Behavioral Genetics and the Genetic Lottery
According to the behavioral geneticist Kathryn Paige Harden, we don’t really deserve where we end up in life. In her book The Genetic Lottery: Why DNA Matters for Social Equality, she argues that since we don’t really deserve what happens to us in life, we have a moral duty to help raise the boats of the less fortunate. We don’t choose the environment we’re raised in. But every human is also the outcome of a genetic lottery too: the possible combinations of your mother and father’s genomes are 70-trillion, which makes you a one-in-70-trillion event. Each of your parents was likewise a one-in-70-trillion event—and this chain of dice rolls extends back into your lineage and throughout human history. In a real sense, then, the fact that anyone exists as they do in their idiosyncratic form is incredibly improbable (at least if you ignore the possibility that we exist in an infinite multiverse where the initial conditions of individual universes can and do vary in every possible way, in which case, strangely, your particular existence is not so surprising).
One thing atypical about Harden’s book is that it incorporates some apposite philosophical reflection that has relevance to the findings of behavioral genetics. In the book, she draws from the philosophy of John Rawls, for instance, to help support her argument that we have a duty to the less fortunate. She also discusses the relevance of counterfactual thinking to understanding the sense in which genetic differences can be said to cause psychological and behavioral outcomes.
In the book, she takes aim at eugenicists who, in the past, have tried to justify social hierarchies on the presumption that those hierarchies were simply a matter of natural sorting. But she also takes aim at people who argue that we should ignore the influence of genetics on psychological traits and social outcomes. In her view, this is wrongheaded for a few reasons. For one thing, the effects that genes have on lives will exist whether or not we choose to ignore those effects. And according to Harden, if we do ignore the influence of genes, then we’ll assume that the social outcomes we observe are fair and deserving or merit, rather than partly the outcome of the unfair genetic lottery. Another reason why Harden thinks ignoring heritability is a bad idea is because social scientists don’t really know how to effectively intervene in order to improve people’s outcomes. She discusses how all of the large reviews of intervention attempts in education and beyond have found tiny to no effects:
“The conclusion either that most interventions don’t work, or that no one has ever even studied whether they work, extends beyond academic performance. The developmental psychologist Larry Steinberg reviewed the effects of school-based intervention programs designed to reduce teenagers’ alcohol and drug use, condomless sex, and other behavioral risks. An estimated 90 percent of American adolescents have been forced to sit through at least one such program. Steinberg concluded: ‘Even the best programs are successful mainly at changing adolescents’ knowledge but not in altering their behavior.” He went on to note that failure isn’t free: ‘Most taxpayers would be surprised—and rightly angry—to learn that vast expenditures of their dollars are invested in … programs that either do not work … or are, at best, of unproven or unstudied effectiveness.’ (pp. 196-197)
It’s only by accounting for the role of genes that we can disentangle them from all of the other possible effects on people’s outcomes. Because social science almost universally doesn’t bother to control for the possible effects of heritability when it studies social phenomena, every correlation between two variables that it finds is always potentially confounded by genetics—something that’s sometimes called the “sociologist’s fallacy”. In other words, if, for instance, you find, statistically speaking, that parents that read a lot to their children have children that get better grades in school, it could be that the parents’ reading to their children caused them to get better grades. But it could also be that parents have genes that dispose them to read a lot to their children and those same genes cause better scholastic performance too; and if those genes are passed on to their children, it would explain the connection between the parents reading to their children and their children’s better grades. But until you check to see whether genes are actually causing the association between the two, you simply cannot know whether the parents reading to their kids exerts an effect on their children’s school grades.
Another problem with failing to account for the possible confounding effects of heritability is that it makes it much easier to simply dismiss the findings of the social sciences. At any rate, in contrast to the academics who would like to ignore the force of heritability, a study actually found that, among the general public, their estimates of the heritability of traits like depression, personality, and eye color are quite similar to what behavioral geneticists have actually found—and this is especially true of mothers with more than one child.
As anyone familiar with the study designs of behavioral genetics knows, behavioral geneticists can examine the effect of genes by comparing different types of siblings raised in the same home, such as identical and fraternal twins, biological siblings, and adoptive siblings—and they can also compare twins raised apart. (Moreover, people’s psychological traits and social outcomes can also be compared with their biological or adoptive parents.) For example, biological siblings share about half their genes on average (sometimes more, sometimes less), whereas identical twins share all their genes in common (barring any new mutations in either twin).
A measure of heritability for some trait or outcome is a measure of how much variation in that trait or outcome is due genetic differences between people. A heritability measure is also specific to a given population at a given time and place. So, in principle, if the heritability of educational attainment were, say, 45% in Norway in 2024, it doesn’t necessarily mean that it was 45% heritable in Norway in 1935, or that it will necessarily be 45% heritable there in 2055. Nor does it necessarily mean that educational attainment is 45% heritable anywhere else. If you want to know the precise heritability of something in a certain population and at a certain point in history, you must measure it.
On the other hand, critics such as Richard Lewontin say that heritability is a useless measure since, strictly speaking, a measure of heritability for a trait, in a population at a certain time, cannot necessarily be imputed to a different population or to another time. But as Harden correctly notes, this is an absurd standard to use, since there are many such measures that are, strictly speaking, spatiotemporally limited, but that are nonetheless interesting and valuable. Harden uses the example of the Gini coefficient, which is used to measure income and wealth inequality within a nation. A specific Gini coefficient—say, of South Korea—is limited to South Korea. But that of course doesn’t render it uninteresting or useless. Likewise, heritability does in fact tell us how much variation between people in a specific trait or outcome is caused by genetic variation in a specific population at a specific time.
Harden also explains what behavioral geneticists mean when they talk about genes causing traits or outcomes. First, as I already mentioned, a measure of heritability is a measure of how much variation in something is caused by genetic variation between people. Genetic differences between people are differences that make a difference to their traits and outcomes. But having certain genes is not a sufficient condition for ending up with a certain trait or outcome. Heritability tells us how much variation in a trait or outcome, within a population, is caused by genes, but it typically can’t tell us how much of this or that specific person’s traits or outcomes are caused by genes. In any specific person, their genes are probabilistic causes of complex traits and outcomes, not destiny. A massive meta-analysis of decades of heritability studies, encompassing 17 804 traits and 2 748 studies of millions of pairs of twins, found that the average heritability of all traits is about 50% (some traits have higher heritability than this, while some traits have lower heritability). Siblings vary in how similar they are depending on how genetically similar they are. For instance, since identical (monozygotic) twins reared in the same home share all of their genes in common as well as the same home environment and much of their environment outside the home, then, if they receive the same parenting, any differences between them are a result of environmental differences that they experience outside of the home or developmental noise—or just plain luck and randomness (or measurement error). (There’s a lot more to be said about the traditional methods and study designs of behavioral genetics, including the “equal environments assumption”, and the effects of randomness. In the latter case, the neurogeneticist Kevin Mitchell has written extensively on the random effects of genes on brain development.) With the rise of cheap and powerful genomic sequencing, behavioral geneticists have now turned to actually looking for the specific genetic variants—in the genome, at the molecular level—that underlie the heritability of traits and outcomes.
Harden’s discussion of genome-wide association studies (GWAS) and polygenic scores is also illuminating for anyone who hasn’t followed their development (she uses the term “polygenic indices”, but I’ll use the term polygenic score). As Harden points out, Eric Turkheimer once was skeptical of what polygenic scores could tell us about complex traits and outcomes like general intelligence and educational attainment. And while it’s common to hear, for instance, that family income predicts educational outcomes, polygenic scores are already as predictive of educational outcomes, if not more so, than family income—polygenic scores can already explain at least 10% of the variation in an outcome like educational attainment.
A polygenic score is a number based on past research that has found genetic variants—that is to say, DNA differences between people—that are associated with a certain outcome, like wealth. For example, if someone has a lot of genetic variants associated with being taller, the more likely they are to be taller than average—and thus their polygenic score will be higher than average. Each genetic variant associated with a trait or outcome also affects a trait to varying degrees; if one genetic variant probabilistically affects a trait or outcome more heavily than another genetic variant, then that genetic variant will add more to a polygenic score. Another important finding of genome-wide association studies is that complex traits and outcomes like intelligence and health are affected by a huge number of genes—hundreds of thousands, perhaps millions—with each genetic variant making a very tiny difference to a trait or outcome. Nonetheless, all of those tiny differences add up, and people differ in the genetic variants they carry. So many small genetic differences between people can produce differences between them, large and small.
Harden discusses a study of retired White Americans aged 65 to 75 that looked at how much genetic differences between them could explain their differences in wealth. The researchers used polygenic scores associated with the length of time spent in school (which is a proxy for educational attainment). They found that those who had the lowest polygenic score for schooling length were, on average, about half-a-million-dollars less wealthier than those who had the highest polygenic score for schooling length (again, keep in mind that these were all retired White Americans between the ages of 65 and 75). However, people with a high polygenic score for schooling length don’t always have more years of schooling. As Harden points out, even among people with the same length of schooling, if they happen to have a polygenic score that’s one standard-deviation above the average, they still statistically tend to have 8% more wealth. So, even comparing apples with apples—retired White people between the ages of 65 and 75 with the same length of schooling—different polygenic scores still predict differences in wealth.
Of course, there are non-genetic differences between retired White Americans aged 65 to 75 that might explain at least some of the wealth differences between them. To this end, Harden discusses a study by Dan Belsky and his colleagues that, among other things, looked at about 2-thousand sibling pairs. What they found was that the sibling with a higher polygenic score for schooling length was also wealthier than their other sibling at retirement.
So, we clearly know that heritable differences between people can deeply affect who they are and what happens in their life. But it doesn’t necessarily follow that we should do nothing to counteract this fact. (Perhaps you think we should do nothing, or perhaps you think that we should.) In Harden’s words, “both things can be true at the same time: genetics can be causes of stratification in society, and measures to address systematic social forces can be effective at enacting social change”. Harden discusses some studies that have looked at how environmental interventions can help buffer the adverse effects that genetic differences can have. For example, a randomized, controlled study looked at an intervention program that (among other things) instructs parents on how to track who their teen’s friends are and how to enforce certain rules, such as curfews: one set of parents received the training, while another set did not. The researchers found that for the parents who were given the intervention instructions, their teens, on average, experienced less drinking and alcohol problems compared to the teens whose parents did not receive the intervention instructions.
A second study looked at the same teens involved in the first study but this time looked at their polygenic scores for alcohol problems. As expected, teens with higher polygenic scores for alcohol problems were in fact more likely to actually have alcohol problems. However, for the teens whose parents received the intervention instructions, there was no correlation between the teens’ polygenic score for alcohol problems and actually having alcohol problems. In other words, the intervention program actually worked to nullify any effect that genetic differences had on the propensity to have alcohol problems. On the other hand, whether the teens with higher polygenic scores for alcohol problems will end up having alcohol problems when they’re older, long after this intervention, is another question. It is worth bearing this possibility in mind, because an unfortunate finding of many studies that have tracked the efficacy of intervention programs, even massive ones like Head Start in the US, is that intervention effects, where they exist, often disappear after an intervention program has ended or as children and youth get older.
On another note, Harden points out that, in places like the US, having a college degree increasingly predicts having better health, happiness, wealth, and income, among other outcomes. And these differences between people with and without a college degree continue to widen. And as we know, differences in educational attainment are partly explained by genetic differences between people, which means that some of these disparities in health, happiness, wealth, and income are thus caused by those same genetic differences. (In other words, genetic differences make some people more likely to get a college degree, which in turn leads to better life outcomes.) Whether a society accepts disparities in life outcomes like these or does something to ameliorate them is, of course, a political question.