Richard Haier on IQ Tests, the G Factor, Genetics and the Controversies of Intelligence Research

High-Level Overview

Lex Fridman sits down with Richard Haier, a psychologist and neuroscientist who has spent his career studying human intelligence, author of The Neuroscience of Intelligence and editor of the journal Intelligence. Lex opens by calling this "a highly controversial topic, but a critically important one" and asks listeners to approach it with empathy, rigor and grace. The conversation covers what researchers mean by intelligence (the general factor, or g), how IQ tests are built, how stable scores are across a lifetime, what they predict about education, careers and even survival, and the fraught history of research on group differences, from Arthur Jensen's 1969 paper to The Bell Curve.

The tone is calm, careful and often philosophical. Haier speaks plainly as an empiricist who follows the data, while Lex repeatedly voices his own discomfort with the idea that something so consequential could be stable and partly genetic, and presses on the scientist's responsibility for how findings get misused. The conversation moves from definitions and test design into genetics and enhancement, then the race and Bell Curve controversies, life outcomes, the Flynn effect, nature versus nurture, consciousness and machine intelligence, before closing on advice for young scientists and mortality.

The Discussion Explores:

• What the g factor is, where it came from, and why Haier calls it psychology's most replicated finding
• How IQ tests are constructed, which subtests best capture g, and why scores are stable from childhood to old age
• What IQ predicts: school, jobs, income, military trainability and even survival decades later
• Whether intelligence can be enhanced, and why brain training, Mozart and compensatory education have failed so far
• The history and backlash around Arthur Jensen and The Bell Curve, and what the science does and does not say about group differences
• The Flynn effect, nature versus nurture, and what evidence would change Haier's mind
• Links between consciousness and intelligence, brain efficiency, and how to test intelligence in machines
• Advice on following the data, communicating with compassion, and facing hard truths like mortality

Deep Dive into Main Ideas

The G Factor: What Intelligence Research Actually Measures

Haier starts with the everyday meaning of intelligence (how well you reason, figure things out and handle situations where you don't know what to do) and then the scientific one. Researchers use the general factor, g: the mental ability common to virtually all tests of mental ability. The idea goes back more than a century to Charles Spearman, who noticed that students' scores across different subjects were all positively correlated and inferred a common factor behind them. Anecdotes about someone who is great at one thing but weaker at another, Haier says, do not contradict this, because g describes a general tendency, not a perfect match.

He is emphatic about how solid the finding is: "it is the most replicated finding in all of psychology," and even with psychology's replication crisis, there is none for g. Give any large, diverse group any battery of mental tests and factor analysis will extract g; pick 12 tests at random and another 12 and the two g factors correlate above 0.9. "It is not a statistical artifact," he says. G typically accounts for about half the variance across tests, alongside specific abilities such as math, memory and test-taking skill, but it is the most reliable and stable part, strongly influenced by genetics and very hard to change with training or drugs. An IQ score is an estimate of g, not g itself, and IQ numbers are relative, like percentiles or chess ratings: an IQ of 140 is not twice as smart as 70, because the scale has no absolute zero. Lex connects the generality of g to Miyamoto Musashi's line from The Book of Five Rings: "if you know the way broadly, you will see it in everything."

Inside the IQ Test and the Scottish Study

Haier explains that a good test item is defined less by its content than by its empirical relationship to valid outcomes. He contrasts intelligence tests with the "dust bowl empiricism" of personality tests like the MMPI, whose odd true/false items have no obvious meaning yet predict diagnoses. IQ subtests look like what they measure: general information, vocabulary (the word regatta was removed for being culturally unfair), digit span forward and backward, and reaction-time tasks. Surprisingly, vocabulary and digit span backward are among the subtests most correlated with g. In the Hick paradigm, simple reaction time barely correlates with IQ, but reaction time on a slightly more complex choice (press the light farthest from the other two) correlates highly, pointing to speed of information processing; that is why good tests are timed.

Lex describes his own test anxiety on the SAT verbal section, and Haier explains that anxiety, illness or a 102-degree fever make a score a poor estimate of g without changing g itself, which is why admissions decisions should use multiple sources of information. He adds that the SAT is essentially a highly g-loaded reasoning test, and that math-verbal "tilt" also predicts college outcomes. On stability, he describes Scotland's 1930s effort to test every 11-year-old in the country. His friend Ian Deary rediscovered the records at the University of Edinburgh and followed up the participants; scores at age 11 correlated highly with scores when the same people were retested around age 70 (in Haier's telling), and about twice as many people from the top IQ quartile were alive 70 years later as from the bottom quartile, in both men and women, despite universal healthcare. Why remains unclear.

Life Is One Long Intelligence Test

For Haier, intelligence matters because of what it reflects about navigating daily life. "It has been said that life is one long intelligence test," he says, arguing that academics who criticize intelligence research are usually bright people who struggle to understand how different everyday life is at an IQ of 75 or 80. He notes that the U.S. military, which he considers unmatched at training diverse people, will not accept recruits below roughly an IQ of 83 because it has found it cannot train them for military jobs. G is closely tied to how fast and how much people learn, and he jokes that academics who dismiss intelligence talk about little else in faculty hiring meetings.

On outcomes, he says IQ correlates with income, life expectancy and even reproductive success in some studies; it is not the only factor, but it is the single one that matters most, and matters more as jobs grow more complex. He worked at the start of the Johns Hopkins Study of Mathematically Precocious Youth, which gave the SAT math test to 12-year-olds. Among children already in the top 1%, the top quartile went on, over a 50-year follow-up, to earn more patents and publications and more university tenure than the bottom quartile. Lex finds this stability painful, and Haier asks whether he also finds it painful that charisma, a relatively stable trait, drives success. He argues that the modern idea that "if you work hard, you can be anything you want to be" differs from the original promise that hard work brings success: "This is completely unrealistic," he says, noting he could never become a successful theoretical physicist.

Better People? Enhancement and the IQ Pill

Lex asks whether intelligence research implies some people are better than others, invoking the ideal that all men are created equal. Haier is clear: there is "zero evidence that smart people are better with respect to important aspects of life" such as honesty or kindness. Lex, having reread The Rise and Fall of the Third Reich, points out that some of history's worst atrocities were enabled by highly intelligent people. Still, Haier says "more intelligence is better than less" all things being equal, and poses a thought experiment: given a pill to make you smarter and one to make you dumber, who would choose the latter? Lex pushes back that intelligence may not bring happiness and that there could be a sweet spot, citing Flowers for Algernon; Haier agrees more intelligence clearly helps at the lower end, and that science should study the question.

On enhancement, the record is bleak. N-back training "doesn't work," memory training improves memory without touching g, and the Mozart effect has been refuted by hundreds of studies and meta-analyses. Yet Haier argues that shifting the whole distribution even half a standard deviation would be world-shaking, since many social problems are worsened by limited reasoning ability. He believes nearly all proposed fixes rest on a blank-slate assumption and that the real opportunity is neuroscience: funding research into the molecular biology of learning and memory with a focus on individual differences. He recalls a Nobel laureate in memory research telling him about individual differences, "don't go there," and predicts the IQ pill might emerge from a pharmaceutical company developing an Alzheimer's drug.

Arthur Jensen, The Bell Curve and Group Differences

Haier describes The Bell Curve (1994, by Richard Herrnstein and Charles Murray, whom he knows) as an empirical book about the importance of intelligence in everyday life, with the controversy stemming from one chapter on the average gap of about one standard deviation (15 points) in IQ scores between Black and white Americans. He stresses that the authors said they were agnostic about whether the gap had any genetic cause, and that it would not matter for how individuals are treated, since every person must be judged as an individual. He believes many critics never read the book. The deeper history goes to Arthur Jensen's 1969 Harvard Educational Review article, which reviewed early compensatory education programs, found no durable IQ gains, and then took the "fateful step" of suggesting a genetic influence on group differences be considered. Haier calls it perhaps the most infamous paper in psychology; Jensen faced bomb threats and campus security, and intelligence research became, in Haier's word, radioactive.

Haier recounts a Nightline producer asking him about the race chapter; when he said that if a difference were strongly genetic that would be a good thing, because biology can be studied and fixed, the invitation evaporated. Asked whether such research could fuel racism, he says there are vast reservoirs of hate unrelated to any data, and he does not want racists to decide what scientists may study. He notes that almost no one disputes that score gaps exist; the argument is over causes and remedies, and he points to lead in gasoline, paint and water as an environmental factor society can eliminate. He relays Jensen's answer when asked if he was racist: "I've thought about that a lot. And I've concluded it doesn't matter," which Haier explains meant that Jensen's "default hypothesis" was a testable scientific claim, and the scientific method would correct any bias. Very little new research has been done since, he says; what exists suggests gaps persist and underachievement remains unsolved.

The Scientist's Responsibility and the Policy Gap

Lex pushes back, arguing that psychology is full of gray areas, the media distort findings, and scientists bear some responsibility to anticipate misinterpretation, as he tries to do even with tweets and his work on semi-autonomous vehicles. Haier accepts that deciding not to publish out of concern for harm is not unreasonable, but says his role is to explain carefully, in books and conversations, what the data mean and do not mean. He has turned down debate formats, such as a BBC request, where his role would be to justify "race science."

On policy, Haier says the word intelligence rarely appears in education policy documents about achievement gaps. He cites the Coleman report and later replications finding that school and teacher variables together account for only about 10% of variance in student achievement, and says the Gates Foundation found its teacher-quality investments had little impact. He criticizes a California math curriculum effort led by an education professor who denied mathematical talent exists. "If you want to fix the problem, you have to know what the problem is," he says, noting that about 16% of people, roughly 51 to 52 million Americans including 14 million children, have IQs under 85, which makes gainful employment very difficult. He dismisses Stephen Jay Gould's The Mismeasure of Man as political rather than scientific and points readers to scientific reviews of it, and to Murray's later books Human Diversity and Facing Reality for readers who want to judge the arguments themselves.

The Flynn Effect and Nature Versus Nurture

The Flynn effect, documented by the late James Flynn, is a rise in IQ scores of about three points per decade, a cohort effect that forces test makers to renorm. It has been used to argue intelligence cannot be genetic, but Haier says that does not follow, since genes explain perhaps 50% of variance (maybe up to 80% in adults), leaving room for environmental change. Researchers increasingly attribute the rise to nutrition and healthcare, and there is evidence it is slowing or even reversing.

On nature versus nurture, Haier's short answer is that it is all both, and many behavioral geneticists think the interactions can never be disentangled. He disagrees, because a stable g suggests the interactions can be studied. He cites correlations of IQ with cortical thickness and glucose metabolism, the similarity of identical twins reared apart, and adoption studies where adult IQ tracks biological rather than adoptive parents, while noting some adoption data show movement toward adoptive parents. Asked the key question of what evidence would change his mind, he says a clear, population-level effect from decades of well-funded interventions like compensatory education; after 50 or 60 years, nothing has worked convincingly. He also cautions that correlations between socioeconomic status and IQ (like the number of books at home) cannot show causation without genetically informed designs, which tend to show genes carrying most of the predictive weight.

Consciousness, Brain Efficiency and Machine Intelligence

Haier describes early PET studies with an anesthesiologist colleague, scanning volunteers awake, drowsy and fully anesthetized to see which brain region turns off last when consciousness is lost; across drugs, the thalamus appeared to be involved. He wonders whether some people are more conscious than others and whether that relates to intelligence, and whether people with higher IQs need more or less anesthetic, a question he could never answer because doses are recorded by weight. His early imaging work also found an unexpected inverse correlation between brain glucose use and test performance, which led to the brain efficiency hypothesis, now supported by growing evidence. Lex playfully proposes that consciousness and g might trade off.

On machines, Haier once suggested giving IQ test questions to Siri and Alexa, but they simply search the internet. Lex describes Francois Chollet's IQ-style tests for AI, which pose pattern problems about symmetry, counting and objectness that are trivial for humans but hard for machines, and notes that novelty is central to difficulty, which Haier connects to fluid intelligence as opposed to crystallized knowledge. Haier says he is "supremely confident" an intelligence-raising intervention is possible precisely because he is ignorant of neurobiology's complexities, and he hopes that 51 million Americans with IQs under 85 will motivate scientists to try.

Advice, Compassion and Mortality

Lex reminds listeners that intelligence is only one part of the human condition and that creating beauty does not depend on it. Haier agrees, clarifying he only said more is better than less, all else equal. His advice to young people drawn to science is to tackle the big questions of consciousness, intelligence and brain illness, follow the data wherever it leads, recruit experts from other fields, and communicate controversial results with skill and compassion. On the meaning of life, he jokes that he is not going to tell, saying we do the best we can to get through a finite number of days, and that mortality, like g, is a hard truth people often walk past.

He closes on the point Lex hoped to hear: nothing in the science argues against compassion. Everything he has learned points to treating people as individuals, respectfully and with compassion, because "through no fault of their own some people are not as capable as others," and "everybody is deserving of love." Lex ends with Albert Einstein's remark that he was not so smart, but stayed with the questions much longer.

Conclusion

Richard Haier presents intelligence research as a robust, uncomfortable science: g is real, measurable, stable over a lifetime, substantially genetic and predictive of education, work and health, while efforts to raise it through training and schooling have so far failed. He argues that ignoring these data, especially in education policy and the debate over group differences, blocks progress, and that the most hopeful path is neuroscience aimed at the biology of learning and memory. Lex supplies the counterweight, pressing on misuse, happiness and human worth, and the two land in the same place: intelligence does not measure a person's goodness or value, and the facts call for more compassion, not less. It is a demanding, nuanced conversation, best heard in full by anyone curious about the hard edges of the science of the mind.

This is a summary of "#302 – Richard Haier: IQ Tests, Human Intelligence, and Group Differences" from Lex Fridman Podcast, giving you the GIST so you can decide if you want the REST. Listen to the full episode.