about which embodied characteristics and which skills are being rewarded, and which aren’t, in the way that we’ve currently constructed the social system.

0:52:29.3 PH: I think when GWAS first started, maybe people thought we would have really nice, pretty biological approximate phenotypes to study, but it turns out that the things that we collect data on on a million people is how far they’ve gone to school and whether or not they own their home. And so we’re actually working backwards. We’ve jumped eight levels of analysis, and now how can we use those associations as a trail of bread crumbs to follow back to figure out, what are the intervening processes here?

0:53:05.0 SC: Can we give the listeners some intuition for this quantitative size of these effects? Like if you are able to map out someone’s genome very effectively, to what extent does that predict something like income or educational attainment? Is it a 1% effect? Is it almost all of it? How do you know what variables to use to quantify it?

0:53:30.6 PH: It’s such a good question, and I think part of the reason why it’s such a good question is because the tools that scientists use, at least social scientists use, to quantify effect size, most commonly something with an R squared, which is a percent variation accounted for. We often don’t have a good intuition for how the relationships that we see play out around us would be quantified on that kind of R squared metric.

0:53:57.8 PH: So just to put some more concrete numbers on it, if you look at Americans and you wanna say, “Okay, there’s all this variation in whether or not someone completes college. We know that affluent children are more likely to complete college than children raised in less affluent families. What is the percent variation accounted for in college completion rates by family income?”

0:54:26.5 PH: And zero would be, everyone has an equal chance of graduating from college regardless of their family income, and one would be, I can predict with absolute certainty your rate of college completion, your likelihood of college condition, by knowing how much money your parents made. So the best estimates for that in the US today are around 11 to 15%. Okay?

0:54:50.6 SC: Sorry. For what?

0:54:52.3 PH: For family income and rate of college completion. We look around and we say, the relationship we observe between being richer and being more likely to complete college is about, let’s round up, let’s say around 15% of the variation. So that’s 85% that’s not related to that. Which, we know rich kids who slack off from school, and we know kids who were raised in poor families who did really, really well.

0:55:23.8 PH: Any college professor can look at the vast… I teach at UT and my students come from vastly different economic circumstances, and I can see that students who were coming from poor families have more challenges, but I also know that it’s not destiny, it’s not deterministic. So that is about the same effect size that we see for a polygenic score in relation to completing college.

0:55:48.0 PH: So I know about as much about a person’s chance of graduating from college if I know their polygenic score, if they’re of European genetic ancestry, and so likely to identify as White, and that’s a very important caveat, as I would from knowing how much money their parents made in the years before they went off to school.

0:56:12.5 PH: So is that a big effect size or a small effect size? I think that depends on your, what are you trying to do? Am I trying to make forecasts about this person’s fate? Then it’s not very good. Am I trying to explain the broad dynamics of how people in American society differ in a really key outcome in their life? Then I do think that matters.

0:56:38.5 PH: And so it’s that kind of middle ground of, it’s neither… I don’t think it’s… I don’t think it’s genetic astrology, I don’t think it’s worthless, but I don’t think it’s deterministic either. It’s in between.

0:56:50.5 SC: Well, and the next obvious question then is, if you know both their polygenic score and their family’s income, does that make you predict things better? Or is it kind redundant information?

0:57:01.6 PH: Yeah, so it’s not redundant information, which, really common question is, “Why would we do genetics when we can do the environment?” I think that’s kind of a false “either/or”. This is not the layer of information about someone, this is not the most important layer of information about someone, but it’s an additional layer of information about someone, that’s telling us something that would be otherwise hard to quantify and see, and it’s giving us a new ability to capture variation in people’s lives.

0:57:40.9 PH: And we see that non-redundancy even if we’re looking at other levels of analysis. So for instance, I can say, “What do I know about you if I know you’re polygenic score above your socioeconomic status?” But I can also say, “Well, what’s the average SES in this school? What’s the concentration of affluent or poor children in this high school? What’s the average polygenic score in this school?” And those are correlated, but only at around 0.4, 0.5. Right? So even our sense of how students are clustered in different educational context, we get this kind of different piece of information about that too.

0:58:23.8 SC: Well, let me ask basically the same question, except instead of considering completely external factors, consider things we can measure about people other than their genetics. Right?

0:58:36.1 PH: Yeah, yeah.

0:58:37.2 SC: So the example I have in mind, I think this is in your book, again, I think you call it the “leaky genetic pipeline”, where if you know certain things about people’s polygenic scores, then you can predict whether or not they will keep taking Math classes later later in school. And that’s completely plausible to me.

0:58:54.1 SC: But also, I remember when I was in junior high school, I would have been able to predict pretty well which of my classmates would have gone on to take the higher Math classes and which ones wouldn’t have. So in that case, just from talking to people, getting an impression of what they’re like, thinking about their test scores and their grades and things like that, is the genetic information is still new, or is it redundant with that kind of thing?

0:59:19.5 PH: I think it’s still new, and I would say in three different respects. So the specific study that you’re talking about is when we are looking at a sample of American high school students, and we’re looking at which Math class where they tracked to in the 9th grade, and then how did they move through the Math curriculum over the course of high school.

0:59:46.2 PH: This was in the 1990s, when math was only compulsory for about two years in most US states. So you could take Algebra and then Geometry and then drop out, or you could take Geometry and Algebra 2 and Pre-Calculus and Calculus, and that’s actually the modal route taken by most Science PhDs, probably some people took Calculus in high school.

1:00:05.6 PH: What we saw is that the educational attainment polygenic score predicted both… And by “predict” I mean it captured non-negligible variation in which Math class people were assigned to in the 9th grade and their likelihood of dropping out, from year to year. What’s interesting about that analysis is the polygenic score predicted math drop out, even controlling for people’s grades in their previous Math class.

1:00:34.2 PH: So if you’re thinking about an observable characteristic that a school district would have or a local high school would have, you’re looking at kids who both had made B-pluses in their Geometry class, and who both have the same level of family SES, and the polygenic score is still predicting which one drops out of math versus not. So I think that speaks to the potential power of some genetic measures and some genetic context or some research context, having some extra bang for our buck. It’s giving us information that would be hard to see just from say, someone’s transcript.

1:01:12.6 PH: The other thing that’s interesting about the DNA measures is that they have two special characteristics that most psychological characteristics don’t have, like test scores or self-reported interest in taking math. And the first is that your DNA sequence doesn’t change. Its association what things might change, but it doesn’t change itself. It’s not reciprocally affected by the experiences that you’re going through in your life.

1:01:45.3 PH: So one metaphor my colleagues and I often use is, if you are going to a radiologist and you’re doing an imaging study, they give you a molecular tracer that is not metabolised the way that your body usually metabolises something if you drink it, so that you can see the structure, its inertness is what allows you to see the structure as it moves through, because it’s not being changed by the structure.

1:02:11.1 PH: And that is something that is very hard to come by with any of our normal psychological variables. My interest in math is affected by whether or not I had a shitty math teacher last year, but my DNA sequence is not. And the second characteristic that I have is that, and again this is a very important caveat, conditional on your parents’ genetics, your genetics is randomly assigned. Right?

1:02:38.1 PH: So conditional on your knowing your parents’ DNA, which gene that you have is random. And we have almost no variables like that in observational social science. So there’s nothing… There’s nothing about someone’s test scores or self-reported interest or motivation that I can say, “If I measure something about your parents, I can treat variation in this as reasonably exogenous to other things, as randomly assigned,” and that allows for different set of analyses and inferences about what’s causing what.

1:03:12.4 PH: I think it’s, just to repeat, that you’re getting observing predictive power over things you ordinarily measure, it’s inert temporally, and it’s randomly assigned, contingent on parents.

1:03:29.5 SC: Okay. Yeah, it’s extremely helpful. So what are we gonna do with it?

[laughter]

1:03:34.1 SC: Other than just judge people? And part of the reason why it’s a tricky question, right, is because people do sort of count things redundantly. Like if they think that a certain group is unlikely to be good at math, but one person within that group turns out to be really good at math, there are those who will still judge them negatively, like, “You can’t be good at math, you’re in that group.”

1:04:03.8 SC: So how can we flip that script a little bit and use this new information to help people, like you say, achieve what they’re capable of achieving and live the lives that we would like them to be able to live?

1:04:15.6 PH: You started off this interview by reading the very, very eloquent pithy publisher copy that Princeton University Press wrote for my book, and I think it’s a good thing that they wrote it. Because I think if I wrote it, it would probably be something like, “Paige Harden wants to make genetics boring.”

[laughter]

1:04:36.9 PH: And what I mean by that is, my goal would be for genetics to be, using polygenic scores or incorporating siblings or twin designs as part of your research, to become a really, really routine part of the everyday workings of developmental clinical and educational psychologists. I want it to be like propensity score matching or interminable variables analysis, like a routine technique that we teach to refine the rigour of someone’s inferences.

1:05:18.7 PH: I think if we did that, we could do a lot better more quickly, at identifying bright spots, identifying the features of people’s environments, particularly of children’s environments, that are most efficacious at actually producing the outcomes that as psychologist, we say we wanna produce, produce as outcomes in children.

1:05:49.3 PH: So currently, all social scientists are really faced with this problem of messy, free-range humans are hard to do experiments with, and everything is correlated with everything else, and we can look to our data and we can say, “Parents who eat dinner as a family by 05:30 PM have kids that do better in school.” But that doesn’t tell us that actually intervening on family dinner time would actually be the best, most cost-effective way of improving children’s academic performance.

1:06:27.6 PH: And that’s kind of a silly example, is family dinner, but that is actually the case with almost all the variables we study. And genetics gives us another way of seeing, how are people who are similar in this one-measure capacity, but who happen to find themselves in different environments, how do they differ? So that we can identify what are the most promising environmental levers for change. That’s what I want people to do with genetics.

1:07:02.1 PH: A lot of times, people are like, “Well, I’m not interested in gene,” and I’m like, “Yeah, but you’re interested in kids. And kids get their genes from the same people who give them their environments.” So if you’re interested in kids and you’re interested in figuring out which environments help kids succeed, you kind of have to be a little bit interested in genes, if only to get it out of the way from messing up what you’re trying to do.

1:07:24.9 SC: The obvious analogy, which maybe I got from your book and I’m now forgetting, but if there were a gene that said that you are much more susceptible to sudden heart attacks, but it’s preventable if you do the right thing, then of course, you’d want to know whether or not you are susceptible to that, so you could prevent it.

1:07:44.6 SC: And presumably, there are similar stories to be told about, “Oh, this student would benefit from this kind of educational environment or plan,” or something like that, that we can learn, ideally, the goal would be that we could learn from their genetic information.

1:08:01.1 PH: People often go, when they’re thinking about that…. Heterogeneity in people’s outcomes and matching people to interventions, it’s easy to go straight to where you went, which is kind of the more personalized medicine route or the more personalized education route, like can we use this to identify people who are very at risk for outcomes, for poor outcomes.

1:08:23.3 PH: I think there’s a more basic level even before we get there, which is that we know in almost all randomised control trials of psychological or educational interventions, that there are vast differences between people in how they respond to that intervention. Right? So if I… I used to be a practicing therapist, I’m not anymore, but if someone comes to me for therapy and I say, “Okay, you’re gonna do CBT for depression.” Some people respond to that and some people don’t.

1:08:59.4 PH: If you do a tutoring program that has a small average treatment effect for kids math skills, that small average treatment effect can mask enormous range in some kids benefiting hugely and some not at all. Often, what you see in educational interventions and psychological interventions is what we call a “Matthew effect”, in which, not only is there variation how people respond, but it’s the people who are least at risk that get the most help.

1:09:31.9 PH: It’s the students who are already doing well, who benefit the fastest. It’s the fact that rich kids learn more vocabulary from Sesame Street than poor kids do. So I think even if we don’t start matching kids to interventions, just knowing whether our current slate of interventions, who was being served by those. Are we helping people who are most at risk for poor outcomes, or are we helping the people who are, is it a “rich get richer” effect.

1:10:00.4 PH: Those sorts of heterogeneity studies can be really difficult to do, and genetics doesn’t solve all the problems, but by adding this ‘nother layer of information, particularly a layer information that’s again, invariant, your DNA sequence can’t be changed by the intervention, I think we can have a better tool for seeing who is being served by which of our policies and interventions, Just knowing that, I think would be useful information.

1:10:26.0 PH: If we could know, “This statin drug, it works on average, but it really doesn’t work for people who are most at genetic risk for heart attacks,” we would consider that a problem, we would wanna know that. I think it’s the same thing for, “Okay, well, this educational intervention works on average, but for the people who are most… From what we can see from their genetics are most at risk for bad outcomes, it’s not serving them,” why wouldn’t we wanna know that information?

1:11:00.2 SC: Can we bring this back to the notion of the lottery and luck? I mean, yes, our genes are in some sense random, so what does this teach us about the question of, “What do we do about that?” Or, how do we conceptualize that? You know what I mean? Does it change the way in which we assign merit or achievement to people?

1:11:22.5 PH: I think that’s a difficult question. Because….

1:11:25.7 SC: It is the title of your book. [chuckle]

1:11:25.8 PH: On one hand, I don’t think the science commits, commits anyone to a certain set of moral or political beliefs.

1:11:32.3 SC: Sure.

1:11:34.6 PH: I think people can take that information, the observation about the world, but people are born different in ways that matter for their lives, and run with that in lots of different directions. For me personally, it’s helped me clarify some of my intuitions about what makes a social structure or society good.

1:12:02.7 PH: When I think about a good society or a just society, I’m thinking about one that is more like a meadow, and less like a monoculture, one in which people who have genetic diversity can all have a place to thrive and participate, and not a place in which one genetically influenced set of skills and traits is favoured to the expense of everyone else.

1:12:37.0 PH: So it’s just thinking about the arbitrariness of what I pass on to my own kids, what kind of society do I wanna leave for them, in their difference, in their genetic difference, and how I think that’s really intuitive for parents. They might look at their own kids and they can see how they’re different from one another, and they want structures, their schools in their neighborhoods that accommodate those differences and allow all of their children to succeed.

1:13:08.4 PH: Thinking about social justice, I know that’s a heated word, “social justice”, from that perspective, that a just society is one, that kind of scales up that vision of accommodation of difference that I would want for my own family. That’s a big part of what I’m trying to articulate in this book.

1:13:27.6 SC: Yeah, no, I think it’s a very good vision. But we’re past the hour mark at the podcast now, so we’re allowed to let our hair down a little bit and ask the crazier questions.

[chuckle]

1:13:38.9 SC: We’ll go back to that very articulate thing you just said, but okay, if I think about other podcasts I’ve done, like with Fyodor Urnov, who is an expert on CRISPR and so forth, when are we gonna reach the point where we just identify the bad genes and fix them before they get propagated down to the next generation? And then everyone will be smart and beautiful.

[chuckle]

1:14:01.6 PH: Well, there’s so much in that question there. I think there’s a couple of things I wanna respond to, and first is back to this small effects size. Even leaving aside something as controversial as educational attainment or income, if you just wanted to CRISPR your baby to be taller, that would be… There’s not a gene that’s gonna do that for you.

1:14:29.2 PH: The stuff that’s being identified is, again, thousands or hundreds of thousands of genetic variants. And so no one is really talking about CRISPR in that context. They might be talking about egg selection for egg donation or embryo selection, where you’re selecting a polygenic score, but that’s a different context in CRISPR, which is I think really more about these more monogenic, single locus variations of large effect. At least to my understanding of it. I don’t think anyone’s really proposing CRISPRing 100,000 variants in your genome at this point in time. [chuckle]

1:15:08.1 PH: The second thing is this idea of “good gene and bad gene”. We have socially valued traits and socially dis-valued traits, and the genetics don’t really conform all the time to our really, our intuitions about that. So going back to that example of education and schizophrenia, if you have a genetic variant that makes it more likely for you to get a STEM PhD, but also more likely to become schizophrenic, is that a good gene or a bad gene? I don’t think we have a sense of that.

1:15:41.8 PH: The classic example is, if you have one copy of this gene, you’re more resistant to malaria. If you have two, you have sickle cell disease. Is that a good gene or bad gene? I think that there’s a lot of different ways that we can define eugenics, but eugenics literally means “good genes”, it’s projecting our social values down into the biology, which I think, our biology tends to confound those kind of neat distinctions that we have.

1:16:13.8 SC: Sure. [chuckle]

1:16:13.9 PH: So those are the major things that I would respond to about that. And then going back to what I said earlier about, I want a society that’s a meadow, not a monoculture. When I think about this idea of everyone being smart and beautiful, smart in the way that, physicist PhDs are smart. I actually think that would be a really, really boring culture, if everyone was smart in that particular way.

1:16:42.9 SC: Oh my God. I would vote against that, don’t worry.

[laughter]

1:16:46.3 PH: Beautiful in the sense that runway models are beautiful. If we think about our revealed preferences, where do people often wanna live, where are vibrant communities, there is not communities in which everyone is kind of narrowly phenotypically the same.

1:17:06.3 SC: Sure. I guess the reason why I ask, mostly in jest. I don’t think that… I’m not advocating for making sure everyone is smart and beautiful. But I kind of think, because it is very complicated, whenever you talk to the biologist about this, they will instantly say, “Yeah, look, it’s much more complicated than you think. You’re not gonna be able to go in there and tinker with DNA and get the babies you want, it just doesn’t work that way.”

1:17:25.7 SC: But I also think people are gonna try. Sure, people don’t wanna live in a monoculture, but if you ask them, do they want their babies to be smart and beautiful, they’re gonna say yes. [chuckle] I don’t know what to do about that. I don’t know what the social policy should be. And I think that the scientific impulse to say, “Look, it’s much more complicated than you think,” is gonna run up into the untrained impulse to say, “I’m gonna do it anyway.” We already see that happening.

1:18:04.2 PH: Yeah, I agree with you about that. I think we’ve already seen evidence of that in direct-to-consumer genetic testing companies where it’s, “We’re gonna match you to the wine that you’re gonna like the most, or your Spotify playlist based on your DNA,” and that’s obviously scientific, but that doesn’t mean that there aren’t consumers for it.

1:18:22.9 PH: There’s a couple of different conflicting intuitions here. One thing that I wanna make salient, something that’s often lost in these conversations, is that… For me, personally, I value women having autonomy over their reproduction choices, even reproductive choices that I don’t agree with for personal reasons, and I think that that value needs to be salient in all of our conversations about the uses of reproductive technology.

1:19:00.5 PH: Not losing sight of the fact that there are women who are making choices about their bodies and the babies that they bring into the world, and that needs to be… I just wanna center that as a consideration. It’s also interesting for me ’cause I live in a fairly conservative part of the world, and my own experiences being pregnant have been that even routine genetic testing that’s not controversial amongst the scientific community is approached with great great delicacy, because many women where I live in Texas don’t agree with the idea of any sort of prenatal genetic testing.

1:19:39.4 PH: When I was pregnant with my first child, I remember my OB was suggesting that I do like the standard 20-week scan to see if there’s any sort of fetal abnormalities, and then she was like, “And then if there is, we could get a genetic test based on this,” and she said it so delicately as if I was gonna be offended by the idea of the genetic testing.

1:20:04.9 PH: And so I think oftentimes our conversations around people adopting embryo selection seem to me a little bit divorced from the other aspects of how reproductive politics play out in this country, in which there are a lot of women who are really, really skeptical of anything that smacks of that. So, that’s not so much an answer, it’s just like a response of two factors that I think are often lost in conversations around this topic.

1:20:38.2 SC: No, I think I’m very glad you said those things, because even though I am trying to be provocative about the Gattaca future that we’re walking into, like you say, there are much more direct and immediate issues that we have to worry about, and we can get distracted by some of these other shiny things to worry about. I think the important… I wanna worry on all life scales.

1:21:00.1 PH: I think it’s a good question. I don’t think it’s being distracted about Gattaca future. My ultimate goal would be to try to empower as many women as possible with as much accurate science about what the genetics is and isn’t saying about what they could do with these reproductive choices, but also honour their autonomy in making those choices. That would be my personal sort of broad brush strokes approach to thinking about this problem.

1:21:29.8 SC: And I can’t think of any other better place to end, than that. That’s an extremely admirable goal. So Paige Harden, thanks so much for being on the Mindscape Podcast.

1:21:37.0 PH: Thank you so much for having me. This is a great conversation.

[music][/accordion-item][/accordion]