The g Factor: What 100 Years of Intelligence Research Actually Proves

The most replicated finding in a century of intelligence research is surprisingly simple: people who perform well on one type of mental test tend to perform well on all types of mental tests. This observation — now supported by hundreds of independent studies across dozens of cultures — is the foundation of what scientists call the general factor of intelligence, or g. Understanding what g actually is, what it predicts, and where it falls short explains both why IQ scores remain scientifically useful and why they are an incomplete picture of human capability.

What the g Factor Actually Is

When researchers administer batteries of cognitive tests — measuring verbal reasoning, spatial ability, working memory, processing speed, numerical reasoning, and other mental capacities — they consistently find that the scores are positively correlated. Someone who scores in the top quarter on verbal reasoning is more likely than chance to score in the top quarter on spatial rotation, on working memory tasks, and on processing speed assessments.

The statistical technique used to identify this pattern is called factor analysis, and it extracts from these correlations a single underlying dimension along which most of the variation in test performance can be explained. That dimension is g. It is not a specific mental skill — it does not represent vocabulary size or mathematical knowledge or spatial visualization ability alone. It represents something that underlies performance across all of these, something that cognitive scientists interpret as the general efficiency or capacity of information processing in the brain.

ScienceBlog.com noted in 2026 that despite ongoing debates about what intelligence truly is, this pattern of positive correlation among mental tests "is the most replicated finding in a century of research." The consistency across cultures, testing instruments, and research groups is what makes g one of the most solid empirical findings in psychology.

What IQ Tests Actually Measure

Standard IQ tests are designed to estimate g. They do this by sampling performance across several different types of cognitive tasks, then aggregating those scores. A high IQ score means that across the sampled tasks, performance was consistently above average — which, given the positive correlations among tasks, suggests a high level of g.

The IQ score itself is a standardized number with a population mean of 100 and a standard deviation of 15. A score of 115 means the test-taker performed better than approximately 84 percent of the reference population. A score of 130 means better than approximately 97.5 percent. These numbers are meaningful precisely because g is stable across testing instruments and correlates predictably with outcomes measured years later.

What IQ Predicts

The predictive validity of IQ scores is substantial and well-documented. Higher IQ scores correlate with higher academic achievement, higher occupational attainment, higher earned income, better health outcomes on average, and lower rates of involvement in the criminal justice system. These correlations hold even after controlling for socioeconomic factors, though socioeconomic conditions influence IQ development — a distinction that matters enormously for policy.

What IQ Does Not Predict

Here is where the picture becomes more complicated. High IQ does not protect against cognitive biases, partisan motivated reasoning, confirmation bias, or reactive thinking. Research published in 2026 confirms what earlier work established: a high g score predicts ability to reason when motivated to reason correctly, but not the tendency to apply that ability in situations where motivated reasoning offers a shortcut. People with high IQ scores hold wrong beliefs, make poor decisions, and fall for emotionally appealing but logically flawed arguments — sometimes at higher rates than average, because their cognitive ability makes them better at rationalizing what they already believe.

The Growing Case for Thinking Style Over IQ Score

Travis Mark, writing for Forbes in February 2026, made the case that how you tend to think — your characteristic approach to problems — predicts real-world success more reliably than your IQ score. The research behind this argument identifies what psychologists call "thinking dispositions": traits like actively open-minded thinking, the tendency to seek disconfirming evidence, comfort with uncertainty, and the practice of updating beliefs when evidence changes.

These dispositions are separable from raw cognitive ability. Someone can have the processing power to reason clearly and yet habitually avoid the kinds of thinking that produce accurate beliefs. Conversely, someone of average IQ who has cultivated strong thinking dispositions may navigate complex decisions more reliably than a high-IQ person who leans on motivated reasoning.

Critical Thinking as a Distinct Skill

Researchers at Lincoln Park Therapy Group summarized the evidence this way: while IQ correlated with academic success, critical thinking skills had a stronger connection to real-life decision-making and problem-solving. Critical thinking encompasses skills that are neither captured by g nor automatically associated with it: identifying logical fallacies, evaluating source credibility, distinguishing between correlation and causation, recognizing when a question requires evidence rather than opinion.

These skills can be taught and practiced, which is part of their appeal. Unlike g, which is largely stable after early adulthood, thinking dispositions and critical thinking skills are amenable to deliberate development across the lifespan.

The Question of Whether Global IQ Is Declining

The Quantum Record documented in 2026 the emerging concern about global IQ trends and what has been called the "paradox of AI-assisted thinking." Several countries that showed rising IQ scores through most of the 20th century — a phenomenon called the Flynn Effect — have begun to show score reversals in recent decades. Researchers propose multiple explanations, including changes in education, nutrition, environmental factors, and the cognitive demands of modern information environments.

The AI concern is more recent. If individuals increasingly outsource the cognitive work of reasoning, calculating, navigating, and remembering to AI tools, the argument goes, those individuals may exercise the neural pathways underlying g less frequently — similar to how a previously literate population might show declining reading fluency if reading was replaced by listening. This remains a hypothesis rather than established finding, but it is being actively investigated.

What This Means for Personal Cognitive Maintenance

The Flynn Effect itself was driven in part by environmental changes that provided richer cognitive stimulation — more education, more complex media, more problem-solving in daily life. If the AI-assisted environment removes cognitive challenge from daily life, similar dynamics might work in reverse. Deliberately seeking out cognitively demanding activities, maintaining the practice of working through problems before reaching for AI assistance, and engaging with complex text that requires effort to understand are all defensible cognitive maintenance strategies.

The Skepticism About Single-Number Intelligence

Cognitive scientists have grown increasingly skeptical of treating g as the entirety of intelligence, or of treating a single IQ number as an adequate summary of a person's cognitive capacity. Several frameworks have proposed that intelligence encompasses distinct abilities that g does not fully capture.

Howard Gardner's theory of multiple intelligences identified domains like musical-rhythmic, bodily-kinesthetic, and interpersonal intelligence that sit outside the g measurement space. Robert Sternberg's triarchic theory distinguishes among analytical, creative, and practical intelligence — arguing that real-world success depends heavily on practical intelligence (the ability to apply knowledge to actual situations), which IQ tests do not measure well.

Emotional intelligence — the ability to accurately perceive, understand, and regulate emotions in oneself and others — has attracted substantial research interest as a component of human capability that is separate from g and that predicts important outcomes including relationship quality, leadership effectiveness, and mental health resilience. Recent research shows emotional intelligence peaks in the mid-40s and continues to develop through lived experience, unlike g which stabilizes in early adulthood.

What the Field Actually Agrees On

Despite these disputes, the scientific consensus holds several positions clearly. g exists as a real, measurable dimension that explains meaningful variance in cognitive performance. IQ tests measure g reasonably well. IQ scores predict important life outcomes with genuine validity. And IQ scores are nevertheless incomplete: they miss dimensions of human capability that matter, and they measure a potential for performance that other factors — motivation, opportunity, character, thinking dispositions — translate into actual outcomes. That is why the Real World IQ test assesses dispositions and applied reasoning alongside raw processing — for a more complete picture of where you actually stand.

The Real World IQ Framework

The concept of a "real world IQ" attempts to bridge the gap between laboratory cognitive measurement and the messy, judgment-intensive demands of actual life. Real-world problem solving requires not just the processing efficiency that g represents, but also accurate knowledge, the ability to recognize relevant information, comfort with ambiguity, ethical reasoning, and the willingness to act on conclusions rather than endlessly defer.

A person might score at the 50th percentile on a standard IQ test and yet demonstrate exceptional real-world judgment because they have developed strong thinking dispositions, deep domain expertise, and effective decision habits. Another person scoring at the 99th percentile might make poor life decisions because their cognitive power is deployed in service of motivated reasoning and confirmation of existing beliefs. The score is the starting line, not the finish line. To see where your real-world cognitive performance actually lands across reasoning, dispositions, and applied judgment, take the Real World IQ test.

FAQ

Is IQ fixed, or can it change?

IQ scores are relatively stable from about age 10 onward for most people, meaning a person who scores high at 10 is likely to score high at 40. However, environmental factors — nutrition, education, stimulation, stress, and health — influence IQ development in childhood, and scores can be raised meaningfully through interventions in early childhood. For adults, raw g scores are largely stable, but the skills, knowledge, and thinking dispositions built on top of g continue to develop and can improve substantially with deliberate effort.

Do IQ tests measure intelligence fairly across different groups?

This is a genuine and complex scientific question. IQ tests have predictive validity across diverse groups — the scores predict academic and occupational outcomes in documented ways across different demographic groups. But group average score differences exist and are substantially explained by socioeconomic and environmental factors rather than inherent differences. Well-designed tests work to minimize culturally specific content, but no test achieves complete cultural neutrality because all testing occurs in a cultural context.

What is the difference between IQ and EQ?

IQ (intelligence quotient) estimates the general cognitive processing capacity (g) through performance on varied mental tasks. EQ (emotional quotient or emotional intelligence) measures the ability to perceive, understand, and regulate emotions — in oneself and in others. These two dimensions are largely independent: having a high IQ does not predict high EQ, and vice versa. Both predict important outcomes, but in different domains. IQ predicts academic and technical performance more strongly; EQ predicts relationship quality, leadership, and certain aspects of occupational performance.

What does it mean that AI now scores higher than humans on EQ tests?

A 2026 study found that AI language models scored an average of 82 percent on standard emotional intelligence assessments, compared to 56 percent for human participants. This does not mean AI systems experience or process emotions in the way humans do. It means AI systems have learned from enormous training data how to recognize emotionally intelligent language patterns and produce them in response to test questions. The gap between producing emotionally intelligent-sounding text and developing genuine emotional understanding through lived experience is significant.

How does critical thinking relate to IQ?

Critical thinking skills and IQ are correlated but separable. A high IQ provides the cognitive horsepower to reason carefully, but critical thinking dispositions — the tendency to actively seek disconfirming evidence, to question assumptions, to evaluate source credibility — determine whether that horsepower is actually applied in a given situation. Someone can be capable of excellent critical thinking without regularly engaging in it; developing the habit of critical thinking practice strengthens the actual skill regardless of baseline IQ.