Intelligence is a broad term, and the nature of human (and nonhuman) intelligence continues to be debated. Remembering information, understanding context and meaning, mentally manipulating objects, retrieving and applying knowledge learned from experience, and solving abstract problems in novel ways are all aspects of intelligence. Most scientific work on intelligence has focused on devising ways to test how quickly, accurately, and flexibly people can do these mental tasks and on understanding individual differences in performance on those tests. Intelligence is not a synecdoche for all human talent, but it does make a difference for people’s school grades, job performance, physical health, and even how long they live. In the modern psychological study of intelligence, there is some consensus about the broad dimensions of intelligence and how they change over the lifespan, but there are still controversies and questions regarding the interpretation of intelligence test scores and their heritability. Research on intelligence is deeply connected with research on cognition in machines and nonhuman animals.

History

Two French psychologists, Theodore Simon and Alfred Binet, originated the modern study of intelligence and intelligence testing at the turn of the 20th century (Ritchie, 2016). Tasked with identifying schoolchildren who would benefit from specialized education, they developed a set of problems that could be given to children as young as 3, such as naming body parts or making change. A child with a higher “intelligence quotient” could do tasks that most children, in French society, typically could not do until they were older. From the beginning, then, the study of intelligence has been conceptualized and measured with reference to activities that are valued in Western, industrialized societies, where children spend much of their time in formal schooling—and that might be irrelevant to what is understood as intelligent behavior in other cultural contexts (Cole, 2015).

Intelligence testing jumped from children to adults during World War I, when the U.S. Army adapted the Simon–Binet method to identify new recruits who might perform well in officer positions. Tragically, the uses of intelligence testing in the 20th century were not limited to special education or officer training. In the United States and Europe, eugenicists targeted individuals and groups for involuntary sterilization, institutionalization, and, in Nazi Germany, genocide based on their alleged “feeblemindedness” (Kevles, 1998). In the United States, Virginia’s involuntary sterilization law was upheld in an infamous 1924 Supreme Court case, Buck v. Bell, which centered on Carrie Buck, an unmarried teenager who was sterilized after being assessed with a Simon–Binet test. The pseudoscientific misuse of cognitive testing to justify eugenic violence continues to generate controversy about the study of human intelligence (Klancher Merchant, 2024) [see Eugenic Thinking and the Cognitive Sciences].

The following video is the voice of Carrie Buck describing her pregnancy.

Core concepts

Intelligence encompasses multiple related abilities

Mental tasks range from basic operations, like pressing a button as quickly as possible in response to a sound, to complex problem solving. These tasks can be organized into two broad domains: (1) fluid intelligence, which requires effortful processing in the moment and includes processing speed, working memory, and visuospatial reasoning (Figure 1), and (2) crystallized intelligence, which requires applying previously acquired knowledge, including vocabulary and general information about the world (for a history of this distinction, see Brown, 2016). Although everyone will have relative strengths and weaknesses, someone who performs better than other people of similar age on any one intelligence test tends to perform better on all others (e.g., Johnson et al., 2008).

Intelligence changes over the lifespan

Intelligence continues to improve from early childhood through early adulthood, but then, fluid abilities start to decline: cognitive aging begins before 30 (Salthouse, 2009). Crystallized intelligence, on the other hand, is preserved or even improves through the 60s (Tucker-Drob, 2019). Neuroscience studies have identified brain characteristics that are correlated with higher intelligence into older adulthood, including cortical size and thickness, the integrity of white matter (neural pathways within the brain), and how strongly different brain regions are connected (Deary et al., 2022; Michel et al., 2024). One of the biggest protective factors against dementia in later life, then, is developing higher levels of intelligence in early life. In contrast, there is little evidence that “brain training” programs, which purport to preserve cognitive abilities through deliberate, gamified practice, effectively slow aging-related declines in intelligence (Simons et al., 2016).

Intelligence matters beyond the classroom

Intelligence test performance strongly predicts children’s future school grades and exam scores, even when controlling for measures of family background like income (Kriegbaum et al., 2018). Beyond the classroom, general intelligence test scores predict job performance, particularly as the job involves fewer rote activities (Kuncel & Hezlett, 2010). Finally, intelligence predicts future health and mortality (Calvin et al., 2011). A study of one million military recruits found that 18-year-old men in the top 10% of intelligence test scores were three times less likely to die in the next 20 years than men in the bottom 10% (Batty et al., 2009). This intelligence–mortality relationship was not due to intelligent men having been raised in higher social class homes; rather, intelligent men were healthier largely because they went further in school.

Intelligence depends on the environment

Contrary to claims that intelligence is best understood as an innate biological capacity, the development of intelligence is highly dependent on environmental input and context. Children who are raised by well-resourced families have higher average intelligence than children who remain in institutional care or with lower-income birth parents (Humphreys et al., 2022; Kendler et al., 2015). When children’s schooling is improved or lengthened, their intelligence, on average, increases (Ritchie & Tucker-Drob, 2018). Average intelligence scores are also higher in more recent birth cohorts (a historical trend that has been called the Flynn effect), perhaps because of improved child nutrition and expanded educational opportunity (Flynn, 1984; Trahan et al., 2014). Exposure to neurotoxins like lead decreases intelligence (Reuben et al., 2017). Intelligence tests are useful, therefore, for indexing social and environmental exposures that affect brain health (Washington, 2019).

Questions, controversies, and new developments

Factor analysis and the interpretation of “general intelligence”

Positive correlations among scores on different intelligence tests can be analyzed using a statistical method called factor analysis. To use an analogy from Earl Hunt (1995), scores on various tests can be imagined as pimentos stuck in a hot dog. If you wanted to identify where any pimento was, the most informative first approximation of its location would be how far it is from the end of the hot dog. A skewer down the length of the hot dog, then, would represent the first dimension of variation in the scores, also know as the “first factor.”

When factor analysis is applied to intelligence test scores, the first factor is traditionally labeled general intelligence, or “g” (Spearman, 1904). What, if anything, is being measured by the g factor—or any latent factor—is an enduring controversy (Borsboom et al., 2003). Some argue that the g factor is purely a statistical artefact (e.g., Gould, 1981), whereas others argue that it represents an innate biological quality (e.g., Jensen, 1998).

Most contemporary experts eschew both extreme views and take a constructivist view of g: it is a statistically convenient composite measure that—like other composite measures such as well-being or socioeconomic status—is consistently associated with a wide variety of developmental antecedents and subsequent life outcomes but does not reflect anything singular and inherent about an individual (Clapp Sullivan et al., 2024; Cronbach & Meehl, 1955; Dickens, 2007).

The heritability of intelligence

Identical twins, even when they have been separated at birth, have very similar intelligence, whereas adoptive siblings raised together scarcely resemble each other (Plomin & von Stumm, 2018). Twin studies typically estimate the heritability of general intelligence to be 50% to 80%, meaning that at least half of the differences within a group of people (within a given society) in how they perform on intelligence tests is linked with genetic differences.

Heritability is commonly misinterpreted. The heritability of intelligence does not mean there is a single gene “for” intelligence, that intelligence is fixed or genetically determined, or that any differences between races, nations, ethnicities, genders, social classes, or other socially constructed groups have a genetic basis (Harden, 2021). Instead, genetic influences on intelligence operate through children’s dynamic interactions with culturally specific environmental experiences, including interactions with caregivers and formal schooling (Tucker-Drob & Harden, 2012). Consistent with the theory that genetic influences are maximized when children are embedded in varied and enriched environments, children raised in higher-income families in the United States show greater heritability of intelligence (Tucker-Drob & Bates, 2016).

More recently, DNA-based polygenic scores, which add up information about which genetic variants a person has inherited, have been correlated with intelligence test scores at about r = 0.3 (von Stumm & Plomin, 2021). The identification of specific genetic variants associated with intelligence has animated debates about preimplantation genetic screening of in vitro fertilized embryos (Johnston & Matthews, 2022).

Broader connections

From the beginning, the study of intelligence has been entwined with the study of child development, a field that continues to explore how child cognition changes with age, schooling, and culture (Legare, 2019). How to conceptualize and measure intelligence in nonhuman animals has been a topic of interest since the beginning of psychology as a discipline (Thorndike, 1898), and researchers have developed cognitive test batteries to measure individual differences in “g” in multiple species, including rodents, dogs, birds, and nonhuman primates (Shaw & Schmelz, 2017) [see Animal Cognition]. Artificial intelligence can now perform very well on many standard intelligence tests, and research on the similarities, differences, and potential complementarities between artificial and human intelligence is rapidly advancing (Lake et al., 2017) [see Large Language Models].

Acknowledgments

K. Paige Harden is a faculty research associate of the Population Research Center at the University of Texas at Austin, which is supported by grant P2CHD042849 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development.

Further reading

• Calvin, C. M., Deary, I. J., Fenton, C., Roberts, B. A., Der, G., Leckenby, N., & Batty, G. D. (2011). Intelligence in youth and all-cause-mortality: Systematic review with meta-analysis. International Journal of Epidemiology, 40(3), 626–644. https://doi.org/10.1093/ije/dyq190

• Humphreys, K. L., King, L. S., Guyon-Harris, K. L., Sheridan, M. A., McLaughlin, K. A., Radulescu, A., Nelson, C. A., Fox, N. A., & Zeanah, C. H. (2022). Foster care leads to sustained cognitive gains following severe early deprivation. Proceedings of the National Academy of Sciences, 119(38), e2119318119. https://doi.org/10.1073/pnas.2119318119

• Plomin, R., & von Stumm, S. (2018). The new genetics of intelligence. Nature Reviews Genetics, 19(3), 148–159. https://doi.org/10.1038/nrg.2017.104

• Tucker-Drob, E. M. (2019). Cognitive aging and dementia: A life-span perspective. Annual Review of Developmental Psychology, 1, 177–196. https://doi.org/10.1146/annurev-devpsych-121318-085204