Fluid vs Crystallized Intelligence: The Two Halves of IQ

The distinction, precisely

Raymond Cattell first proposed the split in 1943 and tested it formally in 1963. John Horn extended it, and John Carroll's 1993 survey of hundreds of factor-analytic studies folded it into what is now the Cattell–Horn–Carroll model — the closest thing the field has to a consensus taxonomy of human abilities.

How the two are connected: investment theory

Cattell did not think of these as two unrelated abilities sitting side by side. He proposed that crystallized intelligence is the historical product of fluid intelligence.

The idea, called investment theory, is that a person invests their fluid reasoning capacity in learning situations over years, and that what precipitates out of that investment is knowledge — a store of facts, procedures, vocabulary and skill. A young person with high Gf will, given the opportunity, accumulate more Gc than a young person with lower Gf, because they extract more from each encounter with something new.

The phrasing matters. Gf is a rate. Gc is an accumulation. That is part of why the two can follow different courses, and why they can diverge sharply in the same person: a sixty-year-old professor may have a large crystallized store and a fluid reasoning capacity noticeably below what it was at twenty-five.

In practice the two are linked. A large study of cognitive ageing found that people whose fluid abilities decline more tend to gain less crystallized knowledge, or even lose some — so growing knowledge does not simply make up for the decline.

Is fluid intelligence just g?

This is the most interesting open question in the area, and the answer is genuinely unresolved.

When researchers extract a general factor from a battery of diverse cognitive tests, the resulting g tends to correlate extraordinarily highly with fluid intelligence — so highly that some have argued they are the same construct measured two ways.

Kvist and Gustafsson tested this in 2008 and reached a sharper conclusion. In about 3,600 people in Sweden — native Swedes, European immigrants and non-European immigrants — the correlation between Gf and g was close to 1 within each group, where people had broadly the same opportunities to learn, but 0.83 in the pooled, culturally mixed sample. The authors read this as support for investment theory: if everyone has had the same chance to convert fluid ability into knowledge, then knowledge is a clean proxy for fluid ability, and the two factors collapse together. Where opportunity differs, the correlation drops.

How each changes with age

This is where the distinction earns its keep, because the two abilities have almost opposite life courses.

Fluid reasoning peaks early. Processing speed, which underpins much of it, peaks around the end of the teens. Working memory peaks later, between the early twenties and about thirty. From there, fluid measures decline across the adult lifespan, and the decline is one of the most reliably replicated findings in cognitive ageing — at least in cross-sectional comparisons; studies that follow the same people over time sometimes show smaller declines.

Crystallized ability goes the other way. Hartshorne and Germine analysed data from nearly 50,000 online participants alongside standardisation samples from the Wechsler tests, and found that cognitive abilities do not peak together at all. They rise and fall asynchronously, with peaks ranging from the late teens to the forties or later. In their web data, vocabulary peaked at around 65, later than the roughly 50 in the older WAIS-III norms — a difference the authors attribute partly to generational change.

How tests measure each

The distinction is not merely theoretical. It is wired into the structure of every major intelligence battery.

• The WAIS-5 reports a Fluid Reasoning Index alongside a Verbal Comprehension Index. The first is close to Gf; the second is close to Gc.
• Raven's Progressive Matrices is a near-pure fluid measure: it contains no words and no arithmetic, and keeps cultural content to a minimum.
• Vocabulary and general-information subtests are near-pure crystallized measures. They ask what you know. See verbal reasoning for why vocabulary holds up longer with age than almost any other measure.
• Working memory and processing speed sit underneath fluid reasoning and are usually reported separately.

This has a practical consequence for anyone comparing scores. A short online matrix test mainly measures Gf. A full-scale IQ aggregates both. Someone whose strengths are verbal and knowledge-based will score lower on the first than on the second, and neither number is wrong.

Can fluid intelligence be increased?

The commercial answer is yes. The evidential answer is that nobody has convincingly shown it.

Working-memory training became the flagship claim, on the reasonable premise that working memory sits close to fluid reasoning. Trainees do reliably improve at the trained task, and often at closely similar tasks. What has repeatedly failed to appear is far transfer — improvement on measures of fluid intelligence that share no surface features with the training.

Melby-Lervåg, Redick and Hulme's meta-analytic review found no convincing evidence that working-memory training improves performance on tests of intelligence or other far-transfer measures. Getting better at n-back makes you better at n-back.

The Flynn effect is mostly a fluid effect

Scores on IQ tests rose substantially across the twentieth century. The gains were not evenly spread.

They were largest on the most fluid, most abstract measures — Raven's Matrices above all — and considerably smaller on vocabulary and general knowledge. A meta-analysis of 271 samples put the average gain at about 0.41 IQ points a year on fluid tests against 0.21 on crystallized ones. Flynn noted that learned content had held gains back rather than driving them, and that extra schooling explained little of the Dutch rise.

The pattern is a genuine puzzle, and it is part of why Flynn himself doubted the gains represented a straightforward rise in general intelligence. Something changed about how readily modern people take up abstract, hypothetical, classificatory reasoning. What that something is remains argued over.

Myths and facts