To find out, at Clearer Thinking, we ran a massive study.

We tested thousands of people performing random subsets of 62 diverse cognitive tasks (vocab, math, logic, pattern recognition, reaction time, games, memorization, mental rotation, language learning, etc.)

We successfully replicated a classic finding: performance on nearly all cognitive tasks correlates positively with performance on the other tasks—a phenomenon known as the “positive manifold,” foundational to IQ.

IQ scores explained ~45% of variation across our diverse cognitive tasks, aligning with previous research. That’s very substantial for a single number (IQ score), but also far from capturing everything.

Some of the remaining 55% variance is pure noise; the rest likely comes from task skill (developed through practice) and task-specific aptitudes (likely influenced by both genetics and childhood experiences).

It’s also worth noting that while IQ is predictive of a diverse range of intelligence tasks, it doesn’t necessarily capture ALL that’s meant by intelligence. It’s unclear if it includes “street smarts,” social skills, or deep nature skills (like hunter-gatherers have).

We confirmed IQ predicts interesting outcomes:

• Actively open-minded thinking (r=0.43)

• Household income (weakly, r=0.15)

• Self-reported job performance (for lower IQ ranges only, r=0.48)

• Celebrity worship (substantially negatively correlated with IQ, r=-0.42)

• No link to happiness or life satisfaction.

IQ captures something—but WHAT? There’s no consensus. Theories include that IQ is…

• A single cognitive resource

• Working memory + executive control

• A measure of brain integration

• The result of overlapping cognitive processes

• The top of a hierarchy of abilities

Guess what’s a BETTER predictor of major life outcomes than IQ?

The Big Five personality traits.

In our data, openness, conscientiousness, extraversion, agreeableness, and neuroticism (when all 5 are used together) outperformed IQ at predicting most outcomes.

IQ clearly matters—it unavoidably jumps out from the data when testing people on a diverse range of cognitive tasks.

And yet…

IQ is far from being all that matters and is definitely not destiny.

We used the data from our study to create a new cognitive assessment, which analyzes ability across 7 dimensions and aims to provide you with useful insights about your mind.

If you’d like to take it, you can do so here:

https://programs.clearerthinking.org/cognitive-test-intro.html

(Proceeds support our mission.)

This piece was first written on March 25, 2025, and first appeared on my website on May 15, 2025.

Below I list all the different theories and interpretations of IQ that I’m aware of.

Theory 1: Multiple Intelligences

IQ tests measure just a limited part of intelligence while missing (or at least mostly missing) other important parts of intelligence.

For instance, IQ tests may do an adequate job at measuring these forms of intelligence:

• linguistic/verbal
• mathematical/logical
• visual/spatial

But they don’t substantially capture other parts of intelligence (e.g., social skills/interpersonal, self-reflection/introspection, bodily/kinesthetic like dancers might have, naturalistic like hunters and farmers might have, existential/spiritual, etc.).

We can use knowledge of neurology, evolution, and so on to help us identify the distinct types of intelligence that exist in humans.

The claims made about IQ predictiveness, Heritability, and so on may well be true, but IQ is still missing a lot because of its narrow scope.

This theory was popularized by Howard Gardner.

—

Theory 2: Hierarchical Factors

Intelligence tasks form a hierarchy. At the top is the general factor, g, which is what IQ tests seek to measure when they give just one number (IQ) as an output.

But g can be subdivided into “broad” components (e.g., fluid intelligence, crystallized intelligence, and perhaps others like processing speed, motor speed, sensory processing, reading/writing, and memory).

These components are positively correlated with each other but distinct, and they all feed into g.

These “broad” parts can be further subdivided into “narrow” components that reflect specialized abilities.

For instance, if “reading/writing” is a broad part, then it might contain narrow components within it, like “word decoding,” “reading comprehension,” “reading speed,” and so on, which a person may well be able to practice and get better at. Each of these narrow parts is mainly related to just one broad part, which is what determines the hierarchical structure.

Your ability at a broad part is determined by your ability at each narrow part that composes it, and your overall intelligence (g) is determined by your combined ability at the broad components.

Hence, IQ seeks to measure g, the most important single factor in intelligence. It is a mathematical approximation that attempts to be the single number that BEST reflects intelligence if you are forced to represent intelligence as just one number.

However, intelligence cannot be captured fully in a single number. So while a well-measured IQ score might capture something like 30%-70% of what intelligence is overall and thus serves as a useful approximation, we must introduce more numbers to capture the rest (e.g., one per broad factor). For an even more accurate model that captures just about all of intelligence, we’d need to measure the narrow factors as well.

This is my attempt to explain Cattell–Horn–Carroll theory in simple terms.

—

Theory 3: Independent Functions

There are at least a few dozen (and perhaps hundreds) of distinct functions and capabilities that the mind can perform, and people can be better or worse at each of these.

For instance, “keeping numbers in your working memory,” “noticing a pattern in an image,” and “rotating an object in your mind’s eye” might be three distinct capabilities, and there is nothing that prevents a person from being good at any of these while being bad at any others.

That being said, there tend to be positive correlations between capabilities at these functions due to facts about the world (e.g., in childhood, if you experience lead poisoning, lack of verbal communication from parents, or malnutrition, it may reduce ability at a number of different functions simultaneously). Some functions may depend on other functions as well for their operation.

It also seems that genes influence one’s aptitude at these different functions. The contribution of genes to these functions can also increase the correlation between them (genes can be linked, which increases correlation between functions), as well as explaining the high IQ heritability numbers that intelligence researchers estimate.

Another factor that may explain high correlations between different IQ tests is that many of them capture multiple functions simultaneously. If IQ test A measures functions 1 through 6, and IQ test B measures functions 3 through 8, then, of course, the scores from tests A and B will be significantly correlated because there is overlap in what functions they measure.

What IQ is capturing is something like an average capability across many of the most often used of these functions. So IQ can be interpreted as saying something like, “how good are your many mental functions, on average.” Some IQ questions work better than others because they tap more of these functions simultaneously or because they tap more frequently used functions rather than ones that have little relevance to life.

However, this means IQ is fundamentally arbitrary because there is a subjective choice of how much to weight each of the many different mental functions when calculating a person’s average. In practice, this arbitrariness problem is lessened (though not eliminated) by the positive correlations between most mental functions. Someone could have a relatively low IQ score but be incredibly good at some of the mental functions, especially if they are ones that particular IQ test doesn’t involve.

Overall, since there are so many mental functions, it is silly to think that a single number like IQ could capture a person’s full intelligence well. Someone with a higher IQ will tend to be better at a number of mental functions, yet not all of them.

Individual mental functions are almost all trainable with practice (e.g., with practice, you can likely improve at rotating objects in your mind’s eye or at keeping numbers in working memory). But to make your “IQ” go up appreciably, as measured by IQ tests that have a diverse range of questions, you’d likely have to train in many different mental functions, which explains why, in practice, IQ (broadly speaking) is hard to increase. 

If you improve at a few mental functions, that still won’t budge your IQ on IQ tests that measure different mental functions—and improving at one intelligence function generally won’t increase your ability at other functions unless they happen to rely on each other.

Many of the tasks we do with our minds apply our mental function along with the specific skills and knowledge we’ve trained in. For instance, doing mental algebra requires the learned skill of how to solve algebra problems, certain memorized knowledge (such as that (1/2)*(2)=1), and certain intelligence functions like keeping numbers in working memory. 

These capabilities at the low-level mental functions are combined with memorized knowledge and trained skill to produce a person’s total skill level at solving algebra problems (but solving algebra problems is not, by itself, an intelligence function).

—

Theory 4: Impairment

High IQ doesn’t measure anything important, but low IQ does measure cognitive impairment (e.g., mental retardation, some kinds of brain damage, etc.)

What IQ actually measures is something like “the ability to do school-like tasks,” which matter very little in the real world outside of school and outside of “school-like” jobs.

However, people with substantial cognitive impairment will struggle to be able to do these sorts of tasks, hence why IQ becomes a meaningful measure on the low end of IQ.

The most famous proponent of this view is Nassim Taleb. In my view, it is pretty strongly contradicted by the empirical evidence.

—

Theory 5: IQ is all of intelligence

The general intelligence factor, g, measures most of what matters when it comes to intelligence (say, at least 75% of what we might reasonably call “intelligence”). IQ tests simply attempt to measure g, and they can be better or worse (e.g., depending on their level of accuracy/noise).

Since g captures the substantial majority of intelligence, and because a well-made IQ test accurately measures g, loosely speaking, it’s reasonable to speak of IQ as simply being a person’s “intelligence level.” Any aspects of “intelligence” this way of thinking misses are minor, and they don’t have much effect on outcomes anyway.

There are other factors that may matter significantly for a person’s success (like tendency to work hard, curiosity, and ability to read emotions on someone’s face), but these are simply not part of intelligence – these are non-intelligence-based skills.

My view is that IQ definitely doesn’t capture everything that matters when it comes to what we mean by intelligence broadly (and that we have a lot of evidence against the idea that IQ is all that matters).

—

Theory 6: Dominant Component

There are a bunch of different parts of human intelligence, but there is one bigger one, which is “g”, or general intelligence. The other aspects of intelligence are pretty much uncorrelated with g, and g accounts for substantially more of what we think of as “intelligence” than any one of the other factors does.

Most intelligence questions that might be found on an IQ test can be thought of as partly tapping this g factor, and partly tapping one or more other (non-g) aspects of intelligence. 

What IQ tests are primarily trying to do is measure this factor g, which is one big part of intelligence. If you look at what many different types of intelligence questions have in common, it is precisely this g factor, and it can be extracted statistically using statistical methods (e.g., principal component analysis). 

One could design tests to measure the other parts of intelligence too (beyond g), and many IQ tests in fact try to do that as well. These other aspects of intelligence may be very important for some specific intelligence-related skills, even though most of the time they are not as important as g.

This theory differs from theory 6 because it acknowledges the meaningfulness of other aspects of intelligence, and it differs from theory 3 because it claims that g is mostly independent from the other, more narrow aspects of intelligence, whereas theory 3 says that g is actually composed of the more narrow aspects.

—

Theory 7: Bullshit

IQ doesn’t measure anything that matters about a person’s mind. When differences in IQ scores are found to predict outcomes, insofar as those relationships are real, they may be the result of IQ tests measuring class (e.g., by using vocabulary words that are linked to class). Or these results may be due to confounding variables like education or income (e.g., elite schools might teach information that is then regurgitated on IQ tests, and kids from higher-income families may be trained for IQ-like tests, such as college admission exams).

The apparent hereditary nature of IQ is either just the result of badly performed studies or because of unfair and unfortunate links between class and ethnic group. In any given country at any particular time, some ethnic groups may be in a worse position and others in a better position due to factors like immigration patterns, prejudice, institutional racism, and so on.

IQ tests do not measure true intelligence. Their usage tends to reinforce existing power structures.

In my view, while this view is not totally meritless (prejudice and test bias are real, problematic factors), its overall conclusion is strongly contradicted by the empirical evidence.

—

Theory 8: Hardware

IQ is an approximation of general intelligence (g). An IQ test is analogous to measuring the performance of the CPU of a computer by running it through a battery of performance tests. Some computers have faster processors just like some people have higher g factors. 

Of course, a computer that has a faster processor is not better at doing everything than a slower computer, because software also matters a great deal. A faster computer with no accounting software installed won’t be able to help you file your taxes. More generally, better software on a slower processor may well outperform worse software on a faster computer. In this analogy, software corresponds to the skills and knowledge we learn. The processing speed on a computer matters, but it is very far from ALL that matters.

So, according to this perspective, some people have brains that have more g (corresponding in the analogy to computers with a faster CPU). This would map to physical neurological parameters, perhaps ones like rate of information transfer, neural density, level of connectivity, and so on. To a significant degree, these neurological characteristics appear to be set by genetics (perhaps 45% to 75% of the variance in g seems to be attributable to genetics rather than environment). But g can also be severely reduced by environmental factors (such as poisoning, brain injury, aging, disease, and malnutrition). As of now, the best way known to make your g higher seems to be to avoid all environmental factors that are known to reduce g.

Some people start at an advantage and others at a disadvantage when it comes to intelligence. Empirically, IQ seems to roughly follow a normal distribution in the population. Unfortunately, the width of this distribution is wide enough that, on average, those in the bottom 20th percentile have meaningfully different experiences than those in the top 20th percentile for life outcomes like income, educational attainment, and job performance (as rated by supervisors). Traits like conscientiousness and hard work also matter substantially when it comes to life outcomes, but g is a factor that is as big, and perhaps even bigger, depending on the outcome.

Of course, this doesn’t mean that people with higher g will be “smarter” at all tasks than those with lower g. Learning and practice give a person a tremendous advantage over someone who knows less or has practiced less (a 50th percentile IQ experienced chess player will beat a 90th percentile IQ person who has just started learning chess just about every time). But someone with higher g will have an advantage at almost all intelligence-related tasks when all else is held equal (i.e., when comparing people with equal knowledge, equal practice, equal experience, etc.) 

—

As you can see, there are many different theories about the meaning of IQ. Which of these is most accurate? It’s obviously a very complex question, and reasonable people can certainly disagree.

Anecdotally, I’ve been noticing a seeming increase in people posting on social media who appear to adhere either to “IQ is bullshit” or “IQ is all of intelligence”. I find this seeming polarization into nearly opposite extreme views concerning it, especially given that these two extreme views are strongly contradicted by the empirical evidence. 

This piece was first written on October 20, 2020, and first appeared on my website on August 17, 2026.

There are difficult questions like the blue eyes logic puzzle, which require strong thinking skills to solve. And then there are TRULY difficult questions, that require an extremely honed thinking process, like:

• How can I make so much money in the next five years that I never have to work again?
• Which job should I choose right now to maximize my average lifetime happiness?
• Which charities help the world the most per thousand dollars donated to them?
• What can I do to change our society into a significantly more ideal place?
• What is the greatest threat to humanity as a species, and what can we do now to reduce this threat?

Notice that these important questions are somewhat ambiguous. They also can’t be addressed without gathering and combining information from many sources, and involve factors that are extremely hard to predict. And they can’t be solved unless you can do a reasonable job of estimating difficult to estimate quantities. They all require truly excellent thinking to tackle. Truth discernment becomes a super power when it leaves the realm of puzzles and enters the realm of extremely valuable real world questions.

At high levels, truth discernment ability is not just IQ, though surely is somewhat correlated with it. It is not just the ability to apply logic, though logic is an important component. It also requires certain habits of mind, like regularly asking yourself what the evidence is for your own beliefs. It requires using techniques, like thinking in terms of probabilities rather than thinking in black and white. It requires caring more about getting the right answer than about seeming right or believing what it is convenient psychologically to believe.

The exciting thing about truth discernment is that it is not just a genetic ability (though genetics is surely a contributing factor). It is an ability that we can improve a great deal if we take the time to hone our thinking. We can familiarize ourselves with the logical fallacies to the point where our brain notices them occurring in real-time. We can learn the cognitive biases so that we know the ways our brains are likely to fail us and can consciously correct for these failures. We can learn to prevent unhelpful emotions from derailing our logic too often. We can practice our reasoning by drilling in LSAT questions. And we can learn the methods of rationality, or at least read about another person using them.

Many comic book writers have realized that intelligence is a legitimate super power. Lex Luthor was superman’s nemesis despite being physically inferior to him in every capacity. He could still compete with superman only because of the power of his thinking. But it is not a very high level of intelligence, per se, that is so powerful. Being able to solve the hardest logic puzzles won’t automatically let you solve problems that matter. Truth discernment is the essential skill, the ability to figure out the likely answer to truth claims regarding important real world questions. Will taking this action bring me more happiness in the long-term than not taking this action? Will this regulation, if it passes, improve poverty? Will this business venture make large quantities of money? Is string theory an accurate description of reality?

Being able to accurately discern the truth can be a super power, and with study and practice, it is an ability that we can very substantially improve.

Influences: Eliezer Yudkowsky