What IQ Tests Actually Measure—And What They Don't

In 1994, a group of 52 prominent intelligence researchers published a statement in the Wall Street Journal titled "Mainstream Science on Intelligence," asserting that IQ is a valid, well-established measure with robust predictive power for academic achievement, job performance, and even health outcomes. That same year, The Bell Curve by Herrnstein and Murray ignited a firestorm by claiming IQ differences explain social stratification. The debate that followed revealed how politically charged—and scientifically complex—the study of human intelligence remains.

The science itself is clearer than the culture war. IQ tests measure something real and important. But they measure a narrow slice of a broader construct, their predictive power has limits, and questions about group differences and heritability remain deeply contested.

Key takeaways

• IQ predicts important outcomes: Despite limitations, IQ scores are among the best predictors of academic achievement, occupational performance, income, and health—with correlations that rival or exceed those of most other psychological measures.
• The g-factor is real: Statistical analyses of diverse cognitive tests consistently reveal a general factor (g) that explains much of the variance in performance, suggesting a real underlying construct.
• Multiple intelligences theories lack psychometric support: Howard Gardner's theory is educationally influential but has not been validated by standard psychometric methods; most proposed "intelligences" correlate significantly with g.
• IQ has been rising: The Flynn effect documents massive IQ gains across the 20th century—roughly 3 points per decade in many countries—which cannot be genetic and highlight environmental factors.
• Heritability is not destiny: Twin studies show high heritability of IQ in adults (around 0.6-0.8), but heritability is a population statistic that cannot explain group differences or negate the importance of environment.
• The brain correlates of intelligence are diffuse: No single brain region is uniquely responsible for high IQ; neuroimaging research links intelligence to processing speed, working memory capacity, and neural efficiency across frontal and parietal networks.

What Is Intelligence? Defining the Construct

Intelligence is one of the oldest and most contested concepts in psychology. In 1921, Journal of Educational Psychology asked 14 leading researchers to define it. They gave 14 different answers. A century later, the definitions have converged somewhat: most researchers now accept something like the definition proposed by the Mainstream Science on Intelligence statement: "a very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly, and learn from experience."

The modern scientific study of intelligence began with Francis Galton in the 1880s (who tried, unsuccessfully, to measure intelligence via reaction time and sensory acuity) and with Alfred Binet, who developed the first practical intelligence test in 1905 at the French government's request to identify children who needed educational support. Binet's test was practical and empirical; he explicitly warned against interpreting the scores as measuring fixed, innate intelligence—a caution that was largely ignored when his test was imported to the United States.

The g-Factor: General Intelligence

The most important finding in the science of intelligence is the existence of a general factor (g), first identified by Charles Spearman in 1904 through factor analysis. When you give a battery of diverse cognitive tests—verbal, spatial, mathematical, memory—people who score well on one tend to score well on all of them, to a degree that exceeds what chance would predict. This positive manifold can be statistically extracted as a single general factor.

What does g represent? Several theories compete:

• Mental speed: g correlates with reaction time, inspection time, and neural conduction velocity—suggesting that faster information processing underlies it.
• Working memory capacity: g correlates strongly with working memory measures, suggesting that the capacity to hold and manipulate information in mind is central.
• Neural efficiency: PET and fMRI studies (led by Richard Haier) show that higher-IQ individuals often show less brain activation while performing cognitive tasks—suggesting more efficient neural processing.

The PFIT (Parieto-Frontal Integration Theory), proposed by Rex Jung and Richard Haier based on a review of neuroimaging literature, identifies a network of frontal and parietal regions as the biological substrate of g. This network includes areas supporting working memory, attention, and abstract reasoning, connected by white matter tracts that vary in integrity with IQ.

IQ Tests: What They Measure

Modern IQ tests like the Wechsler Adult Intelligence Scale (WAIS-IV) and the Cattell-Horn-Carroll (CHC) model-based tests assess multiple cognitive abilities that together index g:

• Fluid intelligence (Gf): Problem-solving with novel information, pattern recognition, inductive reasoning. Relatively independent of accumulated knowledge.
• Crystallized intelligence (Gc): Accumulated knowledge and verbal skills. Increases through education and experience into middle adulthood.
• Working memory index: Span tasks, digit recall, arithmetic.
• Processing speed: Reaction time, visual scanning speed.
• Visuospatial ability: Mental rotation, visual pattern recognition.

The test yields an IQ score (mean 100, standard deviation 15) through norm-referenced comparison to a representative population sample. Scores are not absolute; they indicate where you fall relative to others in your age cohort.

IQ scores are highly predictive of academic achievement (r ≈ 0.5-0.6), occupational performance (r ≈ 0.3-0.5, with higher validity for complex jobs), income (r ≈ 0.3-0.4), and even health and longevity (the Scottish Mental Survey follow-up studies by Ian Deary found IQ at age 11 predicted survival to age 70+).

The Flynn Effect: IQ Scores Rising Rapidly

One of the most striking findings in the intelligence literature is the Flynn effect, documented by political scientist James Flynn in the 1980s: raw IQ scores have risen dramatically across the 20th century in virtually every country studied, at roughly 3 IQ points per decade.

This rise is far too rapid to be genetic. It must reflect environmental changes. Proposed causes include:

• Improved nutrition (particularly iodine sufficiency and reduced lead exposure)
• Increased education and familiarity with abstract, out-of-context thinking
• More cognitively stimulating environments (greater media, technology)
• Better public health reducing early developmental insults

The Flynn effect is strongest for fluid intelligence and weakest for crystallized intelligence—consistent with the hypothesis that it reflects improved abstract, non-verbal reasoning skills driven by environmental exposure to novel cognitive challenges.

Notably, the Flynn effect appears to have reversed or plateaued in some high-income countries since roughly the 1990s-2000s (Norway, Denmark, Finland, UK). The reasons are debated—possible contributors include changes in education, increased screen time, or simply reaching a biological ceiling.

Howard Gardner's Multiple Intelligences: Theory vs. Evidence

Howard Gardner proposed in 1983 that there are seven (later expanded to 9 or more) distinct intelligences: linguistic, logical-mathematical, spatial, musical, bodily-kinesthetic, interpersonal, intrapersonal, naturalistic, and existential. The theory has been enormously influential in education.

However, its psychometric evidence base is weak. When researchers factor-analyze tests of Gardner's proposed intelligences, they find the scores correlate significantly with each other and with g—which is the opposite of what distinct, independent intelligences would predict. The "intelligences" look more like facets of general intelligence plus specific skills and talents than genuinely independent cognitive systems. Critics, including Gould, Waterhouse, and Klein, have argued that Gardner's framework conflates intelligence with talent, personality, and motor skill.

This is not to say the pedagogical implications are worthless—recognizing different learning styles and strengths has value—but the theory should not be cited as established neuroscience.

Genetics, Environment, and the Heritability Debate

Twin studies consistently show that adult IQ has a heritability of approximately 0.6-0.8 in high-income Western samples—meaning 60-80% of the variance in IQ scores can be attributed to genetic variation. This is among the highest heritabilities of any complex trait.

But heritability is profoundly misunderstood. Three crucial points:

1. Heritability is a population statistic, not an individual one. Saying IQ heritability is 0.7 does not mean "70% of your IQ comes from genes." It means that, in the population studied, 70% of the variation in IQ scores is associated with genetic variation.

2. Heritability changes with the environment. Eric Turkheimer demonstrated that IQ heritability is low (≈0.1) in children raised in poverty and high (≈0.7) in children raised in affluent environments. In impoverished environments, environmental differences overwhelm genetic ones; in enriched environments, environmental variation is reduced, allowing genetic differences to express more fully.

3. Heritability cannot explain group differences. The heritability of height within groups tells us nothing about why average heights differ between groups. Similarly, high within-group heritability of IQ does not explain gaps between demographic groups—which are far better explained by historical and ongoing inequalities in education, health, nutrition, and social opportunity.

FAQ

Q: Is IQ fixed, or can it change?

A: IQ is relatively stable across adulthood but is not fixed. Early childhood interventions (like high-quality preschool programs) can produce meaningful IQ gains that persist. Environmental factors—nutrition, education, lead exposure, iodine deficiency—can substantially affect IQ, particularly in developing brains. Within adulthood, IQ scores are quite stable (test-retest correlations of 0.8-0.9 over decades), but crystallized intelligence (knowledge and verbal ability) continues to grow into the 50s and 60s, while fluid intelligence typically peaks in the late 20s and declines slowly thereafter.

Q: Do IQ tests measure innate intelligence or acquired knowledge?

A: Modern IQ tests assess both. Fluid intelligence tests (matrix reasoning, novel problem-solving) aim to minimize the role of acquired knowledge. Crystallized intelligence tests explicitly measure knowledge. The g-factor extracted from tests is a blend of both. Importantly, no test can fully separate "innate" from "acquired" ability, because experience shapes cognitive development from conception onward. The distinction between nature and nurture in cognition is a false dichotomy.

Q: Does emotional intelligence (EQ) predict success better than IQ?

A: This claim, popularized by Daniel Goleman's 1995 book, was substantially overstated. The research literature shows that IQ remains the single strongest predictor of academic and occupational performance. Emotional intelligence—the ability to perceive, understand, and manage emotions—does predict additional variance in outcomes beyond IQ, particularly for leadership, social, and interpersonal domains. But the predictive validity of EQ is considerably smaller than Goleman claimed, and many EQ measures suffer from validity issues.

Q: Why are some people with high IQs unsuccessful in life?

A: High IQ is a probabilistic predictor, not a guarantee. Many other factors determine life outcomes: conscientiousness, emotional regulation, social skills, motivation, health, circumstance, and luck. IQ scores predict performance on cognitively demanding tasks but have weaker relationships with qualities like persistence, creativity, emotional resilience, and wisdom. Above a threshold (roughly IQ 120), additional IQ increments have diminishing returns for many real-world outcomes.

Q: Are IQ tests culturally biased?

A: IQ tests do show some cultural bias—performance is affected by familiarity with the testing context, language, and the specific type of abstract reasoning emphasized in Western education. Attempts to create "culture-fair" tests (like Raven's Progressive Matrices) have reduced but not eliminated this issue. Some researchers argue that the predictive validity of IQ tests (how well they predict academic and occupational outcomes) is broadly similar across demographic groups, which suggests they measure something real and relevant; others argue that if the institutions doing the predicting are themselves biased, predictive validity doesn't establish freedom from cultural bias.

Q: What is the relationship between brain size and IQ?

A: There is a modest positive correlation between brain size and IQ (approximately r = 0.3), but the relationship is far from deterministic, and brain size is a crude measure. More informative are measures of white matter integrity, cortical thickness in frontal and parietal regions, and neural processing speed. Large population studies (like UK Biobank analyses by Stuart Ritchie) have confirmed that intelligence correlates with multiple structural and functional brain measures, but no single anatomical feature captures it.

---

Read next:

• How Human Memory Works: Encoding, Storage, and Retrieval
• How Language Shapes the Way You Think
• How Pain Works in the Brain: It's Not What You Think