Two Kinds of Thinking, One Distinction
Raymond Cattell proposed in the 1940s, and elaborated with John Horn through the 1960s, that what psychometricians had been calling general intelligence was better described as two broad abilities that behave differently.
Crystallised intelligence (Gc) is what you know and the procedures you have learned to apply: vocabulary, general information, acquired reasoning habits, domain knowledge. It is the sediment left by years of exposure and education.
Fluid intelligence (Gf) is the ability to solve a problem you have never seen before, with nothing relevant stored to draw on. It leans on working memory, pattern detection, and the ability to hold and manipulate several relations at once.
The distinction is not philosophical. The two behave differently across the lifespan, respond differently to training, and are best measured by visibly different kinds of test item — which is exactly why it matters when you read a verbal reasoning score.
Verbal Reasoning Is Predominantly Crystallised
Verbal reasoning is the clearest everyday expression of crystallised intelligence. Every component of it depends on stored material:
- Vocabulary items are pure Gc. There is no way to reason your way to the meaning of a word you have never encountered, beyond guessing from morphology — which is itself stored knowledge about how words are built.
- Analogies need both words known and a stored sense of the kinds of relations that hold between words: part-to-whole, tool-to-user, cause-to-effect. That relational repertoire is learned.
- Reading comprehension depends on syntactic fluency and on familiarity with the register — legal, scientific, financial. A reader who has met the conventions of a genre processes it faster and with more capacity left over for reasoning.
There is a genuine fluid component sitting on top. The "cannot say" judgement in a graduate-style verbal item is a reasoning operation, not a retrieval one, and syllogistic items in prose test relational manipulation directly. But strip the stored language away and nothing remains to reason with.
In Carroll's three-stratum synthesis of 1993, verbal comprehension sits under the crystallised broad ability, and the factor loadings have been consistent about this for as long as the data have existed.
What Sits on the Fluid Side
The contrast is sharpest against abstract reasoning. A matrix item — the format John Raven introduced in 1938 — presents a grid of unfamiliar shapes with one cell missing. Nothing you have ever read helps.
The item is deliberately constructed so that stored knowledge confers no advantage, which is why matrix tests have been the workhorse measure of fluid ability for eighty years and why they are used when a culture-reduced measure is wanted.
Spatial reasoning sits mostly on the fluid side too, and numerical reasoning sits awkwardly in between: the arithmetic operations are learned, but the working-memory load of holding intermediate results while scanning a data table is fluid work. This is why the same person's verbal, numerical, and abstract scores can diverge sharply — they are not three views of one thing.
The Lifespan Pattern Is the Strongest Evidence
If the distinction were an artefact of test construction, the two would age together. They do not. The consistent finding across longitudinal work — the Seattle Longitudinal Study being the most-cited example — is that fluid measures peak in early adulthood and decline gradually thereafter, while crystallised measures continue rising well into middle age and hold up much later.
The everyday version of this is familiar to anyone who has watched an experienced colleague work. Vocabulary and the ability to read a dense document accurately are frequently better at fifty than at twenty-five. Raw speed at an unfamiliar puzzle, with everything having to be held in mind at once, generally is not.
For anyone reading their own profile, this has a direct implication: a verbal score is not a fixed personal constant. It is a measure of accumulated exposure, and accumulation continues.
Why the Distinction Changes What You Should Do About a Score
The most practical consequence concerns improvement. Because verbal reasoning is largely crystallised, the input that moves it is available and unglamorous: sustained reading in registers you find difficult. Not more of what you already read comfortably — the point is exposure to unfamiliar syntax and vocabulary.
A month of reading legal judgments or scientific abstracts does more for a verbal score than any amount of item drilling, because it enlarges the store the test is sampling.
Fluid ability is far less obliging. The working-memory training literature has repeatedly found that people get better at the trained task while transfer to untrained reasoning measures is weak or absent. That asymmetry is the single most useful thing the Gf/Gc distinction tells a test-taker: verbal reasoning genuinely responds to effort, and abstract reasoning mostly does not.
The Flynn Effect Reads the Same Way
James Flynn's documentation of generational score gains through the twentieth century fits the pattern and reinforces it.
The gains have been consistently larger on fluid and abstract measures than on vocabulary and general-information measures — the opposite of what would be expected if rising scores simply reflected better schooling, and part of why the effect is usually attributed to changes in how habitually people handle abstraction rather than to knowledge accumulation.
The practical caution: verbal and abstract norms have drifted at different rates, so comparing an old verbal score against a new abstract one is not a clean comparison of two abilities in the same person.
Is Reading Really the Answer?
Largely, yes, with one qualification. Reading volume builds the store, but reading exclusively inside a comfortable register builds it slowly, because the vocabulary and syntax repeat. The exposure that moves a verbal score is the effortful kind, where the sentence has to be re-read.
The second half is retrieval. Encountering a word is weaker than using it. Anything that forces production — writing, argument, explanation — converts recognition into the fluent access that a timed test rewards. A candidate who reads widely but never writes often shows the signature profile: strong on comprehension, slower than expected on vocabulary items under a clock.
Reading a Split Profile
The most informative result is an uneven one, and the Gf/Gc frame is what makes it legible.
- Verbal high, abstract low. A well-stocked store working with modest novel-problem capacity. Typically thrives where domain knowledge compounds and struggles most with genuinely unprecedented problems.
- Abstract high, verbal low. Strong machinery, less material loaded into it — common in people who reason well but have had limited exposure to formal or academic registers, and the profile most likely to improve substantially with reading.
- Both high, numerical low. Usually points at a specific gap in arithmetic fluency rather than at reasoning ability, and it is the most straightforwardly remediable of the three.
None of these is a verdict. The distinction's value is diagnostic: it tells you which of your scores is a measure of what you have accumulated so far, and therefore which one is most worth working on.