What Is Intelligence?: Beyond the Flynn Effect

By James R. Flynn

10 min read

Why does it matter? Because the debate between nature and nurture has been asking the wrong question all along.

Here is a paradox that should keep you up at night: IQ is roughly 80% heritable within any given generation, yet average scores have climbed nearly 30 points across the twentieth century. Those two facts cannot both be true — except they are. Genes didn't change. Something else is running the machine. James Flynn spent decades working out exactly what, and his answer dismantles the oldest argument in science: that biology sets a ceiling on what human beings can become. It doesn't. It sets a starting position. Culture holds the throttle. And once you understand the feedback mechanism that connects the two — the same mechanism that explains why basketball players became unrecognizable after television arrived, and why your family home stops shaping your mind right when colleges start measuring it — you will never look at the nature-versus-nurture debate the same way again.

How Can Something Be 80% Genetic and Still Rise 30 Points? The Paradox That Breaks the Nature-Nurture Debate

Here is the puzzle in its sharpest form: twin studies consistently show that IQ is around 80% heritable within a given generation, yet across the twentieth century, average IQ scores climbed roughly 30 points — two full standard deviations. If genes explain nearly all the variation in intelligence between people, how did environments lift everyone so dramatically? The two facts seem to call each other liars.

Here is why the contradiction dissolves: within a single generation, genes are what differ between people, so genes appear to explain everything. But between generations, the cultural environment shifts for everyone simultaneously, while genes stay essentially unchanged. High heritability within a generation and large environmental gains between generations aren't contradictions. They're two different mechanisms operating on two different timescales — and Section 2 builds the model that shows exactly how they fit together.

Genes Don't Determine Your Ceiling — They Just Help You Find Better Coaching

Flynn makes this precise with two identical twins — call them Joe and Jerry — separated at birth and raised in different Indiana towns. Same genes, different families, same outcome: both become skilled basketball players. Researchers see that and conclude genes explain everything. Flynn says that reading is exactly backwards. The genes didn't produce the skill; they recruited a potent environment and rode it upward. Joe's slight edge in height and reflexes got him picked for sandlot games early, which sharpened his game, which got him onto a school team, which earned him better coaching. Jerry's identical genes ran the same sequence in a different town. When they meet as adults with nearly identical abilities, the right explanation isn't "genes are destiny" — it's that genes are a magnet for the kind of environment that does the real work. Flynn calls this the individual multiplier.

Now pull back from the individual and look at the culture. After World War II, television made basketball beautiful to watch. Participation surged. The rising average skill level became its own pressure: to stay competitive, everyone had to improve, which raised the mean further, which raised the pressure again. Slam dunks, ambidextrous passing, mid-air body control — none of that required a genetic mutation. It emerged from a feedback loop between cultural enthusiasm and competitive necessity. Flynn calls this the social multiplier, and it runs independently of anyone's chromosomes.

These two mechanisms dissolve the paradox from the previous section. Within a single generation, people differ mainly in their genes, so genes appear to explain the hierarchy of performance — that's heritability doing its legitimate job. But between generations, genes are static while culture can shift for everyone at once. High heritability within a generation and a 30-point IQ rise between generations aren't contradictions. They're the individual multiplier and the social multiplier taking turns. The first makes genes look omnipotent in the snapshot; the second makes environment look omnipotent across time. Both readings are true, and neither tells you what the other is measuring.

The IQ Gains Are Real — And the Argument Against Them Is Logically Incoherent

The objection you've probably heard runs like this: yes, IQ scores went up, but the gains were largest on the least intellectually demanding subtests — things like copying symbols or reading maps — and smallest on the genuinely complex ones like vocabulary or abstract reasoning. Psychometricians measure a subtest's cognitive weight with a statistic called its g loading — a measure of how much a given subtest reflects general reasoning ability rather than a specific skill. Since Flynn Effect gains don't track those g loadings, the argument goes, the gains must be hollow: test sophistication, not real intelligence. Arthur Jensen and J. Philippe Rushton built careers on this claim.

Flynn's answer is two words: wrong ranking. Society assigns cognitive tasks based on what the economy needs, not how subtests score on a g-loading table. There is no reason those two rankings should align.

The real kill shot is arithmetic. Flynn calculated a GQ score: a gain weighted by g loadings, so improvement on high-complexity subtests counts for more. If the hollow-gains argument were correct, GQ should lag far behind raw IQ. In a comparison of two generations on the Wechsler children's test, a 5.00 IQ point gain translated into a 4.72 GQ gain. The "hollow" gains, properly weighted, were 94% as large as the raw gains. Five percent isn't nothing, but it is not the chasm the hollow-gains story requires.

Then comes the logical trap Jensen and Rushton walked into themselves. They argued that the Black-White IQ gap at any given moment is real and meaningful, precisely because it correlates with g loadings. But they simultaneously argued that the gains Black Americans made on White Americans over the following decades are hollow and unreal, because those gains don't correlate with g loadings. Flynn reduces this to three propositions they must simultaneously hold: the gap in 1972 was real; the gap in 2002 was real; the gains that reduced it were hollow. But a hollow cause cannot dissolve one reality and install a different one in its place. If the method of correlated vectors proves the gap real when it's large and proves the gains unreal when they shrink the gap, that's not a methodology — it's a conclusion dressed up as one.

Even Pol Pot Could Not Permanently Damage a Nation's Genetic Intelligence

In 1975, Pol Pot's Khmer Rouge marched two million people out of Phnom Penh at gunpoint and began systematically murdering anyone connected to education, government, or professional life. By the time the regime fell in 1979, roughly a quarter of Cambodia's population was dead. If you wanted to design a catastrophe specifically aimed at destroying a nation's cognitive future, this was it.

So how much damage did it actually do? Flynn ran the arithmetic. Pol Pot killed about 2.1 million people — around 26% of Cambodia's population — using occupational status as a rough death warrant. If he had done the same thing using IQ tests, eliminating the top quarter by score would have lowered the mean IQ of surviving parents by about six points, and their children would have inherited a good chunk of that deficit. But occupation is a poor proxy for genetic intelligence. In a semi-rural society like Cambodia, the correlation between a parent's job and a child's IQ runs around 0.300 — meaning occupation explains only about 9% of the variance in cognitive outcomes. At that correlation, purging the top 26% by occupation drops the next generation's mean IQ by roughly one point. Not six. One.

Then consider what happened next. By 1985 — just six years after the regime collapsed — Cambodia had rebuilt a functioning administrative and technical class from scratch, despite having no schools, no books, no hospitals, and no civil service left standing. The human cost was staggering. The genetic cost, by the numbers, was almost nothing.

The hard intuition to shake is that mass death can permanently diminish what a population is capable of. The numbers say otherwise. Genes are resilient; environments are the variable that actually moves.

Your Family Home Stops Shaping Your Intelligence Around Age 17 — But That's Exactly When It Matters Most

The family home stops shaping your intelligence long before you think it does — but it hangs on just long enough to decide whether you get into a good university.

Flynn measured how quickly that influence fades by comparing children's IQ samples against adult samples across decades. Adults have shed their family environments and sorted themselves into circumstances that match their genes: the sharp ones end up in cognitively demanding jobs, the less sharp in less demanding ones. Children don't have that freedom. A kid in the 98th percentile for ability is almost certainly being raised in a family that sits much lower on the cognitive ladder — not because parents are cruel, but because genuine 98th-percentile households are rare. That family pulls the child's measured score down. A kid at the 2nd percentile is statistically likely to live in a family environment above that floor, which nudges their score up. Track how those deviations shrink with age, and you can read off exactly how much of the family's distorting hand is still visible.

The numbers drop sharply. Family environment accounts for roughly a third of IQ variance at age 12. By 17, that share has fallen by more than half. By the early twenties, it has nearly vanished. That trajectory sounds like progress — the environment loosening its grip, genes settling into place. And eventually it is. But notice where the knife is still halfway through the cut: age 17, precisely when American teenagers sit the SAT.

Flynn estimated that a student whose family environment runs one standard deviation below their own ability level — a genuinely bright child raised in a household with limited books, limited conversation, limited cognitive friction — pays a penalty of roughly 30 points on the SAT Reading section alone. Extend that across all three sections and the total deficit approaches 90 points. As Flynn put it, every American parent will recognize that gap as something that often separates joy from tears: the difference between a school that expands your world and one that merely confirms the ceiling your zip code already suggested.

The argument isn't that genes don't matter. It's about timing. The model that showed IQ gains were real, that genes ride environments rather than replace them, quietly reveals that the moment family environment releases its grip is the very moment society asks a teenager to prove what they're worth. The injustice isn't abstract. It's a number on a score report.

The Same Cognitive Shift That Made Us Smarter Also Made Us Less Violent

In 1474, a rooster was hauled before a church court in Basel, tried with full legal proceedings, and convicted of laying an egg fathered by Satan. We find this hilarious. Our ancestors found it terrifying. Flynn's point is that the distance between those two reactions is not merely a matter of accumulated facts — it is a cognitive distance, and it is measurable.

Flynn argues that the same shift in thinking that drove IQ gains across the twentieth century also drove the long retreat of human violence. The key capacity is hypothetical reasoning: the ability to step outside your own position and apply abstract logic to an unfamiliar case. Pinker's data shows violent deaths fell from roughly 15% of all deaths in hunter-gatherer societies to around 1% in the twentieth century. The link to Flynn's test data is direct: the IQ gains he tracked were driven specifically by abstract and categorical thinking — the same mental operation required to entertain a question like 'What if you woke up tomorrow and had turned black?' and follow the logic wherever it leads. Luria's subjects in 1930s rural Russia could not do this. They treated it as an absurd game. Modern anti-racism depends on it absolutely.

The cautious hope in all of this is structural. If the cognitive tools that reduced violence emerged from social conditions — mass education, scientific culture, economic independence for women — then they are not fixed gifts from our genes. They are reproducible. The same reasoning capacity that makes the rooster trial absurd to us is the capacity we need to look at climate projections and act as if the logic actually applies to us. The tools are already in our hands.

The Frozen Mind Thesis Is an Ideology, Not a Finding

What is the actual ceiling on human cognitive potential? The honest answer is that nobody knows — and the people who claimed to know were doing ideology dressed up as psychometrics.

Virginia Woolf and E.M. Forster — the literary aristocracy of the era — found the self-taught working man somewhere between pitiable and nauseating; Forster thought the masses should be stripped of education and returned to breeding sturdy agricultural stock. This wasn't eccentric cruelty. It was the mainstream view, and it had data behind it. If you surveyed the vocabulary hierarchy of 1900 and found it heavily shaped by genetic inheritance, you might reasonably conclude that doubling or tripling years of formal education could never transform a nation of laborers into a nation of professionals. The static fact seemed to close the door on the dynamic possibility.

In 1900, professionals were 3% of the American population. By 2010, they were 35%. The average person's vocabulary, projected back to 1900, would place them at the 98th percentile of that era. No genetic analysis of the 1900 workforce could have predicted this, because the social multiplier that produced it wasn't running yet. Flynn's sharpest point is simply this: a snapshot of how genes rank people within a generation tells you nothing about what a cultural shift — mass education, a new economy, the professional class becoming the norm — can do to lift the entire distribution. A static fact cannot speak to a dynamic potential. That transformation from 3% to 35% is the answer to everyone who thought the door was closed.

What We Owe the Next Multiplier

The multiplier is still running. Every feedback loop Flynn traced — the one that turned a regional sport into midair poetry, the one that moved a continent's vocabulary two standard deviations in a century, the one that made a rooster's trial go from terrifying to absurd — was set in motion by choices nobody made consciously: build schools, expand who counts as a professional, let women leave bad men. Nobody voted for the Flynn Effect. The GI Bill wasn't a cognitive enhancement program. Universal primary education wasn't sold as a plan to raise abstract reasoning scores by thirty points. But that's what happened, because institutions shape the minds that use them. Climate change and nuclear proliferation are not puzzles beyond our cognitive reach. They are puzzles we have not yet decided to treat the way Indiana decided to treat basketball. The same hypothetical reasoning that ended open racism and made violence statistically rare is sitting right there, available, waiting. Flynn's final question isn't really about intelligence. It's about whether we've noticed that we already built the machine once — accidentally, incrementally, without a single vote on the matter — and whether that realization arrives before we need it to.