Charles Spearman was not supposed to become a psychologist. He spent fifteen years as an officer in the British Army before resigning to study experimental psychology in Leipzig — and by the time he published his landmark paper in 1904, he was already in his forties. That paper, with the unwieldy title "General Intelligence, Objectively Determined and Measured," introduced an idea that has shaped intelligence research for over a century: the g factor.
The idea is deceptively simple. Spearman noticed that children who did well in one school subject tended to do well in others — and not just in related subjects. A child strong in Latin was also, on average, likely to be strong in mathematics, in music pitch discrimination, even in simple sensory tasks like distinguishing the weight of two objects. The correlations were everywhere, and they were all positive.
The positive manifold
This pattern — where performance across seemingly unrelated mental tasks tends to be positively correlated — is now called the positive manifold. It's one of the most robust findings in all of psychology. Give a large group of people a diverse battery of cognitive tests, and you'll almost never find someone who is genuinely good at one type and genuinely bad at all others. The abilities hang together.
Spearman's insight was mathematical. If all these tests correlate positively with each other, perhaps they all tap into some single underlying factor — a general mental energy or capacity that contributes to performance on every cognitive task. He called it g, for "general factor." Each individual test would then measure two things: a bit of g, plus some skill specific to that particular task (which he called s, for "specific").
How do you measure something invisible?
Here's where it gets clever. You can't point to g directly — it isn't a thing in the brain you can weigh or photograph. It's a latent variable, inferred from patterns in data. Spearman developed an early form of what we now call factor analysis to extract it: a statistical technique that takes a big table of correlations and asks, "What's the smallest number of hidden factors that could produce this pattern?"
When you run factor analysis on a diverse battery of cognitive tests, a large first factor consistently emerges — one that loads positively on every test. That first factor is g. Depending on the battery, it typically accounts for somewhere between 40% and 50% of the total variance in test scores. In plain terms: roughly half of why people differ on cognitive tests can be traced to this one general factor.
One century of scrutiny
Spearman's g has been attacked, refined, and re-tested more thoroughly than almost any concept in psychology — and it has survived. Not unchanged, but intact at its core. Critics like L.L. Thurstone argued in the 1930s that intelligence was better described by several independent "primary mental abilities" rather than one general factor. But when researchers analyzed Thurstone's own data more carefully, the abilities he identified were themselves correlated — and a general factor re-emerged from behind them.
Every published factor analysis based on a large and diverse battery of cognitive tests has produced a substantial general factor. The positive manifold is simply a fact about the data. — A common summary in modern psychometric textbooks
Today, the dominant framework — the Cattell–Horn–Carroll theory — places g at the apex of a three-level hierarchy, with broad abilities (fluid reasoning, crystallized knowledge, processing speed, working memory) in the middle, and dozens of narrow skills at the base. g didn't get overthrown; it got a more detailed map of everything beneath it.
What g is — and isn't
It's worth being careful here, because g is widely misunderstood. A few clarifications:
g is a statistical regularity, not a substance. It describes a pattern in how abilities cluster. It does not, by itself, tell you what causes that pattern — whether it's a property of neural efficiency, working memory capacity, or something else entirely. Those are active research questions.
g is about differences between people, not a complete description of one mind. Factor analysis works on variation across a population. Saying someone has "high g" is a statement about how they compare to others, not an inventory of everything their mind can do.
g is not the same as a single IQ score. A good IQ test is g-loaded — it correlates strongly with the general factor — but no single test is a pure measure of g. Every test also taps task-specific skills, cultural knowledge, test-taking familiarity, and momentary factors like fatigue.
When people talk about "IQ," they're usually pointing, however loosely, at g — the well-replicated finding that cognitive abilities tend to cluster together. It's real, it's measurable, and it predicts certain outcomes. But it's one dimension of the mind, captured imperfectly by any single test, and it says nothing about creativity, wisdom, character, or worth.
Spearman, the late-blooming army officer, gave us a tool for seeing a pattern that had always been there in the data. More than a century later, that pattern is still one of the most reliable findings we have about the human mind — and also one of the most frequently overstated. Holding both of those truths at once is the beginning of understanding what intelligence testing can and cannot do.
This article is for educational purposes. Cortextest assessments are not clinical instruments and do not provide a clinical measure of g or IQ. For a formal cognitive assessment, consult a licensed psychologist.