What Intelligence Tests Try to Measure

Give someone a vocabulary test, a spatial puzzle, and a math problem, and their scores tend to rise and fall together. Someone who's good at one is statistically more likely to be good at the others, even though the tasks look unrelated. That pattern is the starting clue for general intelligence, or g: the idea that one underlying mental ability contributes to performance across many different kinds of cognitive tasks.

Charles Spearman pulled g out of test data using factor analysis, a statistical method that looks at a big tangle of correlations between many different tests and asks what common factor could explain why they all move together. If scores on ten unrelated tests all correlate positively, factor analysis can extract a single shared factor driving much of that overlap, rather than treating each test as measuring something totally separate. This is the main argument for one general ability over a collection of fully independent abilities: the positive correlations themselves are the evidence, and specific abilities still exist on top of g but don't erase the common thread. An IQ score, though, is not a measurement of a fixed quantity sitting inside someone's skull. It's a snapshot of performance on a particular test, compared to a norm group, shaped by motivation, culture, practice, and the test's content.

Intelligence also isn't one flat number across a lifetime. Crystallized intelligence, accumulated facts, vocabulary, and learned skills, tends to hold steady or grow through adulthood. Fluid intelligence, the ability to reason through brand-new problems quickly without relying on prior knowledge, tends to peak earlier and decline somewhat with age. This split connects general intelligence theory to development: the same person's g-related abilities can shift in different directions as crystallized knowledge keeps building while fluid processing speed slows.