Heritability as Stalking Horse for Mutability

something's missing, even if it's not heritability

note: this is a Trojan horse, not a stalking horse, because there’s no stalking horse meme generators

CORRECTION: I goofed! This post attributes a post written by Sasha Gusev to Eric Hoel. Hoel shared the piece on Substack Notes and Gusev’s name does not appear on the piece, and I was confused. (Also I haven't slept more than three or four hours in a night since late March.) It's my bad and I regret the error: apologies to both. As Hoel has explicitly endorsed the piece, all of the arguments here stand.

For about a decade now I’ve followed the research about the negative effects of preterm birth on later academic achievement. I’m interested in part because I’ve wondered why, exactly, there doesn’t appear to be much public knowledge about the topic. It would appear to be the kind of phenomenon that prospective parents obsess over, and yet most people seem to have no idea that the association exists.

The effect has been replicated in dozens of papers, which have been combined in several high-quality meta-analyses, and in the context of educational research where effect sizes tend to be quite small the negative impact of prematurity is very large. Most of the research specifically concerns severe prematurity, so this next part is a little less certain, but the effects do indeed scale with the severity of prematurity at the extremes of preterm birth and the best evidence is that it has some impact on babies who are much less premature but still premature. Intuitively, that is to say, you’d expect the effect to scale even with prematurity that does not reach the medical threshold of severity. (Of course, the “correct” amount of gestational time varies naturally and babies born a little early would have small effects that could be easily canceled out by other factors that influence cognition.) In any event, we know that severely premature babies tend (tend!) to grow into children that face serious academic hurdles, and the degree of this impairment is as large or larger than those thought to be the result of many behaviors we discourage in pregnant women.

What kind of effect sizes are we seeing here? With the inevitable caveat that there’s lots of individual variability (plenty of premature babies go on to be high-achieving children), and bearing in mind that different studies look at different DVs, the most prominent meta-analyses report negative academic effects of severe prematurity that are comparable to those of children born with fetal alcohol syndrome. That is to say, having a baby who’s severely preterm tends to result in academic disadvantages that are on the order of those whose mothers drank alcohol heavily throughout pregnancy. These are not small effects.

All of this seems like it should be widely-discussed public knowledge. I have an eight-month-old baby, and the relentless tide of warnings and advice from pregnancy is still fresh in my mind. Every possible factor that could lead to bad outcomes for our fetus was thrust into our face at every turn. Alcohol, smoking, drugs, of course, no no no. But also caffeine, deli meats, heavy lifting, seemingly every medicine known to man, taking a hot bath, cleaning the litter box, X-rays, too much sitting, too much standing, cleaning products, bean sprouts…. A pregnant woman knows all about these prohibitions. You can’t help but absorb them during pregnancy; the world will bludgeon you with them whether you asked for anybody’s input or not. Emily Oster has built a career out of pointing out how the evidence for some of these harms is very thin, not that that changes most people’s minds. To have a baby on the way is to live in constant and totalizing fear of how the fetus might be harmed. And yet, on the lasting cognitive effects of prematurity, I heard nothing. We jokingly call it our “long pregnancy” the way people speak of “the long 19th century” or whatever, because we were steeped in this stuff for the three years it took us to conceive. And I literally did not encounter a single popular source that referred to this major effect that has been confirmed again and again in high-quality research.

I thought about all of this when I saw that

Erik Hoel

has declared the “missing heritability” question over. “Missing heritability” refers to the gap between how much we think genes should influence quantitative traits based on family and twin studies and how little of that genetic influence we can actually pinpoint in specific genes using modern DNA methods like genome-wide association studies. Twin studies tend to suggest that examined traits are around twice as heritable as we can find with current statistical association studies.¹ Hoel says that the very idea of “missing heritability” has been more or less disproven, dealt with, disposed of. I don’t trust myself to accurately summarize his argument, so please do read it yourself. If he’s right, well, there’s no more heritability to find! The twin study estimates were always seriously inflated, in this view. And the people who thought they were accurate, well, let’s just try not to think about their dark motivations, wouldn’t want to speculate, Hoel isn’t saying they’re all motivated by pure racism…. Plus ça change.

Personally, I suspect that we’re probably still in the relative Dark Ages with this stuff, and who knows what we’ll learn in the future. Certainly, the experience of going from Human Genome Project maximalism to understanding the immense polygenicity of human outcomes should remind us that there’s always so much more complication than we can imagine. The future is long. As I’ve said before, the day will likely come when we’ll be able to genetically engineer embryos with (prospective) genes that increase educational aptitude, we’ll assign that manipulation randomly and have a control group that does not receive this engineering, wait for the babies to be born and grow and go through school and compare their outcomes in test scores and grades and similar, and then we’ll know. Then there will be no more controversy (lol jk). But that will be long after I’m dead, and anyway what do I know of such things?

I will note that Hoel has a frustrating tendency to act as though blank slatism is some reviled niche perspective - the anti-hereditarians all tend to talk that way - when in fact in education policy the blank slate represents the bipartisan and cross-ideological assumption that has dominated debate since before I was born, and in particular in the 21st century. Perhaps the reason I am so regularly exasperated with people like Hoel or

Eric Turkheimer

or any number of others in this conversation is because I have been marinated in the ed policy world for a long, long time, and in that world there is no debate at all: every child is capable of any academic outcome. There is no limit to what education can do. Go to your average charter school conference and the idea that individual students have any inherent or intrinsic academic tendency isn’t just unpopular, it will get you shouted out of the room. Try being a public school teacher of a 25th percentile student, telling his parents that it’s not realistic to hope to turn him into a 90th percentile student, and see how that goes for you professionally. Even if you can get people to concede that our goal shouldn’t be equality of educational outcomes, or even that students perhaps don’t all have perfectly equal potentials, such concessions will remain in the realm of the vague and the general.

(When I ask people if they think we should try to establish summative educational equality of outcomes, they say of course not, that’s a straw man, we just want every child to reach their full potential. When I ask if they think it would be alright if, say, 15% of students were a full standard deviation or more below the mean, they say that’s outrageous, we should never condone that level of failure! And it’s like, guys….)

The two most important American educational bills in the 21st century have been No Child Left Behind (bipartisan, signed by a Republican president) and the Every Student Succeeds Act (bipartisan, signed by a Democratic president). Those names are not coincidental or empty; they express exactly what the politicians who drafted them believed was possible. They reflect a cross-ideological and remarkably durable assumption in our education politics that all students can be pushed through the college-to-affluence pipeline. I wish people on the genetics research side of this debate would stop talking as though there’s some rigid hereditarian consensus when, in the ed policy world, there is in fact the exact opposite. There’s a debate about the heritability of cognitive and behavioral traits in the world of population genetics/behavioral genetics, even if it’s one with a ton of bad faith and facile accusations of racism. There is no such debate in the Schools and Departments of Education in this country’s universities, no debate in almost any major thinktanks or foundations, no debate among pretty much any prominent politicians, no debate on your average municipal school board. In those climes, every student is a potential valedictorian and future Ivy Leaguer, and any other perspective is bigotry.

And that is, indeed, the pointy end of the stick here: while the actual scientific details are of course of great importance in many dimensions, the debate we’re having here is not about heritability, not really. Hoel may very well care very much about heritability as such. But heritability, I’m here to tell you, is for most people a stand-in, a totem, an effigy. What most people really care about is mutability or plasticity of educational outcomes; they want to know the degree to which a given student’s academic performance can be changed. As I’ve said, in much of polite society the assumption is that these traits are permanently and entirely mutable. The behavioral genetics research, the twin studies, have represented potential limits to that mutability. That’s why they’ve been controversial! But that’s what most people actually care about, the downstream effect of mutability rather than the upstream cause of heritability. Let’s use the pull quote feature to adequately emphasize this point. Italics, in this instance, just aren’t gonna cut it.

In most public debates about genetic influence on intelligence or academic aptitude, discussion of heritability is really about mutability or plasticity - that is, about how much we think a trait can be changed, how much ground we imagine environment and policy can realistically make up, and how willing we are to accept limits that feel politically or morally uncomfortable.

The fixation on heritability is really a way to assert different claims about how mutable academic outcomes are. The “human biodiversity” people want the heritability of academic ability to be significant because they want it to be immutable; they want to insist not just that white people are more academically accomplished than Black or Hispanic now but that white people will forever be more academically accomplished. The blank slate perspective, which is at least publicly held by most people and almost all liberals, is that these traits are not at all heritable, because if they are heritable there are limits to human possibility and to the social potential of schooling, and that’s uncomfortable. What both of these groups have in common is an absolutely dogged refusal to confront a likely potential future: a future where we do close the racial achievement gap and other group gaps, precisely because those gaps are not genetic in origin, but where there will still be a wide and consequential distribution of ability between individuals - and the people at the bottom of that distribution will suffer professionally, economically, and socially. The HBD people don’t think a post-achievement gap future is possible; the blank slaters don’t think any educational distribution is ever fixed.

I have myself, through a lot of dogged effort, tried to carve out another space, one perhaps shared by

Kathryn Paige Harden

and a small number of others: that group differences in academic outcomes are not genetic or otherwise intrinsic, but that there are intrinsic differences between individuals, thanks in part to genetics, and our complete lack of attention to these intrinsic differences leaves us with an unfair and capricious meritocratic system where some are inherently disadvantaged and pay the price for that disadvantage. This was the subject of my first book. The Cult of Smart is an argument that our failure to account for intrinsic individual differences amounts to “leaving children behind” because we expect students with different levels of natural talent to succeed in the same system, then blame them when they fail within that system. My concern, like Harden’s, is fundamentally for those who suffer because of their lack of natural talent. There is no pragmatic reason to take that disadvantage any less seriously than we take the deprivations of racism or poverty. And yet because of their dogged insistence on blank slate ideology, most progressives have nothing to say about disadvantages of inherent talent at all.

I have always said that my claims about education are ultimately made on three levels, in descending order of confidence:

1. At scale, the relative academic performance hierarchy is remarkably static, with very few students significantly moving to higher or lower positions of educational success over the course of academic life

2. This powerful consistency in student performance over time, even in the face of immense investment and relentless pedagogical and policy efforts to alter student performance, strongly suggests some individual attribute that constitutes an inherent or innate academic potential, predilection, or tendency

3. The most direct and parsimonious explanation for this attribute is genes

I still suspect that these are all true. But it’s the first that I consider indisputable from any honest reading of the research record, as well as personal experience and common sense. (The most ardent blank slater is still someone who has spent most of their lives understanding that some people are smarter than others and that this status doesn’t change much; they just don’t apply that intuitive knowledge to their political arguments.) I have, as you know, invested an immense amount of time in aggregating research that demonstrates that the relative distribution of students in the academic performance spectrum is largely static. Here’s 18,000 words with links to nearly 200 sources, if you’d like to consider the evidence yourself. I’m not equipped to scientifically assess the heritability of intelligence, but I can certainly tell you that decade after decade of education research has demonstrated that students gravitate to a level of academic performance very early in life and tend to stay there, regardless of environment, school type, pedagogy, policy, or intervention. This is what the anti-hereditarians have worked relentlessly to avoid, the reality that regardless of cause, we all have academic constraints that we operate under that schooling cannot alter.

In the comments of his post, I pointed out to Hoel that he was fixating on the mechanism rather than the outcome - that he was too focused on the potential cause of heritability and not focused enough on that actually socially-relevant consequences - and fellow anti-hereditarian

Sasha Gusev

popped up to call me a racist. That’s how this stuff usually goes, unfortunately. But setting my personal character aside, we’re left with the same issue: kids just don’t move around in the academic hierarchy much, no matter how hard we try, and the social consequences of this reality is profound. If there’s an inherent or intrinsic or genetic aspect to academic aptitude, influencing our meritocratic system (where performance in school leads to economic reward), what does Gusev want to do about it? What does Hoel want to do about it? I have no idea.

So let’s return at last to preterm birth. Here we have a condition that is not under the control of either parents or their children, has major implications for future academic outcomes, and yet it is not a heritable condition, has nothing to do with genetics. (Well, perhaps a tendency to have preterm births is genetically mediated, but this is several steps away from what we’re talking about here.) As the review I cited above states, “Early developmental difficulties in children born extremely preterm do not resolve with age and are not improving over time despite advancements in neonatal care.” And note that as medical science advances, we will have more and more severely premature babies who survive, meaning that the salience of this issue is only going to increase. The effects of prematurity on educational outcomes is not a genetic influence, and yet it’s not one we have any control over, either. In this, our inability to ameliorate its lasting academic influence, prematurity is similar to exposure to lead, traumatic brain injury, various developmental or cognitive disabilities, and more. If you’re ardently anti-hereditarian, these are influences that might be called environmental… but do they make you want to celebrate? Such lately-immutable negative influences make the heritability conversation seem rather less central.

People trying to chip the etched slate back to a blank state constantly point out that just because a trait is heritable, that does not make it immutable. Perhaps! But of course the obverse is just as true: just because a trait is not genetically mediated does not make it changeable. And there are so many potential influences on cognition…. Maybe there really is a critical period during which brain development is most fungible and the potential for lifelong influence most pronounced - but maybe it’s from birth to three months, instead of anytime before age five as has been asserted without evidence so many times, and maybe instead of being subject to change by Head Start or a linguistically-rich environment, eventual academic performance is most influenced by exposure to atmospheric pollutants or incidental radiation or who knows what. The point is that, while the heritability of intelligence is certainly a scientifically, morally, and politically essential question, if we’re willing to consider the possibility that educational potential is influenced by uncontrollable factors and fixed at an early age, we might be compelled to ask, well… what’s the difference?

1. The term “heritability” is among the most tediously lawyered in human history; anti-hereditarians tend to be relentlessly exacting about its use, often as a way to derail the uncomfortable conversations that spring up around discussion of human genetics. Well, let me give you the most stock answer I can to avoid having to engage in these theatrics again: is a statistical measure that tells you what proportion of the differences between the individuals in a population can be explained by genetic differences. It ranges from 0 to 1 (or 0–100%), and it applies to variation in a specific trait within a specific environment - not to any one person, and not to how “genetic” a trait is in some absolute sense. If you don’t like this definition, go fight it out somewhere else.