Intelligence is one of the most studied and most argued-about traits in all of psychology, yet its definition remains surprisingly slippery. The closest thing researchers have to a consensus centers on what is called the g factor, a statistical pattern showing that people who do well on one kind of cognitive test tend to do well on most others. That single observation, replicated for over a century, anchors nearly everything scientists claim to know about cognitive ability. But how the brain produces it, how much of it is inherited versus shaped by environment, and whether it can be meaningfully improved are all questions with answers that are more complicated, and more interesting, than popular accounts suggest.
What the g Factor Actually Tells Us
If you give a large group of people a battery of very different mental tests, something odd happens: scores on virtually all of them are positively correlated. People who score high on vocabulary tests tend to score above average on spatial reasoning, mental arithmetic, and memory tasks too. The g factor is the statistical summary of those correlations. It is highly stable across different statistical methods, different test batteries, and different populations.1PubMed. The g factor: psychometrics and biology
That stability does not mean any single test is a perfect stand-in for g. Raven’s Progressive Matrices, widely considered one of the best “pure” measures of general intelligence, actually shares only about half its variance with g. Roughly a quarter of what Raven’s measures is specific to the test itself and has nothing to do with general cognitive ability.2Intelligence. Raven’s is not a pure measure of general intelligence: Implications for g factor theory and the brief measurement of g This is an important point for anyone interpreting a single IQ subtest score as though it were a definitive readout of how smart someone is. It isn’t. The g factor emerges from the pattern across many tests; individual tests are only partially g-loaded.3PubMed. Distributed brain sites for the g-factor of intelligence
How Intelligence Changes Across a Lifetime
Intelligence is not one fixed thing that peaks and then declines. Researchers break it into at least two broad categories. Fluid intelligence covers reasoning, problem-solving, and pattern recognition in novel situations. Crystallized intelligence is the accumulated knowledge and skill you build over years of learning and experience. The two follow strikingly different life-course trajectories: fluid abilities develop early, peak in early adulthood, and then decline, while crystallized abilities keep growing well into middle age.4PubMed. Transformations in the couplings among intellectual abilities and constituent cognitive processes across the life span
What makes the picture more nuanced is that these two forms of intelligence are not independent tracks. Longitudinal data from multiple large cohorts show a strong dependency between how quickly your fluid abilities change and how your crystallized abilities change alongside them. Around age 50, where the divergence between the two is most visible, people whose fluid abilities are declining faster also tend to see slower growth in crystallized ability.5PubMed Central. A strong dependency between changes in fluid and crystallized abilities in human cognitive aging In other words, your capacity to reason in new ways seems to constrain how much new knowledge you can effectively acquire, even though accumulated knowledge can continue to expand for decades.
What Is Happening in the Brain
Neuroimaging research has tried for decades to locate intelligence in the brain, and the honest answer is that it is not located in any single region. Studies consistently find a widely distributed pattern of associations between cortical thickness and g scores, spanning frontal, parietal, and temporal areas.6PubMed Central. Associations between cortical thickness and general intelligence in children, adolescents and young adults Intelligence appears to depend on the coordinated functioning of a network, not on one brain region being bigger or better.
Even more interesting is that the relationship between brain structure and intelligence is not static. In children around age ten, higher intelligence is actually associated with slightly thinner cortex. As those children grow, the cortex thins faster in higher-IQ individuals. But by the early forties, the relationship flips: thicker cortex is now linked to higher intelligence. Surface area follows a similarly dynamic pattern, expanding more in childhood and contracting faster afterward in brighter individuals.7Cerebral Cortex. Changes in Thickness and Surface Area of the Human Cortex and Their Relationship with Intelligence The correlations between cortical thickness and IQ are generally modest, and the brain regions where those correlations are strongest shift substantially over childhood and adolescence.8Cerebral Cortex. The Dynamic Associations Between Cortical Thickness and General Intelligence are Genetically Mediated Intelligence seems to be more about the timing and trajectory of brain development than about any snapshot of brain size or shape.
There is also the idea of neural efficiency: the observation that people with higher cognitive ability tend to use less brain energy, not more, when performing cognitive tasks. Early PET scan studies found negative correlations between intelligence scores and regional metabolic rates in the brain, meaning brighter participants showed lower energy consumption during problem-solving.9PubMed Central. Neural efficiency as a function of task demands Think of it like an efficient engine: the same output, less fuel. Though the picture is more complicated on demanding tasks, neural efficiency has become one of the more durable findings linking intelligence to brain function.
Genetics and Heritability
Intelligence has substantial heritability, estimated at around 0.54 in large samples, meaning that roughly half the variation in cognitive ability across a population can be attributed to genetic differences.10PubMed Central. Genome-wide association meta-analysis of 78,308 individuals identifies new loci and genes influencing human intelligence But “heritable” does not mean “fixed” or “determined at birth.” Heritability is a population-level statistic, not a verdict on any one person’s potential. It tells you how much of the variation in a trait within a group is associated with genetic variation in that group, under the conditions that group lives in. Change the conditions, and the heritability estimate can change too.
Intelligence is profoundly polygenic. That same large meta-analysis of over 78,000 people identified hundreds of associated genetic variants across 18 regions of the genome, fifteen of them newly discovered. Each individual variant has a tiny effect. There is no single “smart gene,” and attempts to predict an individual’s intelligence from their DNA remain far less accurate than simply knowing their parents’ education level. The genetics of intelligence is an active and fast-moving field, but anyone promising a genomic IQ test is getting well ahead of the science.
How Environment Shapes Cognitive Development
The other half of the variance in intelligence is environmental, and there is now a detailed picture of how material conditions get under the skull. Poverty shapes the developing brain through multiple overlapping pathways: nutrient deficiencies, reduced cognitive stimulation, chronic stress, and exposure to environmental toxins each contribute.11PubMed Central. Poverty and the Developing Brain The effects are not just correlational. Animal models confirm that many of these factors directly alter neural plasticity and epigenetic processes.
The interaction between nutrition and socioeconomic status is a vivid example. In a longitudinal study of children in Guatemala City, those who were better nourished (measured by height-for-age) scored roughly a third of a standard deviation higher on IQ tests than the most undernourished children. But in the poorest homes, nutritional status made almost no difference to cognitive scores; the overwhelming deprivation of their home environment drowned out the nutritional signal.12PubMed. Interaction of nutritional and socioeconomic status as determinants of cognitive development in disadvantaged urban Guatemalan children A parallel study of schoolchildren in Kolkata found the same pattern: cognitive test scores dropped with each grade of undernutrition and with lower socioeconomic status, with both factors independently predicting performance.13PubMed. Grades of undernutrition and socioeconomic status influence cognitive development in school children of Kolkata
These findings matter because they place a hard limit on what genetic studies can tell us. High heritability in a well-fed, well-educated population tells you nothing about what would happen if environmental conditions changed radically. Genes set ranges, not destinies, and where someone lands within that range depends on nutrition, stimulation, toxin exposure, and stress during critical periods of brain development.
The Flynn Effect and Its Reversal
One of the most striking facts about intelligence is that average test scores rose dramatically throughout the twentieth century. This phenomenon, known as the Flynn effect, amounts to about two to three standard score points per decade according to a meta-analysis of 285 studies.14PubMed Central. The Flynn effect: a meta-analysis A broader meta-analysis spanning data from 1909 to 2013, covering 31 countries and nearly four million participants, found gains of varying magnitude depending on the domain: fluid intelligence gains were the largest (roughly 0.41 IQ points per year), followed by spatial, full-scale, and crystallized abilities. Gains were stronger in adults than in children, and they have slowed in more recent decades.15PubMed. One Century of Global IQ Gains: A Formal Meta-Analysis of the Flynn Effect (1909-2013)
No one argues that humans got genetically smarter in a century. The gains are too fast and too large for genetic change. Better nutrition, more years of schooling, smaller family sizes, reduced exposure to lead and other neurotoxins, and greater familiarity with abstract test-like thinking have all been proposed as contributors. In some countries, particularly in Scandinavia, the effect has already reversed: recent cohorts score lower than those born a decade or two earlier. Data from Norwegian military conscripts show that the reversal occurs within families, ruling out immigration or differential fertility as explanations and pointing instead to changing environmental factors.16PubMed Central. Flynn effect and its reversal are both environmentally caused The cause of the reversal is still debated, but the fact that it is happening in some of the world’s wealthiest nations is a challenge for anyone who assumed the gains would continue indefinitely.
Cultural Bias in Intelligence Testing
A persistent question in the field is whether intelligence tests measure the same thing across cultures. Even so-called “culture-free” nonverbal tests can produce severe misclassification when norms from one country are applied to children in another. A study of healthy Moroccan children given a nonverbal intelligence test normed in Britain found that up to about 16% of the children were classified as “intellectually impaired” and up to 62.5% as “below average” when British norms were used.17Archives of Clinical Neuropsychology. Cultural Bias in Intelligence Assessment Using a Culture-Free Test in Moroccan Children These were healthy, normally developing children. The misclassification was worst at older ages, suggesting that cultural exposure increasingly shapes even nonverbal cognitive performance as children grow up.
The lesson is sobering. Stripping words from a test does not strip culture from it. Familiarity with abstract visual patterns, experience with timed test formats, and the very concept of selecting the “correct” answer among options are all culturally loaded. Any responsible use of intelligence tests requires local norms, not off-the-shelf scores exported from another society.
Can You Improve Your Intelligence?
This is the question people care about most, and the evidence is a mix of real promise and frustrated expectations. Working memory training, the kind where you practice holding and manipulating information in mind under increasing difficulty, has shown some ability to improve fluid intelligence scores. In one controlled study, 30 hours of working memory training led to significant gains on all measures of fluid intelligence, while a control group receiving communication training showed no improvement.18PubMed. Working memory training: improving intelligence–changing brain activity The training also changed brain activity patterns, suggesting genuine neural reorganization rather than simple test familiarity.
Exercise appears to play a complementary role, especially when combined with cognitive training. In a study of chronic stroke patients, a group that received combined aerobic and cognitive training improved their fluid intelligence scores by nearly 48% compared to baseline, while a control group that received only recreational activity and games actually declined slightly. The aerobic-plus-cognitive combination was the only group significantly better than control, and a growth factor linked to brain plasticity predicted a large share of the cognitive improvement.19PubMed. Synergistic Benefits of Combined Aerobic and Cognitive Training on Fluid Intelligence and the Role of IGF-1 in Chronic Stroke That said, this was a clinical population recovering from brain injury, so extrapolating those effect sizes to healthy adults is risky. The broader literature on cognitive training in healthy people shows smaller and more contested gains, with ongoing debate about whether improvements transfer beyond the trained tasks.
Why “Multiple Intelligences” Has Not Held Up
Howard Gardner’s theory of multiple intelligences has been enormously influential in education. The idea that people possess distinct, independent intelligences, including linguistic, logical, musical, spatial, bodily, interpersonal, intrapersonal, and naturalistic, has shaped curriculum design in schools worldwide. The problem is that the theory has not survived empirical scrutiny. No researcher has directly identified a brain basis for the proposed intelligences, factor analyses have not confirmed that they are independent of one another, and studies of classroom interventions based on the theory suffer from serious methodological flaws.20PubMed Central. Why multiple intelligences theory is a neuromyth
A review of intervention studies that tried to apply multiple intelligences theory in classrooms found small sample sizes, missing control groups, vague measurement tools, and signs of reporting bias throughout.21Intelligence. A valid evaluation of the theory of multiple intelligences is not yet possible: Problems of methodological quality for intervention studies This does not mean that people do not have different strengths. They clearly do. But calling those strengths separate “intelligences” implies a degree of independence that the data do not support. The g factor sits stubbornly at the center: someone who excels in language is more likely than chance to excel in math, spatial reasoning, and most other cognitive domains. The correlations are not perfect, and individual profiles vary, but the idea of entirely separate intelligences has become something closer to a neuromyth than a useful scientific framework.
Intelligence and Resilience Against Cognitive Decline
Higher cognitive ability earlier in life appears to buffer the brain against the effects of Alzheimer’s disease pathology. Research on cognitive reserve suggests that people who enter old age with greater cognitive and brain reserve can tolerate more disease-related brain changes before showing clinical symptoms. A recent study found that both cognitive reserve (built through education, occupational complexity, and intellectual engagement) and brain reserve (physical measures like cortical volume) helped modify the vulnerability to early Alzheimer’s-related cognitive decline.22PubMed Central. Cognitive and Brain Reserve as Modifiers of Early Alzheimer Disease-Related Cognitive Vulnerability The practical implication is straightforward: intellectually engaged lives may not prevent Alzheimer’s pathology from developing, but they can delay its clinical expression, buying years of functional independence.
Collective Intelligence Is Not the Sum of Individual IQs
Most discussions of intelligence focus on individuals, but groups have their own measurable cognitive ability. Research has identified a “c factor” for groups, analogous to the g factor for individuals, that predicts how well a group will perform across a wide range of tasks. The surprising finding is that this collective intelligence factor is not strongly correlated with the average IQ of the group’s members, or even with the smartest person in the room. Instead, it correlates with the average social sensitivity of group members, how equally conversational turns are distributed, and the proportion of women in the group.23PubMed. Evidence for a collective intelligence factor in the performance of human groups
Follow-up work has reinforced these findings, showing that group collaboration process matters more than individual skill in predicting collective intelligence, and that the factor is robust across diverse task types.24PubMed Central. Quantifying collective intelligence in human groups For workplaces and teams, this means that hiring the smartest individuals is less important than creating conditions where members listen to one another and share speaking time. A team of moderately bright people who communicate well can outperform a team of geniuses who talk over one another.
Intelligence Beyond Brains
The very concept of intelligence gets stranger the further you look from humans. Complex brains and high intelligence have evolved independently many times across the animal kingdom, appearing in insects, octopuses, certain fish, corvid and parrot-family birds, cetaceans, elephants, and primates.25PubMed Central. Convergent evolution of complex brains and high intelligence The octopus is perhaps the most alien example: its nervous system is radically decentralized, with the majority of neurons located in the arms rather than the central brain. A severed octopus arm continues to exhibit complex behaviors nearly identical to those of an intact animal, suggesting that decision-making happens at many levels of the body simultaneously.26PubMed Central. Where Is It Like to Be an Octopus?27Canadian Journal of Zoology. The cephalopod specialties: complex nervous system, learning, and cognition
Even organisms without any nervous system at all show behaviors that look like intelligence under some definitions. Plants acquire information from their environment, store memories of past stresses (through a process called priming), and adjust their growth and defense strategies accordingly.28PubMed Central. The foundations of plant intelligence Researchers have pointed out that the molecular signaling networks within plant cells share structural similarities with neural connectomes, suggesting that neural systems may not be necessary for intelligent behavior at all.29PubMed Central. Extended cognition in plants: is it possible? Whether this qualifies as “intelligence” or merely as adaptive complexity is still debated, but it underscores how much our understanding of the concept remains tethered to the human case.
What AI Borrows from Biological Intelligence
The relationship between human intelligence and artificial intelligence is often framed as a race, but the more scientifically interesting question is what AI researchers are learning from studying brains. Biological brains perform lifelong learning, adaptive control, and flexible reasoning using orders of magnitude less energy than current AI systems require for comparable tasks. Researchers are now actively studying the architectural constraints, signaling strategies, and learning algorithms used by brains to inform the design of more energy-efficient artificial systems.30PubMed. Looking to the brain to improve energy efficiency of AI The neural efficiency findings discussed earlier, where brighter brains use less energy rather than more, turn out to have direct engineering relevance. If the goal is to build AI that can reason flexibly without consuming the energy budget of a small city, studying how biological brains manage the trick is not a philosophical exercise but a design constraint.
One of the more humbling aspects of this comparison is that primates may have evolved their large brains not primarily for tool use or environmental problem-solving, but for managing complex social relationships. The social brain hypothesis proposes that primate brain size correlates with social group size because the cognitive demands of tracking alliances, predicting behavior, and navigating social hierarchies are the primary selective pressure on brain expansion.31PubMed. The social brain hypothesis and its implications for social evolution If true, social cognition is not a secondary use of intelligence but the evolutionary reason it exists in the first place, a point that resonates with the collective intelligence research showing that groups outperform individuals not through raw brainpower but through social perceptiveness and turn-taking.