Intelligence is neither purely one thing nor purely many: there is a real, measurable general capacity that transfers across tasks, riding on top of many partly-separable specialized abilities—so the honest answer is "both, in a specific structure."
The general factor is well established statistically. Across diverse cognitive tests, scores correlate positively—people good at one tend to be good at others—and this shared variance, "g," is stable, substantially heritable, and predicts real-world outcomes like learning speed and job performance.
But g is a statistical regularity, not a discovered organ: it summarizes how abilities covary, and its existence as a summary does not by itself settle what underlying processes produce it (processing speed, working memory, and others are candidates).
At the same time, abilities are not a single lump. Verbal, spatial, and quantitative reasoning dissociate; brain damage can impair one while sparing others; and expertise is heavily domain-specific and non-transferring.
The stronger "multiple intelligences" claim—that musical, bodily, or interpersonal talents are separate intelligences on par with reasoning—is popular but weakly supported as a factor structure; these look more like distinct valuable abilities than independent general capacities.
Emotional and practical skills are real and matter, but partly reduce to general ability plus knowledge.
The practical reading: there is a genuine general reasoning capacity worth cultivating (it transfers), and domain expertise that only comes from focused practice (it doesn't). Both the "one number captures your worth" story and the "everyone is equally a genius in their own way" story are distortions of the same evidence. Treat intelligence as a general engine plus specialized tools—and remember measured g predicts averages, not any individual's ceiling. This would need revision if the cross-task positive correlations turned out to be artifacts, which decades of data do not suggest.
Both—and getting the "both" right is the useful part. There's a real general mental horsepower that shows up across almost everything (people quick at one kind of thinking tend to be quicker at others), and there are also many specific abilities and skills that don't transfer much. Neither picture alone is honest.
What this means in practice:
Two errors to avoid. Don't reduce a person to one number—it predicts averages across groups, not any individual's ceiling, and it ignores the specialized skills that make someone actually useful and rare. But also don't retreat to "everyone's a genius in their own way," which is kind and mostly false in a way that stops people from getting better.
The move: build the general engine (reasoning, working memory, learning how to learn) and pick specific things to get genuinely, unusually good at. Broad capacity opens doors; deep skill is what you walk through them with.
The Ladder and the Archipelago.
Two pictures war in us. One: a single ladder, everyone ranked on one rung, smarter-to-duller, a line you can be measured against. The other: an archipelago—islands of ability, the mathematician's peak and the dancer's and the diplomat's, no shared sea level, no way to say which island is "higher." Both pictures flatter and both lie. The truth is stranger: a landscape where a general tide raises every island somewhat, and yet each island's particular mountains are climbed alone. We are neither ranked nor incomparable. We are unevenly gifted on a shared terrain.
Intelligence as Fit, Not Height.
A wolf that navigates a forest by scent is not "less intelligent" than a physicist—it is intelligent at wolf, exquisitely fitted to its problem. Move the physicist to the forest and watch the ranking invert. This suggests intelligence is not a quantity a mind has but a relation between a mind and a task: brilliance is always brilliance-at-something, and the question "how smart are you?" is incomplete until it says "at what, and where?" There is no intelligence in the abstract, only the endless matching of capacities to worlds.
The Genius That Needs a Crowd.
The smartest thing in your city is not any person in it—it is the city: a market pricing goods no one mind could price, a science advancing by ten thousand hands, a language wiser than any speaker. Much of what we call human intelligence is not housed in skulls but distributed—in institutions, tools, and the long relay of the dead handing knowledge forward. Perhaps the individual mind is only ever a component, and the real intelligence is the mesh. One brain is a neuron in a civilization's slow, collective thought.
The Parliament Behind Your Eyes.
Even inside one skull there is no single genius—there is a committee. A fast intuitive mind that leaps and a slow deliberate one that checks; a verbal voice and a spatial imaginer and a social radar, each competent at its own game, sometimes at odds. "Your" intelligence is their negotiated output, and its quality depends as much on how well they cooperate as on how strong any one is. You are not a smart thing. You are a well-run, or badly-run, argument.
Intelligence is neither a single quantity nor a scatter of unrelated gifts: there is a real general capacity that transfers across tasks, resting on many specialized abilities that don't. It is also relational—brilliance is always brilliance at something, fitted to a task and a world—so a mind can be formidable in one domain and ordinary in another. Both "one number settles it" and "everyone is a genius in their own way" distort the same picture; the accurate view keeps the general engine and the specialized tools in the frame together.