Cognitive Abilities: The Structure Behind a Single IQ Score

What counts as a cognitive ability

A cognitive ability is a mental capacity that can be measured consistently, varies between people, and predicts performance on tasks that need it. That definition does real work: it excludes things that are unstable from day to day, things everyone can do equally, and things that sound impressive but predict nothing.

It also excludes a great deal that gets called an ability in everyday speech. Learning styles, left- versus right-brain thinking and most personality typologies fail one or more of these tests — see neuromyths and left brain vs right brain for why the most popular of them do not survive measurement.

The hierarchy: narrow skills, broad abilities, and g

In 1904 Charles Spearman noticed something that still shapes the field: children's marks across unrelated school subjects were all positively correlated. Good performance on one mental task predicted good performance on another, however different the two looked. This pattern is called the positive manifold, and it is the central empirical fact about cognitive abilities.

Spearman's explanation was a single general factor, g, plus a specific factor for each task. That turned out to be too simple — abilities cluster into families before they converge on g — but the positive manifold itself has never gone away. Every large battery of mental tests, in every decade since, has reproduced it.

The structure that resolved the argument came from John Carroll, who in 1993 reanalysed more than 460 datasets of cognitive test scores using a consistent method. His three-stratum model describes what he found: a very large number of narrow abilities at the bottom, a small number of broad abilities in the middle, and a single general factor at the top.

Carroll's model was later merged with Raymond Cattell and John Horn's work on fluid and crystallized intelligence to produce the Cattell–Horn–Carroll (CHC) framework, integrated by Kevin McGrew in 2009. CHC is now the taxonomy most modern test batteries are built around, including the Stanford–Binet and, in large part, the WAIS.

The broad abilities, one at a time

These are the abilities most often reported separately on a test result, and the ones IQCognify covers in depth. Each links to a full guide. Two entries below are not broad abilities in their own right — quantitative reasoning is a narrow ability nested under fluid reasoning, and executive function sits slightly apart from the CHC scheme altogether — and each is flagged where it appears, because the distinction changes how a subscore should be read.

Fluid reasoning

Solving problems you have not seen before, using no specific prior knowledge — spotting the rule in a sequence, completing a matrix, drawing an inference from unfamiliar premises. It is the ability most closely identified with g, and the one that declines earliest with age. See fluid vs crystallized intelligence and pattern recognition.

Quantitative reasoning (a narrow ability, not a broad one)

Reasoning with numbers and mathematical relations — extending a proportion, inverting a percentage, converting a duration into a rate. CHC lists it as a narrow ability under fluid reasoning rather than as a broad ability of its own, which is why it appears here directly beneath it. It is distinct from quantitative knowledge, the acquired store of mathematics built by schooling; the two are easy to confuse because nearly every numerical item draws on both. See numerical reasoning.

Crystallized ability and verbal reasoning

The knowledge and verbal skill you have accumulated, and your ability to reason with it. Measured largely through vocabulary and verbal analogies. It is unusually stable — stable enough that clinicians use vocabulary to estimate someone's pre-illness ability level. See verbal reasoning.

Visual-spatial ability

Representing and transforming objects in space: mental rotation, folding a flat net into a solid, reading a cross-section. It predicts achievement in science and engineering beyond what verbal and mathematical scores capture, and almost no admissions test measures it. See spatial reasoning.

Working memory

Holding information in mind while doing something with it — not simply storing it. It is strongly related to fluid reasoning without being identical to it. See working memory.

Processing speed

How quickly you carry out simple, well-understood cognitive operations accurately. It peaks earlier than any other ability and is often the first to show age-related change. See processing speed.

Executive function

Inhibition, updating and shifting — the control processes that direct the others. It sits slightly apart from the CHC broad abilities and is measured with notoriously disagreeable instruments: laboratory tasks and rating scales of the same construct correlate only weakly. See executive function.

How much do the separate scores actually tell you?

This is the part most score reports skip. Broad abilities are real and distinguishable, but when a test's own standardisation data are analysed, the general factor dominates and the individual indices add surprisingly little on top of it.

The practical consequence: treat a profile of ability scores as a rough picture of relative strengths, not as six precise measurements. A gap of a few points between two indices is usually measurement error, not a discovered talent. This is doubly true of short online tests, where each sub-score rests on a handful of items.

It does not follow that the separate abilities are fictions. Spatial ability in particular predicts real outcomes that the overall score misses. It follows that the difference between two of your own sub-scores is a much weaker signal than either score on its own.

Abilities do not age together

If cognitive abilities were simply facets of one thing, they would rise and fall together. They do not, and this is one of the strongest pieces of evidence that the distinctions are worth drawing.

Because tests are normed within age bands, this divergence is mostly invisible in the score you receive: a 60-year-old is compared with other 60-year-olds. The raw performance changes; the standardised score need not. Average IQ by age explains why an 'IQ by age' table is a misunderstanding of how scoring works.

What IQCognify measures

The IQCognify test samples several of these abilities with dedicated items and reports them as a profile alongside your overall score. The strands are measured directly rather than inferred from unrelated questions, and every strand's score comes from accuracy on its own items.

It is a short test, which sets a hard limit on what the profile can bear. Each strand rests on a small number of items and carries considerably more measurement error than the overall score — so the profile is best read as a rank order of relative strengths, not as a set of separate IQ scores. Our methodology sets out exactly how the test is built and scored, including what response time does and does not affect.

If you want the underlying structure rather than a score, how IQ tests work covers norming and reliability, and psychometrics covers why a test is never 'valid' in the abstract — only an interpretation of it can be.

Myths and facts about cognitive abilities